duplicated block id: 1 size: 388 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (104:788) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (449:1132) duplicated block id: 2 size: 312 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2788:3273) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3403:3892) duplicated block id: 3 size: 266 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (374:811) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1526:1963) duplicated block id: 4 size: 129 cleaned lines of code in 2 files: - maga_transformer/config/gpt_init_model_parameters.py (106:234) - maga_transformer/ops/libth_transformer.pyi (132:260) duplicated block id: 5 size: 121 cleaned lines of code in 2 files: - bazel/arch_select.bzl (1:138) - open_source/bazel/arch_select.bzl (1:138) duplicated block id: 6 size: 114 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (113:299) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1236:1421) duplicated block id: 7 size: 103 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.h (245:432) - maga_transformer/cpp/rocm/hip_utils.h (101:286) duplicated block id: 8 size: 99 cleaned lines of code in 2 files: - maga_transformer/_ft_pickler.py (112:214) - maga_transformer/utils/meta_pickler.py (121:223) duplicated block id: 9 size: 88 cleaned lines of code in 2 files: - maga_transformer/_ft_pickler.py (5:110) - maga_transformer/utils/meta_pickler.py (5:110) duplicated block id: 10 size: 76 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_logn_attention.h (5:109) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2098:2202) duplicated block id: 11 size: 76 cleaned lines of code in 2 files: - bazel/defs.bzl (187:265) - def.bzl (55:133) duplicated block id: 12 size: 58 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (5:99) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (321:415) duplicated block id: 13 size: 57 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_to_float.h (12:109) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2355:2452) duplicated block id: 14 size: 56 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_sum_dot_zero.h (5:130) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1968:2093) duplicated block id: 15 size: 53 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (5:94) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1137:1226) duplicated block id: 16 size: 51 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (57:109) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (59:111) duplicated block id: 17 size: 50 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (43:123) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (23:104) duplicated block id: 18 size: 48 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (783:854) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (887:958) duplicated block id: 19 size: 47 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_to_fp8.h (6:104) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2625:2723) duplicated block id: 20 size: 47 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (390:463) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1005:1078) duplicated block id: 21 size: 47 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (91:159) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (699:771) duplicated block id: 22 size: 46 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (580:650) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (683:753) duplicated block id: 23 size: 46 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (616:699) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (609:691) duplicated block id: 24 size: 45 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (239:307) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (852:924) duplicated block id: 25 size: 45 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (297:390) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (257:350) duplicated block id: 26 size: 45 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (543:616) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1157:1230) duplicated block id: 27 size: 44 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaActOp.cc (11:61) - maga_transformer/cpp/devices/rocm_impl/ROCmActOp.cc (10:60) duplicated block id: 28 size: 43 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (624:678) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (828:882) duplicated block id: 29 size: 42 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/minicpmv.py (64:107) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (122:165) duplicated block id: 30 size: 40 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (727:778) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (932:983) duplicated block id: 31 size: 40 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (75:122) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (102:149) duplicated block id: 32 size: 38 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_cast_to_int8.h (18:72) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2730:2783) duplicated block id: 33 size: 38 cleaned lines of code in 2 files: - bazel/defs.bzl (148:185) - def.bzl (16:53) duplicated block id: 34 size: 38 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (5:78) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (56:129) duplicated block id: 35 size: 38 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (170:234) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (153:217) duplicated block id: 36 size: 37 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (137:179) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (100:142) duplicated block id: 37 size: 36 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (357:397) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (230:269) duplicated block id: 38 size: 36 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (148:234) - maga_transformer/tokenizer/tokenization_chatglm3.py (240:327) duplicated block id: 39 size: 35 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (433:484) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (509:560) duplicated block id: 40 size: 35 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaPrefillAttention.cc (17:52) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (17:52) duplicated block id: 41 size: 34 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (226:262) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (318:354) duplicated block id: 42 size: 33 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (150:185) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (543:578) duplicated block id: 43 size: 32 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (105:149) - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (301:345) duplicated block id: 44 size: 32 cleaned lines of code in 2 files: - maga_transformer/tools/quant/fp8_quanter.py (24:55) - maga_transformer/tools/quant/fp8_quanter.py (68:99) duplicated block id: 45 size: 32 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (456:527) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (422:493) duplicated block id: 46 size: 32 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (256:323) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (243:310) duplicated block id: 47 size: 31 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (269:305) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (334:370) duplicated block id: 48 size: 31 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (55:94) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (179:215) duplicated block id: 49 size: 31 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (141:178) - maga_transformer/device/device_impl.py (275:312) duplicated block id: 50 size: 30 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (228:268) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (453:493) duplicated block id: 51 size: 30 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_fp8.h (100:150) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2566:2617) duplicated block id: 52 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (467:519) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1082:1133) duplicated block id: 53 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (137:233) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (101:194) duplicated block id: 54 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/gen.py (907:937) - maga_transformer/cpp/cutlass/gen.py (940:970) duplicated block id: 55 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (398:434) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (832:870) duplicated block id: 56 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (314:366) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (930:981) duplicated block id: 57 size: 29 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (401:450) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (369:418) duplicated block id: 58 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (36:91) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (184:239) duplicated block id: 59 size: 28 cleaned lines of code in 2 files: - maga_transformer/_ft_pickler.py (216:249) - maga_transformer/utils/meta_pickler.py (225:258) duplicated block id: 60 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (775:826) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1082:1131) duplicated block id: 61 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (954:1005) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1106:1157) duplicated block id: 62 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (622:673) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (930:979) duplicated block id: 63 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (646:699) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (799:852) duplicated block id: 64 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (314:363) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (622:673) duplicated block id: 65 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (467:516) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (775:826) duplicated block id: 66 size: 28 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (338:390) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (491:543) duplicated block id: 67 size: 27 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (796:828) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (809:841) duplicated block id: 68 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (16:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (314:361) duplicated block id: 69 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (164:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (775:824) duplicated block id: 70 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (164:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (467:514) duplicated block id: 71 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (16:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (930:977) duplicated block id: 72 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (16:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (622:671) duplicated block id: 73 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (164:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1082:1129) duplicated block id: 74 size: 27 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (234:263) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (362:391) duplicated block id: 75 size: 27 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (483:526) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (470:513) duplicated block id: 76 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (43:68) - maga_transformer/ops/libth_transformer.pyi (105:130) duplicated block id: 77 size: 26 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (102:139) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (37:74) duplicated block id: 78 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (105:130) - rtpllm_master_py/stub/librtpllm_master.pyi (42:67) duplicated block id: 79 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (330:355) - rtpllm_master_py/stub/librtpllm_master.pyi (42:67) duplicated block id: 80 size: 26 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (449:477) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (291:317) duplicated block id: 81 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (105:130) - maga_transformer/ops/libth_transformer.pyi (330:355) duplicated block id: 82 size: 26 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (162:219) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (163:220) duplicated block id: 83 size: 26 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (103:179) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (302:377) duplicated block id: 84 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (43:68) - maga_transformer/ops/libth_transformer.pyi (330:355) duplicated block id: 85 size: 26 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (303:331) - maga_transformer/openai/renderers/custom_renderer.py (562:590) duplicated block id: 86 size: 26 cleaned lines of code in 2 files: - maga_transformer/ops/libth_transformer.pyi (43:68) - rtpllm_master_py/stub/librtpllm_master.pyi (42:67) duplicated block id: 87 size: 25 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (293:318) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (252:277) duplicated block id: 88 size: 25 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (225:249) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (354:378) duplicated block id: 89 size: 25 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaEmbeddingLookup.cc (12:39) - maga_transformer/cpp/devices/rocm_impl/ROCmDevice.cc (295:322) duplicated block id: 90 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:26) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:26) duplicated block id: 91 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (86:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (385:419) duplicated block id: 92 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (341:367) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (477:503) duplicated block id: 93 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmEmbeddingLookup.cc (12:42) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (80:110) duplicated block id: 94 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmActOp.cc (11:41) - maga_transformer/cpp/devices/arm_impl/ArmEmbeddingLookup.cc (12:42) duplicated block id: 95 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmActOp.cc (11:41) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (80:110) duplicated block id: 96 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:26) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:26) duplicated block id: 97 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:26) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:26) duplicated block id: 98 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (694:727) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1000:1034) duplicated block id: 99 size: 24 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:626) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:788) duplicated block id: 100 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:25) duplicated block id: 101 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (245:268) - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (314:337) duplicated block id: 102 size: 23 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/openvino.py (148:170) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (161:183) duplicated block id: 103 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (91:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1005:1034) duplicated block id: 104 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:25) duplicated block id: 105 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:25) duplicated block id: 106 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:25) duplicated block id: 107 size: 23 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (159:185) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (162:188) duplicated block id: 108 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:25) duplicated block id: 109 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:25) duplicated block id: 110 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:25) duplicated block id: 111 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (390:419) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (699:727) duplicated block id: 112 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:25) duplicated block id: 113 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (88:110) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (65:87) duplicated block id: 114 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (37:77) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (138:178) duplicated block id: 115 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:25) duplicated block id: 116 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:25) duplicated block id: 117 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (180:209) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (672:701) duplicated block id: 118 size: 23 cleaned lines of code in 2 files: - maga_transformer/config/gpt_init_model_parameters.py (236:258) - maga_transformer/ops/libth_transformer.pyi (261:283) duplicated block id: 119 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:25) duplicated block id: 120 size: 23 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:25) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:25) duplicated block id: 121 size: 23 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (690:729) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (649:687) duplicated block id: 122 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:24) duplicated block id: 123 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (458:488) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (828:858) duplicated block id: 124 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (51:102) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (51:102) duplicated block id: 125 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (458:488) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (624:654) duplicated block id: 126 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 127 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 128 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 129 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaWeights.cc (12:42) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (57:87) duplicated block id: 130 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:24) duplicated block id: 131 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 132 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 133 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 134 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 135 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 136 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 137 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 138 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 139 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 140 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 141 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 142 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 143 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 144 size: 22 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm3.py (271:327) - maga_transformer/tokenizer/tokenization_chatglm4.py (169:224) duplicated block id: 145 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (847:880) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1152:1186) duplicated block id: 146 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 147 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:24) duplicated block id: 148 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 149 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmWeights.cc (13:43) - maga_transformer/cpp/devices/cuda_impl/CudaWeights.cc (12:42) duplicated block id: 150 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2582:2603) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2623:2644) duplicated block id: 151 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 152 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (234:267) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (538:572) duplicated block id: 153 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 154 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 155 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 156 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 157 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 158 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 159 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:24) duplicated block id: 160 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmWeights.cc (13:43) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (57:87) duplicated block id: 161 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 162 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (265:293) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (226:254) duplicated block id: 163 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 164 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 165 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 166 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 167 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 168 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 169 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 170 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 171 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 172 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 173 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 174 size: 22 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (613:642) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (372:401) duplicated block id: 175 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:24) duplicated block id: 176 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 177 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 178 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 179 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 180 size: 22 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (179:234) - maga_transformer/tokenizer/tokenization_chatglm4.py (169:224) duplicated block id: 181 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 182 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 183 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 184 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 185 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 186 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 187 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 188 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 189 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 190 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 191 size: 22 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (118:147) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (119:148) duplicated block id: 192 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 193 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 194 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) duplicated block id: 195 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 196 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 197 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:24) duplicated block id: 198 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:24) duplicated block id: 199 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:24) duplicated block id: 200 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:24) duplicated block id: 201 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 202 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:24) duplicated block id: 203 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:24) duplicated block id: 204 size: 22 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:24) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:24) duplicated block id: 205 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (41:65) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (9:33) duplicated block id: 206 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (41:65) - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (74:98) duplicated block id: 207 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (239:267) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1157:1186) duplicated block id: 208 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) duplicated block id: 209 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) duplicated block id: 210 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) duplicated block id: 211 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (26:49) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (78:102) duplicated block id: 212 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (400:424) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (754:778) duplicated block id: 213 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) duplicated block id: 214 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) duplicated block id: 215 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 216 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) duplicated block id: 217 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) duplicated block id: 218 size: 21 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (105:146) - maga_transformer/tokenizer/tokenization_chatglm3.py (177:218) duplicated block id: 219 size: 21 cleaned lines of code in 2 files: - maga_transformer/tools/fake_glm_v2.py (31:52) - maga_transformer/tools/fake_model_base.py (32:53) duplicated block id: 220 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) duplicated block id: 221 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) duplicated block id: 222 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 223 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 224 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (122:166) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (264:308) duplicated block id: 225 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) duplicated block id: 226 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (38:74) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (125:161) duplicated block id: 227 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) duplicated block id: 228 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) duplicated block id: 229 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) duplicated block id: 230 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (576:602) duplicated block id: 231 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) duplicated block id: 232 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (534:563) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (727:756) duplicated block id: 233 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (123:149) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (271:297) duplicated block id: 234 size: 21 cleaned lines of code in 2 files: - maga_transformer/tools/fake_bloom.py (7:27) - maga_transformer/tools/fake_model_base.py (13:33) duplicated block id: 235 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (534:563) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (932:961) duplicated block id: 236 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (423:449) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 237 size: 21 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/minicpmv.py (211:235) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (269:293) duplicated block id: 238 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) duplicated block id: 239 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (884:910) duplicated block id: 240 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1038:1064) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 241 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (731:757) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1190:1216) duplicated block id: 242 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (543:572) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (852:880) duplicated block id: 243 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (74:98) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (9:33) duplicated block id: 244 size: 21 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (406:427) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (519:540) duplicated block id: 245 size: 20 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (104:126) - bazel/tf_http_archive.bzl (172:194) duplicated block id: 246 size: 20 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (762:783) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (795:816) duplicated block id: 247 size: 20 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (328:370) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (315:357) duplicated block id: 248 size: 20 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (137:161) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (229:252) duplicated block id: 249 size: 20 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (52:71) - maga_transformer/cpp/kernels/quantize_weight.cu (79:98) duplicated block id: 250 size: 20 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (259:278) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (157:176) duplicated block id: 251 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage.h (51:98) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (51:98) duplicated block id: 252 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (667:693) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (827:853) duplicated block id: 253 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (217:235) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (97:115) duplicated block id: 254 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (304:333) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1435:1464) duplicated block id: 255 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (481:511) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (605:634) duplicated block id: 256 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (241:265) - maga_transformer/cpp/cuda/cuda_utils.cc (288:312) duplicated block id: 257 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage.h (51:98) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (51:98) duplicated block id: 258 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (680:698) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (744:762) duplicated block id: 259 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (680:698) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (876:894) duplicated block id: 260 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (30:67) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (29:68) duplicated block id: 261 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (744:762) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (876:894) duplicated block id: 262 size: 19 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (582:600) - maga_transformer/cpp/kernels/activation_kernels.cu (613:631) duplicated block id: 263 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (458:484) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (932:958) duplicated block id: 264 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (458:484) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (727:753) duplicated block id: 265 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (164:183) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (255:274) duplicated block id: 266 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (548:586) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (510:548) duplicated block id: 267 size: 18 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (135:155) - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (190:210) duplicated block id: 268 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (581:608) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (784:811) duplicated block id: 269 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (611:631) - maga_transformer/cpp/kernels/layernorm_kernels.cu (653:673) duplicated block id: 270 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (581:608) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (888:915) duplicated block id: 271 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (491:508) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (536:553) duplicated block id: 272 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (35:52) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (300:317) duplicated block id: 273 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (637:654) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (657:674) duplicated block id: 274 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (27:46) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (380:399) duplicated block id: 275 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (727:753) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (828:854) duplicated block id: 276 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (534:560) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (828:854) duplicated block id: 277 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (534:560) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (624:650) duplicated block id: 278 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (684:711) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (888:915) duplicated block id: 279 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (684:711) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (784:811) duplicated block id: 280 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (429:447) - maga_transformer/cpp/kernels/rmsnormKernels.cu (133:151) duplicated block id: 281 size: 18 cleaned lines of code in 2 files: - maga_transformer/models/gpt_neox.py (37:55) - maga_transformer/models/gpt_neox.py (98:116) duplicated block id: 282 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_vector_abs_max.h (25:69) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (217:261) duplicated block id: 283 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (624:650) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (932:958) duplicated block id: 284 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (93:130) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (188:225) duplicated block id: 285 size: 18 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (687:707) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (740:760) duplicated block id: 286 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (1:18) duplicated block id: 287 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (136:168) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (122:155) duplicated block id: 288 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (1:18) duplicated block id: 289 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (1:18) duplicated block id: 290 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (1:18) duplicated block id: 291 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (161:204) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (281:324) duplicated block id: 292 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (355:377) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (785:808) duplicated block id: 293 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (303:338) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (406:441) duplicated block id: 294 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (102:135) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (66:99) duplicated block id: 295 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (1:18) duplicated block id: 296 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaPrefillAttention.cc (54:74) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (54:74) duplicated block id: 297 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (346:374) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1491:1519) duplicated block id: 298 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (398:419) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (672:693) duplicated block id: 299 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (79:97) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (120:138) duplicated block id: 300 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (110:151) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (230:271) duplicated block id: 301 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (180:201) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (832:853) duplicated block id: 302 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (180:201) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (398:419) duplicated block id: 303 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (204:226) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (429:451) duplicated block id: 304 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (205:236) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (389:419) duplicated block id: 305 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (654:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (902:918) duplicated block id: 306 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (1:18) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (1:18) duplicated block id: 307 size: 17 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (256:273) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (404:421) duplicated block id: 308 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (44:60) - maga_transformer/cpp/api_server/ChatService.cc (109:125) duplicated block id: 309 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (397:415) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (488:506) duplicated block id: 310 size: 16 cleaned lines of code in 2 files: - maga_transformer/server/backend_server.py (143:159) - maga_transformer/server/frontend_server.py (179:195) duplicated block id: 311 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (7:22) duplicated block id: 312 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (194:225) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (309:340) duplicated block id: 313 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (735:757) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (909:931) duplicated block id: 314 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/eplb/experts_stats_kernels.cu (43:61) - maga_transformer/cpp/kernels/eplb/experts_stats_kernels.cu (73:91) duplicated block id: 315 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (19:36) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (17:34) duplicated block id: 316 size: 16 cleaned lines of code in 2 files: - maga_transformer/models/gpt_neox_weight.py (27:48) - maga_transformer/models/gpt_neox_weight.py (101:122) duplicated block id: 317 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (1:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (1:16) duplicated block id: 318 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (325:342) - maga_transformer/cpp/kernels/rmsnormKernels.cu (413:430) duplicated block id: 319 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (1:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (1:16) duplicated block id: 320 size: 16 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (104:119) - maga_transformer/model_loader/smooth_quant_weight.py (175:190) duplicated block id: 321 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (99:130) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (309:340) duplicated block id: 322 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (7:22) duplicated block id: 323 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:17) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:17) duplicated block id: 324 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (491:519) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (954:981) duplicated block id: 325 size: 16 cleaned lines of code in 2 files: - maga_transformer/models/llama.py (35:51) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (292:307) duplicated block id: 326 size: 16 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (507:523) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (101:117) duplicated block id: 327 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1001) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1018) duplicated block id: 328 size: 16 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (81:99) - maga_transformer/models/jina_bert/jina_bert_weight.py (83:101) duplicated block id: 329 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1066:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1247:1262) duplicated block id: 330 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (7:22) duplicated block id: 331 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:984) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1001) duplicated block id: 332 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:984) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1018) duplicated block id: 333 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (321:337) - maga_transformer/cpp/cuda/cufmha/cufmha.cc (360:376) duplicated block id: 334 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (135:150) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (154:169) duplicated block id: 335 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/26_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v4.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (7:22) duplicated block id: 336 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (436:451) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (456:471) duplicated block id: 337 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:17) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:17) duplicated block id: 338 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (83:103) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (184:204) duplicated block id: 339 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (406:422) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (273:289) duplicated block id: 340 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1018) duplicated block id: 341 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (7:22) duplicated block id: 342 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1001) duplicated block id: 343 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/12_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/24_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (7:22) duplicated block id: 344 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:967) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1018) duplicated block id: 345 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:967) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:984) duplicated block id: 346 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:967) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1001) duplicated block id: 347 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:17) - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:17) duplicated block id: 348 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/21_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/33_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x16_4_intrawave_v4.cc (7:22) duplicated block id: 349 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (7:22) duplicated block id: 350 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (1:17) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:17) duplicated block id: 351 size: 16 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (349:364) - maga_transformer/openai/renderers/internvl_renderer.py (146:161) duplicated block id: 352 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (7:22) duplicated block id: 353 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (113:128) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (299:314) duplicated block id: 354 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:984) duplicated block id: 355 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:967) duplicated block id: 356 size: 16 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (83:98) - maga_transformer/model_loader/smooth_quant_weight.py (235:250) duplicated block id: 357 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (7:22) duplicated block id: 358 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (1:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1:16) duplicated block id: 359 size: 16 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (189:204) - maga_transformer/model_loader/smooth_quant_weight.py (339:354) duplicated block id: 360 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (56:92) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (56:94) duplicated block id: 361 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (1:17) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (1:17) duplicated block id: 362 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (1:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (1:16) duplicated block id: 363 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/1_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v4.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:22) duplicated block id: 364 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/l1norm_kernels.cu (77:94) - maga_transformer/cpp/kernels/rmsnormKernels.cu (413:430) duplicated block id: 365 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/l1norm_kernels.cu (77:94) - maga_transformer/cpp/kernels/rmsnormKernels.cu (325:342) duplicated block id: 366 size: 16 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (265:284) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (393:412) duplicated block id: 367 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (138:168) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (38:68) duplicated block id: 368 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (128:148) - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (255:275) duplicated block id: 369 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (1:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (1:16) duplicated block id: 370 size: 16 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (82:101) - maga_transformer/model_loader/static_fp8_quant_weight.py (99:118) duplicated block id: 371 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (37:67) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (125:155) duplicated block id: 372 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:17) - maga_transformer/cpp/kernels/int8_utils.cuh (1:17) duplicated block id: 373 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (145:167) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (236:258) duplicated block id: 374 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.h (35:62) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.h (14:40) duplicated block id: 375 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/3_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v4.cc (7:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/4_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v3.cc (7:22) duplicated block id: 376 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (338:366) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1106:1133) duplicated block id: 377 size: 16 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (131:146) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (321:336) duplicated block id: 378 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 379 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 380 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 381 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 382 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 383 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) duplicated block id: 384 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 385 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) duplicated block id: 386 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 387 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:258) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (586:608) duplicated block id: 388 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) duplicated block id: 389 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 390 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) duplicated block id: 391 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (194:223) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (412:441) duplicated block id: 392 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:949) duplicated block id: 393 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) duplicated block id: 394 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 395 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 396 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 397 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (56:72) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (157:173) duplicated block id: 398 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:966) duplicated block id: 399 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:983) duplicated block id: 400 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1000) duplicated block id: 401 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1017) duplicated block id: 402 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) duplicated block id: 403 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 404 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 405 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 406 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 407 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 408 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 409 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 410 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 411 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (1:15) duplicated block id: 412 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (316:331) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (684:698) duplicated block id: 413 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (242:262) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (202:222) duplicated block id: 414 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) duplicated block id: 415 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (8:23) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (17:32) duplicated block id: 416 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 417 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) duplicated block id: 418 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 419 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 420 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) duplicated block id: 421 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 422 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (1:15) duplicated block id: 423 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 424 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (113:151) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (164:204) duplicated block id: 425 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 426 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 427 size: 15 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (53:69) - maga_transformer/device/device_impl.py (149:165) duplicated block id: 428 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (166:180) - maga_transformer/cpp/kernels/rotary_position_embedding.h (230:244) duplicated block id: 429 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) duplicated block id: 430 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 431 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 432 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 433 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (974:991) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1024:1041) duplicated block id: 434 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (532:550) - maga_transformer/cpp/kernels/rotary_position_embedding.h (590:608) duplicated block id: 435 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (99:128) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (412:441) duplicated block id: 436 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:258) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (893:915) duplicated block id: 437 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 438 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 439 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 440 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 441 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 442 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 443 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (656:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1248:1262) duplicated block id: 444 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 445 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 446 size: 15 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (53:69) - maga_transformer/device/device_impl.py (283:299) duplicated block id: 447 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 448 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 449 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 450 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 451 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) duplicated block id: 452 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 453 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:258) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (689:711) duplicated block id: 454 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (172:194) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (421:443) duplicated block id: 455 size: 15 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (84:98) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (142:156) duplicated block id: 456 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:258) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (789:811) duplicated block id: 457 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 458 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 459 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 460 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 461 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 462 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 463 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 464 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 465 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) duplicated block id: 466 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasAlgoMap.cc (206:220) - maga_transformer/cpp/rocm/hipblasAlgoMap.h (86:100) duplicated block id: 467 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 468 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) duplicated block id: 469 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/l1norm_kernels.cu (78:94) - maga_transformer/cpp/kernels/layernorm_kernels.cu (657:673) duplicated block id: 470 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) duplicated block id: 471 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/l1norm_kernels.cu (78:94) - maga_transformer/cpp/kernels/layernorm_kernels.cu (615:631) duplicated block id: 472 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 473 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 474 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 475 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 476 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 477 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.h (119:136) - maga_transformer/cpp/rocm/hip_utils.h (44:60) duplicated block id: 478 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (71:87) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (485:501) duplicated block id: 479 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 480 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 481 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 482 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (363:377) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (414:428) duplicated block id: 483 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 484 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) duplicated block id: 485 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 486 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 487 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 488 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (491:516) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (646:673) duplicated block id: 489 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) duplicated block id: 490 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) duplicated block id: 491 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 492 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) duplicated block id: 493 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) duplicated block id: 494 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 495 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (656:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1067:1081) duplicated block id: 496 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (8:23) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (19:34) duplicated block id: 497 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 498 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 499 size: 15 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (57:74) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (60:77) duplicated block id: 500 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (657:673) - maga_transformer/cpp/kernels/rmsnormKernels.cu (326:342) duplicated block id: 501 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 502 size: 15 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (328:342) - maga_transformer/utils/smooth_quant_convert/qwen/hf_qwen_convert.py (87:101) duplicated block id: 503 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 504 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (431:446) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (782:797) duplicated block id: 505 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (657:673) - maga_transformer/cpp/kernels/rmsnormKernels.cu (414:430) duplicated block id: 506 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 507 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 508 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 509 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 510 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 511 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 512 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.h (37:51) - maga_transformer/cpp/kernels/sampling_topp_kernels.h (60:74) duplicated block id: 513 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (799:826) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (954:979) duplicated block id: 514 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 515 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (615:631) - maga_transformer/cpp/kernels/rmsnormKernels.cu (414:430) duplicated block id: 516 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) duplicated block id: 517 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (479:498) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (527:546) duplicated block id: 518 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (615:631) - maga_transformer/cpp/kernels/rmsnormKernels.cu (326:342) duplicated block id: 519 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:19) - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:19) duplicated block id: 520 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 521 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 522 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) duplicated block id: 523 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 524 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) duplicated block id: 525 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 526 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 527 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 528 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (164:204) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (233:271) duplicated block id: 529 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 530 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) duplicated block id: 531 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 532 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (265:279) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (283:297) duplicated block id: 533 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 534 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 535 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (724:748) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (716:740) duplicated block id: 536 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 537 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (62:101) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (182:221) duplicated block id: 538 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 539 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 540 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 541 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) duplicated block id: 542 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 543 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 544 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 545 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 546 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 547 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 548 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) duplicated block id: 549 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 550 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 551 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (113:151) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (284:324) duplicated block id: 552 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (233:271) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (284:324) duplicated block id: 553 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 554 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) duplicated block id: 555 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 556 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 557 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 558 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 559 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (1:15) duplicated block id: 560 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 561 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 562 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 563 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 564 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (188:205) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (602:619) duplicated block id: 565 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 566 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (228:254) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (193:220) duplicated block id: 567 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 568 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) duplicated block id: 569 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (121:135) - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (142:156) duplicated block id: 570 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 571 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 572 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) duplicated block id: 573 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) duplicated block id: 574 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 575 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (212:232) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (492:511) duplicated block id: 576 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 577 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 578 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 579 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 580 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 581 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 582 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 583 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 584 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (245:259) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (265:279) duplicated block id: 585 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) duplicated block id: 586 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 587 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (245:259) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (283:297) duplicated block id: 588 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 589 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (1:15) duplicated block id: 590 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 591 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 592 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (646:673) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1106:1131) duplicated block id: 593 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 594 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 595 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:167) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (893:915) duplicated block id: 596 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 597 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 598 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 599 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 600 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 601 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 602 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 603 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 604 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 605 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) duplicated block id: 606 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 607 size: 15 cleaned lines of code in 2 files: - maga_transformer/openai/openai_endpoint.py (166:181) - maga_transformer/openai/renderers/custom_renderer.py (898:913) duplicated block id: 608 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 609 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 610 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 611 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (904:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1067:1081) duplicated block id: 612 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:167) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (586:608) duplicated block id: 613 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 614 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 615 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:167) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (789:811) duplicated block id: 616 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (18:32) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (79:93) duplicated block id: 617 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 618 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:167) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (689:711) duplicated block id: 619 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 620 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 621 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (142:163) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (623:644) duplicated block id: 622 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 623 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1:15) duplicated block id: 624 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 625 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (96:119) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (138:160) duplicated block id: 626 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (1:15) duplicated block id: 627 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 628 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 629 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 630 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 631 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 632 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (1:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (1:15) duplicated block id: 633 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (1:15) duplicated block id: 634 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 635 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 636 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 637 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/int8_utils.cuh (1:15) duplicated block id: 638 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 639 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 640 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) duplicated block id: 641 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 642 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (338:363) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (799:826) duplicated block id: 643 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 644 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (1:15) duplicated block id: 645 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 646 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 647 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) duplicated block id: 648 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 649 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) duplicated block id: 650 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) duplicated block id: 651 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/cuda/memory_utils.cu (1:15) duplicated block id: 652 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) duplicated block id: 653 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 654 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 655 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (76:90) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (235:249) duplicated block id: 656 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 657 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (1:15) - maga_transformer/cpp/kernels/quantize_weight.cu (1:15) duplicated block id: 658 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) duplicated block id: 659 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 660 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (1:15) duplicated block id: 661 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (1:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (1:15) duplicated block id: 662 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (1:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (1:15) duplicated block id: 663 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 664 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (1:15) duplicated block id: 665 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 666 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 667 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (1:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (1:15) duplicated block id: 668 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 669 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:983) duplicated block id: 670 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:966) duplicated block id: 671 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1017) duplicated block id: 672 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:1000) duplicated block id: 673 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:949) duplicated block id: 674 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (1:15) duplicated block id: 675 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (1:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (1:15) duplicated block id: 676 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (1:15) duplicated block id: 677 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (1:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (1:15) duplicated block id: 678 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (904:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1248:1262) duplicated block id: 679 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (1:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (1:15) duplicated block id: 680 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (1:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (1:15) duplicated block id: 681 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (1:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (1:15) duplicated block id: 682 size: 15 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (844:858) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (106:120) duplicated block id: 683 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (36:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (491:514) duplicated block id: 684 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (449:462) - maga_transformer/cpp/kernels/rmsnormKernels.cu (152:165) duplicated block id: 685 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (42:56) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (43:57) duplicated block id: 686 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2051:2065) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2102:2116) duplicated block id: 687 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:17) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:18) duplicated block id: 688 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (184:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (646:671) duplicated block id: 689 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (3:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (3:16) duplicated block id: 690 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (36:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (799:824) duplicated block id: 691 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/propose_executor/DeterministicExecutor.cc (71:86) - maga_transformer/cpp/speculative_engine/propose_executor/DeterministicExecutor.cc (120:135) duplicated block id: 692 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (184:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (954:977) duplicated block id: 693 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (36:61) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1106:1129) duplicated block id: 694 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (516:529) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (534:547) duplicated block id: 695 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (295:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (329:342) duplicated block id: 696 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (3:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (3:16) duplicated block id: 697 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (62:99) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (110:145) duplicated block id: 698 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:17) - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:16) duplicated block id: 699 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (86:99) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (102:115) duplicated block id: 700 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (110:145) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (182:219) duplicated block id: 701 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (229:257) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:118) duplicated block id: 702 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (207:222) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (299:314) duplicated block id: 703 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (905:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1369:1382) duplicated block id: 704 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1068:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1369:1382) duplicated block id: 705 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (184:209) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (338:361) duplicated block id: 706 size: 14 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (751:764) - maga_transformer/openai/renderers/custom_renderer.py (817:830) duplicated block id: 707 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (182:219) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (230:265) duplicated block id: 708 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (3:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (3:16) duplicated block id: 709 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:169) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (223:236) duplicated block id: 710 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:16) - maga_transformer/cpp/kernels/int8_utils.cuh (3:17) duplicated block id: 711 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:145) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (203:217) duplicated block id: 712 size: 14 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/resampler.py (56:81) - maga_transformer/models/qwen_vl_vit.py (52:77) duplicated block id: 713 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (407:420) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (456:469) duplicated block id: 714 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (3:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (3:16) duplicated block id: 715 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (559:576) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (629:646) duplicated block id: 716 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (3:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (3:16) duplicated block id: 717 size: 14 cleaned lines of code in 2 files: - maga_transformer/utils/model_weight.py (157:170) - maga_transformer/utils/model_weight.py (180:194) duplicated block id: 718 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (103:116) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (311:324) duplicated block id: 719 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (593:606) - maga_transformer/cpp/trt_plugins/mixtureOfExperts/mixtureOfExpertsPlugin.h (66:79) duplicated block id: 720 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (166:181) - maga_transformer/cpp/devices/rocm_impl/ROCmFfnLayer.cc (217:232) duplicated block id: 721 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:17) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:18) duplicated block id: 722 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.h (32:53) - maga_transformer/cpp/rocm/quantizePreprocessors.h (12:34) duplicated block id: 723 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (3:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (3:16) duplicated block id: 724 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:236) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (244:257) duplicated block id: 725 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (62:99) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (230:265) duplicated block id: 726 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (718:733) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (892:907) duplicated block id: 727 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (657:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1369:1382) duplicated block id: 728 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1249:1262) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1369:1382) duplicated block id: 729 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:17) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:17) duplicated block id: 730 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (214:232) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (189:207) duplicated block id: 731 size: 14 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (63:79) - maga_transformer/model_loader/smooth_quant_weight.py (134:150) duplicated block id: 732 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (91:104) - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (168:181) duplicated block id: 733 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (277:300) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (446:469) duplicated block id: 734 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:17) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:17) duplicated block id: 735 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (422:442) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (613:634) duplicated block id: 736 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:17) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:17) duplicated block id: 737 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:17) - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:16) duplicated block id: 738 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (422:442) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (491:511) duplicated block id: 739 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:18) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:17) duplicated block id: 740 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:17) - maga_transformer/cpp/kernels/int8_utils.cuh (3:17) duplicated block id: 741 size: 14 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (3:16) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (3:16) duplicated block id: 742 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 743 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 744 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 745 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 746 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 747 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 748 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 749 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 750 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 751 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 752 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:774) duplicated block id: 753 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 754 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 755 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 756 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 757 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 758 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 759 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1064:1076) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1142:1154) duplicated block id: 760 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 761 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 762 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 763 size: 13 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (385:399) - maga_transformer/openai/renderers/llama_template.py (413:427) duplicated block id: 764 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 765 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 766 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 767 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:947) duplicated block id: 768 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 769 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 770 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 771 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 772 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 773 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 774 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) duplicated block id: 775 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 776 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 777 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) duplicated block id: 778 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 779 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1043:1055) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1120:1132) duplicated block id: 780 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 781 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 782 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 783 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/mixtureOfExperts/mixtureOfExpertsPlugin.cpp (145:157) - maga_transformer/cpp/trt_plugins/mixtureOfExperts/mixtureOfExpertsPlugin.h (66:78) duplicated block id: 784 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 785 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 786 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 787 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 788 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 789 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 790 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 791 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 792 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 793 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 794 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 795 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 796 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 797 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 798 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 799 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 800 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) duplicated block id: 801 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 802 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 803 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 804 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 805 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) duplicated block id: 806 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 807 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:612) duplicated block id: 808 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 809 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 810 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 811 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 812 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 813 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 814 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 815 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 816 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (331:345) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (70:84) duplicated block id: 817 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 818 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 819 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 820 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 821 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 822 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 823 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 824 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (3:20) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (28:45) duplicated block id: 825 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 826 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 827 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 828 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 829 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 830 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 831 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 832 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 833 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (866:887) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (998:1018) duplicated block id: 834 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 835 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 836 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 837 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 838 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 839 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 840 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 841 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) duplicated block id: 842 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 843 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 844 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) duplicated block id: 845 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 846 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 847 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 848 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) duplicated block id: 849 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 850 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 851 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 852 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) duplicated block id: 853 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 854 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 855 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 856 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 857 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (300:313) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (670:683) duplicated block id: 858 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 859 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 860 size: 13 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (116:144) - maga_transformer/tokenizer/tokenization_chatglm4.py (95:123) duplicated block id: 861 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 862 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 863 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 864 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 865 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 866 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (23:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (23:51) duplicated block id: 867 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 868 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 869 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 870 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (989:1001) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1054:1066) duplicated block id: 871 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 872 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 873 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 874 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (103:115) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (154:166) duplicated block id: 875 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 876 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 877 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 878 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 879 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 880 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 881 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 882 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 883 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 884 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 885 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 886 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 887 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 888 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 889 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 890 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 891 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) duplicated block id: 892 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 893 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 894 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 895 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 896 size: 13 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (23:35) - maga_transformer/models/whisper.py (55:67) duplicated block id: 897 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (54:66) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (69:81) duplicated block id: 898 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 899 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (64:92) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (102:130) duplicated block id: 900 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 901 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 902 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 903 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 904 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 905 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 906 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_float.h (83:114) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2302:2333) duplicated block id: 907 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 908 size: 13 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/resampler.py (84:107) - maga_transformer/models/qwen_vl_vit.py (80:102) duplicated block id: 909 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) duplicated block id: 910 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) duplicated block id: 911 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 912 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) duplicated block id: 913 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 914 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 915 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 916 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 917 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 918 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 919 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 920 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 921 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 922 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 923 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 924 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 925 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 926 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 927 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (110:123) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (113:126) duplicated block id: 928 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 929 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 930 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 931 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 932 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) duplicated block id: 933 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) duplicated block id: 934 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 935 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 936 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 937 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 938 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 939 size: 13 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2.py (44:56) - maga_transformer/models/llava.py (93:105) duplicated block id: 940 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 941 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 942 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 943 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 944 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) duplicated block id: 945 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 946 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 947 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 948 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 949 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 950 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 951 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 952 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 953 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1015) duplicated block id: 954 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 955 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:998) duplicated block id: 956 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 957 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 958 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 959 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 960 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 961 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 962 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 963 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 964 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 965 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 966 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 967 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 968 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 969 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 970 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 971 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 972 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (87:99) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (154:166) duplicated block id: 973 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (696:711) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (865:880) duplicated block id: 974 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 975 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 976 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 977 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 978 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 979 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 980 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 981 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_float.h (36:61) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2243:2269) duplicated block id: 982 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 983 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) duplicated block id: 984 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 985 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 986 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 987 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 988 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 989 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 990 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 991 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 992 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 993 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 994 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 995 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:964) duplicated block id: 996 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 997 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (302:314) - maga_transformer/cpp/kernels/vec_dtypes.cuh (335:347) duplicated block id: 998 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:981) duplicated block id: 999 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1000 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1001 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:998) duplicated block id: 1002 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1015) duplicated block id: 1003 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:947) duplicated block id: 1004 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1005 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1006 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1007 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1008 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1009 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1010 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (70:82) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (135:147) duplicated block id: 1011 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1012 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1013 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (342:354) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (68:80) duplicated block id: 1014 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1015 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1016 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:774) duplicated block id: 1017 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1018 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1019 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1020 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (70:82) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (103:115) duplicated block id: 1021 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (70:82) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (87:99) duplicated block id: 1022 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1023 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1024 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) duplicated block id: 1025 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1026 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1027 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1028 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1029 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1030 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1031 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1032 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1033 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1034 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1035 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:612) duplicated block id: 1036 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1037 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1038 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1039 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1040 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1041 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1042 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1043 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1044 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1045 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) duplicated block id: 1046 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1047 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1048 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1049 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1050 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1051 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1052 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1053 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1054 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1055 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1056 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1057 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1058 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1059 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (207:219) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (251:263) duplicated block id: 1060 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (88:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (120:132) duplicated block id: 1061 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1062 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1063 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) duplicated block id: 1064 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1065 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1066 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1067 size: 13 cleaned lines of code in 2 files: - example/perf_test/defs.bzl (30:42) - example/perf_test/defs.bzl (86:98) duplicated block id: 1068 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1069 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1070 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1071 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1072 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1073 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (32:47) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (452:468) duplicated block id: 1074 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1075 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1076 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1077 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1078 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1122:1134) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1144:1156) duplicated block id: 1079 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1080 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1081 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1082 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1083 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (81:102) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (182:203) duplicated block id: 1084 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1085 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (434:446) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (90:102) duplicated block id: 1086 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1087 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (25:37) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (50:62) duplicated block id: 1088 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1253:1265) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1276:1288) duplicated block id: 1089 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1090 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1091 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1092 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1093 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1094 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1095 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1096 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1097 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) duplicated block id: 1098 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1099 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1100 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1101 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1102 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1103 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1104 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1105 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cuda/memory_utils.cu (3:15) duplicated block id: 1106 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1107 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1108 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1109 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1110 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1111 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1112 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1113 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1114 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1115 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1116 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1117 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1118 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1119 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1120 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1121 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1122 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1123 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) duplicated block id: 1124 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1125 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1126 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1127 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1128 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 1129 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1130 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1131 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1132 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1133 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1134 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1135 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1136 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (188:201) - maga_transformer/cpp/devices/rocm_impl/ROCmOps.cc (10:23) duplicated block id: 1137 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1138 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1139 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cuda/memory_utils.cu (3:15) duplicated block id: 1140 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1141 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1142 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1143 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1144 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1145 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1146 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1147 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1148 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (551:563) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (247:259) duplicated block id: 1149 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1150 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1151 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1152 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1153 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1154 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1155 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1156 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1157 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1158 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1159 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1160 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1161 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1162 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1163 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1164 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1165 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1166 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1167 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1168 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) duplicated block id: 1169 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1170 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (87:99) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (135:147) duplicated block id: 1171 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1172 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1173 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1174 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (7:19) - maga_transformer/cpp/rocm/int4_gemm_kernels/34_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v4.cc (7:19) duplicated block id: 1175 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1176 size: 13 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (268:284) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (337:353) duplicated block id: 1177 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1178 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1179 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1180 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1181 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1182 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1183 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1184 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) duplicated block id: 1185 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1186 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1187 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1188 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1189 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1190 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1191 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1192 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1193 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (369:383) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (472:484) duplicated block id: 1194 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:964) duplicated block id: 1195 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1196 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1197 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1198 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1199 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1200 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1201 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1202 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1203 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1204 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1205 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (70:82) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (154:166) duplicated block id: 1206 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1207 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 1208 size: 13 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (337:353) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (396:412) duplicated block id: 1209 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) duplicated block id: 1210 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1015) duplicated block id: 1211 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:998) duplicated block id: 1212 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1213 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1214 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1215 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1216 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (121:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:981) duplicated block id: 1217 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1218 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1219 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1220 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1221 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1222 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1223 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1224 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1225 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1226 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (42:54) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (58:70) duplicated block id: 1227 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1228 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1229 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1230 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1231 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1232 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1233 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1234 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1235 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1236 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1237 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1238 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1239 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1240 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1241 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1242 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1243 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1244 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1245 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1246 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1247 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1248 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1249 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1250 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1251 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1252 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1253 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1254 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1255 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1256 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1257 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1258 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1259 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1260 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1261 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1262 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1263 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (135:149) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (70:84) duplicated block id: 1264 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1265 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1045:1057) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1066:1078) duplicated block id: 1266 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1267 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1268 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1269 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1270 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1271 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1272 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1273 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1274 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1275 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1276 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1277 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1278 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1279 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1280 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1281 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1282 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1283 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1284 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1285 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1286 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (651:663) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (705:717) duplicated block id: 1287 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1288 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1289 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1290 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1291 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1292 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/triton/aot_triton_kernel.bzl (128:140) - maga_transformer/cpp/kernels/triton/aot_triton_kernel.bzl (163:176) duplicated block id: 1293 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1294 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1295 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1296 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1297 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1298 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1299 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1300 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1301 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1302 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1303 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1304 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1305 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1306 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (408:421) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (547:560) duplicated block id: 1307 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) duplicated block id: 1308 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1309 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1310 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1311 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1312 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1313 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) duplicated block id: 1314 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1315 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1316 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) duplicated block id: 1317 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1318 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1319 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1320 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1321 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1322 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1323 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) duplicated block id: 1324 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1325 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1326 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1327 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1328 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1329 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1330 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1331 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) duplicated block id: 1332 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1333 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1334 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1335 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1336 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1337 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1338 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1339 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1340 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1341 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1342 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1343 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1344 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1345 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1346 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1347 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1348 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1349 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1350 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1351 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1352 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1353 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1354 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1355 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1356 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1357 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1358 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1359 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (172:186) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (586:600) duplicated block id: 1360 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1361 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1362 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1363 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1364 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1365 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1366 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1367 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1368 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1369 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1370 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1371 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1372 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1373 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1374 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1375 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1376 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) duplicated block id: 1377 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1378 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1379 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) duplicated block id: 1380 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1381 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1382 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1383 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1384 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1385 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1386 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1387 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1388 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (656:675) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (712:731) duplicated block id: 1389 size: 13 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm3.py (188:216) - maga_transformer/tokenizer/tokenization_chatglm4.py (95:123) duplicated block id: 1390 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1391 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1392 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1393 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1394 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1395 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1396 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1397 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1398 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1399 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1400 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1401 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1402 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (229:253) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (232:258) duplicated block id: 1403 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1404 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1405 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1406 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1407 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1408 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1409 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1410 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1411 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1412 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1413 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1414 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (389:402) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (456:468) duplicated block id: 1415 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1416 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1417 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1418 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1419 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1420 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1421 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1422 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1423 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1424 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1425 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1426 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (593:605) - maga_transformer/cpp/trt_plugins/mixtureOfExperts/mixtureOfExpertsPlugin.cpp (145:157) duplicated block id: 1427 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (84:97) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (67:80) duplicated block id: 1428 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1429 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1430 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1431 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1432 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1433 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1434 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1435 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1436 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1437 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1438 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1439 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) duplicated block id: 1440 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) duplicated block id: 1441 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1442 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1443 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1444 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1445 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1446 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1447 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1448 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1449 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1450 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1451 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1452 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1453 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1454 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1455 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1456 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1457 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1458 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1459 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1460 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1461 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 1462 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1463 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1464 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1465 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.h (38:50) - maga_transformer/cpp/kernels/activation_kernels.h (55:67) duplicated block id: 1466 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1467 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1468 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1469 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1470 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1471 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1472 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:612) duplicated block id: 1473 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1474 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1475 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1476 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1477 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1478 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1479 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) duplicated block id: 1480 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1481 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) duplicated block id: 1482 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1483 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1484 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1485 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1486 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1487 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1488 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1489 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1490 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1491 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1492 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1493 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1494 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1495 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1496 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1497 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1498 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1499 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1500 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:981) duplicated block id: 1501 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1502 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:964) duplicated block id: 1503 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1504 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1505 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1506 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1507 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1508 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) duplicated block id: 1509 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1510 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1511 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1512 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:947) duplicated block id: 1513 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1514 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1515 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) duplicated block id: 1516 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1517 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1518 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1519 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:774) duplicated block id: 1520 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1521 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 1522 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1523 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1524 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1525 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1526 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1527 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1528 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1529 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1530 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (103:115) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (135:147) duplicated block id: 1531 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1532 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1533 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1534 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1535 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1536 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1537 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1538 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1539 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1540 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1541 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1542 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1543 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1544 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1545 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1546 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (1744:1758) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (1861:1875) duplicated block id: 1547 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1548 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1549 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1550 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1551 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1552 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1553 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1554 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1555 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) duplicated block id: 1556 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1557 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1558 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1559 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1560 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1561 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1562 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1563 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1564 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1565 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1566 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1567 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1568 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) duplicated block id: 1569 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1570 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1571 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1572 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1573 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1574 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1575 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1576 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1577 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1578 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) duplicated block id: 1579 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1580 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1581 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1582 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) duplicated block id: 1583 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1584 size: 13 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (65:79) - maga_transformer/model_loader/smooth_quant_weight.py (291:305) duplicated block id: 1585 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1586 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1587 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1588 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1589 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) duplicated block id: 1590 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1591 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1592 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1593 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1594 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1595 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1596 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1597 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1598 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1599 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) duplicated block id: 1600 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1601 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1602 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (3:15) duplicated block id: 1603 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) duplicated block id: 1604 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) duplicated block id: 1605 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1606 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1607 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1608 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1609 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (3:15) duplicated block id: 1610 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1611 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (3:15) - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) duplicated block id: 1612 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1613 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1614 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1615 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1616 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1617 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1618 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/gpt_kernels.cu (3:15) duplicated block id: 1619 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1620 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1621 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) duplicated block id: 1622 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1623 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1624 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1625 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (273:292) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (738:757) duplicated block id: 1626 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) duplicated block id: 1627 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (3:15) duplicated block id: 1628 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1629 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1630 size: 13 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (136:150) - maga_transformer/model_loader/smooth_quant_weight.py (291:305) duplicated block id: 1631 size: 13 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (209:225) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (337:353) duplicated block id: 1632 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1633 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1634 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1635 size: 13 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (209:225) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (396:412) duplicated block id: 1636 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1637 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (3:15) duplicated block id: 1638 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) duplicated block id: 1639 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1640 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (3:15) duplicated block id: 1641 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) duplicated block id: 1642 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1643 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/rmsnormKernels.cu (3:15) duplicated block id: 1644 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (3:15) duplicated block id: 1645 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1646 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1647 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1648 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1649 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1650 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1651 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1652 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1653 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1654 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (3:15) duplicated block id: 1655 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1656 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1657 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1658 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1659 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (3:15) - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) duplicated block id: 1660 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) duplicated block id: 1661 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (273:292) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (912:931) duplicated block id: 1662 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (3:15) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (4:16) duplicated block id: 1663 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (3:15) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (4:16) duplicated block id: 1664 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (3:15) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (4:16) duplicated block id: 1665 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (3:15) - maga_transformer/cpp/kernels/quantization_tensor.cu (3:15) duplicated block id: 1666 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (3:15) - maga_transformer/cpp/kernels/quantize_weight.cu (3:15) duplicated block id: 1667 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (3:15) - maga_transformer/cpp/kernels/logprob_kernels.cu (3:15) duplicated block id: 1668 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (3:15) duplicated block id: 1669 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (3:15) - maga_transformer/cpp/kernels/banRepeatNgram.cu (3:15) duplicated block id: 1670 size: 13 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (209:225) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (268:284) duplicated block id: 1671 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (3:15) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (3:15) duplicated block id: 1672 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (3:15) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (3:15) duplicated block id: 1673 size: 13 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (3:15) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (3:15) duplicated block id: 1674 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:658) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:727) duplicated block id: 1675 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/1_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v4.cc (7:18) - maga_transformer/cpp/rocm/int4_gemm_kernels/3_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v4.cc (7:18) duplicated block id: 1676 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1142:1153) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1320:1331) duplicated block id: 1677 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (53:65) - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (82:94) duplicated block id: 1678 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (137:149) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (59:71) duplicated block id: 1679 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1298) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1333) duplicated block id: 1680 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (534:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1187:1198) duplicated block id: 1681 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (574:586) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (298:310) duplicated block id: 1682 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (782:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1187:1198) duplicated block id: 1683 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (488:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1142:1153) duplicated block id: 1684 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (488:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1320:1331) duplicated block id: 1685 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:548) - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:596) duplicated block id: 1686 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (133:144) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (171:182) duplicated block id: 1687 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (639:657) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (803:821) duplicated block id: 1688 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:709) - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:749) duplicated block id: 1689 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/eplb/ExpertBalancer.cc (102:113) - maga_transformer/cpp/eplb/ExpertBalancer.h (73:84) duplicated block id: 1690 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (488:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (735:746) duplicated block id: 1691 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (940:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1104:1115) duplicated block id: 1692 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (333:345) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (59:71) duplicated block id: 1693 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (488:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (975:986) duplicated block id: 1694 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (127:140) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (433:446) duplicated block id: 1695 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (30:43) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (76:89) duplicated block id: 1696 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/image_processing_qwen2_vl.py (424:435) - maga_transformer/models/qwen2_vl/image_processing_qwen2_vl.py (446:457) duplicated block id: 1697 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaActOp.cc (66:80) - maga_transformer/cpp/devices/rocm_impl/ROCmActOp.cc (65:79) duplicated block id: 1698 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (582:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (830:841) duplicated block id: 1699 size: 12 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (493:504) - maga_transformer/openai/renderers/custom_renderer.py (710:721) duplicated block id: 1700 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (135:147) - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (167:179) duplicated block id: 1701 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (236:254) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (220:238) duplicated block id: 1702 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (172:184) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (262:274) duplicated block id: 1703 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (123:134) - maga_transformer/cpp/kernels/ban_bad_words.cu (151:162) duplicated block id: 1704 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (592:608) duplicated block id: 1705 size: 12 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (305:316) - maga_transformer/utils/smooth_quant_convert/qwen/hf_qwen_convert.py (57:68) duplicated block id: 1706 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (448:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1104:1115) duplicated block id: 1707 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (164:198) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (185:219) duplicated block id: 1708 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:501) - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:596) duplicated block id: 1709 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (695:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1104:1115) duplicated block id: 1710 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (468:479) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (169:180) duplicated block id: 1711 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (243:258) duplicated block id: 1712 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (695:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (940:951) duplicated block id: 1713 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:501) - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:548) duplicated block id: 1714 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (185:219) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (284:318) duplicated block id: 1715 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (51:62) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (35:46) duplicated block id: 1716 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (448:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1285:1296) duplicated block id: 1717 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:173) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (197:221) duplicated block id: 1718 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (448:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (695:706) duplicated block id: 1719 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (21:33) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (37:49) duplicated block id: 1720 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (152:167) duplicated block id: 1721 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (448:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (940:951) duplicated block id: 1722 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (28:42) - maga_transformer/utils/model_weight.py (434:447) duplicated block id: 1723 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1013:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1187:1198) duplicated block id: 1724 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fpA_intB_gemm.h (358:375) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cutlass_kernel.h (335:352) duplicated block id: 1725 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:462) - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:596) duplicated block id: 1726 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (65:99) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (164:198) duplicated block id: 1727 size: 12 cleaned lines of code in 2 files: - maga_transformer/server/backend_app.py (92:104) - maga_transformer/server/frontend_app.py (97:109) duplicated block id: 1728 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (158:176) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (372:390) duplicated block id: 1729 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (782:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1013:1024) duplicated block id: 1730 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (695:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1285:1296) duplicated block id: 1731 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_heuristic.cc (232:246) - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_heuristic.cc (428:442) duplicated block id: 1732 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (387:405) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (398:416) duplicated block id: 1733 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (78:89) - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (54:65) duplicated block id: 1734 size: 12 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (98:113) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (99:114) duplicated block id: 1735 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:954) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:989) duplicated block id: 1736 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (940:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1285:1296) duplicated block id: 1737 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:954) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1026) duplicated block id: 1738 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:462) - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:501) duplicated block id: 1739 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:462) - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:548) duplicated block id: 1740 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (899:915) duplicated block id: 1741 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (695:711) duplicated block id: 1742 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (318:336) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (326:344) duplicated block id: 1743 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (22:37) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (795:811) duplicated block id: 1744 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1104:1115) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1285:1296) duplicated block id: 1745 size: 12 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (296:322) - maga_transformer/tokenizer/tokenization_chatglm4.py (95:121) duplicated block id: 1746 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (59:71) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (72:84) duplicated block id: 1747 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:18) - maga_transformer/cpp/rocm/int4_gemm_kernels/4_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v3.cc (7:18) duplicated block id: 1748 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (735:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (975:986) duplicated block id: 1749 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (355:366) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (717:728) duplicated block id: 1750 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/1_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v4.cc (7:18) - maga_transformer/cpp/rocm/int4_gemm_kernels/4_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v3.cc (7:18) duplicated block id: 1751 size: 12 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (296:322) - maga_transformer/tokenizer/tokenization_chatglm2.py (116:142) duplicated block id: 1752 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (167:179) - maga_transformer/models/qwen2_vl/modeling_qwen2_vl.py (190:203) duplicated block id: 1753 size: 12 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tokenization_qwen.py (123:134) - maga_transformer/tokenizer/tokenization_qwen.py (139:150) duplicated block id: 1754 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:749) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:843) duplicated block id: 1755 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:749) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:796) duplicated block id: 1756 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (534:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1013:1024) duplicated block id: 1757 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1003:1014) duplicated block id: 1758 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (952:963) duplicated block id: 1759 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (986:997) duplicated block id: 1760 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (969:980) duplicated block id: 1761 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (935:946) duplicated block id: 1762 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (534:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (782:793) duplicated block id: 1763 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (55:66) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (154:165) duplicated block id: 1764 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (341:352) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (355:366) duplicated block id: 1765 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (330:341) duplicated block id: 1766 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (296:307) duplicated block id: 1767 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (762:773) duplicated block id: 1768 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (77:101) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:173) duplicated block id: 1769 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (87:100) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (423:436) duplicated block id: 1770 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (89:100) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (600:611) duplicated block id: 1771 size: 12 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (296:322) - maga_transformer/tokenizer/tokenization_chatglm3.py (188:214) duplicated block id: 1772 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:796) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:843) duplicated block id: 1773 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (735:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1320:1331) duplicated block id: 1774 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (735:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1142:1153) duplicated block id: 1775 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:709) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:843) duplicated block id: 1776 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:709) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:796) duplicated block id: 1777 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (386:397) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (423:434) duplicated block id: 1778 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (808:822) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (887:901) duplicated block id: 1779 size: 12 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (260:273) - maga_transformer/device/device_impl.py (330:343) duplicated block id: 1780 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (975:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1320:1331) duplicated block id: 1781 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (55:66) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (87:98) duplicated block id: 1782 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (108:121) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (184:197) duplicated block id: 1783 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (55:66) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (103:114) duplicated block id: 1784 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (55:66) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (135:146) duplicated block id: 1785 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (264:275) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (639:650) duplicated block id: 1786 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (293:307) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (399:413) duplicated block id: 1787 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (975:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1142:1153) duplicated block id: 1788 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (17:31) - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (116:130) duplicated block id: 1789 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1117) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1200) duplicated block id: 1790 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1117) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1155) duplicated block id: 1791 size: 12 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/resampler.py (117:131) - maga_transformer/models/qwen_vl_vit.py (110:124) duplicated block id: 1792 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1346:1357) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1368:1379) duplicated block id: 1793 size: 12 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (355:366) - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (381:392) duplicated block id: 1794 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (65:99) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (284:318) duplicated block id: 1795 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:989) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1026) duplicated block id: 1796 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1155) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1200) duplicated block id: 1797 size: 12 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:18) - maga_transformer/cpp/rocm/int4_gemm_kernels/3_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v4.cc (7:18) duplicated block id: 1798 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (29:39) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (43:53) duplicated block id: 1799 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (700:714) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1112:1126) duplicated block id: 1800 size: 11 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (761:794) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (774:805) duplicated block id: 1801 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (176:198) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:171) duplicated block id: 1802 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (768:782) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1045:1059) duplicated block id: 1803 size: 11 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (61:74) - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (100:113) duplicated block id: 1804 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (161:171) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (373:383) duplicated block id: 1805 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (147:157) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (163:173) duplicated block id: 1806 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:916) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:982) duplicated block id: 1807 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (700:714) - maga_transformer/cpp/kernels/_fma.h (768:782) duplicated block id: 1808 size: 11 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (21:31) - maga_transformer/model_loader/smooth_quant_weight.py (236:246) duplicated block id: 1809 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/group_gemm/group_gemm.h (22:33) - maga_transformer/cpp/trt_plugins/GroupGemmPlugin/GroupGemmPlugin.h (10:20) duplicated block id: 1810 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (23:33) - maga_transformer/cpp/cuda/cufmha/cufmha.h (16:26) duplicated block id: 1811 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (585:595) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (985:995) duplicated block id: 1812 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (718:728) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (844:854) duplicated block id: 1813 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (366:376) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (144:154) duplicated block id: 1814 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (784:794) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (887:897) duplicated block id: 1815 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:53) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (300:320) duplicated block id: 1816 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:53) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (247:267) duplicated block id: 1817 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (460:470) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (473:483) duplicated block id: 1818 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (460:470) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (486:496) duplicated block id: 1819 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (168:178) - maga_transformer/cpp/devices/utils/DebugUtils.cc (214:224) duplicated block id: 1820 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (473:483) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (486:496) duplicated block id: 1821 size: 11 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_renderer.py (62:75) - maga_transformer/openai/renderers/qwen_renderer.py (85:98) duplicated block id: 1822 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (238:248) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (977:987) duplicated block id: 1823 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (260:270) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (384:395) duplicated block id: 1824 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2151:2163) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2202:2214) duplicated block id: 1825 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (287:297) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (810:820) duplicated block id: 1826 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (587:598) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (122:133) duplicated block id: 1827 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (623:637) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:916) duplicated block id: 1828 size: 11 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (372:382) - maga_transformer/openai/renderers/llama_template.py (401:411) duplicated block id: 1829 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (604:614) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (265:275) duplicated block id: 1830 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (421:431) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (434:444) duplicated block id: 1831 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (421:431) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (447:457) duplicated block id: 1832 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (104:116) - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (127:139) duplicated block id: 1833 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (532:545) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (606:619) duplicated block id: 1834 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (29:40) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (29:40) duplicated block id: 1835 size: 11 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (98:108) - bazel/tf_http_archive.bzl (248:258) duplicated block id: 1836 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (271:281) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (70:80) duplicated block id: 1837 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (106:116) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (137:147) duplicated block id: 1838 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (180:200) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:53) duplicated block id: 1839 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (112:124) - maga_transformer/cpp/kernels/quantization_tensor.cu (186:198) duplicated block id: 1840 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (127:147) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:53) duplicated block id: 1841 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (219:229) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (127:137) duplicated block id: 1842 size: 11 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (592:602) - maga_transformer/openai/renderers/qwen_renderer.py (425:435) duplicated block id: 1843 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (258:274) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (268:284) duplicated block id: 1844 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (30:46) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (51:67) duplicated block id: 1845 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:600) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:726) duplicated block id: 1846 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:600) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:657) duplicated block id: 1847 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1045:1059) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1112:1126) duplicated block id: 1848 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/score_executor/ScoreExecutor.cc (10:22) - maga_transformer/cpp/speculative_engine/score_executor/ScoreExecutor.cc (43:55) duplicated block id: 1849 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:571) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:982) duplicated block id: 1850 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (218:230) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (600:612) duplicated block id: 1851 size: 11 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/internvl_renderer.py (42:52) - maga_transformer/openai/renderers/llava_renderer.py (73:83) duplicated block id: 1852 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (118:131) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (83:96) duplicated block id: 1853 size: 11 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (61:74) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (101:114) duplicated block id: 1854 size: 11 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (668:679) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (622:633) duplicated block id: 1855 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (295:310) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (256:271) duplicated block id: 1856 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:171) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (296:318) duplicated block id: 1857 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:171) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (245:265) duplicated block id: 1858 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (718:728) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (137:147) duplicated block id: 1859 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (718:728) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (106:116) duplicated block id: 1860 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:544) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:726) duplicated block id: 1861 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:544) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:600) duplicated block id: 1862 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:544) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:657) duplicated block id: 1863 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (861:873) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1112:1124) duplicated block id: 1864 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (408:418) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (447:457) duplicated block id: 1865 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (408:418) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (421:431) duplicated block id: 1866 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (408:418) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (434:444) duplicated block id: 1867 size: 11 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (64:77) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (101:114) duplicated block id: 1868 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1045:1055) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1144:1154) duplicated block id: 1869 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:571) - maga_transformer/cpp/kernels/_fma.h (623:637) duplicated block id: 1870 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (125:145) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:171) duplicated block id: 1871 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (66:76) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (166:176) duplicated block id: 1872 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (267:277) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (151:161) duplicated block id: 1873 size: 11 cleaned lines of code in 2 files: - maga_transformer/models/downstream_modules/embedding/api_datatype.py (19:31) - maga_transformer/openai/api_datatype.py (56:68) duplicated block id: 1874 size: 11 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (21:31) - maga_transformer/model_loader/smooth_quant_weight.py (84:94) duplicated block id: 1875 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (250:261) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (627:638) duplicated block id: 1876 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:53) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (81:101) duplicated block id: 1877 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (227:239) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (189:201) duplicated block id: 1878 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (104:116) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (62:74) duplicated block id: 1879 size: 11 cleaned lines of code in 2 files: - maga_transformer/distribute/gang_info.py (18:28) - maga_transformer/distribute/gang_info.py (51:61) duplicated block id: 1880 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:53) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (153:173) duplicated block id: 1881 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:53) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (201:221) duplicated block id: 1882 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (499:509) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (512:522) duplicated block id: 1883 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/propose_executor/DeterministicExecutor.cc (36:46) - maga_transformer/cpp/speculative_engine/propose_executor/DeterministicExecutor.cc (97:107) duplicated block id: 1884 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1066:1076) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1122:1132) duplicated block id: 1885 size: 11 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (100:113) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (64:77) duplicated block id: 1886 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (844:854) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (137:147) duplicated block id: 1887 size: 11 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (434:444) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (447:457) duplicated block id: 1888 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) duplicated block id: 1889 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) duplicated block id: 1890 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (253:263) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (77:87) duplicated block id: 1891 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (240:251) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (288:299) duplicated block id: 1892 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1893 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1894 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1400:1412) - maga_transformer/openai/renderers/conversation.py (1414:1426) duplicated block id: 1895 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) duplicated block id: 1896 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1897 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1898 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (7:16) duplicated block id: 1899 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:25) duplicated block id: 1900 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 1901 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (11:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (11:20) duplicated block id: 1902 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 1903 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 1904 size: 10 cleaned lines of code in 2 files: - maga_transformer/tools/fake_bloom.py (37:47) - maga_transformer/tools/fake_glm_v2.py (42:52) duplicated block id: 1905 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 1906 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 1907 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (533:544) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (401:413) duplicated block id: 1908 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (533:544) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (390:402) duplicated block id: 1909 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_float.h (5:30) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2209:2235) duplicated block id: 1910 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (63:76) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (88:100) duplicated block id: 1911 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:16) duplicated block id: 1912 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (588:597) - maga_transformer/cpp/cuda/cuda_utils.h (501:510) duplicated block id: 1913 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (20:30) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (475:485) duplicated block id: 1914 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1915 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 1916 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (197:208) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (173:184) duplicated block id: 1917 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (187:196) - maga_transformer/cpp/kernels/logprob_kernels.cu (200:209) duplicated block id: 1918 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (20:30) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (783:793) duplicated block id: 1919 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) duplicated block id: 1920 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 1921 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 1922 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 1923 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (181:190) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (301:310) duplicated block id: 1924 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (161:190) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (214:243) duplicated block id: 1925 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1926 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (246:255) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (275:284) duplicated block id: 1927 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (465:478) - maga_transformer/cpp/kernels/rotary_position_embedding.h (499:512) duplicated block id: 1928 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (499:512) - maga_transformer/cpp/kernels/rotary_position_embedding.h (557:570) duplicated block id: 1929 size: 10 cleaned lines of code in 2 files: - maga_transformer/server/backend_app.py (127:136) - maga_transformer/server/frontend_app.py (115:124) duplicated block id: 1930 size: 10 cleaned lines of code in 2 files: - maga_transformer/tools/fake_bloom.py (6:15) - maga_transformer/tools/fake_qwen.py (5:14) duplicated block id: 1931 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1932 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (465:478) - maga_transformer/cpp/kernels/rotary_position_embedding.h (557:570) duplicated block id: 1933 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1934 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1935 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1936 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1937 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1938 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1939 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 1940 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1941 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/quantization_rocm.cu (108:119) - maga_transformer/cpp/kernels/rocm/quantization_rocm.cu (533:544) duplicated block id: 1942 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1943 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:914) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (52:61) duplicated block id: 1944 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (103:112) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (158:167) duplicated block id: 1945 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:438) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:725) duplicated block id: 1946 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (565:574) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (122:131) duplicated block id: 1947 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1039:1050) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1102:1113) duplicated block id: 1948 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (7:16) duplicated block id: 1949 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (288:297) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (660:669) duplicated block id: 1950 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (274:283) - maga_transformer/cpp/cuda/cufmha/cufmha.h (96:105) duplicated block id: 1951 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (214:243) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (281:310) duplicated block id: 1952 size: 10 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (33:43) - maga_transformer/device/device_impl.py (94:104) duplicated block id: 1953 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:438) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:543) duplicated block id: 1954 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:438) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:490) duplicated block id: 1955 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:438) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:656) duplicated block id: 1956 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (11:20) duplicated block id: 1957 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:438) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:599) duplicated block id: 1958 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:914) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (36:45) duplicated block id: 1959 size: 10 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (46:56) - maga_transformer/models/jina_bert/jina_bert_weight.py (36:46) duplicated block id: 1960 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (201:219) duplicated block id: 1961 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (153:171) duplicated block id: 1962 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (247:265) duplicated block id: 1963 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (300:318) duplicated block id: 1964 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) duplicated block id: 1965 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) duplicated block id: 1966 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1967 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) duplicated block id: 1968 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1969 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) duplicated block id: 1970 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) duplicated block id: 1971 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 1972 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (588:597) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (601:610) duplicated block id: 1973 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 1974 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 1975 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1976 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1977 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1978 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1979 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 1980 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1981 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1982 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1983 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 1984 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 1985 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 1986 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1987 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1988 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 1989 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 1990 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 1991 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (571:582) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (390:402) duplicated block id: 1992 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 1993 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (620:639) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (574:592) duplicated block id: 1994 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (536:545) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 1995 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (783:793) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (938:948) duplicated block id: 1996 size: 10 cleaned lines of code in 2 files: - maga_transformer/tools/quant/awq_quanter.py (9:18) - maga_transformer/tools/quant/fp8_quanter.py (100:110) duplicated block id: 1997 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (475:485) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (630:640) duplicated block id: 1998 size: 10 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (176:185) - maga_transformer/models/qwen_v2.py (142:151) duplicated block id: 1999 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_renderer.py (400:409) - maga_transformer/openai/renderers/qwen_renderer.py (503:512) duplicated block id: 2000 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) duplicated block id: 2001 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2002 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (381:391) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (77:87) duplicated block id: 2003 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (666:675) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (718:727) duplicated block id: 2004 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (11:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (11:20) duplicated block id: 2005 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) duplicated block id: 2006 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (7:16) duplicated block id: 2007 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (475:485) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (938:948) duplicated block id: 2008 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2009 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2010 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (43:56) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (88:100) duplicated block id: 2011 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (43:56) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (63:76) duplicated block id: 2012 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) duplicated block id: 2013 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) duplicated block id: 2014 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (666:675) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (124:133) duplicated block id: 2015 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (666:675) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (137:146) duplicated block id: 2016 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (666:675) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (106:115) duplicated block id: 2017 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (450:459) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2018 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (666:675) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (844:853) duplicated block id: 2019 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (20:30) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (168:178) duplicated block id: 2020 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) duplicated block id: 2021 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (273:289) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (398:414) duplicated block id: 2022 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2023 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:25) duplicated block id: 2024 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2025 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) duplicated block id: 2026 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (81:99) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:51) duplicated block id: 2027 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) duplicated block id: 2028 size: 10 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (51:61) - maga_transformer/model_loader/static_fp8_quant_weight.py (71:80) duplicated block id: 2029 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (565:574) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (587:596) duplicated block id: 2030 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2031 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (565:574) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (576:585) duplicated block id: 2032 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (443:454) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (682:693) duplicated block id: 2033 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2034 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2035 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSoftmaxOp.cc (21:34) - maga_transformer/cpp/devices/rocm_impl/ROCmSoftmaxOp.cc (19:32) duplicated block id: 2036 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (139:148) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (172:181) duplicated block id: 2037 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2038 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/core/QBuffer.cc (103:113) - maga_transformer/cpp/core/QBuffer.cc (119:129) duplicated block id: 2039 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:490) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:656) duplicated block id: 2040 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (181:190) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (54:63) duplicated block id: 2041 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (319:329) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (77:87) duplicated block id: 2042 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (125:134) - maga_transformer/cpp/devices/utils/DebugUtils.cc (168:177) duplicated block id: 2043 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:490) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:725) duplicated block id: 2044 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (125:134) - maga_transformer/cpp/devices/utils/DebugUtils.cc (214:223) duplicated block id: 2045 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (106:115) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (124:133) duplicated block id: 2046 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (301:310) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (54:63) duplicated block id: 2047 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (168:178) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (322:332) duplicated block id: 2048 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (554:563) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (122:131) duplicated block id: 2049 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2050 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (70:79) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (180:189) duplicated block id: 2051 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2052 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2053 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (548:558) - maga_transformer/cpp/cuda/cuda_utils.cc (662:673) duplicated block id: 2054 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (188:198) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (579:589) duplicated block id: 2055 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2056 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (168:178) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (630:640) duplicated block id: 2057 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (168:178) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (938:948) duplicated block id: 2058 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2059 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (194:204) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (77:87) duplicated block id: 2060 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2061 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2062 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2063 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:490) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:599) duplicated block id: 2064 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:490) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:543) duplicated block id: 2065 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2066 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (75:88) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (42:54) duplicated block id: 2067 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (11:20) duplicated block id: 2068 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (776:787) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (851:862) duplicated block id: 2069 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2070 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2071 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2072 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1386:1398) - maga_transformer/openai/renderers/conversation.py (1400:1412) duplicated block id: 2073 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1386:1398) - maga_transformer/openai/renderers/conversation.py (1414:1426) duplicated block id: 2074 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 2075 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2076 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2077 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2078 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2079 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2080 size: 10 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (171:181) - maga_transformer/model_loader/static_fp8_quant_weight.py (169:179) duplicated block id: 2081 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2082 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2083 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2084 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (37:47) - maga_transformer/openai/renderers/custom_renderer.py (89:99) duplicated block id: 2085 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2086 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2087 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (319:337) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (379:395) duplicated block id: 2088 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2089 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (124:133) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (137:146) duplicated block id: 2090 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (718:727) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (124:133) duplicated block id: 2091 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2092 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (938:948) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1090:1100) duplicated block id: 2093 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2094 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2095 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2096 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (539:548) - maga_transformer/cpp/cuda/cuda_utils.cc (648:657) duplicated block id: 2097 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2098 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2099 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2100 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.h (64:73) - maga_transformer/cpp/kernels/rmsnormKernels.h (46:55) duplicated block id: 2101 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (322:332) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (783:793) duplicated block id: 2102 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2103 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (423:432) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (233:242) duplicated block id: 2104 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2105 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2106 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2107 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2108 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (316:331) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (453:467) duplicated block id: 2109 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 2110 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2111 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2112 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (7:16) duplicated block id: 2113 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2114 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (194:204) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (381:391) duplicated block id: 2115 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2116 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2117 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (194:204) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (319:329) duplicated block id: 2118 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (322:332) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (475:485) duplicated block id: 2119 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (194:204) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (253:263) duplicated block id: 2120 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (503:514) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (541:552) duplicated block id: 2121 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2122 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 2123 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2124 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2125 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2126 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2127 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2128 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) duplicated block id: 2129 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (737:746) duplicated block id: 2130 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (784:793) duplicated block id: 2131 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2132 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (554:563) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (565:574) duplicated block id: 2133 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (554:563) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (576:585) duplicated block id: 2134 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2135 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (554:563) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (587:596) duplicated block id: 2136 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (147:160) - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (171:184) duplicated block id: 2137 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2138 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (322:332) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1090:1100) duplicated block id: 2139 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2140 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (576:585) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (587:596) duplicated block id: 2141 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2142 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (490:499) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 2143 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2144 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2145 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2146 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) duplicated block id: 2147 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (697:706) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2148 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1189:1198) duplicated block id: 2149 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (28:40) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (53:65) duplicated block id: 2150 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (661:670) duplicated block id: 2151 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (92:101) - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (68:77) duplicated block id: 2152 size: 10 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (31:42) - maga_transformer/tokenizer/tokenization_chatglm3.py (50:61) duplicated block id: 2153 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1084:1093) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1269:1278) duplicated block id: 2154 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2155 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (20:30) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1090:1100) duplicated block id: 2156 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1322:1331) duplicated block id: 2157 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (104:114) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (504:514) duplicated block id: 2158 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (180:198) duplicated block id: 2159 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:51) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (127:145) duplicated block id: 2160 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (630:640) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (783:793) duplicated block id: 2161 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (65:89) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (91:113) duplicated block id: 2162 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2163 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (909:918) duplicated block id: 2164 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (942:951) duplicated block id: 2165 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) duplicated block id: 2166 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1015:1024) duplicated block id: 2167 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/26_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v4.cc (7:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (7:16) duplicated block id: 2168 size: 10 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (773:782) - maga_transformer/openai/renderers/custom_renderer.py (834:843) duplicated block id: 2169 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (571:582) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (401:413) duplicated block id: 2170 size: 10 cleaned lines of code in 2 files: - maga_transformer/utils/model_weight.py (249:258) - maga_transformer/utils/model_weight.py (311:320) duplicated block id: 2171 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2172 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (584:593) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2173 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (84:93) - maga_transformer/cpp/kernels/add_residual_kernels.cu (101:110) duplicated block id: 2174 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (630:640) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1090:1100) duplicated block id: 2175 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (319:329) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (381:391) duplicated block id: 2176 size: 10 cleaned lines of code in 2 files: - maga_transformer/tools/convert/weights_convert.py (110:119) - maga_transformer/tools/convert/weights_convert.py (148:157) duplicated block id: 2177 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1106:1115) duplicated block id: 2178 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (977:986) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1072:1081) duplicated block id: 2179 size: 10 cleaned lines of code in 2 files: - maga_transformer/models/chat_glm_v4_vision.py (30:40) - maga_transformer/models/cogvlm2.py (32:42) duplicated block id: 2180 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (576:585) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (122:131) duplicated block id: 2181 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1253:1262) duplicated block id: 2182 size: 10 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (253:263) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (319:329) duplicated block id: 2183 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1287:1296) duplicated block id: 2184 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (832:841) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2185 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1144:1153) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1373:1382) duplicated block id: 2186 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (844:853) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (124:133) duplicated block id: 2187 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (889:899) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (962:972) duplicated block id: 2188 size: 10 cleaned lines of code in 2 files: - bazel/defs.bzl (121:130) - bazel/defs.bzl (137:146) duplicated block id: 2189 size: 10 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.h (10:19) - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.h (23:32) duplicated block id: 2190 size: 10 cleaned lines of code in 2 files: - maga_transformer/tools/fake_bloom.py (37:47) - maga_transformer/tools/fake_model_base.py (43:53) duplicated block id: 2191 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/resampler.py (155:164) - maga_transformer/models/qwen_vl_vit.py (141:150) duplicated block id: 2192 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (587:595) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1575:1583) duplicated block id: 2193 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (71:85) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (681:693) duplicated block id: 2194 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (71:85) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (834:846) duplicated block id: 2195 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/models/GptModel.cc (624:632) - maga_transformer/cpp/models/GptModel.cc (726:734) duplicated block id: 2196 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (153:164) - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (280:291) duplicated block id: 2197 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (755:766) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (910:921) duplicated block id: 2198 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/10_int4_dequant_gemm_128x128x16x128_16_16x16_4x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/12_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) duplicated block id: 2199 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (301:309) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (369:377) duplicated block id: 2200 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (92:100) - maga_transformer/models/qwen_vl.py (80:88) duplicated block id: 2201 size: 9 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (45:57) - maga_transformer/tokenizer/tokenization_chatglm3.py (75:87) duplicated block id: 2202 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (401:412) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (109:120) duplicated block id: 2203 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (612:623) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (816:826) duplicated block id: 2204 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (901:911) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1005:1015) duplicated block id: 2205 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (313:322) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (451:460) duplicated block id: 2206 size: 9 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (134:142) - bazel/tf_http_archive.bzl (202:210) duplicated block id: 2207 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.h (483:491) - maga_transformer/cpp/cuda/cuda_utils.h (493:501) duplicated block id: 2208 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/WorkerStatusService.h (18:26) - maga_transformer/cpp/disaggregate/load_balancer/HeartbeatSynchronizer.h (31:39) duplicated block id: 2209 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (333:341) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (699:707) duplicated block id: 2210 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (344:352) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (708:716) duplicated block id: 2211 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (219:233) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (834:846) duplicated block id: 2212 size: 9 cleaned lines of code in 2 files: - maga_transformer/openai/openai_endpoint.py (147:155) - maga_transformer/openai/renderers/custom_renderer.py (884:892) duplicated block id: 2213 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (181:189) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (272:280) duplicated block id: 2214 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (179:187) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (232:240) duplicated block id: 2215 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (288:296) - maga_transformer/cpp/devices/rocm_impl/ROCmLoraLinearWithActOP.cc (146:154) duplicated block id: 2216 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (181:189) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (369:377) duplicated block id: 2217 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/10_int4_dequant_gemm_128x128x16x128_16_16x16_4x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) duplicated block id: 2218 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (224:233) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (266:275) duplicated block id: 2219 size: 9 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (367:375) - maga_transformer/openai/renderers/custom_renderer.py (636:644) duplicated block id: 2220 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (219:233) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (681:693) duplicated block id: 2221 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (181:189) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (241:249) duplicated block id: 2222 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2.py (35:44) - maga_transformer/models/qwen_vl.py (38:47) duplicated block id: 2223 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.h (53:62) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (50:59) duplicated block id: 2224 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (150:158) - maga_transformer/cpp/devices/rocm_impl/ROCmFfnLayer.cc (202:210) duplicated block id: 2225 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/21_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:19) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (11:19) duplicated block id: 2226 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (23:31) - maga_transformer/models/llama.py (35:43) duplicated block id: 2227 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (132:140) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (144:152) duplicated block id: 2228 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1029:1038) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1106:1115) duplicated block id: 2229 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (68:82) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (51:65) duplicated block id: 2230 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (246:255) - maga_transformer/cpp/cuda/memory_utils.cu (257:266) duplicated block id: 2231 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (76:84) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (100:108) duplicated block id: 2232 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (369:377) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (54:62) duplicated block id: 2233 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1067:1075) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1113:1121) duplicated block id: 2234 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:587) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:709) duplicated block id: 2235 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:587) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:644) duplicated block id: 2236 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/llama.py (35:43) - maga_transformer/models/whisper.py (55:63) duplicated block id: 2237 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (342:353) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (109:120) duplicated block id: 2238 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.h (58:66) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (70:78) duplicated block id: 2239 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/resampler.py (136:145) - maga_transformer/models/qwen_vl_vit.py (126:135) duplicated block id: 2240 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:620) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:677) duplicated block id: 2241 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (713:721) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (376:385) duplicated block id: 2242 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:25) duplicated block id: 2243 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (35:44) - maga_transformer/cpp/kernels/activation_kernels.cu (44:53) duplicated block id: 2244 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:598) duplicated block id: 2245 size: 9 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/qwen/convert.py (206:215) - maga_transformer/utils/smooth_quant_convert/qwen/convert.py (222:231) duplicated block id: 2246 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (716:726) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (816:826) duplicated block id: 2247 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (942:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:169) duplicated block id: 2248 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:542) duplicated block id: 2249 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:489) duplicated block id: 2250 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (942:950) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:150) duplicated block id: 2251 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (716:726) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (920:930) duplicated block id: 2252 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (387:395) - maga_transformer/cpp/cuda/cufmha/cufmha.h (140:148) duplicated block id: 2253 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:437) duplicated block id: 2254 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (269:278) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (337:347) duplicated block id: 2255 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/0_int4_dequant_gemm_256x128x128x128_32_32x32_2x2_16x16x1_4x64x1_32_1x32x1x8_8_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (16:24) duplicated block id: 2256 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (998:1006) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (196:204) duplicated block id: 2257 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (66:75) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (222:231) duplicated block id: 2258 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (127:144) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (108:126) duplicated block id: 2259 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (510:523) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (784:797) duplicated block id: 2260 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (487:495) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (524:532) duplicated block id: 2261 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (525:537) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (987:999) duplicated block id: 2262 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (525:537) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1139:1151) duplicated block id: 2263 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (563:611) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (557:604) duplicated block id: 2264 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (510:523) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (684:697) duplicated block id: 2265 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (380:394) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (811:823) duplicated block id: 2266 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (446:457) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (589:600) duplicated block id: 2267 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (370:379) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (507:516) duplicated block id: 2268 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:724) duplicated block id: 2269 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:655) duplicated block id: 2270 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (510:523) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (888:901) duplicated block id: 2271 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (993:1001) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:169) duplicated block id: 2272 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (993:1001) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:150) duplicated block id: 2273 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/InferenceDataType.h (11:22) - maga_transformer/cpp/openai/ApiDataType.cc (3:16) duplicated block id: 2274 size: 9 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (615:625) - maga_transformer/openai/renderers/llama_template.py (749:759) duplicated block id: 2275 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:25) duplicated block id: 2276 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (716:727) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (869:880) duplicated block id: 2277 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (612:623) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (920:930) duplicated block id: 2278 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (525:533) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (554:562) duplicated block id: 2279 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (525:533) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (565:573) duplicated block id: 2280 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (525:533) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (576:584) duplicated block id: 2281 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (525:533) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (587:595) duplicated block id: 2282 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (71:79) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (272:280) duplicated block id: 2283 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (241:249) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (301:309) duplicated block id: 2284 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (687:695) - maga_transformer/cpp/kernels/vec_dtypes.cuh (822:830) duplicated block id: 2285 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:19) - maga_transformer/cpp/rocm/int4_gemm_kernels/33_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x16_4_intrawave_v4.cc (11:19) duplicated block id: 2286 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (687:695) - maga_transformer/cpp/kernels/vec_dtypes.cuh (774:782) duplicated block id: 2287 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (687:695) - maga_transformer/cpp/kernels/vec_dtypes.cuh (727:735) duplicated block id: 2288 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (277:285) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (651:659) duplicated block id: 2289 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (192:201) - maga_transformer/models/qwen.py (140:149) duplicated block id: 2290 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/TokenizerService.cc (9:19) - maga_transformer/cpp/openai/ApiDataType.cc (3:16) duplicated block id: 2291 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:644) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:709) duplicated block id: 2292 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (716:727) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1175:1186) duplicated block id: 2293 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1023:1034) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1175:1186) duplicated block id: 2294 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:655) duplicated block id: 2295 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:724) duplicated block id: 2296 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (322:330) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (220:228) duplicated block id: 2297 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (774:782) - maga_transformer/cpp/kernels/vec_dtypes.cuh (822:830) duplicated block id: 2298 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:598) duplicated block id: 2299 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (339:347) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (354:362) duplicated block id: 2300 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (271:279) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (58:66) duplicated block id: 2301 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (79:87) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (182:190) duplicated block id: 2302 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (79:87) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (131:139) duplicated block id: 2303 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (26:36) - maga_transformer/cpp/devices/rocm_impl/ROCmDevice.cc (174:184) duplicated block id: 2304 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (104:156) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (104:160) duplicated block id: 2305 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (220:231) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (514:525) duplicated block id: 2306 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (71:79) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (71:79) duplicated block id: 2307 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (434:447) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (888:901) duplicated block id: 2308 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (42:50) - maga_transformer/cpp/devices/utils/DebugUtils.cc (82:90) duplicated block id: 2309 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (434:447) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (684:697) duplicated block id: 2310 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (434:447) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (784:797) duplicated block id: 2311 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (214:225) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (109:120) duplicated block id: 2312 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (181:189) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (71:79) duplicated block id: 2313 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (434:447) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (581:594) duplicated block id: 2314 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_template.h (62:73) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (83:94) duplicated block id: 2315 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:19) - maga_transformer/cpp/rocm/int4_gemm_kernels/21_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (11:19) duplicated block id: 2316 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (372:384) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1139:1151) duplicated block id: 2317 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1010:1018) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:169) duplicated block id: 2318 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (200:208) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (58:66) duplicated block id: 2319 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (976:984) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:169) duplicated block id: 2320 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (976:984) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:150) duplicated block id: 2321 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1010:1018) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:150) duplicated block id: 2322 size: 9 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (317:325) - maga_transformer/utils/smooth_quant_convert/qwen/hf_qwen_convert.py (75:83) duplicated block id: 2323 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (921:929) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (943:951) duplicated block id: 2324 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (921:929) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (932:940) duplicated block id: 2325 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (372:384) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (987:999) duplicated block id: 2326 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/triton/aot_triton_kernel_compiler.py (8:21) - maga_transformer/cpp/kernels/triton/aot_triton_kernels_linker.py (6:19) duplicated block id: 2327 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (510:523) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (581:594) duplicated block id: 2328 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/converter.h (36:55) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/converter.h (60:79) duplicated block id: 2329 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (408:419) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1175:1186) duplicated block id: 2330 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1096:1104) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1134:1142) duplicated block id: 2331 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1096:1104) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1179:1187) duplicated block id: 2332 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.cpp (74:82) - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.h (45:53) duplicated block id: 2333 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (932:940) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (943:951) duplicated block id: 2334 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (241:249) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (54:62) duplicated block id: 2335 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (55:65) - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (150:160) duplicated block id: 2336 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (408:419) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (869:880) duplicated block id: 2337 size: 9 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (273:284) - maga_transformer/aios/kmonitor/python_client/flume/ttypes.py (109:120) duplicated block id: 2338 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.cpp (126:134) - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.cpp (144:152) duplicated block id: 2339 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (69:77) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (422:430) duplicated block id: 2340 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (129:138) - maga_transformer/models/qwen_v2.py (158:167) duplicated block id: 2341 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (237:245) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (361:369) duplicated block id: 2342 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (164:172) - maga_transformer/cpp/cuda/cufmha/cufmha.h (70:78) duplicated block id: 2343 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (408:419) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (561:572) duplicated block id: 2344 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/InferenceDataType.h (11:22) - maga_transformer/cpp/api_server/TokenizerService.cc (9:19) duplicated block id: 2345 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/12_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:25) duplicated block id: 2346 size: 9 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (164:172) - maga_transformer/model_loader/smooth_quant_weight.py (202:210) duplicated block id: 2347 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (11:19) - maga_transformer/cpp/rocm/int4_gemm_kernels/33_int4_dequant_gemm_256x16x256x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x16_4_intrawave_v4.cc (11:19) duplicated block id: 2348 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (93:109) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (146:162) duplicated block id: 2349 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/rocmFmhaWrapper.cc (21:29) - maga_transformer/cpp/rocm/rocmFmhaWrapper.h (40:48) duplicated block id: 2350 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (63:73) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (30:40) duplicated block id: 2351 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (224:232) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (275:283) duplicated block id: 2352 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (108:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1175:1186) duplicated block id: 2353 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (561:572) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (716:727) duplicated block id: 2354 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (440:448) - maga_transformer/cpp/kernels/vec_dtypes.cuh (574:582) duplicated block id: 2355 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (561:572) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1023:1034) duplicated block id: 2356 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (302:310) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (133:141) duplicated block id: 2357 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (727:735) - maga_transformer/cpp/kernels/vec_dtypes.cuh (774:782) duplicated block id: 2358 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/metrics/RtpLLMMetrics.h (31:39) - maga_transformer/cpp/model_rpc/PrefillGenerateContext.h (31:39) duplicated block id: 2359 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (224:232) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (246:254) duplicated block id: 2360 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (108:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (869:880) duplicated block id: 2361 size: 9 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/convert.py (300:308) - maga_transformer/utils/smooth_quant_convert/qwen/convert.py (292:300) duplicated block id: 2362 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (559:567) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1565:1573) duplicated block id: 2363 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (525:533) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (122:130) duplicated block id: 2364 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (108:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (561:572) duplicated block id: 2365 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (97:105) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (149:157) duplicated block id: 2366 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (579:587) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (189:197) duplicated block id: 2367 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1134:1142) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1179:1187) duplicated block id: 2368 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (131:139) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (182:190) duplicated block id: 2369 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (242:250) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (201:209) duplicated block id: 2370 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (869:880) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1023:1034) duplicated block id: 2371 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (108:119) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (256:267) duplicated block id: 2372 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (501:511) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (770:780) duplicated block id: 2373 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/nccl/nccl_utils_torch.cc (50:58) - maga_transformer/cpp/cuda/nccl/nccl_utils_torch.h (20:28) duplicated block id: 2374 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (45:53) - maga_transformer/models/megatron_bert_weight.py (16:24) duplicated block id: 2375 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (727:735) - maga_transformer/cpp/kernels/vec_dtypes.cuh (822:830) duplicated block id: 2376 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (440:448) - maga_transformer/cpp/kernels/vec_dtypes.cuh (526:534) duplicated block id: 2377 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (440:448) - maga_transformer/cpp/kernels/vec_dtypes.cuh (480:488) duplicated block id: 2378 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (270:280) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (495:505) duplicated block id: 2379 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (526:534) - maga_transformer/cpp/kernels/vec_dtypes.cuh (574:582) duplicated block id: 2380 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (18:28) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (38:48) duplicated block id: 2381 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (98:106) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (137:145) duplicated block id: 2382 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (65:84) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (115:134) duplicated block id: 2383 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (111:119) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (163:171) duplicated block id: 2384 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (87:95) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (271:279) duplicated block id: 2385 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (480:488) - maga_transformer/cpp/kernels/vec_dtypes.cuh (526:534) duplicated block id: 2386 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (87:95) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (70:78) duplicated block id: 2387 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (23:31) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (292:300) duplicated block id: 2388 size: 9 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (292:300) - maga_transformer/models/whisper.py (55:63) duplicated block id: 2389 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (87:95) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (58:66) duplicated block id: 2390 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (91:110) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_fpA_intB_traits.h (115:134) duplicated block id: 2391 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (256:267) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (408:419) duplicated block id: 2392 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (120:136) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (239:255) duplicated block id: 2393 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (426:434) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (82:90) duplicated block id: 2394 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (277:288) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/warp/mma_tensorop_dequantizer.h (589:600) duplicated block id: 2395 size: 9 cleaned lines of code in 2 files: - maga_transformer/tools/fake_model_base.py (13:21) - maga_transformer/tools/fake_qwen.py (6:14) duplicated block id: 2396 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (480:488) - maga_transformer/cpp/kernels/vec_dtypes.cuh (574:582) duplicated block id: 2397 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (256:267) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1023:1034) duplicated block id: 2398 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (264:272) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (425:433) duplicated block id: 2399 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (374:382) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (732:740) duplicated block id: 2400 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (256:267) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (716:727) duplicated block id: 2401 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (666:674) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (720:728) duplicated block id: 2402 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (959:967) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:150) duplicated block id: 2403 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (959:967) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:169) duplicated block id: 2404 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:489) duplicated block id: 2405 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:542) duplicated block id: 2406 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (203:211) - maga_transformer/cpp/kernels/vec_dtypes.cuh (223:231) duplicated block id: 2407 size: 9 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/convert.py (81:91) - maga_transformer/utils/smooth_quant_convert/qwen/convert.py (89:99) duplicated block id: 2408 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:388) duplicated block id: 2409 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/24_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:25) duplicated block id: 2410 size: 9 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:340) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:437) duplicated block id: 2411 size: 9 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (87:96) - bazel/tf_http_archive.bzl (224:233) duplicated block id: 2412 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (143:150) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (140:147) duplicated block id: 2413 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (453:460) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (480:487) duplicated block id: 2414 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/rtpllm_master/cluster/PrefillLoadBalancer.cpp (114:121) - maga_transformer/cpp/disaggregate/rtpllm_master/cluster/PrefillLoadBalancer.cpp (128:135) duplicated block id: 2415 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2416 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:22) duplicated block id: 2417 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (576:584) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (658:666) duplicated block id: 2418 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (35:42) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (158:165) duplicated block id: 2419 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (248:255) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (271:278) duplicated block id: 2420 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (667:676) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (767:776) duplicated block id: 2421 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (176:183) - maga_transformer/cpp/devices/rocm_impl/ROCmDevice.cc (283:290) duplicated block id: 2422 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (186:193) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (72:79) duplicated block id: 2423 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (667:676) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (972:981) duplicated block id: 2424 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaFfnLayer.cc (148:155) - maga_transformer/cpp/devices/cuda_impl/CudaFfnLayer.cc (162:169) duplicated block id: 2425 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (323:333) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (424:434) duplicated block id: 2426 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) duplicated block id: 2427 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (293:300) - maga_transformer/cpp/kernels/vec_dtypes.cuh (326:333) duplicated block id: 2428 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_renderer.py (125:134) - maga_transformer/utils/smooth_quant_convert/qwen/utils.py (72:81) duplicated block id: 2429 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (767:776) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (871:880) duplicated block id: 2430 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.h (68:75) - maga_transformer/cpp/kernels/add_residual_kernels.h (79:86) duplicated block id: 2431 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (201:208) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (221:228) duplicated block id: 2432 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/InferenceDataType.h (15:22) - maga_transformer/cpp/dataclass/GenerateConfig.h (137:144) duplicated block id: 2433 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (427:435) - maga_transformer/openai/renderers/conversation.py (440:448) duplicated block id: 2434 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (21:28) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (8:15) duplicated block id: 2435 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:18) duplicated block id: 2436 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/MlaAttentionLayer.cc (98:105) - maga_transformer/cpp/devices/base_impl/MlaAttentionLayer.cc (122:129) duplicated block id: 2437 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (702:711) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (365:374) duplicated block id: 2438 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) duplicated block id: 2439 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (860:867) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1014:1021) duplicated block id: 2440 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (133:140) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (151:158) duplicated block id: 2441 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/chat_glm_v4_vision.py (33:40) - maga_transformer/models/qwen_vl.py (38:45) duplicated block id: 2442 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (90:99) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (56:65) duplicated block id: 2443 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (242:249) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (286:293) duplicated block id: 2444 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (113:137) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (217:243) duplicated block id: 2445 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:436) duplicated block id: 2446 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/internvl.py (50:57) - maga_transformer/models/qwen_v2.py (142:149) duplicated block id: 2447 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:387) duplicated block id: 2448 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:339) duplicated block id: 2449 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (17:25) duplicated block id: 2450 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (738:745) - maga_transformer/cpp/cuda/cuda_utils.cc (747:754) duplicated block id: 2451 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (474:481) - maga_transformer/cpp/kernels/activation_kernels.cu (502:509) duplicated block id: 2452 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (506:515) - maga_transformer/cpp/kernels/_fma.h (527:536) duplicated block id: 2453 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:541) duplicated block id: 2454 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:488) duplicated block id: 2455 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (510:523) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (747:760) duplicated block id: 2456 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSoftmaxOp.cc (89:96) - maga_transformer/cpp/devices/rocm_impl/ROCmSoftmaxOp.cc (70:77) duplicated block id: 2457 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (799:811) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (944:954) duplicated block id: 2458 size: 8 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (74:85) - maga_transformer/device/device_impl.py (170:181) duplicated block id: 2459 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:18) duplicated block id: 2460 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:597) duplicated block id: 2461 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (949:959) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1028:1038) duplicated block id: 2462 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:723) duplicated block id: 2463 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:250) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:654) duplicated block id: 2464 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cutlass_kernel.h (101:121) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/splitk_gemm_grouped.h (93:113) duplicated block id: 2465 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (707:714) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1166:1173) duplicated block id: 2466 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (319:331) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (36:48) duplicated block id: 2467 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (463:472) - maga_transformer/openai/renderers/conversation.py (1050:1058) duplicated block id: 2468 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (463:472) - maga_transformer/openai/renderers/conversation.py (1038:1046) duplicated block id: 2469 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (707:714) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (860:867) duplicated block id: 2470 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:18) - maga_transformer/cpp/kernels/int8_utils.cuh (9:17) duplicated block id: 2471 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (354:361) - maga_transformer/cpp/devices/rocm_impl/ROCmGemmOp.cc (173:180) duplicated block id: 2472 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/jina_bert/jina_bert_weight.py (12:21) - maga_transformer/utils/model_weight.py (381:390) duplicated block id: 2473 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) duplicated block id: 2474 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (19:28) - maga_transformer/utils/model_weight.py (381:390) duplicated block id: 2475 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (123:135) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (36:48) duplicated block id: 2476 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1128:1139) - maga_transformer/openai/renderers/conversation.py (1143:1154) duplicated block id: 2477 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_heuristic.cc (216:227) - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_heuristic.cc (412:423) duplicated block id: 2478 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/internvl.py (50:57) - maga_transformer/models/qwen.py (176:183) duplicated block id: 2479 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (47:60) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (106:119) duplicated block id: 2480 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:509) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:562) duplicated block id: 2481 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:509) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:619) duplicated block id: 2482 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (66:74) - maga_transformer/cpp/kernels/rocm/quantization_rocm.cu (195:203) duplicated block id: 2483 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:509) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:676) duplicated block id: 2484 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (269:277) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (317:325) duplicated block id: 2485 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/FfnLayerTest.hpp (51:59) - maga_transformer/cpp/devices/base_tests/FfnLayerTest.hpp (154:162) duplicated block id: 2486 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/dataclass/GenerateConfig.h (137:144) - maga_transformer/cpp/openai/ApiDataType.cc (9:16) duplicated block id: 2487 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) duplicated block id: 2488 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (785:792) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (978:985) duplicated block id: 2489 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (269:276) - maga_transformer/cpp/kernels/rotary_position_embedding.h (291:298) duplicated block id: 2490 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (422:431) - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (490:499) duplicated block id: 2491 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (338:345) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (64:71) duplicated block id: 2492 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/jina_bert/jina_bert_weight.py (110:118) - maga_transformer/models/megatron_bert_weight.py (80:88) duplicated block id: 2493 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2446:2453) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2531:2538) duplicated block id: 2494 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/TokenizerService.cc (12:19) - maga_transformer/cpp/dataclass/GenerateConfig.h (137:144) duplicated block id: 2495 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (337:344) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:117) duplicated block id: 2496 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/1_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v4.cc (7:14) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:14) duplicated block id: 2497 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:14) - maga_transformer/cpp/rocm/int4_gemm_kernels/4_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v3.cc (7:14) duplicated block id: 2498 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1337:1344) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1359:1366) duplicated block id: 2499 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:18) - maga_transformer/cpp/kernels/int8_utils.cuh (9:17) duplicated block id: 2500 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (65:91) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (217:243) duplicated block id: 2501 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent_renderer.py (42:49) - maga_transformer/openai/renderers/qwen_renderer.py (175:182) duplicated block id: 2502 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (768:777) - maga_transformer/openai/renderers/conversation.py (782:792) duplicated block id: 2503 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/device/splitk_gemm_grouped.h (386:399) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/device/splitk_gemm_grouped.h (430:443) duplicated block id: 2504 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:14) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:14) duplicated block id: 2505 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/rocmFmhaWrapper.cc (9:17) - maga_transformer/cpp/rocm/rocmMoeWrapper.cc (13:21) duplicated block id: 2506 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (485:492) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (899:906) duplicated block id: 2507 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) duplicated block id: 2508 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (340:348) - maga_transformer/openai/renderers/internvl_renderer.py (132:140) duplicated block id: 2509 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1094:1101) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (1142:1149) duplicated block id: 2510 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:22) duplicated block id: 2511 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (742:749) - maga_transformer/openai/renderers/custom_renderer.py (807:814) duplicated block id: 2512 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (178:185) - maga_transformer/models/qwen_v2.py (142:149) duplicated block id: 2513 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (769:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:168) duplicated block id: 2514 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (769:776) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:149) duplicated block id: 2515 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (641:653) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (593:605) duplicated block id: 2516 size: 8 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/convert.py (125:132) - maga_transformer/utils/smooth_quant_convert/qwen/convert.py (109:116) duplicated block id: 2517 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (70:77) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (233:240) duplicated block id: 2518 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:16) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:18) duplicated block id: 2519 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:18) duplicated block id: 2520 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (100:111) - maga_transformer/openai/renderers/llama_template.py (164:175) duplicated block id: 2521 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (485:492) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (725:732) duplicated block id: 2522 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (71:78) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (88:95) duplicated block id: 2523 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:25) duplicated block id: 2524 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (7:14) - maga_transformer/cpp/rocm/int4_gemm_kernels/3_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_16_1x32x1x8_8_intrawave_v4.cc (7:14) duplicated block id: 2525 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (97:105) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (283:291) duplicated block id: 2526 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1097:1108) - maga_transformer/openai/renderers/conversation.py (1128:1139) duplicated block id: 2527 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1097:1108) - maga_transformer/openai/renderers/conversation.py (1143:1154) duplicated block id: 2528 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (21:28) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (17:24) duplicated block id: 2529 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2530 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1014:1021) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1166:1173) duplicated block id: 2531 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (390:398) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (745:752) duplicated block id: 2532 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1097:1108) - maga_transformer/openai/renderers/conversation.py (1112:1124) duplicated block id: 2533 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (298:305) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (414:421) duplicated block id: 2534 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (458:465) - maga_transformer/cpp/cuda/cuda_utils.h (463:470) duplicated block id: 2535 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (317:325) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (337:345) duplicated block id: 2536 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:16) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:18) duplicated block id: 2537 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (870:880) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1028:1038) duplicated block id: 2538 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:18) duplicated block id: 2539 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (870:880) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (949:959) duplicated block id: 2540 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (71:78) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (59:66) duplicated block id: 2541 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (217:243) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (233:257) duplicated block id: 2542 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (143:153) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (188:195) duplicated block id: 2543 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/internvl_vit.py (739:747) - maga_transformer/models/llava_vit.py (872:880) duplicated block id: 2544 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (527:536) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (851:860) duplicated block id: 2545 size: 8 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (92:102) - maga_transformer/tokenizer/tokenization_chatglm3.py (164:174) duplicated block id: 2546 size: 8 cleaned lines of code in 2 files: - maga_transformer/async_decoder_engine/async_model.py (61:68) - maga_transformer/async_decoder_engine/backend_rpc_server_visitor.py (19:26) duplicated block id: 2547 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (144:151) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (174:181) duplicated block id: 2548 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:18) - maga_transformer/cpp/kernels/int8_utils.cuh (9:17) duplicated block id: 2549 size: 8 cleaned lines of code in 2 files: - maga_transformer/model_loader/ffn_weight.py (128:135) - maga_transformer/model_loader/ffn_weight.py (229:236) duplicated block id: 2550 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (399:406) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1166:1173) duplicated block id: 2551 size: 8 cleaned lines of code in 2 files: - maga_transformer/server/backend_app.py (47:54) - maga_transformer/server/frontend_app.py (49:56) duplicated block id: 2552 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (134:142) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (166:174) duplicated block id: 2553 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:18) duplicated block id: 2554 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (24:32) - maga_transformer/cpp/kernels/activation_kernels.cu (34:42) duplicated block id: 2555 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1112:1124) - maga_transformer/openai/renderers/conversation.py (1143:1154) duplicated block id: 2556 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (888:895) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (978:985) duplicated block id: 2557 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1112:1124) - maga_transformer/openai/renderers/conversation.py (1128:1139) duplicated block id: 2558 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2559 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (40:50) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (23:33) duplicated block id: 2560 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:562) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:676) duplicated block id: 2561 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/jina_bert/jina_bert_weight.py (36:43) - maga_transformer/models/megatron_bert_weight.py (17:24) duplicated block id: 2562 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1038:1046) - maga_transformer/openai/renderers/conversation.py (1050:1058) duplicated block id: 2563 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:478) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:531) duplicated block id: 2564 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:18) - maga_transformer/cpp/kernels/int8_utils.cuh (9:17) duplicated block id: 2565 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2433:2442) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2515:2524) duplicated block id: 2566 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:562) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:619) duplicated block id: 2567 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:478) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:643) duplicated block id: 2568 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (286:293) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (221:228) duplicated block id: 2569 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (286:293) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (201:208) duplicated block id: 2570 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:478) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:586) duplicated block id: 2571 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:531) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:708) duplicated block id: 2572 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (178:185) - maga_transformer/models/qwen.py (176:183) duplicated block id: 2573 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/llama_weight.py (62:69) - maga_transformer/models/llama_weight.py (119:126) duplicated block id: 2574 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:478) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:708) duplicated block id: 2575 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:531) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:643) duplicated block id: 2576 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:531) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:586) duplicated block id: 2577 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (399:406) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (552:559) duplicated block id: 2578 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent_renderer.py (192:199) - maga_transformer/openai/renderers/qwen_agent_renderer.py (234:241) duplicated block id: 2579 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (197:205) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (350:359) duplicated block id: 2580 size: 8 cleaned lines of code in 2 files: - maga_transformer/model_loader/model_weight_info.py (72:80) - maga_transformer/model_loader/model_weight_info.py (96:104) duplicated block id: 2581 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (230:239) - maga_transformer/cpp/kernels/rmsnormKernels.cu (377:386) duplicated block id: 2582 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (451:459) - maga_transformer/openai/renderers/conversation.py (893:902) duplicated block id: 2583 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/bert_weight.py (19:28) - maga_transformer/models/jina_bert/jina_bert_weight.py (12:21) duplicated block id: 2584 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2585 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (399:406) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (860:867) duplicated block id: 2586 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (871:880) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (972:981) duplicated block id: 2587 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (567:575) - maga_transformer/cpp/kernels/gpt_kernels.cu (657:665) duplicated block id: 2588 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (181:188) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (88:95) duplicated block id: 2589 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (238:245) - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (268:275) duplicated block id: 2590 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1138:1147) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1176:1185) duplicated block id: 2591 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:18) duplicated block id: 2592 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (316:323) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (480:487) duplicated block id: 2593 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (423:430) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (180:187) duplicated block id: 2594 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (221:229) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (237:245) duplicated block id: 2595 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:654) duplicated block id: 2596 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (316:323) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (381:388) duplicated block id: 2597 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (99:106) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (552:559) duplicated block id: 2598 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:723) duplicated block id: 2599 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/resampler.py (121:128) - maga_transformer/models/minicpmv_embedding/resampler.py (138:145) duplicated block id: 2600 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:541) duplicated block id: 2601 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/resampler.py (207:214) - maga_transformer/models/minicpmv/resampler.py (406:413) duplicated block id: 2602 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/gpt_neox.py (20:27) - maga_transformer/models/gpt_neox.py (81:88) duplicated block id: 2603 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (530:542) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (523:535) duplicated block id: 2604 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (496:503) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (572:579) duplicated block id: 2605 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:597) duplicated block id: 2606 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:16) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:18) duplicated block id: 2607 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (99:106) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (247:254) duplicated block id: 2608 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:387) duplicated block id: 2609 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:436) duplicated block id: 2610 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (552:559) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (707:714) duplicated block id: 2611 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (247:254) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1014:1021) duplicated block id: 2612 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:488) duplicated block id: 2613 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (147:154) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (173:180) duplicated block id: 2614 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:16) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:18) duplicated block id: 2615 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/AttentionOpTest.hpp (102:113) - maga_transformer/cpp/devices/base_tests/AttentionOpTest.hpp (204:216) duplicated block id: 2616 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/resampler.py (121:128) - maga_transformer/models/qwen_vl_vit.py (128:135) duplicated block id: 2617 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (17:25) duplicated block id: 2618 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:294) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:339) duplicated block id: 2619 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (552:559) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1014:1021) duplicated block id: 2620 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (21:29) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (5:13) duplicated block id: 2621 size: 8 cleaned lines of code in 2 files: - maga_transformer/tools/fake_bloom.py (17:24) - maga_transformer/tools/fake_glm_v2.py (23:30) duplicated block id: 2622 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2623 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (242:249) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (221:228) duplicated block id: 2624 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (21:28) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (19:26) duplicated block id: 2625 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (247:254) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (707:714) duplicated block id: 2626 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (381:388) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (453:460) duplicated block id: 2627 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1100:1109) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1176:1185) duplicated block id: 2628 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1100:1109) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1138:1147) duplicated block id: 2629 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (17:25) duplicated block id: 2630 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (99:106) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1166:1173) duplicated block id: 2631 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (21:29) - maga_transformer/cpp/kernels/layernorm_kernels.cu (25:33) duplicated block id: 2632 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (843:850) - maga_transformer/models/llava_vit.py (923:930) duplicated block id: 2633 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (178:185) - maga_transformer/models/internvl.py (50:57) duplicated block id: 2634 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:18) duplicated block id: 2635 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2570:2577) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2611:2618) duplicated block id: 2636 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (623:630) - maga_transformer/cpp/kernels/gpt_kernels.cu (633:640) duplicated block id: 2637 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (62:69) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/response_converter.py (63:70) duplicated block id: 2638 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (99:106) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (860:867) duplicated block id: 2639 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (247:254) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (399:406) duplicated block id: 2640 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (381:388) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (480:487) duplicated block id: 2641 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (204:211) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (275:282) duplicated block id: 2642 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (204:211) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (246:253) duplicated block id: 2643 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (204:211) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (224:231) duplicated block id: 2644 size: 8 cleaned lines of code in 2 files: - maga_transformer/tools/fake_glm_v2.py (23:30) - maga_transformer/tools/fake_model_base.py (23:30) duplicated block id: 2645 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2646 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (541:548) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (572:579) duplicated block id: 2647 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (181:188) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (59:66) duplicated block id: 2648 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (260:267) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (725:732) duplicated block id: 2649 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (500:507) - maga_transformer/cpp/cuda/cuda_utils.h (472:479) duplicated block id: 2650 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (5:13) - maga_transformer/cpp/kernels/layernorm_kernels.cu (25:33) duplicated block id: 2651 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (260:267) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (899:906) duplicated block id: 2652 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (403:410) - maga_transformer/cpp/cuda/cuda_utils.h (452:459) duplicated block id: 2653 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (694:703) - maga_transformer/openai/renderers/conversation.py (705:714) duplicated block id: 2654 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (607:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:149) duplicated block id: 2655 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (607:614) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:168) duplicated block id: 2656 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1244:1251) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1267:1274) duplicated block id: 2657 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (506:515) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (872:881) duplicated block id: 2658 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (377:386) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (192:201) duplicated block id: 2659 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (239:246) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (888:895) duplicated block id: 2660 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (239:246) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (785:792) duplicated block id: 2661 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:25) duplicated block id: 2662 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (209:216) - maga_transformer/cpp/cuda/cufmha/cufmha.h (56:63) duplicated block id: 2663 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (851:860) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (872:881) duplicated block id: 2664 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:17) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:18) duplicated block id: 2665 size: 8 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (547:554) - maga_transformer/openai/renderers/llama_template.py (723:730) duplicated block id: 2666 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (383:392) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (465:474) duplicated block id: 2667 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (472:479) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (147:154) duplicated block id: 2668 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (369:376) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (173:180) duplicated block id: 2669 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (81:88) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (99:106) duplicated block id: 2670 size: 8 cleaned lines of code in 2 files: - maga_transformer/models/downstream_modules/embedding/all_embedding_module.py (56:65) - maga_transformer/models/downstream_modules/embedding/dense_embedding_module.py (48:56) duplicated block id: 2671 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.h (45:52) - maga_transformer/cpp/api_server/ChatService.h (54:61) duplicated block id: 2672 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (341:349) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (188:196) duplicated block id: 2673 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (217:243) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (185:211) duplicated block id: 2674 size: 8 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.h (57:64) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.h (48:55) duplicated block id: 2675 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (207:214) - maga_transformer/cpp/devices/rocm_impl/ROCmOps.cc (31:38) duplicated block id: 2676 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1012:1018) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (208:214) duplicated block id: 2677 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:22) duplicated block id: 2678 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 2679 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2680 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 2681 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (158:164) - maga_transformer/cpp/kernels/add_residual_kernels.cu (177:183) duplicated block id: 2682 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 2683 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2684 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 2685 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_template.h (504:513) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_template.h (441:450) duplicated block id: 2686 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (82:88) - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (69:75) duplicated block id: 2687 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 2688 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2689 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (128:140) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (37:49) duplicated block id: 2690 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (160:168) - maga_transformer/cpp/devices/rocm_impl/ROCmDevice.cc (270:279) duplicated block id: 2691 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2692 size: 7 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (386:395) - maga_transformer/tokenizer/tokenization_chatglm3.py (300:309) duplicated block id: 2693 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 2694 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 2695 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (569:575) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (626:632) duplicated block id: 2696 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/int8_utils.cuh (9:15) duplicated block id: 2697 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 2698 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (56:72) duplicated block id: 2699 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (66:81) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (367:380) duplicated block id: 2700 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2701 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) duplicated block id: 2702 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (66:81) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (520:533) duplicated block id: 2703 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 2704 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2705 size: 7 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (74:82) - maga_transformer/device/device_impl.py (304:312) duplicated block id: 2706 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (199:205) - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (224:230) duplicated block id: 2707 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 2708 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2709 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (241:250) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (529:538) duplicated block id: 2710 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2711 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (508:515) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1801:1809) duplicated block id: 2712 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2713 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) duplicated block id: 2714 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (441:448) - maga_transformer/openai/renderers/conversation.py (476:484) duplicated block id: 2715 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2716 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) duplicated block id: 2717 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 2718 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/normal_engine/NormalBatchStreamProcessor.cc (382:388) - maga_transformer/cpp/speculative_engine/score_executor/ScoreBatchStreamProcessor.cc (106:112) duplicated block id: 2719 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (195:201) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (207:213) duplicated block id: 2720 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2721 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2722 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/propose_executor/MTPStream.h (76:84) - maga_transformer/cpp/speculative_engine/propose_executor/VanillaStream.h (43:52) duplicated block id: 2723 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2724 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 2725 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 2726 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaWeights.cc (12:19) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (15:22) duplicated block id: 2727 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/metrics/RtpLLMMetrics.h (42:50) - maga_transformer/cpp/model_rpc/DecodeGenerateContext.h (23:29) duplicated block id: 2728 size: 7 cleaned lines of code in 2 files: - maga_transformer/tools/quant/awq_quanter.py (12:18) - maga_transformer/tools/quant/gptq_quanter.py (14:20) duplicated block id: 2729 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (109:115) - maga_transformer/cpp/api_server/ChatService.h (45:51) duplicated block id: 2730 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2731 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (109:115) - maga_transformer/cpp/api_server/ChatService.h (54:60) duplicated block id: 2732 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2733 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) duplicated block id: 2734 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (105:111) - maga_transformer/model_loader/smooth_quant_weight.py (257:263) duplicated block id: 2735 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaWeights.cc (27:35) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (33:41) duplicated block id: 2736 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (409:417) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (618:626) duplicated block id: 2737 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 2738 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (100:106) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (114:120) duplicated block id: 2739 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) duplicated block id: 2740 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llava_renderer.py (52:58) - maga_transformer/openai/renderers/llava_renderer.py (79:85) duplicated block id: 2741 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) duplicated block id: 2742 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 2743 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (136:143) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (150:157) duplicated block id: 2744 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (123:133) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (57:67) duplicated block id: 2745 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 2746 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (315:321) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (246:252) duplicated block id: 2747 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2748 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 2749 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 2750 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (455:461) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (582:588) duplicated block id: 2751 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) duplicated block id: 2752 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (758:771) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (878:892) duplicated block id: 2753 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2754 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (158:166) - maga_transformer/model_loader/smooth_quant_weight.py (311:319) duplicated block id: 2755 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 2756 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (176:190) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) duplicated block id: 2757 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2758 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:707) duplicated block id: 2759 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 2760 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 2761 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2762 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (874:881) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1801:1809) duplicated block id: 2763 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) duplicated block id: 2764 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2765 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (242:253) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (209:220) duplicated block id: 2766 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 2767 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2768 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 2769 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (44:52) - maga_transformer/models/qwen_v2.py (130:138) duplicated block id: 2770 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (182:188) - maga_transformer/openai/renderers/qwen_agent/llm/text_base.py (17:23) duplicated block id: 2771 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) duplicated block id: 2772 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:585) duplicated block id: 2773 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 2774 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (378:385) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (477:485) duplicated block id: 2775 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 2776 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2777 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (92:98) - maga_transformer/models/whisper.py (55:61) duplicated block id: 2778 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:642) duplicated block id: 2779 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2780 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (189:195) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (246:252) duplicated block id: 2781 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2782 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2783 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) duplicated block id: 2784 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 2785 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (243:254) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (246:258) duplicated block id: 2786 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 2787 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 2788 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (372:378) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (1074:1080) duplicated block id: 2789 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (56:74) duplicated block id: 2790 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (355:361) - maga_transformer/cpp/kernels/rmsnormKernels.cu (87:93) duplicated block id: 2791 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (197:203) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (207:213) duplicated block id: 2792 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 2793 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (15:22) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (57:64) duplicated block id: 2794 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (152:158) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (166:172) duplicated block id: 2795 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (240:247) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (618:625) duplicated block id: 2796 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2797 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (253:261) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (545:553) duplicated block id: 2798 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (424:430) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (88:94) duplicated block id: 2799 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 2800 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 2801 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 2802 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) duplicated block id: 2803 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2804 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2805 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/10_int4_dequant_gemm_128x128x16x128_16_16x16_4x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) duplicated block id: 2806 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (78:93) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (165:180) duplicated block id: 2807 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/12_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:22) duplicated block id: 2808 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2809 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2810 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (652:658) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (754:760) duplicated block id: 2811 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (829:842) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (982:995) duplicated block id: 2812 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (261:268) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (222:229) duplicated block id: 2813 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 2814 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (649:657) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (874:881) duplicated block id: 2815 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:249) duplicated block id: 2816 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (649:657) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (853:860) duplicated block id: 2817 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (166:176) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (811:821) duplicated block id: 2818 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 2819 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (829:842) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1134:1147) duplicated block id: 2820 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:293) duplicated block id: 2821 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 2822 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (652:658) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (959:965) duplicated block id: 2823 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (969:975) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1007:1013) duplicated block id: 2824 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (128:134) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:83) duplicated block id: 2825 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) duplicated block id: 2826 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/oai.py (84:90) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (28:34) duplicated block id: 2827 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (93:99) - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (58:64) duplicated block id: 2828 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:386) duplicated block id: 2829 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (166:176) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (647:657) duplicated block id: 2830 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) duplicated block id: 2831 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2832 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:338) duplicated block id: 2833 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 2834 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (279:285) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (239:245) duplicated block id: 2835 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 2836 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 2837 size: 7 cleaned lines of code in 2 files: - bazel/defs.bzl (108:114) - bazel/defs.bzl (124:130) duplicated block id: 2838 size: 7 cleaned lines of code in 2 files: - bazel/defs.bzl (108:114) - bazel/defs.bzl (140:146) duplicated block id: 2839 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 2840 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/device/gemm_universal_base_compat.h (400:414) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/device/splitk_gemm_grouped.h (493:508) duplicated block id: 2841 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2842 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) duplicated block id: 2843 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 2844 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (44:50) - maga_transformer/cpp/api_server/ChatService.h (45:51) duplicated block id: 2845 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (44:50) - maga_transformer/cpp/api_server/ChatService.h (54:60) duplicated block id: 2846 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (16:22) duplicated block id: 2847 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llava.py (95:101) - maga_transformer/models/whisper.py (57:63) duplicated block id: 2848 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmWeights.cc (13:20) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (15:22) duplicated block id: 2849 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2850 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2851 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2852 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/26_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v4.cc (10:16) duplicated block id: 2853 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent_renderer.py (193:199) - maga_transformer/openai/renderers/qwen_agent_renderer.py (215:221) duplicated block id: 2854 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2855 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_float.h (66:76) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2278:2289) duplicated block id: 2856 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 2857 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2858 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (158:164) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (197:203) duplicated block id: 2859 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (243:254) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (242:253) duplicated block id: 2860 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 2861 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.h (182:197) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.h (201:215) duplicated block id: 2862 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 2863 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) duplicated block id: 2864 size: 7 cleaned lines of code in 2 files: - bazel/defs.bzl (90:96) - bazel/defs.bzl (140:146) duplicated block id: 2865 size: 7 cleaned lines of code in 2 files: - bazel/defs.bzl (90:96) - bazel/defs.bzl (124:130) duplicated block id: 2866 size: 7 cleaned lines of code in 2 files: - bazel/defs.bzl (90:96) - bazel/defs.bzl (108:114) duplicated block id: 2867 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) duplicated block id: 2868 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2.py (46:52) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (294:300) duplicated block id: 2869 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) duplicated block id: 2870 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2871 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:653) duplicated block id: 2872 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2873 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2874 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:722) duplicated block id: 2875 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:391) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (439:449) duplicated block id: 2876 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 2877 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2878 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 2879 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2880 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 2881 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2882 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (628:638) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (791:801) duplicated block id: 2883 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2884 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:391) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (515:525) duplicated block id: 2885 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (651:659) - maga_transformer/cpp/kernels/_fma.h (680:688) duplicated block id: 2886 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 2887 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (53:59) - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (156:162) duplicated block id: 2888 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llama.py (35:41) - maga_transformer/models/qwen_vl.py (80:86) duplicated block id: 2889 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:435) duplicated block id: 2890 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2891 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) duplicated block id: 2892 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 2893 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2894 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 2895 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (490:498) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (629:637) duplicated block id: 2896 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (210:216) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (509:515) duplicated block id: 2897 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2898 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:487) duplicated block id: 2899 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (70:79) - maga_transformer/cpp/api_server/ChatService.cc (143:152) duplicated block id: 2900 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 2901 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (537:543) - maga_transformer/cpp/cuda/cuda_utils.h (481:487) duplicated block id: 2902 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 2903 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 2904 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:540) duplicated block id: 2905 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (315:321) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (579:585) duplicated block id: 2906 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2907 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2908 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (490:498) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (559:567) duplicated block id: 2909 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 2910 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 2911 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (626:632) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (687:693) duplicated block id: 2912 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 2913 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (58:64) - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (69:75) duplicated block id: 2914 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 2915 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 2916 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (232:238) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (1063:1069) duplicated block id: 2917 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:596) duplicated block id: 2918 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (228:234) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (522:528) duplicated block id: 2919 size: 7 cleaned lines of code in 2 files: - maga_transformer/tools/quant/fp8_quanter.py (104:110) - maga_transformer/tools/quant/gptq_quanter.py (14:20) duplicated block id: 2920 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2921 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (229:243) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:163) duplicated block id: 2922 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 2923 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2924 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (304:312) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (973:981) duplicated block id: 2925 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 2926 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/GemmOpTest.hpp (188:194) - maga_transformer/cpp/devices/base_tests/GemmOpTest.hpp (206:212) duplicated block id: 2927 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 2928 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (300:308) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (485:494) duplicated block id: 2929 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2582:2588) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2611:2617) duplicated block id: 2930 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (303:309) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (470:476) duplicated block id: 2931 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/10_int4_dequant_gemm_128x128x16x128_16_16x16_4x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/24_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:22) duplicated block id: 2932 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 2933 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (520:530) - maga_transformer/cpp/kernels/rotary_position_embedding.h (577:588) duplicated block id: 2934 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (177:183) - maga_transformer/model_loader/smooth_quant_weight.py (327:333) duplicated block id: 2935 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (292:298) - maga_transformer/models/qwen2_vl/qwen2_vl.py (92:98) duplicated block id: 2936 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) duplicated block id: 2937 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2938 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (21:27) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (20:26) duplicated block id: 2939 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2940 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 2941 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (128:140) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (125:137) duplicated block id: 2942 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (1279:1285) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1314:1320) duplicated block id: 2943 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 2944 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (487:493) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (566:572) duplicated block id: 2945 size: 7 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (386:395) - maga_transformer/tokenizer/tokenization_chatglm4.py (197:206) duplicated block id: 2946 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (107:122) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (212:227) duplicated block id: 2947 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmWeights.cc (28:36) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (33:41) duplicated block id: 2948 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (304:312) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (668:676) duplicated block id: 2949 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 2950 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (62:69) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (88:96) duplicated block id: 2951 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 2952 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (36:45) - maga_transformer/cpp/kernels/mla_kernels/mla_merge_transpose_kernel.cu (135:144) duplicated block id: 2953 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/load_balancer/WRRLoadBalancer.cpp (32:40) - maga_transformer/cpp/disaggregate/rtpllm_master/cluster/PrefillLoadBalancer.cpp (44:51) duplicated block id: 2954 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (380:390) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (647:657) duplicated block id: 2955 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 2956 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 2957 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2958 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 2959 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2960 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (243:249) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (1074:1080) duplicated block id: 2961 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.h (43:50) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.h (37:44) duplicated block id: 2962 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (234:240) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (88:94) duplicated block id: 2963 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (304:312) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (768:776) duplicated block id: 2964 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (304:312) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (872:880) duplicated block id: 2965 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 2966 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2967 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 2968 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (361:367) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (1063:1069) duplicated block id: 2969 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (147:157) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (361:370) duplicated block id: 2970 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (449:457) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (816:824) duplicated block id: 2971 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (148:163) duplicated block id: 2972 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2973 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 2974 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2975 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (449:457) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (920:928) duplicated block id: 2976 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 2977 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (98:105) - maga_transformer/cpp/kernels/quantization_tensor.cu (178:185) duplicated block id: 2978 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 2979 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (197:211) duplicated block id: 2980 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 2981 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (449:457) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (716:724) duplicated block id: 2982 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llama_weight.py (188:194) - maga_transformer/models/qwen.py (49:55) duplicated block id: 2983 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (449:457) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (612:621) duplicated block id: 2984 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 2985 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (315:321) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (189:195) duplicated block id: 2986 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2987 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (105:113) - maga_transformer/models/qwen_v2.py (130:138) duplicated block id: 2988 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) duplicated block id: 2989 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 2990 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.cu (45:51) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (271:277) duplicated block id: 2991 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (106:112) - maga_transformer/model_loader/smooth_quant_weight.py (327:333) duplicated block id: 2992 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 2993 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) duplicated block id: 2994 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 2995 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 2996 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (402:410) - maga_transformer/openai/renderers/conversation.py (415:423) duplicated block id: 2997 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 2998 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 2999 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 3000 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2395:2401) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2470:2476) duplicated block id: 3001 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3002 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 3003 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (147:157) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (791:801) duplicated block id: 3004 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3005 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3006 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3007 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/weightOnlyQuantMatmulPlugin/weightOnlyQuantMatmulPlugin.cpp (73:79) - maga_transformer/cpp/trt_plugins/weightOnlyQuantMatmulPlugin/weightOnlyQuantMatmulPlugin.h (67:73) duplicated block id: 3008 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3009 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 3010 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/load_balancer/RRLoadBalancer.cpp (27:35) - maga_transformer/cpp/disaggregate/load_balancer/WRRLoadBalancer.cpp (32:40) duplicated block id: 3011 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 3012 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:406) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:561) duplicated block id: 3013 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:406) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:508) duplicated block id: 3014 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 3015 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3016 size: 7 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (50:57) - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (82:90) duplicated block id: 3017 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) duplicated block id: 3018 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 3019 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (282:292) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (510:520) duplicated block id: 3020 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent_renderer.py (96:102) - maga_transformer/openai/renderers/qwen_renderer.py (330:336) duplicated block id: 3021 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 3022 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:406) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:675) duplicated block id: 3023 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (25:31) - maga_transformer/models/llava.py (95:101) duplicated block id: 3024 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (428:435) - maga_transformer/openai/renderers/conversation.py (476:484) duplicated block id: 3025 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:406) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:618) duplicated block id: 3026 size: 7 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (58:64) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (89:95) duplicated block id: 3027 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 3028 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3029 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (219:225) - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.h (85:91) duplicated block id: 3030 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (82:88) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (114:120) duplicated block id: 3031 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 3032 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/basic_renderer.py (25:31) - maga_transformer/tokenizer/tokenization_qwen.py (67:74) duplicated block id: 3033 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 3034 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (429:436) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (270:277) duplicated block id: 3035 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:406) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:455) duplicated block id: 3036 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 3037 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3038 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (226:233) - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (235:242) duplicated block id: 3039 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (30:36) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (47:53) duplicated block id: 3040 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/group_wise_quant_weight.py (23:29) - maga_transformer/model_loader/omni_quant_weight.py (19:25) duplicated block id: 3041 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (337:344) - maga_transformer/openai/renderers/custom_renderer.py (606:613) duplicated block id: 3042 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3043 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 3044 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (319:329) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (57:67) duplicated block id: 3045 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3046 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) duplicated block id: 3047 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (997:1003) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1011:1017) duplicated block id: 3048 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) duplicated block id: 3049 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1085:1092) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1121:1128) duplicated block id: 3050 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (185:191) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (204:210) duplicated block id: 3051 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 3052 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1085:1092) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1161:1168) duplicated block id: 3053 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/internvl_vit.py (191:198) - maga_transformer/models/llava_vit.py (78:85) duplicated block id: 3054 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (128:140) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (138:150) duplicated block id: 3055 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 3056 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 3057 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 3058 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3059 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (10:16) duplicated block id: 3060 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (399:407) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (559:567) duplicated block id: 3061 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3062 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3063 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (66:72) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (190:196) duplicated block id: 3064 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3065 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (234:240) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (272:278) duplicated block id: 3066 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) duplicated block id: 3067 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (56:74) duplicated block id: 3068 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 3069 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2611:2617) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2623:2629) duplicated block id: 3070 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 3071 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1221:1228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1273:1280) duplicated block id: 3072 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3073 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (81:87) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (151:157) duplicated block id: 3074 size: 7 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (9:18) - maga_transformer/tokenizer/tokenization_chatglm3.py (14:23) duplicated block id: 3075 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 3076 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3077 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 3078 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (82:88) - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (93:99) duplicated block id: 3079 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:331) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (439:449) duplicated block id: 3080 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) duplicated block id: 3081 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:455) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:675) duplicated block id: 3082 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3083 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 3084 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (82:90) - maga_transformer/model_loader/smooth_quant_weight.py (56:64) duplicated block id: 3085 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3086 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (676:689) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (982:995) duplicated block id: 3087 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) duplicated block id: 3088 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 3089 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:331) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (515:525) duplicated block id: 3090 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaActOp.cc (103:109) - maga_transformer/cpp/devices/rocm_impl/ROCmActOp.cc (89:95) duplicated block id: 3091 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/banRepeatNgram.cu (9:15) duplicated block id: 3092 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:455) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:561) duplicated block id: 3093 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:455) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:618) duplicated block id: 3094 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3095 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 3096 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 3097 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3098 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) duplicated block id: 3099 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (826:832) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (858:864) duplicated block id: 3100 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3101 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (60:67) - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (171:178) duplicated block id: 3102 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/minicpmv.py (178:189) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (224:233) duplicated block id: 3103 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 3104 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 3105 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3106 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl_vit.py (123:129) - maga_transformer/models/qwen2_vl/qwen2_vl_vit.py (163:169) duplicated block id: 3107 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3108 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (312:318) - maga_transformer/cpp/kernels/rmsnormKernels.cu (45:51) duplicated block id: 3109 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (56:72) duplicated block id: 3110 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) duplicated block id: 3111 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (185:191) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (275:281) duplicated block id: 3112 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3113 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (185:191) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (246:252) duplicated block id: 3114 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (185:191) - maga_transformer/cpp/devices/arm_impl/gemm_opt/arm_common.h (224:230) duplicated block id: 3115 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (508:515) - maga_transformer/cpp/kernels/_mul.h (649:657) duplicated block id: 3116 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) duplicated block id: 3117 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (399:407) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (629:637) duplicated block id: 3118 size: 7 cleaned lines of code in 2 files: - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (133:139) - maga_transformer/aios/kmonitor/python_client/flume/ThriftSourceProtocol.py (152:158) duplicated block id: 3119 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) duplicated block id: 3120 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 3121 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 3122 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/10_int4_dequant_gemm_128x128x16x128_16_16x16_4x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:22) duplicated block id: 3123 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 3124 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuggemm/cuggemm.cc (6:12) - maga_transformer/cpp/cuda/cuggemm/cuggemm.h (29:35) duplicated block id: 3125 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (574:580) - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (604:610) duplicated block id: 3126 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 3127 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (676:689) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1134:1147) duplicated block id: 3128 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (366:377) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (375:386) duplicated block id: 3129 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3130 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (292:298) - maga_transformer/models/qwen_vl.py (80:86) duplicated block id: 3131 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (73:79) - maga_transformer/model_loader/static_fp8_quant_weight.py (92:98) duplicated block id: 3132 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (525:533) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (716:724) duplicated block id: 3133 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (9:15) duplicated block id: 3134 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (151:157) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (184:190) duplicated block id: 3135 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/GroupGemm.cc (17:23) - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (38:44) duplicated block id: 3136 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3137 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 3138 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/int8_utils.cuh (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3139 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3140 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 3141 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (220:228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (872:880) duplicated block id: 3142 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (424:430) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (272:278) duplicated block id: 3143 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (294:304) - maga_transformer/cpp/kernels/_mul.h (308:318) duplicated block id: 3144 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:585) duplicated block id: 3145 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:642) duplicated block id: 3146 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (220:228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (768:776) duplicated block id: 3147 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (525:533) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (612:621) duplicated block id: 3148 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3149 size: 7 cleaned lines of code in 2 files: - bazel/tf_proto.bzl (235:241) - bazel/tf_proto.bzl (352:358) duplicated block id: 3150 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 3151 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (60:67) - maga_transformer/model_loader/static_fp8_quant_weight.py (169:176) duplicated block id: 3152 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (934:940) - maga_transformer/cpp/kernels/vec_dtypes.cuh (1007:1013) duplicated block id: 3153 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 3154 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (934:940) - maga_transformer/cpp/kernels/vec_dtypes.cuh (969:975) duplicated block id: 3155 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1121:1128) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1161:1168) duplicated block id: 3156 size: 7 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (45:51) - maga_transformer/device/device_impl.py (141:147) duplicated block id: 3157 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:707) duplicated block id: 3158 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 3159 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3160 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (220:228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (973:981) duplicated block id: 3161 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (121:129) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (104:112) duplicated block id: 3162 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen_vl.py (80:86) - maga_transformer/models/whisper.py (55:61) duplicated block id: 3163 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3164 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (998:1009) - maga_transformer/openai/renderers/conversation.py (1012:1021) duplicated block id: 3165 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:455) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:508) duplicated block id: 3166 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3167 size: 7 cleaned lines of code in 2 files: - maga_transformer/device/device_impl.py (45:51) - maga_transformer/device/device_impl.py (275:281) duplicated block id: 3168 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.h (72:78) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.h (50:56) duplicated block id: 3169 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/normal_engine/NormalEngine.cc (224:230) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (266:272) duplicated block id: 3170 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/26_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v4.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) duplicated block id: 3171 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (301:307) - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.h (54:60) duplicated block id: 3172 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1878:1884) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1904:1910) duplicated block id: 3173 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 3174 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (525:533) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (920:928) duplicated block id: 3175 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/MessagerClient.cpp (68:74) - maga_transformer/cpp/disaggregate/cache_store/MessagerClient.h (23:29) duplicated block id: 3176 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) duplicated block id: 3177 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 3178 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (279:285) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (416:422) duplicated block id: 3179 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/modeling_navit_siglip.py (837:844) - maga_transformer/models/qwen_v2_audio/modeling_qwen2_audio.py (483:490) duplicated block id: 3180 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (651:659) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1025:1033) duplicated block id: 3181 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3182 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) duplicated block id: 3183 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3184 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:477) duplicated block id: 3185 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:377) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:530) duplicated block id: 3186 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3187 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (525:533) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (816:824) duplicated block id: 3188 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) duplicated block id: 3189 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (13:19) - maga_transformer/cpp/devices/utils/DebugUtils.cc (125:131) duplicated block id: 3190 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (245:257) duplicated block id: 3191 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3192 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (296:310) duplicated block id: 3193 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) duplicated block id: 3194 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (526:536) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (119:129) duplicated block id: 3195 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmKernel.h (34:40) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmKernel.h (49:55) duplicated block id: 3196 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (680:688) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (996:1004) duplicated block id: 3197 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 3198 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (303:309) - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (397:403) duplicated block id: 3199 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) duplicated block id: 3200 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 3201 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (195:203) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (157:165) duplicated block id: 3202 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) duplicated block id: 3203 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (315:321) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (342:348) duplicated block id: 3204 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (220:228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (304:312) duplicated block id: 3205 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (303:309) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.h (171:177) duplicated block id: 3206 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 3207 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (856:862) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (959:965) duplicated block id: 3208 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/26_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v4.cc (10:16) duplicated block id: 3209 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3210 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (245:257) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (47:59) duplicated block id: 3211 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 3212 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (9:15) duplicated block id: 3213 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (82:88) - maga_transformer/models/rotary_embedding/deepseek_rotary_embedding.py (156:162) duplicated block id: 3214 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (627:633) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (636:642) duplicated block id: 3215 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent_renderer.py (215:221) - maga_transformer/openai/renderers/qwen_agent_renderer.py (235:241) duplicated block id: 3216 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.h (151:157) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.h (161:167) duplicated block id: 3217 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2.py (46:52) - maga_transformer/models/whisper.py (57:63) duplicated block id: 3218 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3219 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (21:27) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (18:24) duplicated block id: 3220 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) duplicated block id: 3221 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (193:202) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (280:290) duplicated block id: 3222 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (853:860) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1801:1809) duplicated block id: 3223 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (77:91) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) duplicated block id: 3224 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (220:228) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (668:676) duplicated block id: 3225 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 3226 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 3227 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/quantization_rocm.h (40:47) - maga_transformer/cpp/kernels/rocm/quantization_rocm.h (50:57) duplicated block id: 3228 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (192:199) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (318:325) duplicated block id: 3229 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (10:16) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3230 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (646:652) - maga_transformer/cpp/cuda/cuda_utils.h (491:497) duplicated block id: 3231 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 3232 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (754:760) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (856:862) duplicated block id: 3233 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (715:721) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (477:485) duplicated block id: 3234 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.h (90:96) - maga_transformer/cpp/kernels/sampling_topp_kernels.h (158:164) duplicated block id: 3235 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (99:105) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (184:190) duplicated block id: 3236 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (421:428) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (482:489) duplicated block id: 3237 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (16:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/24_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (16:22) duplicated block id: 3238 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 3239 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaPrefillAttention.cc (231:237) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (253:259) duplicated block id: 3240 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 3241 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (9:15) duplicated block id: 3242 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (128:140) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (38:50) duplicated block id: 3243 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/12_int4_dequant_gemm_128x64x16x128_16_16x16_2x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) duplicated block id: 3244 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 3245 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (361:370) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (628:638) duplicated block id: 3246 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (114:120) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (237:243) duplicated block id: 3247 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/static_fp8_quant_weight.py (145:151) - maga_transformer/model_loader/static_fp8_quant_weight.py (287:293) duplicated block id: 3248 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3249 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3250 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3251 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) duplicated block id: 3252 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (424:430) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (71:77) duplicated block id: 3253 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl_vit.py (113:119) - maga_transformer/models/qwen2_vl/qwen2_vl_vit.py (153:159) duplicated block id: 3254 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (424:430) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (59:65) duplicated block id: 3255 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3256 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3257 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantize_weight.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3258 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (10:16) duplicated block id: 3259 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (234:240) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (59:65) duplicated block id: 3260 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/image_processing_qwen2_vl.py (207:213) - maga_transformer/models/qwen2_vl/image_processing_qwen2_vl.py (328:334) duplicated block id: 3261 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (579:585) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (246:252) duplicated block id: 3262 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (99:105) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (133:139) duplicated block id: 3263 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (234:240) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (71:77) duplicated block id: 3264 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 3265 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (246:258) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (209:220) duplicated block id: 3266 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (244:250) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:137) duplicated block id: 3267 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) duplicated block id: 3268 size: 7 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (386:395) - maga_transformer/tokenizer/tokenization_chatglm2.py (207:216) duplicated block id: 3269 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (23:29) - maga_transformer/models/qwen2_vl/qwen2_vl.py (92:98) duplicated block id: 3270 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuDevice.cc (112:118) - maga_transformer/cpp/devices/cpu_impl/CpuDevice.cc (138:144) duplicated block id: 3271 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (605:611) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (361:367) duplicated block id: 3272 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 3273 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 3274 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 3275 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (128:134) - maga_transformer/model_loader/smooth_quant_weight.py (347:353) duplicated block id: 3276 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (996:1004) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1025:1033) duplicated block id: 3277 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (510:520) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (921:931) duplicated block id: 3278 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3279 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 3280 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (393:399) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (441:447) duplicated block id: 3281 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/stop_criteria_kernels.cu (9:15) duplicated block id: 3282 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (125:137) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (90:104) duplicated block id: 3283 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 3284 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:229) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (203:209) duplicated block id: 3285 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) duplicated block id: 3286 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) duplicated block id: 3287 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (9:15) duplicated block id: 3288 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:477) duplicated block id: 3289 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (480:487) - maga_transformer/cpp/kernels/rotary_position_embedding.h (514:521) duplicated block id: 3290 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2570:2576) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2582:2588) duplicated block id: 3291 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2570:2576) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2623:2629) duplicated block id: 3292 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (529:536) - maga_transformer/cpp/kernels/_mul.h (649:657) duplicated block id: 3293 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:426) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:530) duplicated block id: 3294 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/int8SQ.cu (9:15) duplicated block id: 3295 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3296 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (73:79) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (133:139) duplicated block id: 3297 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (128:134) - maga_transformer/model_loader/smooth_quant_weight.py (197:203) duplicated block id: 3298 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 3299 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (9:15) duplicated block id: 3300 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3301 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 3302 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/25_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v4.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 3303 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) duplicated block id: 3304 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3305 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (643:649) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (667:674) duplicated block id: 3306 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (56:74) duplicated block id: 3307 size: 7 cleaned lines of code in 2 files: - example/perf_test/defs.bzl (54:60) - example/perf_test/defs.bzl (110:116) duplicated block id: 3308 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) duplicated block id: 3309 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (529:536) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1801:1809) duplicated block id: 3310 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/load_balancer/RRLoadBalancer.cpp (27:35) - maga_transformer/cpp/disaggregate/rtpllm_master/cluster/PrefillLoadBalancer.cpp (44:51) duplicated block id: 3311 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (9:15) duplicated block id: 3312 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 3313 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (606:612) - maga_transformer/cpp/kernels/gpt_kernels.cu (614:620) duplicated block id: 3314 size: 7 cleaned lines of code in 2 files: - maga_transformer/server/frontend_worker.py (172:178) - maga_transformer/server/frontend_worker.py (214:222) duplicated block id: 3315 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/logprob_kernels.cu (21:27) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (9:15) duplicated block id: 3316 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (197:205) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_fp8_kernels.cu (289:297) duplicated block id: 3317 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (294:304) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1439:1449) duplicated block id: 3318 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (25:31) - maga_transformer/models/cogvlm2.py (46:52) duplicated block id: 3319 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_kernels.cu (9:15) duplicated block id: 3320 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (9:15) duplicated block id: 3321 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaDeepEPLLFfnLayer.cc (21:29) - maga_transformer/cpp/devices/cuda_impl/CudaFfnLayer.cc (55:61) duplicated block id: 3322 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (295:301) - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (387:393) duplicated block id: 3323 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (268:275) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (399:406) duplicated block id: 3324 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llava.py (95:101) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (294:300) duplicated block id: 3325 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/score_executor/ScoreExecutor.cc (25:36) - maga_transformer/cpp/speculative_engine/score_executor/ScoreExecutor.cc (67:78) duplicated block id: 3326 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/ban_bad_words.cu (9:15) duplicated block id: 3327 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 3328 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) duplicated block id: 3329 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) duplicated block id: 3330 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (214:229) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (367:380) duplicated block id: 3331 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (214:229) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (520:533) duplicated block id: 3332 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/int8_utils.cuh (9:15) duplicated block id: 3333 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (13:19) - maga_transformer/cpp/devices/utils/DebugUtils.cc (168:174) duplicated block id: 3334 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) duplicated block id: 3335 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (26:33) - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (24:31) duplicated block id: 3336 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (22:28) - maga_transformer/cpp/kernels/logprob_kernels.cu (21:27) duplicated block id: 3337 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (9:15) duplicated block id: 3338 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3339 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (13:19) - maga_transformer/cpp/devices/utils/DebugUtils.cc (214:220) duplicated block id: 3340 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2340:2346) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2395:2401) duplicated block id: 3341 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (105:113) - maga_transformer/models/qwen2_vl/qwen2_vl.py (44:52) duplicated block id: 3342 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/5_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_8x16x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (18:25) duplicated block id: 3343 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/add_residual_kernels.cu (9:15) duplicated block id: 3344 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3345 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 3346 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (9:15) duplicated block id: 3347 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llama.py (35:41) - maga_transformer/models/qwen2_vl/qwen2_vl.py (92:98) duplicated block id: 3348 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 3349 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3350 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/8_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_8x16x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (18:25) duplicated block id: 3351 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/llama.py (37:43) - maga_transformer/models/llava.py (95:101) duplicated block id: 3352 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2340:2346) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2470:2476) duplicated block id: 3353 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 3354 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/13_int4_dequant_gemm_128x32x16x128_16_16x16_1x1_16x8x1_8x16x1_16_1x16x1x8_2_intrawave_v3.cc (10:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (10:16) duplicated block id: 3355 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/fp8Gemm.cu (9:15) duplicated block id: 3356 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3357 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/quantization_tensor.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3358 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_renderer.py (147:154) - maga_transformer/utils/smooth_quant_convert/qwen/utils.py (92:99) duplicated block id: 3359 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3360 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (228:234) - maga_transformer/model_loader/static_fp8_quant_weight.py (246:252) duplicated block id: 3361 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3362 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (33:41) - maga_transformer/cpp/devices/rocm_impl/ROCmWeights.cc (72:80) duplicated block id: 3363 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (17:24) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (371:378) duplicated block id: 3364 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (9:15) duplicated block id: 3365 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3366 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/custom_ar_kernels.cu (244:251) - maga_transformer/cpp/kernels/custom_ar_kernels.cu (299:306) duplicated block id: 3367 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (9:15) duplicated block id: 3368 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (342:348) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (189:195) duplicated block id: 3369 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (9:15) duplicated block id: 3370 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3371 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/gen_relative_pos_bias.cu (9:15) duplicated block id: 3372 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cute_util.cuh (10:16) - maga_transformer/cpp/kernels/triton/layernorm_kernels.cu (9:15) duplicated block id: 3373 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3374 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (569:575) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (687:693) duplicated block id: 3375 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cutlass_kernel.h (140:147) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cutlass_kernel.h (222:229) duplicated block id: 3376 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (682:688) - maga_transformer/cpp/cuda/cuda_utils.cc (695:701) duplicated block id: 3377 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/rmsnormKernels.cu (9:15) duplicated block id: 3378 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (267:273) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (971:977) duplicated block id: 3379 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/6_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_16x4x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (18:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/7_int4_dequant_gemm_64x16x16x128_16_16x16_1x1_8x8x1_8x8x1_16_1x16x1x4_4_intrawave_v3.cc (18:25) duplicated block id: 3380 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel.cuh (10:16) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (9:15) duplicated block id: 3381 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3382 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3383 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (342:348) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (579:585) duplicated block id: 3384 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (51:71) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (56:72) duplicated block id: 3385 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2.py (46:52) - maga_transformer/models/llama.py (37:43) duplicated block id: 3386 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3387 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3388 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/logprob_kernels.cu (9:15) duplicated block id: 3389 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) duplicated block id: 3390 size: 7 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (56:64) - maga_transformer/model_loader/static_fp8_quant_weight.py (99:107) duplicated block id: 3391 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (471:478) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (305:312) duplicated block id: 3392 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (19:25) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:26) duplicated block id: 3393 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (9:15) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_traits.cuh (10:16) duplicated block id: 3394 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (201:208) - maga_transformer/openai/renderers/qwen_agent/llm/text_base.py (15:22) duplicated block id: 3395 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (10:16) - maga_transformer/cpp/kernels/activation_fp8_kernels.cu (9:15) duplicated block id: 3396 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/gpt_kernels.cu (9:15) duplicated block id: 3397 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (10:16) - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (9:15) duplicated block id: 3398 size: 7 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (717:725) - maga_transformer/openai/renderers/conversation.py (728:736) duplicated block id: 3399 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/CacheStoreServiceImpl.cpp (12:18) - maga_transformer/cpp/disaggregate/cache_store/MessagerServer.cpp (12:18) duplicated block id: 3400 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/memory_utils.cu (9:15) - maga_transformer/cpp/kernels/vec_dtypes.cuh (9:15) duplicated block id: 3401 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/reduce_kernel_utils.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3402 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) - maga_transformer/cpp/kernels/activation_kernels.cu (9:15) duplicated block id: 3403 size: 7 cleaned lines of code in 2 files: - maga_transformer/models/bert.py (23:29) - maga_transformer/models/qwen_vl.py (80:86) duplicated block id: 3404 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (9:15) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_hopper_input.cu (10:16) duplicated block id: 3405 size: 7 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (342:348) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (246:252) duplicated block id: 3406 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/group_wise_quant_weight.py (51:56) - maga_transformer/model_loader/smooth_quant_weight.py (258:263) duplicated block id: 3407 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1012:1017) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (196:201) duplicated block id: 3408 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen_v2.py (131:138) - maga_transformer/models/starcoder.py (41:48) duplicated block id: 3409 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (301:309) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (35:44) duplicated block id: 3410 size: 6 cleaned lines of code in 2 files: - maga_transformer/server/frontend_server.py (124:131) - maga_transformer/server/frontend_server.py (163:170) duplicated block id: 3411 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (589:595) duplicated block id: 3412 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) duplicated block id: 3413 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (400:406) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (469:475) duplicated block id: 3414 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) duplicated block id: 3415 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (423:428) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (551:556) duplicated block id: 3416 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (213:219) - maga_transformer/cpp/kernels/vec_dtypes.cuh (233:239) duplicated block id: 3417 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3418 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (607:612) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3419 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (449:454) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (524:529) duplicated block id: 3420 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (607:612) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3421 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:245) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) duplicated block id: 3422 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (607:612) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3423 size: 6 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/hf_llama_convert.py (226:232) - maga_transformer/utils/smooth_quant_convert/qwen/hf_qwen_convert.py (282:288) duplicated block id: 3424 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (288:293) duplicated block id: 3425 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tokenization_qwen.py (80:85) - maga_transformer/openai/renderers/qwen_agent/utils/tokenization_qwen.py (108:113) duplicated block id: 3426 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (334:340) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (400:406) duplicated block id: 3427 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3428 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:245) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) duplicated block id: 3429 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (52:57) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (196:201) duplicated block id: 3430 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3431 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (52:57) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (208:213) duplicated block id: 3432 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (188:199) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (197:208) duplicated block id: 3433 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:245) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (439:447) duplicated block id: 3434 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (259:267) - maga_transformer/cpp/devices/rocm_impl/ROCmFfnLayer.cc (170:178) duplicated block id: 3435 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (715:721) duplicated block id: 3436 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cosyvoice_qwen.py (17:23) - maga_transformer/models/qwen_vl.py (101:106) duplicated block id: 3437 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (392:397) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (429:434) duplicated block id: 3438 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (646:652) duplicated block id: 3439 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/jina_bert/jina_bert_weight.py (75:80) - maga_transformer/models/megatron_bert_weight.py (49:54) duplicated block id: 3440 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (334:340) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (469:475) duplicated block id: 3441 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (396:402) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (241:247) duplicated block id: 3442 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (66:71) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (85:90) duplicated block id: 3443 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (489:496) - maga_transformer/openai/renderers/conversation.py (1273:1280) duplicated block id: 3444 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (224:230) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (358:365) duplicated block id: 3445 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/model_weight_info.py (379:384) - maga_transformer/models/megatron_bert_weight.py (49:54) duplicated block id: 3446 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (38:43) - maga_transformer/cpp/kernels/sampling_topk_kernels.h (90:95) duplicated block id: 3447 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen_vl_vit.py (271:276) - maga_transformer/models/qwen_vl_vit.py (284:289) duplicated block id: 3448 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (237:245) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (515:523) duplicated block id: 3449 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (429:434) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (696:701) duplicated block id: 3450 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (720:727) - maga_transformer/cpp/kernels/_fma.h (768:775) duplicated block id: 3451 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 3452 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (80:90) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (290:300) duplicated block id: 3453 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (20:25) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (29:34) duplicated block id: 3454 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (244:249) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:161) duplicated block id: 3455 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1452:1457) - maga_transformer/cpp/kernels/sampling_topp_kernels.cu (1461:1466) duplicated block id: 3456 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (80:90) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (393:403) duplicated block id: 3457 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (102:107) - maga_transformer/models/qwen_v2.py (175:181) duplicated block id: 3458 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (82:87) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (166:171) duplicated block id: 3459 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) duplicated block id: 3460 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (174:179) - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (104:109) duplicated block id: 3461 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (180:185) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (247:252) duplicated block id: 3462 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/model_weight_info.py (379:384) - maga_transformer/models/jina_bert/jina_bert_weight.py (75:80) duplicated block id: 3463 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (161:166) - maga_transformer/cpp/kernels/_add.h (189:194) duplicated block id: 3464 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (205:210) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (267:272) duplicated block id: 3465 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (579:584) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (647:652) duplicated block id: 3466 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (739:747) - maga_transformer/openai/renderers/conversation.py (1348:1356) duplicated block id: 3467 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (161:167) duplicated block id: 3468 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (703:708) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (788:793) duplicated block id: 3469 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (80:90) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (175:185) duplicated block id: 3470 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasAlgoMap.cc (52:58) - maga_transformer/cpp/cuda/cublas/cublasAlgoMap.cc (141:147) duplicated block id: 3471 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (219:229) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1139:1147) duplicated block id: 3472 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (196:201) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (36:41) duplicated block id: 3473 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/BeamSearchOpTest.hpp (16:26) - maga_transformer/cpp/devices/torch_impl/BeamSearchOp.h (10:20) duplicated block id: 3474 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (207:213) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (621:627) duplicated block id: 3475 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (700:707) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1065:1072) duplicated block id: 3476 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (450:456) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (499:504) duplicated block id: 3477 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (78:84) - maga_transformer/openai/renderers/custom_renderer.py (118:124) duplicated block id: 3478 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3479 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (210:215) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (275:280) duplicated block id: 3480 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (286:292) duplicated block id: 3481 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (242:248) duplicated block id: 3482 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 3483 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (571:576) - maga_transformer/cpp/kernels/activation_kernels.cu (602:607) duplicated block id: 3484 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3485 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/cogvlm2_render.py (77:83) - maga_transformer/openai/renderers/qwen_vl_renderer.py (47:53) duplicated block id: 3486 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1163:1168) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1180:1185) duplicated block id: 3487 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (70:75) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (167:172) duplicated block id: 3488 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmDevice.cc (32:37) - maga_transformer/cpp/devices/cuda_impl/CudaDevice.cc (130:135) duplicated block id: 3489 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (19:24) - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 3490 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (571:576) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (640:645) duplicated block id: 3491 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (428:434) duplicated block id: 3492 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (71:81) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (987:995) duplicated block id: 3493 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (379:385) duplicated block id: 3494 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (556:563) - maga_transformer/cpp/kernels/_mul.h (605:612) duplicated block id: 3495 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (556:563) - maga_transformer/cpp/kernels/_mul.h (627:634) duplicated block id: 3496 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (323:328) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (385:390) duplicated block id: 3497 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (331:337) duplicated block id: 3498 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (107:112) - maga_transformer/cpp/kernels/activation_kernels.cu (122:127) duplicated block id: 3499 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (47:52) - maga_transformer/model_loader/smooth_quant_weight.py (258:263) duplicated block id: 3500 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.h (32:38) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (7:13) duplicated block id: 3501 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/normal_engine/NormalEngine.cc (119:124) - maga_transformer/cpp/normal_engine/NormalEngine.cc (140:145) duplicated block id: 3502 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (162:168) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) duplicated block id: 3503 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (71:81) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (1139:1147) duplicated block id: 3504 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/static_fp8_quant_weight.py (136:142) - maga_transformer/model_loader/static_fp8_quant_weight.py (278:284) duplicated block id: 3505 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (533:539) duplicated block id: 3506 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (124:130) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (480:486) duplicated block id: 3507 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (84:89) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (184:189) duplicated block id: 3508 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/embedding_engine/arpc/ArpcServiceCreator.cc (7:13) - maga_transformer/cpp/embedding_engine/arpc/ArpcServiceCreator.h (10:16) duplicated block id: 3509 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (412:418) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (255:261) duplicated block id: 3510 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (203:208) - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (182:187) duplicated block id: 3511 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (219:229) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (987:995) duplicated block id: 3512 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3513 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1213:1220) - maga_transformer/openai/renderers/conversation.py (1228:1236) duplicated block id: 3514 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (716:722) duplicated block id: 3515 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) duplicated block id: 3516 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) duplicated block id: 3517 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (768:775) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1065:1072) duplicated block id: 3518 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 3519 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 3520 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (47:52) - maga_transformer/model_loader/smooth_quant_weight.py (106:111) duplicated block id: 3521 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/oai.py (84:89) - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (26:31) duplicated block id: 3522 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (647:653) duplicated block id: 3523 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (74:80) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (384:390) duplicated block id: 3524 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (565:570) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (763:768) duplicated block id: 3525 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2_weight.py (64:69) - maga_transformer/models/cogvlm2_weight.py (84:89) duplicated block id: 3526 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (15:20) duplicated block id: 3527 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/launchers/fused_moe_gemm_launcher_sm80.h (17:23) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/launchers/fused_moe_gemm_launcher_sm80.inl (31:37) duplicated block id: 3528 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3529 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1030:1038) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1130:1138) duplicated block id: 3530 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (590:596) duplicated block id: 3531 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (976:981) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3532 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (354:359) - maga_transformer/openai/renderers/llama_template.py (429:434) duplicated block id: 3533 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (976:981) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3534 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (976:981) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3535 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (94:99) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (167:172) duplicated block id: 3536 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:909) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) duplicated block id: 3537 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (700:707) - maga_transformer/cpp/kernels/_fma.h (720:727) duplicated block id: 3538 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:909) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) duplicated block id: 3539 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 3540 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (394:399) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (266:271) duplicated block id: 3541 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (534:540) duplicated block id: 3542 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3256:3261) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3864:3869) duplicated block id: 3543 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3544 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (144:149) - maga_transformer/cpp/kernels/sampling_topp_kernels.h (158:163) duplicated block id: 3545 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (639:644) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (670:675) duplicated block id: 3546 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (70:76) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (119:125) duplicated block id: 3547 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (386:392) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (532:538) duplicated block id: 3548 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (146:156) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (57:67) duplicated block id: 3549 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (122:127) - maga_transformer/models/qwen_v2.py (152:157) duplicated block id: 3550 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (481:487) duplicated block id: 3551 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (482:487) - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (503:508) duplicated block id: 3552 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (299:306) - maga_transformer/openai/renderers/llama_template.py (315:322) duplicated block id: 3553 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (19:24) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 3554 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (146:156) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (158:168) duplicated block id: 3555 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_from_fp8.h (50:56) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2510:2516) duplicated block id: 3556 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/openvino.py (93:98) - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (26:31) duplicated block id: 3557 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:910) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1012:1017) duplicated block id: 3558 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (196:201) - maga_transformer/models/qwen_v2.py (175:181) duplicated block id: 3559 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/mpt.py (21:26) - maga_transformer/models/phi.py (19:24) duplicated block id: 3560 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3561 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3562 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (429:435) duplicated block id: 3563 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (332:338) duplicated block id: 3564 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (380:386) duplicated block id: 3565 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (655:664) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (606:615) duplicated block id: 3566 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (386:392) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (606:612) duplicated block id: 3567 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (282:290) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (351:359) duplicated block id: 3568 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (641:646) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (836:841) duplicated block id: 3569 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (288:293) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:161) duplicated block id: 3570 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (159:164) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (176:181) duplicated block id: 3571 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (20:25) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (36:41) duplicated block id: 3572 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (144:149) - maga_transformer/models/qwen2_vl/qwen2_vl.py (102:107) duplicated block id: 3573 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/minicpmv.py (55:61) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (97:103) duplicated block id: 3574 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (287:293) duplicated block id: 3575 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/propose_executor/MTPStream.h (125:133) - maga_transformer/cpp/speculative_engine/propose_executor/VanillaStream.h (71:79) duplicated block id: 3576 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (202:208) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (243:249) duplicated block id: 3577 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 3578 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (274:279) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (287:292) duplicated block id: 3579 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (20:25) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (301:306) duplicated block id: 3580 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/ChatService.cc (99:105) - maga_transformer/cpp/api_server/ChatService.cc (171:176) duplicated block id: 3581 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3582 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/bloom.py (119:125) - maga_transformer/models/starcoder.py (120:126) duplicated block id: 3583 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (479:484) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (738:743) duplicated block id: 3584 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (603:608) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (668:673) duplicated block id: 3585 size: 6 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (130:135) - bazel/tf_http_archive.bzl (281:286) duplicated block id: 3586 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3864:3869) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3875:3880) duplicated block id: 3587 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (128:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3588 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (17:22) duplicated block id: 3589 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:910) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (998:1003) duplicated block id: 3590 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (862:868) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (921:927) duplicated block id: 3591 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (128:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3592 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/utils/RpcErrorCode.h (41:46) - maga_transformer/cpp/utils/RpcErrorCode.h (62:67) duplicated block id: 3593 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (523:529) - maga_transformer/cpp/kernels/no_aux_tc_kernels.cu (668:674) duplicated block id: 3594 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (276:288) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (252:264) duplicated block id: 3595 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (310:317) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (26:33) duplicated block id: 3596 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (725:732) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3597 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.h (41:46) - maga_transformer/cpp/kernels/sampling_topk_kernels.h (62:67) duplicated block id: 3598 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (725:732) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) duplicated block id: 3599 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (725:732) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) duplicated block id: 3600 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (232:237) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (775:780) duplicated block id: 3601 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (233:238) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (247:252) duplicated block id: 3602 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3603 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (846:854) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (878:885) duplicated block id: 3604 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (11:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/34_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v4.cc (11:16) duplicated block id: 3605 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (200:205) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (236:241) duplicated block id: 3606 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (200:205) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (220:225) duplicated block id: 3607 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (597:604) - maga_transformer/cpp/kernels/_fma.h (623:630) duplicated block id: 3608 size: 6 cleaned lines of code in 2 files: - maga_transformer/config/gpt_init_model_parameters.py (625:630) - maga_transformer/openai/openai_endpoint.py (70:75) duplicated block id: 3609 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (285:290) - maga_transformer/cpp/kernels/_fma.h (480:485) duplicated block id: 3610 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3611 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) duplicated block id: 3612 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) duplicated block id: 3613 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (449:454) - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (487:492) duplicated block id: 3614 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_to_float.h (93:102) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2393:2402) duplicated block id: 3615 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) duplicated block id: 3616 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (196:201) - maga_transformer/models/qwen_vl.py (101:106) duplicated block id: 3617 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (343:348) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (68:73) duplicated block id: 3618 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (241:247) - maga_transformer/model_loader/static_fp8_quant_weight.py (298:304) duplicated block id: 3619 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1415:1421) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1673:1679) duplicated block id: 3620 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/comm_buffer/comm_buffer.cc (153:158) - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (180:185) duplicated block id: 3621 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (704:711) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) duplicated block id: 3622 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) duplicated block id: 3623 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (17:22) - maga_transformer/cpp/kernels/hello_world.cu (2:7) duplicated block id: 3624 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (704:711) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) duplicated block id: 3625 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (704:711) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3626 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3627 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (96:101) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (240:245) duplicated block id: 3628 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (296:301) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (836:841) duplicated block id: 3629 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (180:185) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (220:225) duplicated block id: 3630 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (180:185) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (236:241) duplicated block id: 3631 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (597:604) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:975) duplicated block id: 3632 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (264:269) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (477:482) duplicated block id: 3633 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (597:604) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) duplicated block id: 3634 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (638:643) - maga_transformer/cpp/rocm/hipblasMMWrapper.cc (195:200) duplicated block id: 3635 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3636 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (595:600) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (661:666) duplicated block id: 3637 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (725:732) - maga_transformer/cpp/kernels/_mul.h (775:782) duplicated block id: 3638 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (473:478) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (728:733) duplicated block id: 3639 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/hello_world.cu (2:7) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (24:29) duplicated block id: 3640 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (100:105) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (166:171) duplicated block id: 3641 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (597:604) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:909) duplicated block id: 3642 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (36:41) - maga_transformer/cpp/rocm/rocmFmhaWrapper.h (32:37) duplicated block id: 3643 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (187:192) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (207:212) duplicated block id: 3644 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:706) duplicated block id: 3645 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (84:89) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (126:131) duplicated block id: 3646 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (480:485) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (630:635) duplicated block id: 3647 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (88:94) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (201:207) duplicated block id: 3648 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (316:322) - maga_transformer/cpp/kernels/vec_dtypes.cuh (349:355) duplicated block id: 3649 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/group_wise_quant_weight.py (51:56) - maga_transformer/model_loader/omni_quant_weight.py (47:52) duplicated block id: 3650 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (810:815) - maga_transformer/models/qwen_v2_audio/modeling_qwen2_audio.py (463:468) duplicated block id: 3651 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (123:128) - maga_transformer/cpp/kernels/sampling_topk_kernels.cu (132:137) duplicated block id: 3652 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (31:37) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (74:80) duplicated block id: 3653 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (586:594) duplicated block id: 3654 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:43) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (94:104) duplicated block id: 3655 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (673:678) - maga_transformer/cpp/kernels/vec_dtypes.cuh (920:925) duplicated block id: 3656 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3657 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (158:163) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (65:70) duplicated block id: 3658 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/quantizePreprocessors.cc (143:150) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (463:470) duplicated block id: 3659 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (144:149) - maga_transformer/models/qwen_vl.py (101:106) duplicated block id: 3660 size: 6 cleaned lines of code in 2 files: - bazel/tf_http_archive.bzl (139:144) - bazel/tf_http_archive.bzl (291:296) duplicated block id: 3661 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (689:697) duplicated block id: 3662 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (366:372) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (465:471) duplicated block id: 3663 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (217:224) - maga_transformer/model_loader/smooth_quant_weight.py (214:221) duplicated block id: 3664 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/resampler.py (308:313) - maga_transformer/models/minicpmv/resampler.py (322:327) duplicated block id: 3665 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (508:514) - maga_transformer/cpp/cuda/cuda_utils.cc (546:552) duplicated block id: 3666 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (146:156) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (145:155) duplicated block id: 3667 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (273:278) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (288:293) duplicated block id: 3668 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaActOp.cc (82:88) - maga_transformer/cpp/devices/rocm_impl/ROCmActOp.cc (81:87) duplicated block id: 3669 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (146:156) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (58:68) duplicated block id: 3670 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (213:218) - maga_transformer/cpp/api_server/HttpApiServer.cc (347:352) duplicated block id: 3671 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (213:218) - maga_transformer/cpp/api_server/HttpApiServer.cc (366:371) duplicated block id: 3672 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (213:218) - maga_transformer/cpp/api_server/HttpApiServer.cc (328:333) duplicated block id: 3673 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (285:290) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (825:830) duplicated block id: 3674 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (180:185) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (200:205) duplicated block id: 3675 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (893:901) duplicated block id: 3676 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (789:797) duplicated block id: 3677 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (187:198) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (197:208) duplicated block id: 3678 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (69:75) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (106:112) duplicated block id: 3679 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (418:423) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (565:570) duplicated block id: 3680 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (998:1003) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (208:213) duplicated block id: 3681 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.cpp (78:83) - maga_transformer/cpp/trt_plugins/weightOnlyQuantMatmulPlugin/weightOnlyQuantMatmulPlugin.cpp (75:80) duplicated block id: 3682 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (92:97) - maga_transformer/models/qwen2_vl/qwen2_vl_vit.py (103:108) duplicated block id: 3683 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (173:178) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (81:86) duplicated block id: 3684 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (83:88) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (167:172) duplicated block id: 3685 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.h (159:165) - maga_transformer/cpp/rocm/rocmFmhaWrapper.h (9:15) duplicated block id: 3686 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (525:533) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (834:842) duplicated block id: 3687 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.h (100:105) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3325:3330) duplicated block id: 3688 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (693:698) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (780:785) duplicated block id: 3689 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/1_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v4.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (17:22) duplicated block id: 3690 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (214:219) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (173:178) duplicated block id: 3691 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (701:706) duplicated block id: 3692 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/openvino.py (93:98) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (28:33) duplicated block id: 3693 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (718:725) - maga_transformer/openai/renderers/conversation.py (825:832) duplicated block id: 3694 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (220:226) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (375:381) duplicated block id: 3695 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:376) duplicated block id: 3696 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (998:1003) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (52:57) duplicated block id: 3697 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (872:880) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1130:1138) duplicated block id: 3698 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (378:385) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (597:604) duplicated block id: 3699 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:425) duplicated block id: 3700 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:136) duplicated block id: 3701 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (747:752) - maga_transformer/openai/renderers/llama_template.py (767:772) duplicated block id: 3702 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (342:347) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (442:447) duplicated block id: 3703 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaDeepEPFfnLayer.cc (93:98) - maga_transformer/cpp/devices/cuda_impl/CudaDeepEPLLFfnLayer.cc (21:28) duplicated block id: 3704 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (200:205) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (256:261) duplicated block id: 3705 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (24:29) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (150:155) duplicated block id: 3706 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (171:176) - maga_transformer/cpp/cuda/cufmha/cufmha.cc (215:220) duplicated block id: 3707 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:529) duplicated block id: 3708 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (733:738) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (812:817) duplicated block id: 3709 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (50:56) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (119:125) duplicated block id: 3710 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3711 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:476) duplicated block id: 3712 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/models/GptModel.cc (468:473) - maga_transformer/cpp/models/GptModel.cc (1148:1153) duplicated block id: 3713 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (15:20) duplicated block id: 3714 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (105:110) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (148:153) duplicated block id: 3715 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (84:89) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (105:110) duplicated block id: 3716 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (105:110) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (126:131) duplicated block id: 3717 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/DeviceBase.cc (349:354) - maga_transformer/cpp/devices/cuda_impl/CudaOps.cc (245:250) duplicated block id: 3718 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (57:62) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (76:81) duplicated block id: 3719 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (471:476) duplicated block id: 3720 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3721 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (430:435) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (577:582) duplicated block id: 3722 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (84:89) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (148:153) duplicated block id: 3723 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (296:301) - maga_transformer/cpp/kernels/_fma.h (491:496) duplicated block id: 3724 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (524:529) duplicated block id: 3725 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (348:354) - maga_transformer/tokenizer/tokenization_chatglm3.py (263:269) duplicated block id: 3726 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:641) duplicated block id: 3727 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (605:612) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) duplicated block id: 3728 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (605:612) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 3729 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (525:533) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (681:689) duplicated block id: 3730 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/oai.py (107:112) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (47:52) duplicated block id: 3731 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1927:1934) duplicated block id: 3732 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (150:156) - maga_transformer/cpp/kernels/add_residual_kernels.cu (170:176) duplicated block id: 3733 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (29:34) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (81:86) duplicated block id: 3734 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (409:414) - maga_transformer/cpp/kernels/add_residual_kernels.cu (452:457) duplicated block id: 3735 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:584) duplicated block id: 3736 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (466:472) - maga_transformer/cpp/cuda/cuda_utils.cc (546:552) duplicated block id: 3737 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1065:1072) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1112:1119) duplicated block id: 3738 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (527:536) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (605:615) duplicated block id: 3739 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (579:584) duplicated block id: 3740 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_utils.cc (466:472) - maga_transformer/cpp/cuda/cuda_utils.cc (508:514) duplicated block id: 3741 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (605:612) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) duplicated block id: 3742 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/CacheStore.h (30:35) - maga_transformer/cpp/disaggregate/cache_store/NormalCacheStore.h (38:43) duplicated block id: 3743 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (636:641) duplicated block id: 3744 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (337:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3745 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (337:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3746 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3747 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (539:544) - maga_transformer/openai/renderers/llama_template.py (561:566) duplicated block id: 3748 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (753:758) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (828:833) duplicated block id: 3749 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/phi.py (19:24) - maga_transformer/models/whisper_weight.py (29:34) duplicated block id: 3750 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rmsnormKernels.h (49:54) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (61:66) duplicated block id: 3751 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (183:188) - maga_transformer/openai/renderers/qwen_agent/llm/oai.py (84:89) duplicated block id: 3752 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (50:55) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (69:74) duplicated block id: 3753 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (498:505) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (912:919) duplicated block id: 3754 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (347:352) - maga_transformer/cpp/api_server/HttpApiServer.cc (366:371) duplicated block id: 3755 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (65:70) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (171:176) duplicated block id: 3756 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (324:329) duplicated block id: 3757 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (323:328) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (257:262) duplicated block id: 3758 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (269:275) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_gemm_kernels_template.h (469:475) duplicated block id: 3759 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (414:420) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (560:566) duplicated block id: 3760 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/NormalCacheStore.cpp (207:212) - maga_transformer/cpp/disaggregate/cache_store/NormalCacheStore.h (39:44) duplicated block id: 3761 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 3762 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.cu (329:335) - maga_transformer/cpp/kernels/moe_topKSoftmax_kernels.cu (176:182) duplicated block id: 3763 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaDeepEPFfnLayer.cc (93:98) - maga_transformer/cpp/devices/cuda_impl/CudaFfnLayer.cc (55:60) duplicated block id: 3764 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (176:181) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (198:203) duplicated block id: 3765 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (779:784) - maga_transformer/openai/renderers/conversation.py (794:799) duplicated block id: 3766 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (346:351) - maga_transformer/openai/renderers/custom_renderer.py (615:620) duplicated block id: 3767 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:910) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (208:213) duplicated block id: 3768 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (905:910) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (196:201) duplicated block id: 3769 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (420:425) duplicated block id: 3770 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (623:630) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) duplicated block id: 3771 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (623:630) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) duplicated block id: 3772 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (279:284) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (371:376) duplicated block id: 3773 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (392:397) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (85:90) duplicated block id: 3774 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (392:397) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (66:71) duplicated block id: 3775 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3776 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (703:709) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (465:471) duplicated block id: 3777 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1243:1250) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1318:1325) duplicated block id: 3778 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 3779 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (59:66) - maga_transformer/openai/renderers/custom_renderer.py (111:118) duplicated block id: 3780 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (38:43) - maga_transformer/cpp/kernels/sampling_topp_kernels.h (158:163) duplicated block id: 3781 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (477:484) - maga_transformer/openai/renderers/conversation.py (1026:1035) duplicated block id: 3782 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (31:37) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (488:494) duplicated block id: 3783 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (70:75) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (551:556) duplicated block id: 3784 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (143:148) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (159:164) duplicated block id: 3785 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (551:556) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (233:238) duplicated block id: 3786 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rotary_position_embedding.h (220:225) - maga_transformer/cpp/kernels/rotary_position_embedding.h (244:249) duplicated block id: 3787 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen_v2.py (175:181) - maga_transformer/models/qwen_vl.py (101:106) duplicated block id: 3788 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (551:556) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (180:185) duplicated block id: 3789 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (564:569) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (1085:1090) duplicated block id: 3790 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (36:41) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (81:86) duplicated block id: 3791 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (36:41) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (29:34) duplicated block id: 3792 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/eplb/ExpertBalancer.cc (12:17) - maga_transformer/cpp/eplb/ExpertBalancer.h (26:31) duplicated block id: 3793 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (105:113) - maga_transformer/cpp/devices/arm_impl/ArmGemmOptOp.cc (35:44) duplicated block id: 3794 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (193:200) - maga_transformer/cpp/model_rpc/DecodeRpcServer.cc (227:234) duplicated block id: 3795 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (220:225) - maga_transformer/cpp/cutlass/cutlass_kernels/weightOnlyBatchedGemv/cudaCoreGemm.cu (236:241) duplicated block id: 3796 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (577:584) - maga_transformer/openai/renderers/conversation.py (587:594) duplicated block id: 3797 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (838:845) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (973:980) duplicated block id: 3798 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (11:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/34_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v4.cc (11:16) duplicated block id: 3799 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3800 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (94:100) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.h (119:125) duplicated block id: 3801 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:20) duplicated block id: 3802 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (769:774) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3803 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (769:774) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3804 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3805 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (229:236) - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (259:266) duplicated block id: 3806 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (769:774) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3807 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (704:711) - maga_transformer/cpp/kernels/_mul.h (775:782) duplicated block id: 3808 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 3809 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (288:293) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (474:480) duplicated block id: 3810 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (389:396) - maga_transformer/models/llava_vit.py (1008:1015) duplicated block id: 3811 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:161) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (203:208) duplicated block id: 3812 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3813 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (577:584) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (902:909) duplicated block id: 3814 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cosyvoice_qwen.py (17:23) - maga_transformer/models/qwen2_vl/qwen2_vl.py (102:107) duplicated block id: 3815 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (577:584) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) duplicated block id: 3816 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (577:584) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:975) duplicated block id: 3817 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (704:711) - maga_transformer/cpp/kernels/_mul.h (725:732) duplicated block id: 3818 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (114:119) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (185:190) duplicated block id: 3819 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (209:214) - maga_transformer/cpp/kernels/_fma.h (407:412) duplicated block id: 3820 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (114:119) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (134:139) duplicated block id: 3821 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fpA_intB_gemm_dummy_stubs.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 3822 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (114:119) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (152:157) duplicated block id: 3823 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/_mul.h (725:732) duplicated block id: 3824 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/core/TrackerAllocator.cc (70:77) - maga_transformer/cpp/core/TrackerAllocator.cc (98:105) duplicated block id: 3825 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/_mul.h (704:711) duplicated block id: 3826 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3245:3250) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3875:3880) duplicated block id: 3827 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3828 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (161:166) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (438:443) duplicated block id: 3829 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (288:293) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:136) duplicated block id: 3830 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (630:635) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (825:830) duplicated block id: 3831 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (209:214) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (752:757) duplicated block id: 3832 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (71:76) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (209:214) duplicated block id: 3833 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (385:390) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (221:226) duplicated block id: 3834 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (429:434) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (66:71) duplicated block id: 3835 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (174:179) - maga_transformer/cpp/kernels/sampling_topk_kernels.h (90:95) duplicated block id: 3836 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (678:684) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (731:737) duplicated block id: 3837 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (134:141) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (344:351) duplicated block id: 3838 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (301:306) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (29:34) duplicated block id: 3839 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (412:417) - maga_transformer/cpp/kernels/sampling_penalty_kernels.cu (477:482) duplicated block id: 3840 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (45:52) - maga_transformer/models/starcoder.py (41:48) duplicated block id: 3841 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/DeepGemmPlugin.cpp (301:306) - maga_transformer/cpp/deep_gemm/deep_gemm_template.h (81:86) duplicated block id: 3842 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (498:505) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (738:745) duplicated block id: 3843 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (93:103) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (57:67) duplicated block id: 3844 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (683:690) - maga_transformer/cpp/kernels/_mul.h (775:782) duplicated block id: 3845 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (93:103) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_multistage.h (158:168) duplicated block id: 3846 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/cogvlm2_render.py (50:55) - maga_transformer/openai/renderers/cogvlm2_render.py (61:66) duplicated block id: 3847 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (174:179) - maga_transformer/cpp/kernels/sampling_topp_kernels.h (158:163) duplicated block id: 3848 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (367:376) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (547:555) duplicated block id: 3849 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (85:90) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (264:269) duplicated block id: 3850 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (310:315) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (496:501) duplicated block id: 3851 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:136) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:161) duplicated block id: 3852 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1123:1128) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1142:1147) duplicated block id: 3853 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (217:224) - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (249:256) duplicated block id: 3854 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (131:136) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (223:228) duplicated block id: 3855 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1123:1128) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1180:1185) duplicated block id: 3856 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/SoftmaxOpTest.hpp (21:26) - maga_transformer/cpp/devices/base_tests/SoftmaxOpTest.hpp (69:74) duplicated block id: 3857 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (118:124) - maga_transformer/cpp/devices/arm_impl/ArmSoftmaxOp.cc (244:250) duplicated block id: 3858 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl.py (67:73) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (306:311) duplicated block id: 3859 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (44:49) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (85:90) duplicated block id: 3860 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmDevice.h (17:23) - maga_transformer/cpp/devices/cpu_impl/CpuDevice.h (12:18) duplicated block id: 3861 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (685:692) - maga_transformer/openai/renderers/conversation.py (932:939) duplicated block id: 3862 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (567:574) - maga_transformer/openai/renderers/conversation.py (857:864) duplicated block id: 3863 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/activation_macro.h (26:31) - maga_transformer/cpp/devices/arm_impl/gemm_opt/activation_macro.h (79:84) duplicated block id: 3864 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (233:243) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (33:43) duplicated block id: 3865 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (70:75) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (247:252) duplicated block id: 3866 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (1438:1443) - maga_transformer/cpp/kernels/gpt_kernels.cu (1457:1462) duplicated block id: 3867 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/oai.py (84:89) - maga_transformer/openai/renderers/qwen_agent/llm/openvino.py (93:98) duplicated block id: 3868 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (589:598) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (551:560) duplicated block id: 3869 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (78:84) - maga_transformer/tokenizer/tokenization_chatglm3.py (122:128) duplicated block id: 3870 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/moe_cutlass_kernel.h (162:167) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/splitk_gemm_grouped.h (156:161) duplicated block id: 3871 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/gpt_kernels.cu (1633:1638) - maga_transformer/cpp/kernels/gpt_kernels.cu (1688:1693) duplicated block id: 3872 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (532:537) - maga_transformer/cpp/kernels/add_residual_kernels.cu (543:548) duplicated block id: 3873 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (334:339) - maga_transformer/openai/renderers/qwen_renderer.py (350:355) duplicated block id: 3874 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (19:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/37_int4_dequant_gemm_256x16x64x256_32_16x16_1x1_32x8x1_8x32x1_32_1x16x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 3875 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_to_float.h (50:59) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2436:2445) duplicated block id: 3876 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.h (67:72) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.h (61:66) duplicated block id: 3877 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 3878 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (38:43) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (174:179) duplicated block id: 3879 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cosyvoice_qwen.py (17:23) - maga_transformer/models/qwen_v2.py (175:181) duplicated block id: 3880 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2_weight.py (104:109) - maga_transformer/models/cogvlm2_weight.py (114:119) duplicated block id: 3881 size: 6 cleaned lines of code in 2 files: - maga_transformer/utils/model_weight.py (48:53) - maga_transformer/utils/model_weight.py (60:65) duplicated block id: 3882 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/llava_vit.py (178:183) - maga_transformer/models/llava_vit.py (246:251) duplicated block id: 3883 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/downstream_modules/embedding/minicpmv_embedding_module.py (202:207) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (198:203) duplicated block id: 3884 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (429:434) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (264:269) duplicated block id: 3885 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1165:1172) - maga_transformer/openai/renderers/conversation.py (1182:1191) duplicated block id: 3886 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (279:284) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (230:235) duplicated block id: 3887 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (429:435) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (1135:1141) duplicated block id: 3888 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (279:284) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (163:168) duplicated block id: 3889 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (586:594) duplicated block id: 3890 size: 6 cleaned lines of code in 2 files: - bazel/bundle.bzl (468:473) - bazel/bundle.bzl (501:507) duplicated block id: 3891 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (303:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3892 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (303:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3893 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmKernel.h (132:137) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmKernel.h (146:151) duplicated block id: 3894 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (303:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3895 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (443:448) - maga_transformer/cpp/devices/cuda_impl/DeepEPBuffer.cc (564:570) duplicated block id: 3896 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 3897 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (220:225) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (763:768) duplicated block id: 3898 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_logn_attention.h (43:48) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2152:2157) duplicated block id: 3899 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (689:697) duplicated block id: 3900 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen2_vl/qwen2_vl.py (102:107) - maga_transformer/models/qwen_vl.py (101:106) duplicated block id: 3901 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/deepseek_v2.py (196:201) - maga_transformer/models/qwen2_vl/qwen2_vl.py (102:107) duplicated block id: 3902 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2393:2402) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2436:2445) duplicated block id: 3903 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/omni_quant_weight.py (217:224) - maga_transformer/model_loader/static_fp8_quant_weight.py (254:261) duplicated block id: 3904 size: 6 cleaned lines of code in 2 files: - maga_transformer/server/backend_app.py (81:87) - maga_transformer/server/frontend_app.py (86:92) duplicated block id: 3905 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (220:225) - maga_transformer/cpp/kernels/_fma.h (418:423) duplicated block id: 3906 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (15:20) duplicated block id: 3907 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/internvl_renderer.py (62:67) - maga_transformer/openai/renderers/llava_renderer.py (94:99) duplicated block id: 3908 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (348:353) - maga_transformer/cpp/cuda/cufmha/cufmha.h (85:90) duplicated block id: 3909 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/eva2clip_vit.py (85:90) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (622:627) duplicated block id: 3910 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/splitk_gemm_grouped.h (130:137) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/splitk_gemm_grouped.h (210:217) duplicated block id: 3911 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (423:428) - maga_transformer/cpp/kernels/activation_kernels.cu (441:446) duplicated block id: 3912 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (789:797) duplicated block id: 3913 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3914 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (993:998) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 3915 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm3.py (240:261) - maga_transformer/tokenizer/tokenization_chatglm4.py (138:159) duplicated block id: 3916 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (993:998) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 3917 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.cc (231:236) - maga_transformer/cpp/cuda/cufmha/cufmha.cc (257:262) duplicated block id: 3918 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (993:998) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 3919 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (323:328) - maga_transformer/openai/renderers/qwen_agent/llm/function_calling.py (335:340) duplicated block id: 3920 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (743:748) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (820:825) duplicated block id: 3921 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (287:292) - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (306:311) duplicated block id: 3922 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (713:718) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (796:801) duplicated block id: 3923 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (491:496) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (641:646) duplicated block id: 3924 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/bloom.py (119:125) - maga_transformer/models/starcoder2.py (149:155) duplicated block id: 3925 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (893:901) duplicated block id: 3926 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (765:770) - maga_transformer/openai/renderers/custom_renderer.py (783:788) duplicated block id: 3927 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (241:247) - maga_transformer/model_loader/static_fp8_quant_weight.py (124:130) duplicated block id: 3928 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (765:770) - maga_transformer/openai/renderers/custom_renderer.py (795:800) duplicated block id: 3929 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1087:1092) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1180:1185) duplicated block id: 3930 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1104:1109) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1163:1168) duplicated block id: 3931 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1104:1109) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1123:1128) duplicated block id: 3932 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (244:249) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) duplicated block id: 3933 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1045:1052) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1065:1072) duplicated block id: 3934 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2136:2141) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2152:2157) duplicated block id: 3935 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/0_int4_dequant_gemm_256x128x128x128_32_32x32_2x2_16x16x1_4x64x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/36_int4_dequant_gemm_256x128x128x64_32_32x32_4x1_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 3936 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (53:58) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (120:125) duplicated block id: 3937 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (428:433) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.inl (971:976) duplicated block id: 3938 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaPrefillAttention.cc (113:118) - maga_transformer/cpp/devices/cuda_impl/CudaPrefillAttention.cc (150:155) duplicated block id: 3939 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/FfnLayer.cc (59:64) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (167:172) duplicated block id: 3940 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (279:284) - maga_transformer/cpp/kernels/rocm/layernorm_kernels.cu (555:560) duplicated block id: 3941 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (204:209) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (865:870) duplicated block id: 3942 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (211:216) - maga_transformer/model_loader/smooth_quant_weight.py (165:170) duplicated block id: 3943 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1087:1092) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1142:1147) duplicated block id: 3944 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (211:216) - maga_transformer/model_loader/smooth_quant_weight.py (203:208) duplicated block id: 3945 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (11:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (11:16) duplicated block id: 3946 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (563:568) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (633:638) duplicated block id: 3947 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1087:1092) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1104:1109) duplicated block id: 3948 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (143:148) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_launch.h (198:203) duplicated block id: 3949 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (462:467) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (515:520) duplicated block id: 3950 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (250:255) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (288:293) duplicated block id: 3951 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/9_int4_dequant_gemm_128x128x32x128_32_32x32_2x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) duplicated block id: 3952 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 3953 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (998:1003) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (36:41) duplicated block id: 3954 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (340:345) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (264:269) duplicated block id: 3955 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/vec_dtypes.cuh (252:257) - maga_transformer/cpp/kernels/vec_dtypes.cuh (277:282) duplicated block id: 3956 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (221:226) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (257:262) duplicated block id: 3957 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (407:412) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (554:559) duplicated block id: 3958 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (442:448) - maga_transformer/openai/renderers/conversation.py (1026:1035) duplicated block id: 3959 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (190:195) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (237:242) duplicated block id: 3960 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/trt_plugins/weightOnlyGroupwiseQuantMatmulPlugin/weightOnlyGroupwiseQuantMatmulPlugin.h (49:54) - maga_transformer/cpp/trt_plugins/weightOnlyQuantMatmulPlugin/weightOnlyQuantMatmulPlugin.h (69:74) duplicated block id: 3961 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (555:560) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (626:631) duplicated block id: 3962 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (384:390) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (488:494) duplicated block id: 3963 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:564) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) duplicated block id: 3964 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:564) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) duplicated block id: 3965 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:560) duplicated block id: 3966 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (195:200) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (248:253) duplicated block id: 3967 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (587:592) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (654:659) duplicated block id: 3968 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tokenization_qwen.py (199:204) - maga_transformer/tokenizer/tokenization_qwen.py (262:267) duplicated block id: 3969 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (467:472) - maga_transformer/cpp/kernels/rmsnormKernels.cu (220:225) duplicated block id: 3970 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (508:513) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (532:537) duplicated block id: 3971 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen_v2_moe.py (16:22) - maga_transformer/models/qwen_v3_moe.py (13:19) duplicated block id: 3972 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/resampler.py (164:171) - maga_transformer/models/minicpmv_embedding/resampler.py (165:172) duplicated block id: 3973 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (214:221) - maga_transformer/model_loader/static_fp8_quant_weight.py (254:261) duplicated block id: 3974 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (461:466) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (708:713) duplicated block id: 3975 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:454) duplicated block id: 3976 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (400:405) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (446:451) duplicated block id: 3977 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (23:28) - maga_transformer/tokenizer/tokenization_chatglm3.py (29:34) duplicated block id: 3978 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cogvlm2_weight.py (42:47) - maga_transformer/models/cogvlm2_weight.py (53:58) duplicated block id: 3979 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:507) duplicated block id: 3980 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 3981 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (258:263) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (422:427) duplicated block id: 3982 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (372:380) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (834:842) duplicated block id: 3983 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (467:472) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (718:723) duplicated block id: 3984 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_logn_attention.h (59:64) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (2136:2141) duplicated block id: 3985 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (372:380) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmThreadblock.cc (681:689) duplicated block id: 3986 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (35:40) - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (55:60) duplicated block id: 3987 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (144:149) - maga_transformer/models/qwen_v2.py (175:181) duplicated block id: 3988 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/CacheStore.h (21:26) - maga_transformer/cpp/disaggregate/cache_store/NormalCacheStore.h (29:34) duplicated block id: 3989 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (343:348) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (342:347) duplicated block id: 3990 size: 6 cleaned lines of code in 2 files: - maga_transformer/access_logger/access_logger.py (18:23) - maga_transformer/access_logger/access_logger.py (28:33) duplicated block id: 3991 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/0_int4_dequant_gemm_256x128x128x128_32_32x32_2x2_16x16x1_4x64x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/35_int4_dequant_gemm_256x128x128x64_32_32x32_2x2_8x32x1_2x128x1_32_1x32x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 3992 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (488:493) - maga_transformer/cpp/devices/cuda_impl/CudaFlashInfer.cc (510:515) duplicated block id: 3993 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:405) duplicated block id: 3994 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (215:220) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (286:291) duplicated block id: 3995 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/static_fp8_quant_weight.py (124:130) - maga_transformer/model_loader/static_fp8_quant_weight.py (298:304) duplicated block id: 3996 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaQuantizeOp.cc (112:117) - maga_transformer/cpp/devices/cuda_impl/CudaQuantizeOp.cc (128:133) duplicated block id: 3997 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (175:185) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (290:300) duplicated block id: 3998 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (175:185) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (393:403) duplicated block id: 3999 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (93:103) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (58:68) duplicated block id: 4000 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (173:178) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (240:245) duplicated block id: 4001 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/default_splitk_gemm_grouped.h (93:103) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_dq_mma_pipelined.h (145:155) duplicated block id: 4002 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) duplicated block id: 4003 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/30_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (17:22) duplicated block id: 4004 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (783:788) - maga_transformer/openai/renderers/custom_renderer.py (795:800) duplicated block id: 4005 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_logn_attention.h (43:48) - maga_transformer/cpp/kernels/_logn_attention.h (59:64) duplicated block id: 4006 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (203:208) duplicated block id: 4007 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (223:228) duplicated block id: 4008 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/modeling_navit_siglip.py (823:828) - maga_transformer/models/qwen_v2_audio/modeling_qwen2_audio.py (463:468) duplicated block id: 4009 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/group_wise_quant_weight.py (51:56) - maga_transformer/model_loader/smooth_quant_weight.py (106:111) duplicated block id: 4010 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:674) duplicated block id: 4011 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (208:213) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (36:41) duplicated block id: 4012 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_type_utils.cuh (40:45) - maga_transformer/cpp/cuda/cuda_type_utils.cuh (49:54) duplicated block id: 4013 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (49:54) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (165:170) duplicated block id: 4014 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (274:282) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (47:55) duplicated block id: 4015 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1012:1017) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (52:57) duplicated block id: 4016 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/AttentionLayer.cc (95:100) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (209:214) duplicated block id: 4017 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4018 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (64:69) - maga_transformer/cpp/cuda/cuda_fp8_utils.cu (86:91) duplicated block id: 4019 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/eva2clip_vit.py (85:90) - maga_transformer/models/llava_vit.py (668:673) duplicated block id: 4020 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (306:311) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:617) duplicated block id: 4021 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (126:131) - maga_transformer/cpp/cuda/cuda_bf16_fallbacks.cuh (148:153) duplicated block id: 4022 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm2.py (148:169) - maga_transformer/tokenizer/tokenization_chatglm4.py (138:159) duplicated block id: 4023 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1142:1147) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (1163:1168) duplicated block id: 4024 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 4025 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (172:183) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (331:340) duplicated block id: 4026 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (190:200) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (482:487) duplicated block id: 4027 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmGemmKaiOp.cc (114:121) - maga_transformer/cpp/devices/arm_impl/ArmGemmOp.cc (26:33) duplicated block id: 4028 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (462:467) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (687:692) duplicated block id: 4029 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl.py (60:65) - maga_transformer/models/minicpmv/minicpmv.py (243:248) duplicated block id: 4030 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (1012:1017) - maga_transformer/cpp/rocm/hipblasMMWrapper.h (36:41) duplicated block id: 4031 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (671:676) - maga_transformer/openai/renderers/qwen_renderer.py (485:490) duplicated block id: 4032 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (69:74) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (258:263) duplicated block id: 4033 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/group_wise_quant_weight.py (207:212) - maga_transformer/model_loader/per_block_fp8_quant_weight.py (83:89) duplicated block id: 4034 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (942:947) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 4035 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 4036 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (942:947) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 4037 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (942:947) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 4038 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (328:333) - maga_transformer/cpp/api_server/HttpApiServer.cc (366:371) duplicated block id: 4039 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/disaggregate/cache_store/LoadContext.cpp (124:129) - maga_transformer/cpp/disaggregate/cache_store/LoadContext.h (55:60) duplicated block id: 4040 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/HttpApiServer.cc (328:333) - maga_transformer/cpp/api_server/HttpApiServer.cc (347:352) duplicated block id: 4041 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (183:188) - maga_transformer/openai/renderers/qwen_agent/llm/qwenvl_dashscope.py (26:31) duplicated block id: 4042 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (462:467) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (569:574) duplicated block id: 4043 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl_weight.py (79:85) - maga_transformer/models/whisper_weight.py (29:34) duplicated block id: 4044 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_add.h (189:194) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (303:308) duplicated block id: 4045 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm3.py (96:102) - maga_transformer/tokenizer/tokenization_chatglm4.py (12:18) duplicated block id: 4046 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (462:467) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (626:631) duplicated block id: 4047 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/normal_engine/NormalEngine.cc (151:156) - maga_transformer/cpp/speculative_engine/SpeculativeEngine.cc (52:58) duplicated block id: 4048 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (254:259) - maga_transformer/cpp/cutlass/cutlass_kernels/moe_gemm/moe_kernels.h (408:413) duplicated block id: 4049 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/minicpmv.py (257:263) - maga_transformer/models/minicpmv_embedding/minicpmv_embedding.py (317:323) duplicated block id: 4050 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (66:71) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (264:269) duplicated block id: 4051 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (1771:1780) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention_template.h (1890:1899) duplicated block id: 4052 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (515:520) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (569:574) duplicated block id: 4053 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (183:188) - maga_transformer/openai/renderers/qwen_agent/llm/openvino.py (93:98) duplicated block id: 4054 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (423:428) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (247:252) duplicated block id: 4055 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (193:201) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (405:414) duplicated block id: 4056 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 4057 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (515:520) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (626:631) duplicated block id: 4058 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cosyvoice_qwen.py (17:23) - maga_transformer/models/deepseek_v2.py (196:201) duplicated block id: 4059 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1891:1896) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (1921:1926) duplicated block id: 4060 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (515:520) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (687:692) duplicated block id: 4061 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (42:47) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention/decoder_masked_multihead_attention.cu (64:69) duplicated block id: 4062 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_finegrained.h (187:198) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (158:169) duplicated block id: 4063 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:228) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (272:277) duplicated block id: 4064 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 4065 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:228) - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (288:293) duplicated block id: 4066 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4067 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (249:256) - maga_transformer/model_loader/static_fp8_quant_weight.py (254:261) duplicated block id: 4068 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (382:387) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (508:513) duplicated block id: 4069 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (603:608) - maga_transformer/openai/renderers/qwen_renderer.py (444:449) duplicated block id: 4070 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_impl/LoraLinear.cc (96:101) - maga_transformer/cpp/devices/cuda_impl/CudaLoraLinear.cc (81:86) duplicated block id: 4071 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2030:2035) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (2081:2086) duplicated block id: 4072 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (75:80) - maga_transformer/openai/renderers/conversation.py (231:236) duplicated block id: 4073 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined.h (346:364) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (355:373) duplicated block id: 4074 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:154) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (515:523) duplicated block id: 4075 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (556:563) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 4076 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (556:563) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) duplicated block id: 4077 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (468:474) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (532:538) duplicated block id: 4078 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/activation_kernels.cu (74:79) - maga_transformer/cpp/kernels/activation_kernels.cu (90:95) duplicated block id: 4079 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (388:395) - maga_transformer/cpp/devices/arm_impl/ArmAttentionOp.cc (610:618) duplicated block id: 4080 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (556:563) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) duplicated block id: 4081 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1416:1421) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1439:1444) duplicated block id: 4082 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl_weight.py (79:85) - maga_transformer/models/phi.py (19:24) duplicated block id: 4083 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (34:39) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (52:57) duplicated block id: 4084 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (959:964) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 4085 size: 6 cleaned lines of code in 2 files: - maga_transformer/utils/smooth_quant_convert/llama/smoothquant.py (187:192) - maga_transformer/utils/smooth_quant_convert/qwen/smoothquant.py (190:195) duplicated block id: 4086 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (729:736) - maga_transformer/openai/renderers/conversation.py (825:832) duplicated block id: 4087 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (720:727) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1112:1119) duplicated block id: 4088 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (959:964) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 4089 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4090 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (155:161) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (224:230) duplicated block id: 4091 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (959:964) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 4092 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (468:474) - maga_transformer/cpp/kernels/layernorm_fp8_kernels.cu (606:612) duplicated block id: 4093 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (720:727) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1045:1052) duplicated block id: 4094 size: 6 cleaned lines of code in 2 files: - maga_transformer/server/frontend_worker.py (156:161) - maga_transformer/server/frontend_worker.py (181:187) duplicated block id: 4095 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (19:25) - maga_transformer/cpp/rocm/int4_gemm_kernels/37_int4_dequant_gemm_256x16x64x256_32_16x16_1x1_32x8x1_8x32x1_32_1x16x1x8_8_intrawave_v3.cc (19:25) duplicated block id: 4096 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 4097 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/starcoder.py (120:126) - maga_transformer/models/starcoder2.py (149:155) duplicated block id: 4098 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:564) - maga_transformer/cpp/kernels/_fma.h (577:584) duplicated block id: 4099 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (24:29) - maga_transformer/cpp/devices/base_tests/LoraLinearLayerTest.hpp (136:141) duplicated block id: 4100 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/cutlass_preprocessors.cc (455:465) - maga_transformer/cpp/rocm/quantizePreprocessors.cc (442:452) duplicated block id: 4101 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (21:26) - maga_transformer/model_loader/smooth_quant_weight.py (27:32) duplicated block id: 4102 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1010:1015) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (77:82) duplicated block id: 4103 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/gpt_neox_weight.py (20:26) - maga_transformer/models/gpt_weight.py (17:23) duplicated block id: 4104 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (557:564) - maga_transformer/cpp/kernels/_fma.h (597:604) duplicated block id: 4105 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/19_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4106 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/phi.py (19:24) - maga_transformer/models/starcoder2.py (42:48) duplicated block id: 4107 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:975) duplicated block id: 4108 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/_mul.h (605:612) duplicated block id: 4109 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/_mul.h (556:563) duplicated block id: 4110 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:228) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (156:161) duplicated block id: 4111 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (15:20) duplicated block id: 4112 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4113 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaLayernorm.cc (223:228) - maga_transformer/cpp/devices/rocm_impl/ROCmLayernorm.cc (223:228) duplicated block id: 4114 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:154) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (439:447) duplicated block id: 4115 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (166:171) - maga_transformer/cpp/devices/base_tests/LayerNormTest.hpp (185:190) duplicated block id: 4116 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:154) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (320:329) duplicated block id: 4117 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4118 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (146:154) - maga_transformer/cpp/devices/arm_impl/ArmLayerNormOp.cc (381:389) duplicated block id: 4119 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/mpt.py (21:26) - maga_transformer/models/starcoder2.py (42:48) duplicated block id: 4120 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/smooth_quant_weight.py (360:368) - maga_transformer/model_loader/static_fp8_quant_weight.py (293:299) duplicated block id: 4121 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (290:300) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/interleaved_numeric_conversion.h (393:403) duplicated block id: 4122 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (303:308) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (438:443) duplicated block id: 4123 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1779:1786) duplicated block id: 4124 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) duplicated block id: 4125 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) duplicated block id: 4126 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/testing/TestBase.h (208:213) - maga_transformer/cpp/devices/testing/TestBase.h (225:230) duplicated block id: 4127 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_vector_abs_max.h (5:18) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (165:178) duplicated block id: 4128 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 4129 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (723:728) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (804:809) duplicated block id: 4130 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (190:195) - maga_transformer/model_loader/smooth_quant_weight.py (332:337) duplicated block id: 4131 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (605:612) - maga_transformer/cpp/kernels/_mul.h (627:634) duplicated block id: 4132 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/minicpmv/modeling_navit_siglip.py (785:809) - maga_transformer/models/minicpmv/modeling_navit_siglip.py (882:889) duplicated block id: 4133 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/qwen.py (106:113) - maga_transformer/models/starcoder.py (41:48) duplicated block id: 4134 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (535:542) - maga_transformer/cpp/kernels/_mul.h (627:634) duplicated block id: 4135 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (327:332) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.cc (352:357) duplicated block id: 4136 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (1497:1503) - maga_transformer/openai/renderers/conversation.py (1508:1514) duplicated block id: 4137 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (404:410) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (504:510) duplicated block id: 4138 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (308:313) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1416:1421) duplicated block id: 4139 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (340:345) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (429:434) duplicated block id: 4140 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4141 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (340:345) - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (392:397) duplicated block id: 4142 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (429:435) - maga_transformer/openai/renderers/conversation.py (1026:1035) duplicated block id: 4143 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (303:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:166) duplicated block id: 4144 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (303:308) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:147) duplicated block id: 4145 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/layernorm_kernels.cu (291:296) - maga_transformer/cpp/kernels/rmsnormKernels.cu (31:36) duplicated block id: 4146 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3245:3250) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (3256:3261) duplicated block id: 4147 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (144:149) - maga_transformer/cpp/kernels/sampling_topk_kernels.h (90:95) duplicated block id: 4148 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (485:490) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (748:753) duplicated block id: 4149 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_convert_to_float.h (50:59) - maga_transformer/cpp/kernels/_convert_to_float.h (93:102) duplicated block id: 4150 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (561:567) - maga_transformer/cpp/kernels/alpha_layernorm_kernels.cu (616:622) duplicated block id: 4151 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp32.cu (19:24) duplicated block id: 4152 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (627:634) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1757:1764) duplicated block id: 4153 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/banRepeatNgram.cu (17:22) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (24:29) duplicated block id: 4154 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (627:634) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1687:1694) duplicated block id: 4155 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/11_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/31_int4_dequant_gemm_128x16x128x128_32_16x16_1x4_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4156 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (627:634) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1708:1715) duplicated block id: 4157 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/add_residual_kernels.cu (323:330) - maga_transformer/cpp/kernels/add_residual_kernels.cu (440:447) duplicated block id: 4158 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/llm/base.py (183:188) - maga_transformer/openai/renderers/qwen_agent/llm/qwen_dashscope.py (28:33) duplicated block id: 4159 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (455:460) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (698:703) duplicated block id: 4160 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/23_int4_dequant_gemm_128x64x32x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4161 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:43) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (233:243) duplicated block id: 4162 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaAttentionOp.cc (182:187) - maga_transformer/cpp/devices/rocm_impl/ROCmAttentionOp.cc (310:315) duplicated block id: 4163 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (779:784) - maga_transformer/cpp/kernels/unfused_attention_fp8_kernels.cu (860:865) duplicated block id: 4164 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (17:22) - maga_transformer/cpp/rocm/int4_gemm_kernels/28_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (17:22) duplicated block id: 4165 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (128:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:166) duplicated block id: 4166 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int8_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 4167 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (128:133) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:147) duplicated block id: 4168 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (577:582) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (775:780) duplicated block id: 4169 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (942:949) duplicated block id: 4170 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (922:929) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (968:975) duplicated block id: 4171 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/20_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/29_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v4.cc (15:20) duplicated block id: 4172 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/22_int4_dequant_gemm_256x32x256x128_32_32x32_1x2_16x16x1_4x64x1_32_1x16x1x16_8_intrawave_v3.cc (11:16) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (11:16) duplicated block id: 4173 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (669:674) duplicated block id: 4174 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_multistage_percol.h (188:199) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/dq_mma_pipelined_finegrained.h (158:169) duplicated block id: 4175 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/weight_module.py (224:229) - maga_transformer/model_loader/weight_module.py (421:426) duplicated block id: 4176 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (612:617) duplicated block id: 4177 size: 6 cleaned lines of code in 2 files: - maga_transformer/tools/convert/weights_convert.py (42:47) - maga_transformer/tools/quant/weights_quant.py (86:92) duplicated block id: 4178 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4179 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (716:723) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (762:768) duplicated block id: 4180 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (716:723) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (795:801) duplicated block id: 4181 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/mpt.py (21:26) - maga_transformer/models/whisper_weight.py (29:34) duplicated block id: 4182 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1010:1015) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (94:99) duplicated block id: 4183 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/starcoder2.py (42:48) - maga_transformer/models/whisper_weight.py (29:34) duplicated block id: 4184 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (1010:1015) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (110:115) duplicated block id: 4185 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/conversation.py (84:89) - maga_transformer/openai/renderers/conversation.py (251:256) duplicated block id: 4186 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/llama_template.py (32:37) - maga_transformer/openai/renderers/llama_template.py (52:57) duplicated block id: 4187 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cufmha/cufmha.h (62:67) - maga_transformer/cpp/cuda/cufmha/cufmha.h (77:82) duplicated block id: 4188 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (555:560) duplicated block id: 4189 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (502:507) duplicated block id: 4190 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (554:559) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (752:757) duplicated block id: 4191 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_bf16.cu (19:24) duplicated block id: 4192 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (775:782) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1877:1884) duplicated block id: 4193 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (775:782) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1835:1842) duplicated block id: 4194 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_mul.h (775:782) - maga_transformer/cpp/kernels/decoder_masked_multihead_attention_utils.h (1856:1863) duplicated block id: 4195 size: 6 cleaned lines of code in 2 files: - maga_transformer/tokenizer/tokenization_chatglm.py (348:354) - maga_transformer/tokenizer/tokenization_chatglm2.py (171:177) duplicated block id: 4196 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl_weight.py (79:85) - maga_transformer/models/mpt.py (21:26) duplicated block id: 4197 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/custom_renderer.py (514:519) - maga_transformer/openai/renderers/qwen_agent_renderer.py (130:135) duplicated block id: 4198 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma.h (33:43) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/threadblock/default_mma_bf16.h (94:104) duplicated block id: 4199 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (155:161) - maga_transformer/cpp/devices/arm_impl/gemm_opt/ArmGemmPacking.cc (358:365) duplicated block id: 4200 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/16_int4_dequant_gemm_128x16x32x128_32_16x16_1x1_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/18_int4_dequant_gemm_128x32x64x128_32_32x32_1x1_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v3.cc (15:20) duplicated block id: 4201 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (337:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (161:166) duplicated block id: 4202 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.cc (337:342) - maga_transformer/cpp/cuda/cublas/cublasFP8MMWrapper.h (142:147) duplicated block id: 4203 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/rocm/int4_gemm_kernels/17_int4_dequant_gemm_128x16x64x128_32_16x16_1x2_16x8x1_4x32x1_32_1x16x1x8_4_intrawave_v3.cc (15:20) - maga_transformer/cpp/rocm/int4_gemm_kernels/32_int4_dequant_gemm_128x32x128x128_32_32x32_1x2_16x8x1_4x32x1_32_1x16x1x8_8_intrawave_v4.cc (15:20) duplicated block id: 4204 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/cosyvoice_qwen.py (17:23) - maga_transformer/models/qwen.py (144:149) duplicated block id: 4205 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (449:454) duplicated block id: 4206 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (577:584) - maga_transformer/cpp/kernels/_fma.h (597:604) duplicated block id: 4207 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (577:584) - maga_transformer/cpp/kernels/_fma.h (623:630) duplicated block id: 4208 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int8_gemm_fg_scalebias.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24) duplicated block id: 4209 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (351:356) - maga_transformer/cpp/deep_gemm/include/mma_utils.cuh (400:405) duplicated block id: 4210 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (205:210) - maga_transformer/cpp/devices/cuda_impl/CudaGemmOp.cc (223:228) duplicated block id: 4211 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/utils/DebugUtils.cc (71:76) - maga_transformer/cpp/devices/utils/DebugUtils.cc (113:118) duplicated block id: 4212 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/custom_ar/custom_ar_comm.cc (229:240) - maga_transformer/cpp/devices/rocm_impl/custom_ar_comm.cc (191:200) duplicated block id: 4213 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/ArmSampleOp.cc (38:43) - maga_transformer/cpp/devices/cpu_impl/CpuSampleOp.cc (144:149) duplicated block id: 4214 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (753:759) - maga_transformer/cpp/kernels/unfused_attention_kernels.cu (786:792) duplicated block id: 4215 size: 6 cleaned lines of code in 2 files: - maga_transformer/models/internvl_weight.py (79:85) - maga_transformer/models/starcoder2.py (42:48) duplicated block id: 4216 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/cuda_impl/CudaSampleOp.cc (340:345) - maga_transformer/cpp/devices/rocm_impl/ROCmSampleOp.cc (85:90) duplicated block id: 4217 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (951:959) - maga_transformer/cpp/devices/arm_impl/gemm_opt/gemm_microkernel_macro_m8_bf16.h (1130:1138) duplicated block id: 4218 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (156:161) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (179:184) duplicated block id: 4219 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (156:161) - maga_transformer/cpp/cuda/cublas/cublasMMWrapper.h (232:237) duplicated block id: 4220 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/api_server/AccessLogWrapper.cc (69:74) - maga_transformer/cpp/api_server/AccessLogWrapper.cc (94:99) duplicated block id: 4221 size: 6 cleaned lines of code in 2 files: - maga_transformer/model_loader/per_tensor_int8_quant_weight.py (249:256) - maga_transformer/model_loader/smooth_quant_weight.py (214:221) duplicated block id: 4222 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/kernels/_fma.h (232:237) - maga_transformer/cpp/kernels/_fma.h (430:435) duplicated block id: 4223 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (81:88) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/fused_moe_kernel_routine.cuh (486:493) duplicated block id: 4224 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/fp16_int4_gemm_per_col.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_int32.cu (19:24) duplicated block id: 4225 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/device/gemm_universal_base_compat.h (123:133) - maga_transformer/cpp/cutlass/cutlass_extensions/include/cutlass_extensions/gemm/kernel/gemm_with_epilogue_visitor.h (283:293) duplicated block id: 4226 size: 6 cleaned lines of code in 2 files: - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (112:117) - maga_transformer/openai/renderers/qwen_agent/utils/tool_function_converter/request_converter.py (166:171) duplicated block id: 4227 size: 6 cleaned lines of code in 2 files: - maga_transformer/cpp/cutlass/cutlass_kernels/fpA_intB_gemm/bf16_int4_gemm_fg_scaleonly.cu (19:24) - maga_transformer/cpp/cutlass/cutlass_kernels/int8_gemm/int8_gemm_fp16.cu (19:24)