• There are 81 files with units, with 97,743 lines of code.
• 33 very complex files (McCabe index sum 251+; 65,133 lines of code)
• 25 complex files (McCabe index sum 126-250; 24,547 lines of code)
• 10 medium complex files (McCabe index sum 51-125; 6,472 lines of code)
• 3 slightly complex files (McCabe index sum 26-50; 472 lines of code)
• 10 simple files (McCabe index sum 1-25; 1,119 lines of code)
• 3 main files have no units (e.g. languages without unit analysis, or files without functions) and are not classified.

The McCabe index of a file grows with its size: per 100 lines tells a large but plain file from a dense one, and most complex unit whether the complexity sits in one unit or is spread over many.

File	# lines	# units	McCabe index (sum)	per 100 lines	most complex unit
train_gpt.py in records/track_10min_16mb/2026-04-27_SP8192_LQER_SparseGate_BOSSmearFix_9HpStack_1.0611	3349	115	658	20	63
train_gpt.py in records/track_10min_16mb/2026-04-29_SmearGateBOSFix_3Seed_1.06141	3204	110	626	20	63
train_gpt.py in records/track_10min_16mb/2026-04-27_SmearGateBOSFix_PR1787Base_LQERAsym_PhasedTTT	3185	110	626	20	63
train_gpt.py in results/pr1868_rerun_compliance	3204	110	626	20	63
train_gpt.py in records/track_10min_16mb/2026-04-23_SP8192_CaseOps_SparseGate_QuantGate_Loop45_PhasedTTT_PolarNS_MinLR_FusedCE	3116	108	616	20	63
train_gpt.py in records/track_10min_16mb/2026-04-13_VarLenAttn_PhasingTTT2000	3163	104	606	19	56
train_gpt.py in records/track_10min_16mb/2026-04-18_PR1626_CaseOps_Taper	2838	97	584	21	61
train_gpt.py in records/track_10min_16mb/2026-04-19_SP8192_CaseOps_GatedAttn_QuantGate_Loop45_PhasedTTT	2850	96	583	20	63
train_gpt.py in records/track_10min_16mb/2026-04-22_SP8192_CaseOps_GatedAttn_QuantGate_Loop45_PhasedTTT_MLPClip12	2850	96	583	20	63
train_gpt.py in records/track_10min_16mb/2026-04-14_MultiPhaseGlobalSGD_PhasedTTT	2775	96	568	20	60
train_gpt.py in records/track_10min_16mb/2026-04-16_SmearGate_AttentionOutputGate_ScoreFirstTTT	2739	91	549	20	55
train_gpt.py in records/track_10min_16mb/2026-04-10_VarLenAttn	2662	91	531	20	55
train_gpt.py in records/track_10min_16mb/2026-03-25_ValCalib_GPTQ_XSA_BigramHash3072	1997	69	442	22	143
train_mamba3_hybrid.py in records/track_non_record_16mb/2026-04-15_Mamba3Hybrid_SP8192_GPTQ_TTT	1890	69	394	21	184
train_gpt.py in records/track_10min_16mb/2026-03-29_Loader_FullGPTQ_XSA11_BigramHash2816	1907	64	383	20	109
train_gpt_cuda_binary.py in records/track_non_record_16mb/2026-03-24_106M_Binary_Asymmetric_UNet_FP8_15L_8192BPE_YaRN_NeoMuon_Smear	1233	58	357	29	125
train_gpt_cuda_ternary.py in records/track_10min_16mb/2026-03-24_74M_Ternary_UNet_FP8_10L_8192BPE_YaRN_NeoMuon	1231	61	355	29	114
train_gpt.py in records/track_10min_16mb/2026-03-23_LeakyReLU_LegalTTT_ParallelMuon	1766	54	354	20	125
train_gpt_cuda_xnor.py in records/track_non_record_16mb/2026-04-05_118M_XNOR-Net_FP8_1024D_10L	1334	62	341	26	132
train_jepa_ssm.py in records/track_non_record_16mb/2026-03-26_37M_LeWM_Jepa_Mamba2_10L_UNet_INT4FP8QAT_Brotli	1234	55	341	28	134
train_gpt_decode.py in records/track_10min_16mb/2026-04-06_SP8192_HessianSDClip_ProgressiveRecurrence	1541	61	331	21	75
train_gpt.py in records/track_10min_16mb/2026-04-01_Vocab4096_MLPMult4_WD085	1361	67	321	24	47
train_gpt.py in records/track_non_record_16mb/2026-03-23_DGAttention_DavidGao	1458	63	314	22	108
train_gpt_cuda_xnor_eggroll_lora.py in records/track_non_record_16mb/2026-04-05_118M_XNOR-Net_FP8_1024D_10L	1197	64	301	25	99
train_gpt.py in records/track_10min_16mb/2026-03-22_11L_EMA_GPTQ-lite_warmdown3500_QAT015_1.1233	1393	49	297	21	115
train_gpt.py in records/track_10min_16mb/2026-03-21_11L_XSA4_EMA_PartialRoPE_LateQAT_1.1248	1308	46	289	22	119
train_gpt.py in records/track_10min_16mb/2026-03-20_11L_XSA4_EMA_Int6_MLP3x_WD04_1.1271	1275	46	284	22	116
train_gpt_cuda_xnor_eggroll_nlayers.py in records/track_non_record_16mb/2026-04-05_118M_XNOR-Net_FP8_1024D_10L	1155	61	284	25	88
train_gpt_cuda_xnor_eggroll_full.py in records/track_non_record_16mb/2026-04-05_118M_XNOR-Net_FP8_1024D_10L	1145	61	280	24	88
train_gpt.py in records/track_10min_16mb/2026-03-20_11L_EfficientPartialXSA_FA3_SWA120	1260	46	279	22	111
train_gpt_human.py in records/track_10min_16mb/2026-04-05_SP8192_GPTQ-Embeddings_SDClip_Loop45x2	1174	56	268	23	54
train_gpt_legendre.py in records/track_non_record_16mb/2026-03-31_LegendreGPT	1189	44	265	22	120
train_gpt.py in records/track_10min_16mb/2026-03-19_MLP3x_QAT_Int6_SlidingWindow	1150	40	253	22	101
train_gpt.py in records/track_10min_16mb/2026-03-20_10L_Int5MLP_MuonWD04_SWA50	1059	42	237	22	98
train_gpt.py in records/track_10min_16mb/2026-03-20_Int6_MLP3x_SmearGate_BigramHash_MuonWD_SWA	1050	42	230	22	95
train_gpt.py in records/track_non_record_16mb/2026-03-21_DepthRecurrence_MixedPrecisionQuant	1006	44	227	23	82
train_gpt.py in records/track_10min_16mb/2026-03-17_LoRA_TTT	1088	45	216	20	79
train_gpt_v5.py in records/track_10min_16mb/2026-03-19_smeargate_orthoinit_muonwd	1099	40	215	20	84
train_gpt_hnet.py in records/track_non_record_16mb/2026-03-29_HNet_ByteVsSubword_Study	1078	44	212	20	86
train_gpt.py in records/track_10min_16mb/2026-03-19_SlidingWindowEval	1089	39	211	19	85
train_gpt_hnet_byte.py in records/track_non_record_16mb/2026-03-29_HNet_ByteVsSubword_Study	1051	44	210	20	84
train_gpt.py in records/track_10min_16mb/2026-03-19_Seq2048_FP16Emb_TunedLR	1046	37	208	20	87
train_gpt.py in records/track_10min_16mb/2026-03-19_WarmdownQuantization	1007	36	202	20	80
train_gpt.py in records/track_10min_16mb/2026-03-19_MixedQuant_Int6Int8_SlidingWindow	1041	36	201	19	82
train_gpt.py in records/track_10min_16mb/2026-03-19_SlidingWindow_FP16Emb_10L_MuonWD_OvertoneInit	1034	36	199	19	83
train_gpt.py in records/track_non_record_16mb/2026-03-30_ByteJEPA	943	39	199	21	88
train_gpt.py in records/track_non_record_16mb/2026-04-30_Random_Linear_Adapter	1014	38	197	19	78
train_gpt.py in records/track_10min_16mb/2026-03-19_10L_MixedPrecision	917	34	195	21	89
train_gpt_gdn.py in records/track_non_record_16mb/2026-04-04_GDN_Hybrid_LongContext	992	39	192	19	77
train_gpt.py in records/track_10min_16mb/2026-03-18_FP16Embed_WD3600	901	34	187	21	78

• File complexity is the sum of the McCabe indexes (conditional complexity) of the units (methods, functions...) in a file.
• Files are classified in five categories based on that sum: 1-25 (simple files), 26-50 (slightly complex files), 51-125 (medium complex files), 126-250 (complex files), 251+ (very complex files). The default bands are five times the unit complexity bands, as a file is a group of units.
• Only main code files with units are classified. Files without units (languages without unit analysis, or files without functions) have no complexity to measure and are left out rather than counted as simple.
• A very complex file holds many decisions, which often means it has too many responsibilities: a candidate to split, and a place where changes need more testing.