alibaba / ChatLearn - Conditional Complexity

Intro

Conditional complexity (also called cyclomatic complexity) is a term used to measure the complexity of software. The term refers to the number of possible paths through a program function. A higher value ofter means higher maintenance and testing costs (infosecinstitute.com).
Conditional complexity is calculated by counting all conditions in the program that can affect the execution path (e.g. if statement, loops, switches, and/or operators, try and catch blocks...).
Conditional complexity is measured at the unit level (methods, functions...).
Units are classified in four categories based on the measured McCabe index: 1-5 (simple units), 6-10 (medium complex units), 11-25 (complex units), 26+ (very complex units).

Learn more...

Conditional Complexity Overall

There are 1,072 units with 13,076 lines of code in units (83.0% of code).

0 very complex units (0 lines of code)
13 complex units (1,364 lines of code)
61 medium complex units (2,856 lines of code)
132 simple units (3,016 lines of code)
866 very simple units (5,840 lines of code)

Legend:

51+

26-50

11-25

6-10

1-5

Alternative Visuals

3D view of all units

Conditional Complexity per Extension

51+

26-50

11-25

6-10

1-5

Conditional Complexity per Logical Component

primary logical decomposition

51+

26-50

11-25

6-10

1-5

Most Complex Units

Top 50 most complex units

Unit	# lines	McCabe index	# params
def _init_workers_ray() in chatlearn/models/vllm/hooks/vllm_0_6_6/ray_gpu_executor.py	124	47	3
def _validate_params() in chatlearn/utils/arguments.py	118	44	1
def _init_workers_ray() in chatlearn/models/vllm/hooks/vllm_0_6_3/ray_gpu_executor.py	112	43	3
def preprocess_compute() in chatlearn/runtime/decorator.py	94	43	2
def convert_layer_to_lora() in chatlearn/models/megatron/lora/layers.py	72	39	7
def build_rank_mapping_two_stage() in chatlearn/synchronizer/parameter_sync.py	113	37	2
def convert_qwen_state_dict_from_megatron_to_vllm() in chatlearn/utils/vllm_utils.py	211	34	3
def place_gpu_models() in chatlearn/schedule/model_manager.py	58	31	3
def convert_checkpoint_from_megatron_to_transformers() in chatlearn/tools/megatron_to_hf.py	146	30	1
def load_weights() in chatlearn/models/vllm/hooks/vllm_0_6_6/qwen3_moe.py	80	29	3
def _parse_params() in chatlearn/utils/arguments.py	75	28	2
def build_rank_mapping_for_ep() in chatlearn/synchronizer/parameter_sync.py	70	28	2
def broadcast_parameter_two_stage() in chatlearn/models/base_module.py	91	26	8
def map_src_to_dst() in chatlearn/utils/vllm_utils.py	56	25	1
def logging_summary() in chatlearn/runtime/engine.py	46	25	2
def process_model_outputs() in chatlearn/models/vllm_module.py	51	21	3
def check_log_files_and_publish_updates() in chatlearn/utils/log_monitor.py	76	19	1
def _add_request_internal() in chatlearn/models/vllm_module.py	89	19	4
def onload_weights() in chatlearn/models/megatron/memory_manager/trainer_v3.py	69	19	1
def regroup_params_to_sync() in chatlearn/synchronizer/base.py	54	19	5
def sync_broadcast_multi_threads() in chatlearn/synchronizer/parameter_sync.py	59	19	10
def convert_llama_state_dict_from_megatron_to_vllm() in chatlearn/utils/vllm_utils.py	115	18	3
def build_scheduler() in chatlearn/models/vllm_module.py	78	18	1
def generate_vllm() in chatlearn/models/vllm_module.py	54	18	4
def compute_loop_one_model() in chatlearn/runtime/executor.py	39	18	3
def learn() in chatlearn/runtime/engine.py	99	18	1
def regroup_qkv_tp_slices() in chatlearn/synchronizer/megatron_vllm.py	80	18	4
def validate_sync_results() in chatlearn/synchronizer/parameter_sync.py	71	18	6
def onload_weights() in chatlearn/models/megatron/memory_manager/trainer_v1v2.py	62	17	1
def generate_step_one_model() in chatlearn/runtime/executor.py	28	17	12
def get_src_and_dst_dp_ranks() in chatlearn/synchronizer/parameter_sync.py	39	17	2
def _set_sync_param_names() in chatlearn/synchronizer/parameter_sync.py	48	17	7
def update_log_filenames() in chatlearn/utils/log_monitor.py	52	16	1
def log_metrics_stats() in chatlearn/models/vllm_module.py	75	16	2
def offload() in chatlearn/models/torch_module.py	30	16	5
def trace() in chatlearn/runtime/model_flow.py	26	16	3
def forward() in chatlearn/models/megatron/ops/policy_gradient.py	87	15	8
def timeit() in chatlearn/runtime/decorator.py	20	15	2
def __iter__() in chatlearn/data/sampler.py	44	14	1
def convert_llama_state_dict_from_mcore_to_vllm() in chatlearn/utils/vllm_utils.py	96	14	3
def _load_base_checkpoint() in chatlearn/utils/vllm_utils.py	52	14	2
def load_weights() in chatlearn/models/vllm/hooks/vllm_0_6_6/qwen3.py	40	14	3
def build_grad_buffers() in chatlearn/models/megatron/memory_manager/trainer_v1v2.py	54	14	1
def get_merged_data() in chatlearn/runtime/executor.py	34	14	6
def allreduce_word_embedding_grads() in chatlearn/models/megatron/lora/layers.py	28	13	2
def onload_weights() in chatlearn/models/megatron/memory_manager/trainer_v4.py	55	13	1
def onload() in chatlearn/models/torch_module.py	29	13	5
def topological_sort() in chatlearn/runtime/model_flow.py	24	13	1
def remote() in chatlearn/schedule/model_manager.py	43	13	1
def find_model_packing_strategy() in chatlearn/schedule/model_manager.py	31	13	3