huggingface / pytorch-image-models Unit Size

Intro

Unit size measurements show the distribution of size of units of code (methods, functions...).
Units are classified in four categories based on their size (lines of code): 1-20 (small units), 20-50 (medium size units), 51-100 (long units), 101+ (very long units).
You should aim at keeping units small (< 20 lines). Long units may become "bloaters", code that have increased to such gargantuan proportions that they are hard to work with.

Learn more...

Unit Size Overall

There are 4,906 units with 41,972 lines of code in units (52.6% of code).

11 very long units (1,922 lines of code)
35 long units (2,230 lines of code)
249 medium size units (7,133 lines of code)
809 small units (11,648 lines of code)
3,802 very small units (19,039 lines of code)

Legend:

101+

51-100

21-50

11-20

1-10

Unit Size per Extension

101+

51-100

21-50

11-20

1-10

Unit Size per Logical Component

primary logical decomposition

101+

51-100

21-50

11-20

1-10

Alternative Visuals

3D view of all units

Longest Units

Top 50 longest units

Unit	# lines	McCabe index	# params
def main() in train.py	548	90	0
def validate() in validate.py	212	44	1
def main() in inference.py	163	55	0
def create_naflex_loader() in timm/data/naflex_loader.py	151	8	12
def _load_weights() in timm/models/vision_transformer.py	136	42	4
def step() in timm/optim/kron.py	129	26	2
def _register_corrected_decay_optimizers() in timm/optim/_optim_factory.py	121	3	1
def _register_other_optimizers() in timm/optim/_optim_factory.py	119	2	1
def _create_selecsls() in timm/models/selecsls.py	118	6	3
def _decode_block_str() in timm/models/_efficientnet_builder.py	115	33	1
def _register_cautious_optimizers() in timm/optim/_optim_factory.py	110	2	1
def _register_adam_variants() in timm/optim/_optim_factory.py	94	2	1
def run() in benchmark.py	91	11	1
def step() in timm/optim/adabelief.py	80	20	2
def step() in timm/optim/adafactor.py	78	18	2
def __init__() in timm/models/senet.py	78	3	13
def main() in bulk_runner.py	78	7	0
def __init__() in timm/models/levit.py	74	10	5
def _register_bnb_optimizers() in timm/optim/_optim_factory.py	73	2	1
def step() in timm/optim/adafactor_bv.py	72	11	2
def main() in avg_checkpoints.py	72	28	0
def __call__() in timm/models/_efficientnet_builder.py	71	23	3
def step() in timm/optim/adan.py	70	14	2
def step() in timm/optim/madgrad.py	67	15	3
def _gen_mobilenet_edgetpu() in timm/models/efficientnet.py	65	7	5
def forward_features() in timm/models/coat.py	64	24	2
def step() in timm/optim/lamb.py	63	24	2
def generate_readme() in timm/models/_hub.py	60	25	2
def step() in timm/optim/rmsprop_tf.py	59	18	2
def _register_lamb_lars() in timm/optim/_optim_factory.py	59	2	1
def checkpoint_filter_fn() in timm/models/fastvit.py	58	18	2
def checkpoint_filter_fn() in timm/models/naflexvit.py	56	18	3
def forward() in timm/models/mvitv2.py	56	10	3
def step() in timm/optim/adamw.py	54	10	2
def _create_canonical_schedule() in timm/data/naflex_dataset.py	54	18	1
def _prepare_epoch_batches() in timm/data/naflex_dataset.py	54	22	2
def adapt_model_from_string() in timm/models/_prune.py	54	15	2
def _gen_fbnetv3() in timm/models/mobilenetv3.py	54	4	4
def step() in timm/optim/radam.py	53	9	2
def __iter__() in timm/data/naflex_loader.py	53	19	1
def _filter_fn() in timm/models/regnet.py	53	10	2
def run() in benchmark.py	53	10	1
def convert_nest() in convert/convert_nest_flax.py	53	11	2
def main() in validate.py	53	6	0
def switch_to_deploy() in timm/models/ghostnet.py	52	11	1
def _make_block() in timm/models/_efficientnet_builder.py	52	24	4
def _lazy_init() in timm/data/readers/reader_tfds.py	50	12	1
def benchmark() in benchmark.py	50	15	1
def step() in timm/optim/mars.py	49	7	2
def get_attn() in timm/layers/create_attn.py	49	25	1