facebookresearch / fairscale
File Size

The distribution of size of files (measured in lines of code).

Intro
  • File size measurements show the distribution of size of files.
  • Files are classified in four categories based on their size (lines of code): 1-100 (very small files), 101-200 (small files), 201-500 (medium size files), 501-1000 (long files), 1001+(very long files).
  • It is a good practice to keep files small. Long files may become "bloaters", code that have increased to such gargantuan proportions that they are hard to work with.
Learn more...
File Size Overall
  • There are 192 files with 18,003 lines of code.
    • 2 very long files (3,116 lines of code)
    • 2 long files (1,355 lines of code)
    • 17 medium size files (5,205 lines of codeclsfd_ftr_w_mp_ins)
    • 34 small files (4,788 lines of code)
    • 137 very small files (3,539 lines of code)
17% | 7% | 28% | 26% | 19%
Legend:
1001+
501-1000
201-500
101-200
1-100


explore: zoomable circles | sunburst | 3D view
File Size per Extension
1001+
501-1000
201-500
101-200
1-100
pyi51% | 0% | 0% | 12% | 36%
py9% | 9% | 35% | 30% | 15%
cu0% | 0% | 100% | 0% | 0%
cuh0% | 0% | 0% | 100% | 0%
cfg0% | 0% | 0% | 0% | 100%
toml0% | 0% | 0% | 0% | 100%
yml0% | 0% | 0% | 0% | 100%
cpp0% | 0% | 0% | 0% | 100%
h0% | 0% | 0% | 0% | 100%
in0% | 0% | 0% | 0% | 100%
File Size per Logical Decomposition
primary
1001+
501-1000
201-500
101-200
1-100
stubs/torch51% | 0% | 0% | 12% | 36%
fairscale/nn21% | 0% | 25% | 33% | 19%
fairscale/experimental0% | 19% | 35% | 35% | 8%
benchmarks/experimental0% | 54% | 29% | 9% | 6%
fairscale/optim0% | 0% | 69% | 29% | 1%
fairscale/utils0% | 0% | 58% | 12% | 29%
benchmarks0% | 0% | 68% | 17% | 13%
fairscale/clib0% | 0% | 67% | 30% | 2%
benchmarks/models0% | 0% | 0% | 99% | <1%
benchmarks/golden_configs0% | 0% | 0% | 91% | 8%
ROOT0% | 0% | 0% | 0% | 100%
benchmarks/datasets0% | 0% | 0% | 0% | 100%
fairscale0% | 0% | 0% | 0% | 100%
Longest Files (Top 50)
File# lines# units
pyi
__init__.pyi
in stubs/torch
 1856 1232
fully_sharded_data_parallel.py
in fairscale/nn/data_parallel
 1260 74
distributed.py
in fairscale/experimental/nn/data_parallel/gossip
 758 47
experimental_async_approaches.py
in benchmarks/experimental
 597 48
testing.py
in fairscale/utils
 495 35
layer_memory_tracker.py
in fairscale/experimental/tooling
 465 49
async_schedule.py
in fairscale/nn/pipe
 364 24
sharded_ddp.py
in fairscale/nn/data_parallel
 352 20
oss.py
in fairscale/optim
 339 19
offload.py
in benchmarks/experimental
 328 20
mevo.py
in fairscale/experimental/nn
 317 24
ampnet.py
in fairscale/experimental/nn/ampnet_pipe
 303 13
flatten_params_wrapper.py
in fairscale/nn/misc
 303 29
offload.py
in fairscale/experimental/nn
 281 18
oss.py
in benchmarks
 254 6
grad_scaler.py
in fairscale/optim
 247 12
cu
fused_adam_cuda_kernel.cu
in fairscale/clib/fused_adam_cuda
 246 -
pipe.py
in benchmarks
 235 13
adascale.py
in fairscale/optim
 226 23
async_pipe.py
in fairscale/nn/pipe
 225 15
layers.py
in fairscale/nn/model_parallel
 225 11
rpc.py
in fairscale/nn/pipe
 200 18
graph_manager.py
in fairscale/experimental/nn/data_parallel/gossip
 197 45
ssd_offload.py
in fairscale/experimental/nn
 197 24
adam.py
in fairscale/optim
 192 10
pipe.py
in fairscale/nn/pipe
 182 13
gossiper.py
in fairscale/experimental/nn/data_parallel/gossip
 181 13
skippable.py
in fairscale/nn/pipe/skip
 179 13
partition_handler.py
in fairscale/experimental/nn/distributed_pipeline
 178 15
pyi
functional.pyi
in stubs/torch/nn
 177 80
checkpoint_activations.py
in fairscale/nn/checkpoint
 167 14
checkpoint.py
in fairscale/nn/pipe
 165 16
pyi
pooling.pyi
in stubs/torch/nn/modules
 147 50
param_bucket.py
in fairscale/nn/misc
 145 19
layerwise_gradient_scaler.py
in fairscale/optim
 145 13
pipeline.py
in fairscale/nn/pipe
 143 11
fsdp_optim_utils.py
in fairscale/nn/data_parallel
 137 6
sync_batchnorm.py
in fairscale/experimental/nn
 133 8
dynamic_loss_scaler.py
in fairscale/experimental/optim
 124 11
utils.py
in benchmarks
 124 10
auto_shard.py
in fairscale/experimental/nn
 121 7
portal.py
in fairscale/nn/pipe/skip
 119 16
pyi
activation.pyi
in stubs/torch/nn/modules
 117 62
pipeline.py
in fairscale/experimental/nn/distributed_pipeline
 117 8
messages.py
in fairscale/nn/pipe
 115 16
transformer_lm.py
in benchmarks/models
 115 15
graph.py
in fairscale/experimental/nn/distributed_pipeline
 114 13
random.py
in fairscale/nn/model_parallel
 114 12
cuh
multi_tensor_apply.cuh
in fairscale/clib/fused_adam_cuda
 111 -
auto_wrap.py
in fairscale/nn/wrap
 108 11
Files With Most Units (Top 20)
File# lines# units
pyi
__init__.pyi
in stubs/torch
 1856 1232
pyi
functional.pyi
in stubs/torch/nn
 177 80
fully_sharded_data_parallel.py
in fairscale/nn/data_parallel
 1260 74
pyi
activation.pyi
in stubs/torch/nn/modules
 117 62
pyi
container.pyi
in stubs/torch/nn/modules
 81 59
pyi
loss.pyi
in stubs/torch/nn/modules
 100 57
pyi
pooling.pyi
in stubs/torch/nn/modules
 147 50
layer_memory_tracker.py
in fairscale/experimental/tooling
 465 49
experimental_async_approaches.py
in benchmarks/experimental
 597 48
distributed.py
in fairscale/experimental/nn/data_parallel/gossip
 758 47
graph_manager.py
in fairscale/experimental/nn/data_parallel/gossip
 197 45
pyi
__init__.pyi
in stubs/torch/cuda
 69 43
pyi
rnn.pyi
in stubs/torch/nn/modules
 81 40
pyi
module.pyi
in stubs/torch/nn/modules
 57 38
testing.py
in fairscale/utils
 495 35
flatten_params_wrapper.py
in fairscale/nn/misc
 303 29
pyi
__init__.pyi
in stubs/torch/distributed
 68 28
mevo.py
in fairscale/experimental/nn
 317 24
ssd_offload.py
in fairscale/experimental/nn
 197 24
async_schedule.py
in fairscale/nn/pipe
 364 24
Files With Long Lines (Top 20)

There are 30 files with lines longer than 120 characters. In total, there are 275 long lines.

File# lines# units# long lines
pyi
__init__.pyi
in stubs/torch
 1856 1232 225
pyi
__init__.pyi
in stubs/torch/distributed
 68 28 6
pyi
lr_scheduler.pyi
in stubs/torch/optim
 31 17 5
pyi
__init__.pyi
in stubs/torch/autograd
 37 21 4
fully_sharded_data_parallel.py
in fairscale/nn/data_parallel
 1260 74 4
pyi
module.pyi
in stubs/torch/nn/modules
 57 38 3
pyi
loss.pyi
in stubs/torch/nn/modules
 100 57 2
pyi
rnn.pyi
in stubs/torch/nn/modules
 81 40 2
pyi
grad_scaler.pyi
in stubs/torch/cuda/amp
 12 5 2
ssd_offload.py
in fairscale/experimental/nn
 197 24 2
pyi
data_parallel.pyi
in stubs/torch/nn/parallel
 18 4 1
pyi
functional.pyi
in stubs/torch/nn
 177 80 1
pyi
__init__.pyi
in stubs/torch/nn/modules
 48 - 1
pyi
linear.pyi
in stubs/torch/nn/modules
 28 11 1
pyi
activation.pyi
in stubs/torch/nn/modules
 117 62 1
pyi
adam.pyi
in stubs/torch/optim
 4 1 1
pyi
sgd.pyi
in stubs/torch/optim
 3 1 1
pyi
__init__.pyi
in stubs/torch/utils/data
 6 - 1
pyi
__init__.pyi
in stubs/torch/testing
 2 1 1
pyi
functional.pyi
in stubs/torch
 5 3 1