search

hirofumi0810 / neural_sp

End-to-end ASR/LM implementation with PyTorch
Apache License 2.0
589 stars 140 forks source link

Fix DDP launch #320

Closed hirofumi0810 closed 3 years ago

hirofumi0810 commented 3 years ago

Change mp.spawn to torch.distributed.launch to enable num_workers in DataLoader.

codecov-commenter commented 3 years ago

Codecov Report

Merging #320 (26bb0d3) into master (f9c9789) will increase coverage by 0.14%. The diff coverage is 61.90%.

Impacted file tree graph

@@            Coverage Diff             @@
##           master     #320      +/-   ##
==========================================
+ Coverage   67.89%   68.04%   +0.14%     
==========================================
  Files          96       96              
  Lines       10853    10835      -18     
==========================================
+ Hits         7369     7373       +4     
+ Misses       3484     3462      -22
Impacted Files Coverage Δ
neural_sp/datasets/asr/build.py 55.17% <ø> (ø)
neural_sp/datasets/lm.py 86.81% <50.00%> (-0.83%) :arrow_down:
neural_sp/bin/lm/train.py 73.99% <59.37%> (+1.34%) :arrow_up:
neural_sp/bin/asr/train.py 65.92% <60.52%> (+0.51%) :arrow_up:
neural_sp/datasets/asr/sampler.py 85.52% <66.66%> (-0.59%) :arrow_down:
neural_sp/bin/args_common.py 100.00% <100.00%> (ø)
neural_sp/evaluators/word.py 60.93% <0.00%> (+4.68%) :arrow_up:
neural_sp/evaluators/wordpiece.py 76.00% <0.00%> (+4.79%) :arrow_up:

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update f9c9789...26bb0d3. Read the comment docs.