Skip to content

GitLab

Switch branch/tag
  1. 05 Feb, 2019 1 commit
  3. 30 Jan, 2019 1 commit
    • Myle Ott's avatar
      Do distributed init after data loading · ec6f8ef9
      Myle Ott authored 
      Summary:
FACEBOOK

This switches back to torch.multiprocessing.spawn, instead of directly calling fb_train.par using a subprocess.Process. This has the advantage that exceptions are propagated properly. It also moves the distributed_init part to happen after data loading, which gets around the timeout issue.

The downside of this approach is that it's not so easy to pipe stdout to multiple places, which was nice when using the sweep.py scripts. I'm still working on a fix for that.

Reviewed By: rutyrinott, ngoyal2707

Differential Revision: D13873224

fbshipit-source-id: 08d593233b8d23590c01c723363630a79804a8b0
      ec6f8ef9
  5. 25 Jan, 2019 1 commit
  7. 24 Jan, 2019 1 commit
  9. 16 Jan, 2019 1 commit
    • Davide Caroselli's avatar
      FIX: '--user-dir' on multi-gpu (#449) · 7853818c
      Davide Caroselli authored 
      Summary:
On a multi-gpu training scenario, the `train.py` script spawns new processes with `torch.multiprocessing.spawn`. Unfortunately those child processes don't inherit the modules imported with `--user-dir`.

This pull request fixes this problem: custom module import in now explicit on every `main()` function.
Pull Request resolved: https://github.com/pytorch/fairseq/pull/449

Differential Revision: D13676922

Pulled By: myleott

fbshipit-source-id: 520358d66155697885b878a37e7d0484bddbc1c6
      7853818c
  11. 09 Jan, 2019 1 commit
  13. 05 Jan, 2019 1 commit
  15. 28 Dec, 2018 1 commit
  17. 07 Dec, 2018 1 commit
    • Halil Akin's avatar
      Take a dummy train step under OOM to keep multiprocessing in sync · 6c006a34
      Halil Akin authored 
      Summary: This is not a guaranteed solution (since processes may still get out of sync if OOM happens after an all_gather/all_reduce has been done) - but should still make multiprocessing training more robust in practice since it seems we usually OOM early enough.

Reviewed By: myleott

Differential Revision: D13086018

fbshipit-source-id: feb1b01c2eb8818797cfdabc0faac8056ba1b4ee
      6c006a34
  19. 18 Nov, 2018 1 commit
  21. 21 Oct, 2018 1 commit
  23. 30 Sep, 2018 1 commit
    • Myle Ott's avatar
      Merge internal changes (#295) · b87c5366
      Myle Ott authored 
      Summary:
Changelog:
- `90f52a1`: Support loading subsets of the data on each worker with the `--fix-batches-to-gpus` flag. This should fix #217 and #266.
- `6eda0a9`: Update README for replicating the "Scaling Neural Machine Translation" paper
- `b14c7cf`: Fallback to no_c10d backend for pytorch 0.4.1 (fixes #294)
Pull Request resolved: https://github.com/pytorch/fairseq/pull/295

Differential Revision: D10121559

Pulled By: myleott

fbshipit-source-id: 41c84d0ee4cdd113544b5d3aa38ae8b23acc2c27
      b87c5366
  25. 25 Sep, 2018 1 commit
    • Sergey Edunov's avatar
      Switch to DistributedDataParallelC10d and bump version 0.5.0 -> 0.6.0 · 1082ba35
      Sergey Edunov authored 
      - no more FP16Trainer, we just have an FP16Optimizer wrapper
- most of the distributed code is moved to a new wrapper class called DistributedFairseqModel, which behaves like DistributedDataParallel and a FairseqModel at the same time
- Trainer now requires an extra dummy_batch argument at initialization, which we do fwd/bwd on when there's an uneven number of batches per worker. We hide the gradients from these dummy batches by multiplying the loss by 0
- Trainer.train_step now takes a list of samples, which will allow cleaner --update-freq
      1082ba35
  27. 03 Sep, 2018 8 commits
  29. 25 Jul, 2018 1 commit
    • Alexei Baevski's avatar
      Transformer lm · d2e2a1d4
      Alexei Baevski authored 
      This implements transformer based language model. It already obtains better perplexity on wikitext103 without any tuning. I will also train it on gbw where I also expect to get better ppl

Example training command:

python train.py /private/home/abaevski/data/wiki103 —save-dir /tmp —fp16 —max-epoch 80 —save-interval 1 —arch transformer_lm —task language_modeling —optimizer nag —lr 0.008 —lr-scheduler reduce_lr_on_plateau —lr-shrink 0.6 —dropout 0.2 —criterion adaptive_loss —adaptive-softmax-cutoff 10000,50000,200000 —max-tokens 512 —tokens-per-sample 512 —seed 1 —sample-break-mode none —log-format json —log-interval 50 —save-interval-updates 2500 —keep-interval-updates 25
small transformer got to 31.3 ppl on wiki text 103 (compared to 35 with fconv) while @myleott got a big transformer lm to 27 something ppl on wiki text 103
      d2e2a1d4
  31. 21 Jun, 2018 3 commits
  33. 15 Jun, 2018 15 commits