1. 04 Aug, 2019 1 commit
  2. 03 Aug, 2019 2 commits
  3. 02 Aug, 2019 5 commits
    • Ning Dong's avatar
      Avoid cast in PositionalEmbeddings to fix BLEU drop in pytorch native export · 9012e87d
      Ning Dong authored
      Summary:
      Tracing mode doesn't generalize correctly in positional embedding calculation, which caused -5 BLEU at transformer export when using pytorch native.
      
      Details: The original issue was that in ensemble_export, _to_tensor(x) in scripting mode turns integer x into 1-d tensor torch.tensor([x]), not 0-d tensor (scalar x) which is expected in the embedding. So the return value in embedding forward() is actually of wrong shape. When self.weights is of size [x,y], the return value should be (bsz, y, 1) but it was (bsz, 1, y), which caused problem in downstream computation. Tracing only becomes an issue when I used pos = timestep.view(-1)[0] to fix the shape. Then casting the scalar to primary int, to be used as index is not generalizable by tracing mode. Thus I need to convert everything to tensor and replace the advanced indexing with index_select operator.
      
      In summary, less understood features in both scripting&tracing sides caused the bleu drop. :)
      
      Reviewed By: myleott
      
      Differential Revision: D16623025
      
      fbshipit-source-id: 0c7a2c3eafbd774760a5c880c6034009ee084abb
      9012e87d
    • Myle Ott's avatar
      Fewer torch.hub requirements (#959) · 3903f469
      Myle Ott authored
      Summary:
      We will raise exceptions if these are needed and aren't available. Only keep minimum set of reqs
      Pull Request resolved: https://github.com/pytorch/fairseq/pull/959
      
      Differential Revision: D16623304
      
      Pulled By: myleott
      
      fbshipit-source-id: 8e65253742e393b527e8396a9433e64ebec9bb55
      3903f469
    • Myle Ott's avatar
      Add single-models for WMT'19 for hub tutorial · f02f70cc
      Myle Ott authored
      Summary: Pull Request resolved: https://github.com/fairinternal/fairseq-py/pull/800
      
      Differential Revision: D16621509
      
      Pulled By: myleott
      
      fbshipit-source-id: d3e8e97d30bcafbc35c3f67cd8bbc657b6fa5fe7
      f02f70cc
    • Myle Ott's avatar
      Update READMEs for torch.hub · abb7ed4c
      Myle Ott authored
      Summary: Pull Request resolved: https://github.com/fairinternal/fairseq-py/pull/795
      
      Differential Revision: D16620488
      
      Pulled By: myleott
      
      fbshipit-source-id: 1998a9ccd8816fc7f590861fb4898f910a36bc1e
      abb7ed4c
    • Myle Ott's avatar
      Update beam search code to support torch.bool change · 5f342527
      Myle Ott authored
      Summary: Pull Request resolved: https://github.com/fairinternal/fairseq-py/pull/797
      
      Differential Revision: D16617067
      
      Pulled By: myleott
      
      fbshipit-source-id: 52e3aeb98d6e3b55ff9154b784028bf13eabfe38
      5f342527
  4. 01 Aug, 2019 7 commits
  5. 31 Jul, 2019 5 commits
  6. 30 Jul, 2019 4 commits
  7. 29 Jul, 2019 8 commits
  8. 28 Jul, 2019 6 commits
  9. 27 Jul, 2019 2 commits