Commits · v0.6.0 · Simon Will / fairseq

Browse files

Sep 25, 2018
- Online backtranslation module · 864b89d0
  Myle Ott authored Sep 25, 2018
  View commits for tag v0.6.0 v0.6.0
  864b89d0
- Add back secondary set · a4fe8c99
  Sergey Edunov authored Sep 24, 2018
  
  a4fe8c99
- Merge internal changes · 535ca991
  Myle Ott authored Sep 24, 2018
  
  535ca991
- fix issue with truncated dict · 28069cf4
  alexeib authored Sep 21, 2018
  
  28069cf4
- core changes to support latte collab · cfd2a3a0
  Alexei Baevski authored Sep 20, 2018
  
  cfd2a3a0
- Better support for various c10d API changes · fbe8ce65
  Myle Ott authored Sep 17, 2018
  
  fbe8ce65
- Fix type of c10d bucket size · 78071e0f
  Myle Ott authored Sep 12, 2018
  
  78071e0f
- Parallel preprocessing · 862cad11
  Sergey Edunov authored Sep 12, 2018
  
  862cad11
- Fix adaptive loss logging · ee46c63b
  Sergey Edunov authored Sep 10, 2018
  
  ee46c63b
- Add unit test to verify reproducibility after reloading checkpoints · e775877f
  Myle Ott authored Sep 09, 2018
  
  e775877f
- Fix validation loss · 83e08b6f
  Myle Ott authored Sep 09, 2018
  
  83e08b6f
- Pass encoder_input to generator, rather than src_tokens/src_lengths. · bfeb7732
  Stephen Roller authored Sep 08, 2018
  
  bfeb7732
- Update LM test with --no-c10d · 8bd8ec8f
  Myle Ott authored Sep 07, 2018
  
  8bd8ec8f
- Disable c10d for AdaptiveLoss · f66e9cb5
  Myle Ott authored Sep 06, 2018
  
  f66e9cb5
- Switch to DistributedDataParallelC10d and bump version 0.5.0 -> 0.6.0 · 1082ba35
  Sergey Edunov authored Sep 06, 2018
  1082ba35
- Revert sequence generator changes · 311d2c6c
  Myle Ott authored Sep 06, 2018
  
  311d2c6c
- Sequence generator bug fix. · 0714080b
  Stephen Roller authored Sep 05, 2018
  
  0714080b
- Generator: net_input instead of manual src_tokens. · e6d45d5c
  Stephen Roller authored Sep 05, 2018
  
  e6d45d5c
Sep 24, 2018
- Merge pull request #287 from pytorch/oss-master · 25524f19
  Sergey Edunov authored Sep 24, 2018
  25524f19
- Update readme with WMT'18 model (#433) · 86b5cfe4
  Sergey Edunov authored Sep 24, 2018
  
  86b5cfe4
Sep 18, 2018
- Merge pull request #279 from pytorch/oss-master · 5d150856
  Sergey Edunov authored Sep 17, 2018
  5d150856
- Readme fix · 74b3f1e9
  Sergey Edunov authored Sep 17, 2018
  
  74b3f1e9
- Fix docs · fe2d1581
  Sergey Edunov authored Sep 17, 2018
  
  fe2d1581
- Fix readme · 5d944b06
  Sergey Edunov authored Sep 17, 2018
  
  5d944b06
Sep 07, 2018
- modified stories readme to include sample preprocessing code to split stories to 1k tokens · 5d00e8ee
  Angela Fan authored Sep 07, 2018
  
  5d00e8ee
Sep 04, 2018
- Update documentation · 4a47b889
  Myle Ott authored Sep 03, 2018
  
  4a47b889
Sep 03, 2018
- Add documentation · 6381cc97
  Myle Ott authored Sep 03, 2018
  
  6381cc97
- Misc changes to simplify upcoming tutorial · 0e101e9c
  Myle Ott authored Sep 02, 2018
  
  0e101e9c
- Test max_positions · d473620e
  Myle Ott authored Sep 02, 2018
  
  d473620e
- fix cosine lr sched for t_mult=1 with warmup · dfd77717
  alexeib authored Sep 02, 2018
  
  dfd77717
- Further generalize EpochBatchIterator and move iterators into new file · 0a7f9e64
  Myle Ott authored Aug 31, 2018
  
  0a7f9e64
- Fix comment · 75f6ba05
  Myle Ott authored Aug 30, 2018
  
  75f6ba05
- fix max_positions comparison · b3cd43b2
  alexeib authored Aug 30, 2018
  
  b3cd43b2
- Clean up FairseqTask so that it's easier to extend/add new tasks · 2e507d3c
  Myle Ott authored Aug 30, 2018
  
  2e507d3c
- Add --upsample-primary · 6296de82
  Myle Ott authored Aug 28, 2018
  
  6296de82
- Add adaptive softmax changes for lstm model · 5852d3a0
  Li Zhao authored Aug 28, 2018
  
  5852d3a0
- dont send dummy batch when reloading from checkpoint · 343819f9
  Alexei Baevski authored Aug 28, 2018
  343819f9
- Fix FP16 version comparison · b9956a6a
  Myle Ott authored Aug 27, 2018
  
  b9956a6a
- Merge internal changes · 753935ef
  Myle Ott authored Aug 27, 2018
  
  753935ef
- word stats in eval_lm · c7c567a7
  Alexei Baevski authored Aug 26, 2018
  
  c7c567a7