Commits · ce37b8e4819142171b61558e64f7dcb0286e9937 · chenpangpang / transformers

26 Nov, 2018 1 commit
- update cache_dir in readme and examples · 05053d16
  thomwolf authored Nov 26, 2018
  
  05053d16
20 Nov, 2018 2 commits
- fixing weights decay in run_squad example · 6b2136a8
  thomwolf authored Nov 20, 2018
  
  6b2136a8
- fixing param.grad is None in fp16 examples · 2f21497d
  thomwolf authored Nov 20, 2018
  
  2f21497d
17 Nov, 2018 3 commits
- remove unused arguments in example scripts · aa50fd19
  thomwolf authored Nov 17, 2018
  
  aa50fd19
- fix tests · 757750d6
  thomwolf authored Nov 17, 2018
  
  757750d6
- updating examples · 4e46affc
  thomwolf authored Nov 17, 2018
  
  4e46affc
15 Nov, 2018 2 commits
- fix nan in optimizer_on_cpu · cba85a67
  thomwolf authored Nov 15, 2018
  
  cba85a67
- preparing for first release · 1de35b62
  thomwolf authored Nov 15, 2018
  
  1de35b62
12 Nov, 2018 1 commit
- add fp16 training · 66b00908
  thomwolf authored Nov 12, 2018
  
  66b00908
10 Nov, 2018 1 commit
- fixing learning rate schedule when using gradient_accumulation_steps · a81a1ef8
  thomwolf authored Nov 10, 2018
  
  a81a1ef8
09 Nov, 2018 2 commits
- option to perform optimization and keep the optimizer averages on CPU · 5f04aa00
  thomwolf authored Nov 09, 2018
  
  5f04aa00
- remove duplicate accumulate gradient step arguments · 34bdc8b5
  thomwolf authored Nov 09, 2018
  
  34bdc8b5
07 Nov, 2018 2 commits
- cleaning up - speeding up a bit multi-gpu · dbc318a4
  thomwolf authored Nov 07, 2018
  
  dbc318a4
- fixing pre-processing bug - averaging loss for gradient accumulation - no_grad on evaluation · 6bb7510a
  thomwolf authored Nov 07, 2018
  
  6bb7510a
06 Nov, 2018 1 commit
- clean up logits extraction logic · a1126237
  thomwolf authored Nov 06, 2018
  
  a1126237
04 Nov, 2018 13 commits
- Fix `args.gradient_accumulation_steps` used before assigment. · 290633b8
  VictorSanh authored Nov 04, 2018
  
  290633b8
- Fix bug train_batch_size not an int. · 649e9774
  VictorSanh authored Nov 04, 2018
```
Division makes args.train_batch_size becoming a float.
cc @thomwolf
```
  649e9774
- Small logger bug (multi-gpu, distribution) in training · d55c3ae8
  VictorSanh authored Nov 04, 2018
  
  d55c3ae8
- fixes + clean up + mask is long · d69b0b0e
  thomwolf authored Nov 04, 2018
  
  d69b0b0e
- add distributed training · 965b2565
  thomwolf authored Nov 04, 2018
  
  965b2565
- add gradient accumulation · 1ceac85e
  thomwolf authored Nov 04, 2018
  
  1ceac85e
- clean up · 6b0da96b
  thomwolf authored Nov 04, 2018
  
  6b0da96b
- logging + update copyright · 834b485b
  thomwolf authored Nov 04, 2018
  
  834b485b
- multi-gpu cleanup · 1701291e
  thomwolf authored Nov 04, 2018
  
  1701291e
- what's in loss again · 5ee17168
  thomwolf authored Nov 04, 2018
  
  5ee17168
- add tqdm, clean up logging · 0b7a20c6
  thomwolf authored Nov 04, 2018
  
  0b7a20c6
- add numpy import · d4e3cf35
  thomwolf authored Nov 04, 2018
  
  d4e3cf35
- fixing verbose_argument · 26bdef43
  thomwolf authored Nov 04, 2018
  
  26bdef43
03 Nov, 2018 3 commits
- update readme, file names, removing TF code, moving tests · f8276008
  thomwolf authored Nov 03, 2018
  
  f8276008
- Fix imports · 8ec457d3
  Tim Rault authored Nov 03, 2018
  
  8ec457d3
- clean up · 4df60274
  thomwolf authored Nov 03, 2018
  
  4df60274
01 Nov, 2018 1 commit
- Convert indentation from 2 spaces to 4 spaces · 8163baab
  Tim Rault authored Nov 01, 2018
  
  8163baab
31 Oct, 2018 1 commit
- switch to full google code · 13ee61e4
  thomwolf authored Oct 31, 2018
  
  13ee61e4