Commits · 856a63da4d1f0f302633dc73e2d4a1f698bbafda · chenpangpang / transformers

18 Aug, 2019 3 commits
- Fix: save model/model.module · 856a63da
  wangfei authored Aug 18, 2019
  
  856a63da
- Revert "Fix: save model/model.module" · 1ef41b83
  wangfei authored Aug 18, 2019
```
This reverts commit 00e9c4cc.
```
  1ef41b83
- Fix: save model/model.module · 00e9c4cc
  wangfei authored Aug 18, 2019
  
  00e9c4cc
16 Aug, 2019 1 commit
- Fix bug of multi-gpu training in lm finetuning · b8ff5689
  wangfei authored Aug 16, 2019
  
  b8ff5689
18 Jul, 2019 2 commits
- use new API for save and load · a1fe4ba9
  yzy5630 authored Jul 18, 2019
  
  a1fe4ba9
- add parser for adam · a7ba27b1
  yzy5630 authored Jul 18, 2019
  
  a7ba27b1
17 Jul, 2019 2 commits
- change loss and optimizer to new API · d6522e28
  yzy5630 authored Jul 17, 2019
  
  d6522e28
- fix errors for lm_finetuning examples · 123da5a2
  yzy5630 authored Jul 17, 2019
  
  123da5a2
14 Jul, 2019 1 commit
- updating examples and doc · 2397f958
  thomwolf authored Jul 14, 2019
  
  2397f958
05 Jul, 2019 1 commit
- [BIG] name change · 0bab55d5
  thomwolf authored Jul 05, 2019
  
  0bab55d5
03 Jul, 2019 1 commit
- WIP XLM + refactoring · c41f2bad
  thomwolf authored Jul 03, 2019
  
  c41f2bad
11 Jun, 2019 1 commit
- adds the tokenizer + model config to the output · 5c08c8c2
  Oliver Guhr authored Jun 11, 2019
  
  5c08c8c2
09 May, 2019 2 commits
- Division to num_train_optimizer of global_step in lr_this_step is removed. · 00c7fd2b
  burcturkoglu authored May 09, 2019
  
  00c7fd2b
- Division to num_train_optimizer of global_step in lr_this_step is removed. · 5289b4b9
  burcturkoglu authored May 09, 2019
  
  5289b4b9
23 Apr, 2019 1 commit
- fix training schedules in examples to match new API · d94c6b01
  thomwolf authored Apr 23, 2019
  
  d94c6b01
22 Apr, 2019 1 commit
- Made --reduce_memory actually do something in finetune_on_pregenerated · b8e2a9c5
  Matthew Carrigan authored Apr 22, 2019
  
  b8e2a9c5
11 Apr, 2019 1 commit
- finetuning any BERT model - fixes #455 · 4bc4c69a
  thomwolf authored Apr 11, 2019
  
  4bc4c69a
25 Mar, 2019 1 commit
- Corrected the displayed loss when gradient_accumulation_steps > 1 · 8d1d1ffd
  Matthew Carrigan authored Mar 25, 2019
  
  8d1d1ffd
21 Mar, 2019 4 commits
- Added a --reduce_memory option to the training script to keep training · 7d1ae644
  Matthew Carrigan authored Mar 21, 2019
```
data on disc as a memmap rather than in memory
```
  7d1ae644
- Fixed up the notes on a possible future low-memory path · 8a861048
  Matthew Carrigan authored Mar 21, 2019
  
  8a861048
- Reduced memory usage for pregenerating the data a lot by writing it · a8a577ba
  Matthew Carrigan authored Mar 21, 2019
```
out on the fly without shuffling - the Sampler in the finetuning script
will shuffle for us.
```
  a8a577ba
- Reduced memory usage for pregenerating the data a lot by writing it · 0ae59e66
  Matthew Carrigan authored Mar 21, 2019
```
out on the fly without shuffling - the Sampler in the finetuning script
will shuffle for us.
```
  0ae59e66
20 Mar, 2019 5 commits
- PEP8 and formatting cleanups · 7de5c6aa
  Matthew Carrigan authored Mar 20, 2019
  
  7de5c6aa
- Added final TODOs · 1798e98e
  Matthew Carrigan authored Mar 20, 2019
  
  1798e98e
- Fixed embarrassing indentation problem · c64c2fc4
  Matthew Carrigan authored Mar 20, 2019
  
  c64c2fc4
- Fixed logging · 0540d360
  Matthew Carrigan authored Mar 20, 2019
  
  0540d360
- First commit of the new LM finetuning · 976554a4
  Matthew Carrigan authored Mar 20, 2019
  
  976554a4