Commits · 3ae8c8be1e3fc770968cd3fdb3b643e0b166e540 · chenpangpang / transformers

"...lm-evaluation-harness.git" did not exist on "a84c102b3cf2564fd035941beb4f71f3755d5cf4"

29 Apr, 2019 1 commit
- Fix tr_loss rescaling factor using global_step · 87b9ec38
  Mathieu Prouveur authored Apr 29, 2019
  
  87b9ec38
24 Apr, 2019 1 commit
- Update example files so that tr_loss is not affected by args.gradient_accumulation_step · ed8fad73
  Mathieu Prouveur authored Apr 24, 2019
  
  ed8fad73
23 Apr, 2019 1 commit
- fix training schedules in examples to match new API · d94c6b01
  thomwolf authored Apr 23, 2019
  
  d94c6b01
15 Apr, 2019 2 commits
- added best practices for serialization in README and examples · 60ea6c59
  thomwolf authored Apr 15, 2019
  
  60ea6c59
- update example to work with new serialization semantic · 179a2c2f
  thomwolf authored Apr 15, 2019
  
  179a2c2f
21 Mar, 2019 1 commit
- add tqdm to the process of eval · 77944d1b
  Yuqiang Xie authored Mar 21, 2019
```
Maybe better.
```
  77944d1b
15 Mar, 2019 1 commit
- Added missing imports. · 83857ffe
  tseretelitornike authored Mar 15, 2019
  
  83857ffe
06 Mar, 2019 2 commits
- fixing PYTORCH_PRETRAINED_BERT_CACHE use in examples · 994d8660
  thomwolf authored Mar 06, 2019
  
  994d8660
- fix typo - logger info · 5c85fc39
  thomwolf authored Mar 06, 2019
  
  5c85fc39
11 Feb, 2019 1 commit
- clarify and unify model saving logic in examples · eebc8abb
  thomwolf authored Feb 11, 2019
  
  eebc8abb
05 Feb, 2019 2 commits
- python 2 compatibility · 448937c0
  thomwolf authored Feb 06, 2019
  
  448937c0
- more explicit notation: num_train_step => num_train_optimization_steps · 1579c536
  thomwolf authored Feb 05, 2019
  
  1579c536
30 Jan, 2019 1 commit
- make examples consistent, revert error in num_train_steps calculation · 51690699
  Matej Svejda authored Jan 30, 2019
  
  51690699
27 Jan, 2019 1 commit
- fix learning rate/fp16 and warmup problem for all examples · 9c6a48c8
  Matej Svejda authored Jan 27, 2019
  
  9c6a48c8
07 Jan, 2019 1 commit
- remove default when action is store_true in arguments · c9fd3505
  thomwolf authored Jan 07, 2019
  
  c9fd3505
02 Jan, 2019 1 commit
- Adding new pretrained model to the help of the `bert_model` argument. · 186f7534
  Grégory Châtel authored Jan 02, 2019
  
  186f7534
14 Dec, 2018 1 commit
- set encoding to 'utf-8' in calls to open · ae88eb88
  thomwolf authored Dec 14, 2018
  
  ae88eb88
13 Dec, 2018 3 commits
- no fp16 on evaluation · e1eab59a
  thomwolf authored Dec 13, 2018
  
  e1eab59a
- fix reloading model for evaluation in examples · 087798b7
  thomwolf authored Dec 13, 2018
  
  087798b7
- fix swag example for work with apex · 0f544625
  thomwolf authored Dec 13, 2018
  
  0f544625
10 Dec, 2018 1 commit
- Removing the dependency to pandas and using the csv module to load data. · df34f228
  Grégory Châtel authored Dec 10, 2018
  
  df34f228
06 Dec, 2018 7 commits
- Removing old code from copy-paste. · d429c15f
  Grégory Châtel authored Dec 06, 2018
  
  d429c15f
- Finishing the code for the Swag task. · 63c45056
  Grégory Châtel authored Dec 06, 2018
  
  63c45056
- Storing the feature of each choice as a dict for readability. · c45d8ac5
  Grégory Châtel authored Dec 06, 2018
  
  c45d8ac5
- Fixing problems in convert_examples_to_features. · 0812aee2
  Grégory Châtel authored Dec 06, 2018
  
  0812aee2
- convert_examples_to_features code and small improvements. · f2b873e9
  Grégory Châtel authored Dec 06, 2018
  
  f2b873e9
- Adding read_swag_examples to load the dataset. · 83fdbd60
  Grégory Châtel authored Dec 06, 2018
  
  83fdbd60
- SwagExample class. · 7183cded
  Grégory Châtel authored Dec 06, 2018
  
  7183cded