Commits · d6bbcbc4cf79f0d6da6d4753f4d128ff7e3e42a5 · chenpangpang / transformers

21 Aug, 2019 1 commit
- Added finetuning example to documentation · d6bbcbc4
  Lysandre authored Aug 21, 2019
  
  d6bbcbc4
20 Aug, 2019 2 commits
- Sequence special token handling for BERT and RoBERTa · 2d042274
  Lysandre authored Aug 20, 2019
  
  2d042274
- various fix and clean up on run_lm_finetuning · a690edab
  thomwolf authored Aug 20, 2019
  
  a690edab
19 Aug, 2019 1 commit
- Distributed training + tokenizer agnostic mask token · f94f1c60
  Lysandre authored Aug 19, 2019
  
  f94f1c60
16 Aug, 2019 7 commits
- Simplified data generator + better perplexity calculator · 5652f54a
  Lysandre authored Aug 16, 2019
```
GPT-2 now obtains ~20 perplexity on WikiText-2
```
  5652f54a
- BERT + RoBERTa masking tokens handling + GPU device update. · 71553480
  LysandreJik authored Aug 14, 2019
  
  71553480
- CLM for BERT, beginning of CLM fot RoBERTa; still needs a better masking token mechanism. · 339e556f
  LysandreJik authored Aug 09, 2019
  
  339e556f
- Removed dataset limit · 5c18825a
  LysandreJik authored Aug 06, 2019
  
  5c18825a
- Added GPT to the generative fine-tuning. · 3e3e1454
  LysandreJik authored Aug 06, 2019
  
  3e3e1454
- Language Modeling fine-tuning using GPT-2. · 47975ed5
  LysandreJik authored Aug 06, 2019
  
  47975ed5
- Order of strings in AutoModel/AutoTokenizer updated. · ab052806
  LysandreJik authored Aug 16, 2019
  
  ab052806
15 Aug, 2019 6 commits
- Added RoBERTa example to README · 9d0029e2
  LysandreJik authored Aug 15, 2019
  
  9d0029e2
- Added RoBERTa tokenizer to AutoTokenizer · 83dba0b6
  LysandreJik authored Aug 15, 2019
  
  83dba0b6
- Added RoBERTa to AutoModel/AutoConfig · e24e19ce
  LysandreJik authored Aug 15, 2019
  
  e24e19ce
- Release: 1.1.0 · fe02e45e
  LysandreJik authored Aug 15, 2019
  
  fe02e45e
- Merge pull request #964 from huggingface/RoBERTa · 88efc65b
  Lysandre Debut authored Aug 15, 2019
```
RoBERTa: model conversion, inference, tests 🔥
```
  88efc65b
- Warning for RoBERTa sequences encoded without special tokens. · 83081701
  LysandreJik authored Aug 15, 2019
  
  83081701
14 Aug, 2019 2 commits

Doc · 572dcfd1
LysandreJik authored Aug 14, 2019

572dcfd1

Julien Chaumond authored Aug 14, 2019

cf. https://github.com/huggingface/pytorch-transformers/pull/964#discussion_r313574354

Co-Authored-By: Myle Ott <myleott@fb.com>

c4ef1034

13 Aug, 2019 3 commits
- Added special tokens <pad> and <mask> to RoBERTa. · 39f426be
  LysandreJik authored Aug 13, 2019
  
  39f426be
- [RoBERTa] run_glue: correct pad_token + reorder labels · baf08ca1
  Julien Chaumond authored Aug 13, 2019
  
  baf08ca1
- Fixed error with encoding · 3d87991f
  LysandreJik authored Aug 13, 2019
  
  3d87991f
12 Aug, 2019 5 commits
- Added integration tests for sequence builders. · 634a3172
  LysandreJik authored Aug 12, 2019
  
  634a3172
- Added documentation and changed parameters for special_tokens_sentences_pair. · 22ac004a
  LysandreJik authored Aug 12, 2019
  
  22ac004a
- [RoBERTa] Update `run_glue` for RoBERTa · 912fdff8
  Julien Chaumond authored Aug 12, 2019
  
  912fdff8
- Fixup 9d060314 · b3d83d68
  Julien Chaumond authored Aug 12, 2019
  
  b3d83d68
- Update README.md · a7b4cfe9
  carefree0910 authored Aug 11, 2019
```
I assume that it should test the `re-load` functionality after testing the `save` functionality, however I'm also surprised that nobody points this out after such a long time, so maybe I've misunderstood the purpose. This PR is just in case :)
```
  a7b4cfe9
10 Aug, 2019 2 commits
- Merge branch 'master' of https://github.com/huggingface/pytorch-transformers · aaedfc35
  thomwolf authored Aug 10, 2019
  
  aaedfc35
- fix #993 · c683c3d5
  thomwolf authored Aug 10, 2019
  
  c683c3d5
09 Aug, 2019 3 commits
- Corrected logger.error info · 70607664
  Kevin Trebing authored Aug 09, 2019
```
Signed-off-by: Kevin Trebing <Kevin.Trebing@gmx.net>
```
  70607664
- Roberta tokenization + fixed tests (py3 + py2). · 75d5f98f
  LysandreJik authored Aug 09, 2019
  
  75d5f98f
- Tokenization encode/decode class-based sequence handling · 14e970c2
  LysandreJik authored Aug 09, 2019
  
  14e970c2
08 Aug, 2019 8 commits
- Clarified PreTrainedModel.from_pretrained warning messages in documentation. · 3566d279
  LysandreJik authored Aug 08, 2019
  
  3566d279
- Updated test architecture · fbd746bd
  LysandreJik authored Aug 08, 2019
  
  fbd746bd
- Encode and Decode are back in the superclass. They now handle sentence pairs special tokens. · 6c41a8f5
  LysandreJik authored Aug 08, 2019
  
  6c41a8f5
- [RoBERTa] Re-apply 39d72bcc · e367ac46
  Julien Chaumond authored Aug 08, 2019
```
cc @lysandrejik
```
  e367ac46
- [RoBERTa] RobertaForSequenceClassification + conversion · 9d060314
  Julien Chaumond authored Aug 08, 2019
  
  9d060314
- fix #976 · f2b300df
  LysandreJik authored Aug 08, 2019
  
  f2b300df
- fix #971 · 7df303f5
  LysandreJik authored Aug 08, 2019
  
  7df303f5
- Merge branch 'master' into RoBERTa · d2cc6b10
  LysandreJik authored Aug 08, 2019
  
  d2cc6b10