Commits · 80eacb8f16208bcc7ffd8ed5b5750d6fc6854a24 · chenpangpang / transformers · GitLab

13 Dec, 2019 1 commit
- Adding labels mapping for classification models in their respective config. · 80eacb8f
  Morgan Funtowicz authored Dec 13, 2019
  
  80eacb8f
12 Dec, 2019 1 commit
- Expose classification labels mapping (and reverse) in model config. · f69dbecc
  Morgan Funtowicz authored Dec 12, 2019
  
  f69dbecc
11 Dec, 2019 3 commits
- allowing from_pretrained to load from url directly · 6709739a
  thomwolf authored Dec 11, 2019
  
  6709739a
- Add missing DistilBert and Roberta to AutoModelForTokenClassification · c2827379
  Morgan Funtowicz authored Dec 11, 2019
  
  c2827379
- Added supported model to AutoModelTokenClassification · b040bff6
  Morgan Funtowicz authored Dec 11, 2019
  
  b040bff6
10 Dec, 2019 35 commits
- Refactored qa pipeline argument handling + unittests · 9a24e0cf
  Morgan Funtowicz authored Dec 11, 2019
  
  9a24e0cf
- Make sure padding, cls and another non-context tokens cannot appear in the answer. · 63e36007
  Morgan Funtowicz authored Dec 10, 2019
  
  63e36007
- Reuse recent SQuAD refactored data structure inside QA pipelines. · 40a39ab6
  Morgan Funtowicz authored Dec 10, 2019
  
  40a39ab6
- Added QuestionAnsweringPipeline unit tests. · aae74065
  Morgan Funtowicz authored Dec 09, 2019
  
  aae74065
- Remove token_type_ids for compatibility with DistilBert · a7d3794a
  Morgan Funtowicz authored Dec 09, 2019
  
  a7d3794a
- Use attention_mask everywhere. · fe0f552e
  Morgan Funtowicz authored Dec 09, 2019
  
  fe0f552e
- Expose attention_masks and input_lengths arguments to batch_encode_plus · 348e19aa
  Morgan Funtowicz authored Dec 09, 2019
  
  348e19aa
- Enable the Tensorflow backend. · c2407fdd
  Morgan Funtowicz authored Dec 09, 2019
  
  c2407fdd
- Allow hidding frameworks through environment variables (NO_TF, NO_TORCH). · f116cf59
  Morgan Funtowicz authored Dec 09, 2019
  
  f116cf59
- batch_encode_plus generates the encoder_attention_mask to avoid attending over padded values. · 6e61e060
  Morgan Funtowicz authored Dec 09, 2019
  
  6e61e060
- Added batching, topk, chars index and scores. · 02110485
  Morgan Funtowicz authored Dec 06, 2019
  
  02110485
- Added QuestionAnsweringPipeline with batch support. · e1d89cb2
  Morgan Funtowicz authored Dec 06, 2019
  
  e1d89cb2
- Added download command through the cli. · 81babb22
  Morgan Funtowicz authored Dec 03, 2019
```
It allows to predownload models and tokenizers.
```
  81babb22
- updating CLI · 31a3a73e
  thomwolf authored Oct 17, 2019
  
  31a3a73e
- compatibility with sklearn and keras · 7c169756
  thomwolf authored Oct 17, 2019
  
  7c169756
- updating AutoModels and AutoConfiguration - adding pipelines · b81ab431
  thomwolf authored Oct 17, 2019
  
  b81ab431
- add pipeline - train · 2d855973
  thomwolf authored Oct 16, 2019
  
  2d855973
- [WIP] - CLI · 72c36b9e
  thomwolf authored Oct 16, 2019
  
  72c36b9e
- Merge pull request #1984 from huggingface/squad-refactor · e57d00ee
  Thomas Wolf authored Dec 10, 2019
```
[WIP] Squad refactor
```
  e57d00ee
- Merge pull request #2107 from huggingface/encoder-mask-shape · ecabbf6d
  Thomas Wolf authored Dec 10, 2019
```
create encoder attention mask from shape of hidden states
```
  ecabbf6d
- Harmonize `no_cuda` flag with other scripts · 1d189304
  Julien Chaumond authored Dec 10, 2019
  
  1d189304
- clean for release · f7eba090
  Rémi Louf authored Dec 06, 2019
  
  f7eba090
- improve device usage · 2a64107e
  Rémi Louf authored Dec 06, 2019
  
  2a64107e
- add README · c0707a85
  Rémi Louf authored Dec 06, 2019
  
  c0707a85
- integrate ROUGE · ade3cdf5
  Rémi Louf authored Dec 06, 2019
  
  ade3cdf5
- prevent BERT weights from being downloaded twice · 076602bd
  Rémi Louf authored Dec 06, 2019
  
  076602bd
- add py-rouge dependency · 5909f710
  Rémi Louf authored Dec 05, 2019
  
  5909f710
- simplified model and configuration · a1994a71
  Rémi Louf authored Dec 05, 2019
  
  a1994a71
- default output dir to documents dir · 3a9a9f78
  Rémi Louf authored Dec 05, 2019
  
  3a9a9f78
- update the docs · 693606a7
  Rémi Louf authored Dec 05, 2019
  
  693606a7
- remove beam search · c0443df5
  Rémi Louf authored Dec 05, 2019
  
  c0443df5
- give transformers API to BertAbs · 2403a665
  Rémi Louf authored Nov 23, 2019
  
  2403a665
- cast bool tensor to long for pytorch < 1.3 · 4d181999
  Rémi Louf authored Nov 12, 2019
  
  4d181999
- setup training · 9f75565e
  Rémi Louf authored Nov 08, 2019
  
  9f75565e
- tweaks to the BeamSearch API · 4735c2af
  Rémi Louf authored Nov 08, 2019
  
  4735c2af