c-my/transformers
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
1,371 Commits 171 Branches 49 Tags
0ecfd17f49e81cade5ddd7321480a5a3492dd36d
Commit Graph
75 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
LysandreJik f753d4e32b Removed typings for Python 2 2019-08-28 10:15:02 -04:00
LysandreJik 75bc2a03cc Updated article link 2019-08-28 10:05:15 -04:00
LysandreJik 1dc43e56c9 Documentation additions 2019-08-28 09:37:27 -04:00
LysandreJik c513415b19 Dilbert tests from CommonTests 2019-08-27 23:59:00 -04:00
LysandreJik 778a263f09 GilBert added to AutoModels 2019-08-27 23:14:00 -04:00
LysandreJik e08c01aa1a fix #1102 2019-08-26 18:13:06 -04:00
LysandreJik 529a16dec6 Generic encoding implementation. 2019-08-26 15:00:43 -04:00
Lysandre 2f9397139d Added GPT-2 LARGE to Pre-trained Models documentation 2019-08-21 11:29:37 -04:00
Lysandre d6bbcbc4cf Added finetuning example to documentation 2019-08-21 11:22:05 -04:00
Lysandre 2d042274ac Sequence special token handling for BERT and RoBERTa 2019-08-20 14:15:28 -04:00
Lysandre f94f1c6016 Distributed training + tokenizer agnostic mask token 2019-08-19 14:58:50 -04:00
Lysandre c589862b78 Doc: loading from config alone does not load the model weights 2019-08-19 10:17:47 -04:00
Lysandre 5652f54ac2 Simplified data generator + better perplexity calculator 
GPT-2 now obtains ~20 perplexity on WikiText-2
 2019-08-16 13:49:56 -04:00
LysandreJik 7e7fc53da5 Fixing run_glue example with RoBERTa 2019-08-16 11:53:10 -04:00
LysandreJik 715534800a BERT + RoBERTa masking tokens handling + GPU device update. 2019-08-16 10:10:21 -04:00
LysandreJik 339e556feb CLM for BERT, beginning of CLM fot RoBERTa; still needs a better masking token mechanism. 2019-08-16 10:10:20 -04:00
LysandreJik 5c18825a18 Removed dataset limit 2019-08-16 10:10:20 -04:00
LysandreJik 3e3e145497 Added GPT to the generative fine-tuning. 2019-08-16 10:10:20 -04:00
LysandreJik 47975ed53e Language Modeling fine-tuning using GPT-2. 2019-08-16 10:10:20 -04:00
LysandreJik ab05280666 Order of strings in AutoModel/AutoTokenizer updated. 2019-08-16 09:53:26 -04:00
LysandreJik 9d0029e215 Added RoBERTa example to README 2019-08-15 17:17:35 -04:00
LysandreJik 83dba0b67b Added RoBERTa tokenizer to AutoTokenizer 2019-08-15 17:07:07 -04:00
LysandreJik e24e19ce3b Added RoBERTa to AutoModel/AutoConfig 2019-08-15 14:02:11 -04:00
LysandreJik fe02e45e48 Release: 1.1.0 2019-08-15 11:15:08 -04:00
Lysandre Debut 88efc65bac Merge pull request #964 from huggingface/RoBERTa 
RoBERTa: model conversion, inference, tests 🔥
 2019-08-15 11:11:10 -04:00
LysandreJik 8308170156 Warning for RoBERTa sequences encoded without special tokens. 2019-08-15 10:29:04 -04:00
LysandreJik 572dcfd1db Doc 2019-08-14 14:56:14 -04:00
LysandreJik 39f426be65 Added special tokens <pad> and <mask> to RoBERTa. 2019-08-13 15:19:50 -04:00
LysandreJik 3d87991f60 Fixed error with encoding 2019-08-13 12:00:24 -04:00
LysandreJik 634a3172d8 Added integration tests for sequence builders. 2019-08-12 15:14:15 -04:00
LysandreJik 22ac004a7c Added documentation and changed parameters for special_tokens_sentences_pair. 2019-08-12 15:13:53 -04:00
LysandreJik 75d5f98fd2 Roberta tokenization + fixed tests (py3 + py2). 2019-08-09 15:02:13 -04:00
LysandreJik 14e970c271 Tokenization encode/decode class-based sequence handling 2019-08-09 15:01:38 -04:00
LysandreJik 3566d27919 Clarified PreTrainedModel.from_pretrained warning messages in documentation. 2019-08-08 19:04:34 -04:00
LysandreJik fbd746bd06 Updated test architecture 2019-08-08 18:21:34 -04:00
LysandreJik 6c41a8f5dc Encode and Decode are back in the superclass. They now handle sentence pairs special tokens. 2019-08-08 18:20:32 -04:00
LysandreJik f2b300df6b fix #976 2019-08-08 10:38:57 -04:00
LysandreJik 7df303f5ad fix #971 2019-08-08 10:36:26 -04:00
LysandreJik d2cc6b101e Merge branch 'master' into RoBERTa 2019-08-08 09:42:05 -04:00
LysandreJik 39d72bcc7b Fixed the RoBERTa checkpoint conversion script according to the LM head refactoring. 2019-08-07 14:21:57 -04:00
LysandreJik 770043eea2 Sentence-pair tasks handling. Using common tests on RoBERTa. Forced push to fix indentation. 2019-08-07 12:53:19 -04:00
Lysandre Debut 6f05ad72b4 Merge pull request #791 from huggingface/doc 
RestructuredText table for pretrained models.
 2019-08-05 10:18:00 -04:00
LysandreJik 9d381e7be9 Fixed incorrect links in the PretrainedModel 2019-07-17 09:25:38 -04:00
LysandreJik 117ed92992 RestructuredText table for pretrained models. 2019-07-16 11:58:47 -04:00
LysandreJik 3fbceed8d2 Fix layer reference loss + previous attempted fix 2019-07-11 22:29:55 -04:00
LysandreJik 6c2ee16c04 Test suite testing the tie_weights function as well as the resize_token_embeddings function. 
Patched an issue relating to the tied weights I had introduced with the TorchScript addition.
Byte order mark management in TSV glue reading.
 2019-07-11 22:09:16 -04:00
LysandreJik 3821ecbf4a Byte order mark management in TSV glue reading. 2019-07-11 20:16:28 -04:00
LysandreJik e3fb4310d6 From pretrained correct initialization. Unknown token handling for gpt2. 2019-07-11 18:44:29 -04:00
LysandreJik 50e62a4cb4 fix gpt/gpt-2 from pretrained 2019-07-11 16:50:21 -04:00
LysandreJik 6491575fd5 Added TorchScript disclaimer. CSS modifications. 2019-07-11 12:38:21 -04:00
LysandreJik 7fdbc47822 Added the two CLM XLM pretrained checkpoints. 
Fixed file extensions for config/vocab/merges of XLM models.
 2019-07-10 19:37:24 -04:00
LysandreJik dee3e45b93 Fixed XLM weights conversion script. Added 5 new checkpoints for XLM. 2019-07-10 19:04:21 -04:00
LysandreJik c82b74b996 Fixed Sphinx errors and warnings 2019-07-10 15:30:19 -04:00
LysandreJik 5288913bdd All TODOs to be checked by Thom have been added. 2019-07-10 15:16:40 -04:00
LysandreJik f773faa258 Fixed all links. Removed TPU. Changed CLI to Converting TF models. Many minor formatting adjustments. Added "TODO Lysandre filled" where necessary. 2019-07-10 14:45:56 -04:00
LysandreJik 3f56ad5aff Updated CircleCI's config.yml to use a large resource class. 2019-07-09 18:50:59 -04:00
LysandreJik c4bab2dc85 Added footer with social links. 2019-07-09 18:03:01 -04:00
LysandreJik 331db8cc02 Added viewcode plugin for source code visualization within the static website. 2019-07-09 17:01:56 -04:00
LysandreJik 83fb311ef7 Patched warnings + Refactored XLNet's Docstrings 2019-07-09 16:38:30 -04:00
LysandreJik 8fe2c9d98e Refactored Docstrings of BERT, GPT2, GPT, TransfoXL, XLM and XLNet. 2019-07-09 15:55:31 -04:00
LysandreJik e0e5c7faf5 Added requirements.txt file. 2019-07-09 10:16:09 -04:00
LysandreJik 269e73b601 Adding example detailing how to add a new file to the documentation + adding fonts. 2019-07-09 10:11:29 -04:00
LysandreJik 6847e30e1c New page detailing the use of TorchScript. 2019-07-08 17:34:24 -04:00
LysandreJik ab30651802 Hugging Face theme. 2019-07-08 16:05:26 -04:00
LysandreJik a60ae1a505 Docstrings best practice shown in the BERT documentation. 2019-07-08 11:50:32 -04:00
LysandreJik 64fd986376 Tokenizers and Config classes are referenced. 2019-07-05 17:44:59 -04:00
LysandreJik df759114c9 Single file documentation for each model, accompanied by the Documentation overview. 2019-07-05 17:35:26 -04:00
LysandreJik 03de9686a7 Initial folder structure for the documentation. A draft of documentation change has been made in the BertModel class. 2019-07-05 17:11:13 -04:00
LysandreJik b43b130f35 TorchScript flag in config; Tied weights when not running TorchScript; tuple concatenation clean-up. 2019-07-03 16:21:17 -04:00
LysandreJik 4703148f0c TransformerXL can't be exported to TorchScript because of control-flow. Exception added to tests. 2019-07-03 14:50:23 -04:00
LysandreJik 971c24687f XLNET can be exported to TorchScript 2019-07-03 11:03:09 -04:00
LysandreJik be54b16960 GPT can be exported to TorchScript 2019-07-02 18:09:45 -04:00
LysandreJik d8e83de792 GPT2 can be exported to TorchScript 2019-07-02 18:01:09 -04:00
LysandreJik e891bb43d5 BERT can be exported to TorchScript 2019-07-02 17:23:18 -04:00
LysandreJik 6ce1ee04fc TorchScript testing with output_attentions and output_hidden_state 2019-07-02 17:22:59 -04:00