Commit Graph

4087 Commits

Author SHA1 Message Date
Manuel Romero
bac1cc4dc1
Remove MD emojis (#4602) 2020-05-26 16:38:39 -04:00
Patrick von Platen
003c477129
[GPT2, CTRL] Allow input of input_ids and past of variable length (#4581)
* revert convenience  method

* clean docs a bit
2020-05-26 19:43:58 +02:00
ohmeow
5ddd8d6531
Add BART fine-tuning summarization community notebook (#4539)
* adding BART summarization how-to community notebook

* Update notebooks/README.md

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
2020-05-26 16:43:41 +02:00
Bram Vanroy
8cc6807e89
Make transformers-cli cross-platform (#4131)
* make transformers-cli cross-platform

Using "scripts" is a useful option in setup.py particularly when you want to get access to non-python scripts. However, in this case we want to have an entry point into some of our own Python scripts. To do this in a concise, cross-platfom way, we can use entry_points.console_scripts. This change is necessary to provide the CLI on different platforms, which "scripts" does not ensure. Usage remains the same, but the "transformers-cli" script has to be moved (be part of the library) and renamed (underscore + extension)

* make style & quality
2020-05-26 10:00:51 -04:00
Patrick von Platen
c589eae2b8
[Longformer For Question Answering] Conversion script, doc, small fixes (#4593)
* add new longformer for question answering model

* add new config as well

* fix links

* fix links part 2
2020-05-26 14:58:47 +02:00
ZhuBaohe
a163c9ca5b
[T5] Fix Cross Attention position bias (#4499)
* fix

* fix1
2020-05-26 08:57:24 -04:00
ZhuBaohe
1d69028989
fix (#4410) 2020-05-26 08:51:28 -04:00
Sam Shleifer
b86e42e0ac
[ci] fix 3 remaining slow GPU failures (#4584) 2020-05-25 19:20:50 -04:00
Julien Chaumond
365d452d4d
[ci] Slow GPU tests run daily (#4465) 2020-05-25 17:28:02 -04:00
Patrick von Platen
3e3e552125
[Reformer] fix reformer num buckets (#4564)
* fix reformer num buckets

* fix

* adapt docs

* set num buckets in config
2020-05-25 16:04:45 -04:00
Elman Mansimov
3dea40b858
fixing tokenization of extra_id symbols in T5Tokenizer. Related to issue 4021 (#4353) 2020-05-25 16:04:30 -04:00
Suraj Patil
5139733623
LongformerTokenizerFast (#4547) 2020-05-25 16:03:55 -04:00
Oliver Guhr
c9c385c522
Updated the link to the paper (#4570)
I looks like the conference has changed the link to the paper.
2020-05-25 15:29:50 -04:00
Sho Arora
adab7f8332
Add nn.Module as superclass (#4533) 2020-05-25 15:29:33 -04:00
Manuel Romero
8f7c1c7672
Create model card (#4578) 2020-05-25 15:28:30 -04:00
Ali Safaya
4c6b218056
Update README.md (#4556) 2020-05-25 15:12:23 -04:00
Antonis Maronikolakis
50d1ce411f
add DistilBERT to supported models (#4558) 2020-05-25 14:50:45 -04:00
Suraj Patil
03d8527de0
Longformer for question answering (#4500)
* added LongformerForQuestionAnswering

* add LongformerForQuestionAnswering

* fix import for LongformerForMaskedLM

* add LongformerForQuestionAnswering

* hardcoded sep_token_id

* compute attention_mask if not provided

* combine global_attention_mask with attention_mask when provided

* update example in  docstring

* add assert error messages, better attention combine

* add test for longformerForQuestionAnswering

* typo

* cast gloabl_attention_mask to long

* make style

* Update src/transformers/configuration_longformer.py

* Update src/transformers/configuration_longformer.py

* fix the code quality

* Merge branch 'longformer-for-question-answering' of https://github.com/patil-suraj/transformers into longformer-for-question-answering

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
2020-05-25 18:43:36 +02:00
Bharat Raghunathan
a34a9896ac
DOC: Fix typos in modeling_auto (#4534) 2020-05-23 09:40:59 -04:00
Bijay Gurung
e19b978151
Add Type Hints to modeling_utils.py Closes #3911 (#3948)
* Add Type Hints to modeling_utils.py Closes #3911

Add Type Hints to methods in `modeling_utils.py`

Note: The coverage isn't 100%. Mostly skipped internal methods.

* Reformat according to `black` and `isort`

* Use typing.Iterable instead of Sequence

* Parameterize Iterable by its generic type

* Use typing.Optional when None is the default value

* Adhere to style guideline

* Update src/transformers/modeling_utils.py

* Update src/transformers/modeling_utils.py

Co-authored-by: Julien Chaumond <chaumond@gmail.com>
2020-05-22 19:10:22 -04:00
Funtowicz Morgan
996f393a86
Warn the user about max_len being on the path to be deprecated. (#4528)
* Warn the user about max_len being on the path to be deprecated.

* Ensure better backward compatibility when max_len is provided to a tokenizer.

* Make sure to override the parameter and not the actual instance value.

* Format & quality
2020-05-22 18:08:30 -04:00
Patrick von Platen
0f6969b7e9
Better github link for Reformer Colab Notebook 2020-05-22 23:51:36 +02:00
Sam Shleifer
ab44630db2
[Summarization Pipeline]: Fix default tokenizer (#4506)
* Fix pipelines defaults bug

* one liner

* style
2020-05-22 17:49:45 -04:00
Julien Chaumond
2c1ebb8b50 Re-apply #4446 + add packaging dependency
As discussed w/ @lysandrejik

packaging is maintained by PyPA (the Python Packaging Authority), and should be lightweight and stable
2020-05-22 17:29:03 -04:00
Lysandre
e6aeb0d3e8 Style 2020-05-22 17:20:03 -04:00
Alexander Measure
95a26fcf2d
link to paper was broken (#4526)
changed from https://https://arxiv.org/abs/2001.04451.pdf to https://arxiv.org/abs/2001.04451.pdf
2020-05-22 15:17:09 -04:00
HUSEIN ZOLKEPLI
89d795f180
Added huseinzol05/t5-small-bahasa-cased README.md (#4522) 2020-05-22 15:04:06 -04:00
Anthony MOI
35df911485
Fix convert_token_type_ids_from_sequences for fast tokenizers (#4503) 2020-05-22 12:45:10 -04:00
Julien Chaumond
f7677e1623
[model_cards] bart-large-cnn
cc @sshleifer
2020-05-22 12:20:54 -04:00
Patrick von Platen
12e6afe900
Add Reformer colab to community noteboos 2020-05-22 17:03:34 +02:00
Lysandre
ef22ba4836 Re-pin versions 2020-05-22 11:03:07 -04:00
Lysandre
10d72390c0 Revert #4446 Since it introduces a new dependency 2020-05-22 10:49:45 -04:00
Lysandre
e0db6bbd65 Release: v2.10.0 2020-05-22 10:37:44 -04:00
Frankie Liuzzi
bd6e301832
added functionality for electra classification head (#4257)
* added functionality for electra classification head

* unneeded dropout

* Test ELECTRA for sequence classification

* Style

Co-authored-by: Frankie <frankie@frase.io>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>
2020-05-22 09:48:21 -04:00
Lysandre
a086527727 Unused Union should not be imported 2020-05-21 09:42:47 -04:00
Lysandre Debut
9d2ce253de
TPU hangs when saving optimizer/scheduler (#4467)
* TPU hangs when saving optimizer/scheduler

* Style

* ParallelLoader is not a DataLoader

* Style

* Addressing @julien-c's comments
2020-05-21 09:18:27 -04:00
Zhangyx
49296533ca
Adds predict stage for glue tasks, and generate result files which can be submitted to gluebenchmark.com (#4463)
* Adds predict stage for glue tasks, and generate result files which could be submitted to gluebenchmark.com website.

* Use Split enum + always output the label name

Co-authored-by: Julien Chaumond <chaumond@gmail.com>
2020-05-21 09:17:44 -04:00
Tobias Lee
271bedb485
[examples] fix no grad in second pruning in run_bertology (#4479)
* fix no grad in second pruning and typo

* fix prune heads attention mismatch problem

* fix

* fix

* fix

* run make style

* run make style
2020-05-21 09:17:03 -04:00
Julien Chaumond
865d4d595e [ci] Close #4481 2020-05-20 18:27:42 -04:00
Julien Chaumond
a3af8e86cb Update test_trainer_distributed.py 2020-05-20 18:26:51 -04:00
Cola
eacea530c1
🚨 Remove warning of deprecation (#4477)
Remove warning of deprecated overload of addcdiv_

Fix #4451
2020-05-20 16:48:29 -04:00
Julien Plu
fa2fbed3e5
Better None gradients handling in TF Trainer (#4469)
* Better None gradients handling

* Apply Style

* Apply Style
2020-05-20 16:46:21 -04:00
Oliver Åstrand
e708bb75bf
Correct TF formatting to exclude LayerNorms from weight decay (#4448)
* Exclude LayerNorms from weight decay

* Include both formats of layer norm
2020-05-20 16:45:59 -04:00
Rens
49c06132df
pass on tokenizer to pipeline (#4489) 2020-05-20 22:23:21 +02:00
Nathan Cooper
cacb654c7f
Add Fine-tune DialoGPT on new datasets notebook (#4473) 2020-05-20 16:17:52 -04:00
Timo Moeller
30a09f3827
Adjust german bert model card, add new model card (#4488) 2020-05-20 16:08:29 -04:00
Lysandre Debut
14cb5b35fa
Fix slow gpu tests lysandre (#4487)
* There is one missing key in BERT

* Correct device for CamemBERT model

* RoBERTa tokenization adding prefix space

* Style
2020-05-20 11:59:45 -04:00
Manuel Romero
6dc52c78d8
Create README.md (#4482) 2020-05-20 09:45:50 -04:00
Manuel Romero
ed5456daf4
Model card for RuPERTa-base fine-tuned for NER (#4466) 2020-05-20 09:45:24 -04:00
Oleksandr Bushkovskyi
c76450e20c
Model card for Tereveni-AI/gpt2-124M-uk-fiction (#4470)
Create model card for "Tereveni-AI/gpt2-124M-uk-fiction" model
2020-05-20 09:44:26 -04:00