c-my/transformers
1
0 Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
4,933 Commits 171 Branches 49 Tags
a573777901e662ec2e565be312ffaeedef6effec
Commit Graph
4933 Commits
This Branch
This Branch
All Branches
Author SHA1 Message Date
Sylvain Gugger a573777901 Update repo to isort v5 (#6686) 
* Run new isort

* More changes

* Update CI, CONTRIBUTING and benchmarks
 2020-08-24 11:03:01 -04:00
Teven d329c9b05d Fixed DataCollatorForLanguageModeling not accepting lists of lists (#6685) 
* Fixed DataCollatorForLanguageModeling + PermutationLanguageModeling not accepting lists of lists

* Update data_collator.py

* black was grumpy
 2020-08-24 15:31:44 +02:00
sgugger 0a850d210e Missing commit 2020-08-24 09:23:06 -04:00
Sylvain GuggerandTeven b30879fe0c Don't reset the dataset type + plug for rm unused columns (#6683) 
* Don't reset the type of the dataset

* Formatting

* Update trainer.py

Co-authored-by: Teven <teven.lescao@gmail.com>
 2020-08-24 09:22:03 -04:00
Jared T Nielsen 1a779ad7ec Specify config filename (#6626) 2020-08-24 07:27:58 -04:00
Sagor Sarker a622705ef3 added multiple model_cards for below models (#6666) 
* Create README.md

* Update README.md

* Create README.md

* Update README.md

* added multiple codeswitch model
 2020-08-24 05:08:32 -04:00
Patrick von Platen 16e38940bd Add Roberta2Roberta shared 2020-08-23 17:02:22 +02:00
Sam Shleifer f230a64094 new paper bibtex (#6656) 2020-08-23 10:03:41 -04:00
Patrick von Platen f235ee2164 Add Roberta2Roberta model card 2020-08-23 10:01:58 +02:00
Sagor Sarker 068df740bd added model_card for model codeswitch-hineng-lid-lince and codeswitch-spaeng-lid-lince (#6663) 
* Create README.md

* Update README.md

* Create README.md

* Update README.md
 2020-08-22 12:13:21 -04:00
Patrick von Platen 97bb2497ab Correct bug in bert2bert-cnn_dailymail 
Model was trained with the wrong tokenizer. Retrained with correct tokenizer - thanks for spotting @lhoestq !
 2020-08-22 13:44:20 +02:00
Manuel Romero 0f94151dc7 Add model card for electricidad-base-generator (#6650) 
I works like a charm!
Look at the output of the example code!
 2020-08-21 14:18:15 -04:00
Suraj Patil cbda72932c [Doc model summary] add MBart model summary (#6649) 2020-08-21 13:42:59 -04:00
Patrick von Platen 9e8c494da7 Add T5-11B disclaimer 
@julien-c
 2020-08-21 18:11:18 +02:00
Patrick von Platen a4db4e3032 [Docs model summaries] Add pegasus to docs (#6640) 
* add pegasus to docs

* Update docs/source/model_summary.rst
 2020-08-21 16:22:10 +02:00
Suraj Patil d0e42a7bed CamembertForCausalLM (#6577) 
* added CamembertForCausalLM

* add in __init__ and auto model

* style

* doc
 2020-08-21 13:52:54 +02:00
josephrocca bdf7e5de92 Remove accidental comment (#6629) 2020-08-21 05:07:32 -04:00
Manuel Romero efc7460553 model card for Spanish electra base (#6633) 2020-08-21 05:04:29 -04:00
Morgan Funtowicz b105f2c6b3 Update ONNX doc to match the removal of --optimize argument. 
Signed-off-by: Morgan Funtowicz <funtowiczmo@gmail.com>
 2020-08-21 10:37:09 +02:00
Sylvain Gugger e5f452275b Trainer automatically drops unused columns in nlp datasets (#6449) 
* Add a classmethod to easily build a Trainer from nlp dataset and metric

* Fix docstrings

* Split train/eval

* Formatting

* Log dropped columns + docs

* Authorize callable activations

* Poc for auto activation

* Be framework-agnostic

* Formatting

* Remove class method

* Remove unnecessary code
 2020-08-20 16:29:14 -04:00
Sam Shleifer 5bf4465e6c Regression test for pegasus bugfix (#6606) 2020-08-20 15:34:43 -04:00
sgugger 86c07e634f One last threshold to raise 2020-08-20 14:23:09 -04:00
Sylvain Gugger e8af90c052 Move threshold up for flaky test with Electra (#6622) 
* Move threshold up for flaky test with Electra

* Update above as well
 2020-08-20 13:59:40 -04:00
Ivan DolgovandKevin Canwen Xu 953958372a XLNet Bug when training with apex 16-bit precision (#6567) 
* xlnet fp16 bug fix

* comment cast added

* Update modeling_xlnet.py

Co-authored-by: Kevin Canwen Xu <canwenxu@126.com>
 2020-08-21 01:34:23 +08:00
Patrick von Platen 505f2d749e [Tests] fix attention masks in Tests (#6621) 
* fix distilbert

* fix typo
 2020-08-20 13:23:47 -04:00
Denisa Roberts c9454507cf Add tests for Reformer tokenizer (#6485) 2020-08-20 18:58:44 +02:00
Joe Davison f9d280a959 TFTrainer dataset doc & fix evaluation bug (#6618) 
* TFTrainer dataset doc & fix evaluation bug

discussed in #6551

* add docstring to test/eval datasets
 2020-08-20 12:11:36 -04:00
Sylvain Gugger 573bdb0a5d Add tests to Trainer (#6605) 
* Add tests to Trainer

* Test if removing long breaks everything

* Remove ugly hack

* Fix distributed test

* Use float for number of epochs
 2020-08-20 11:13:50 -04:00
Joe Davison 039d8d65fc add intro to nlp lib & dataset links to custom datasets tutorial (#6583) 
* add intro to nlp lib + links

* unique links...
 2020-08-20 10:32:51 -04:00
sgugger b3e54698dd Fix CI 2020-08-20 08:34:02 -04:00
Prajjwal Bhargava 33bf426498 removed redundant arg in prepare_inputs (#6614) 
* removed redundant arg in prepare_inputs

* made same change in prediction_loop
 2020-08-20 08:23:35 -04:00
Romain Rigaux cabfdfafc0 Docs copy button misses ... prefixed code (#6518) 
Tested in a local build of the docs.

e.g. Just above https://huggingface.co/transformers/task_summary.html#causal-language-modeling

Copy will copy the full code, e.g.

for token in top_5_tokens:
     print(sequence.replace(tokenizer.mask_token, tokenizer.decode([token])))

Instead of currently only:

for token in top_5_tokens:


>>> for token in top_5_tokens:
...     print(sequence.replace(tokenizer.mask_token, tokenizer.decode([token])))
Distilled models are smaller than the models they mimic. Using them instead of the large versions would help reduce our carbon footprint.
Distilled models are smaller than the models they mimic. Using them instead of the large versions would help increase our carbon footprint.
Distilled models are smaller than the models they mimic. Using them instead of the large versions would help decrease our carbon footprint.
Distilled models are smaller than the models they mimic. Using them instead of the large versions would help offset our carbon footprint.
Distilled models are smaller than the models they mimic. Using them instead of the large versions would help improve our carbon footprint.

Docs for the option fix:
https://sphinx-copybutton.readthedocs.io/en/latest/
 2020-08-20 17:35:06 +08:00
Stas Bekman 61b5ee11e3 lighter 'make test' (#6512) 2020-08-20 17:24:25 +08:00
Siddharth Jain 3c3c46f563 Typo fix in 04-onnx-export (#6595) 2020-08-20 16:17:16 +08:00
Oren Amsalem 93c5c9a528 [cleanup] remove confusing newline (#6603) 2020-08-20 00:33:36 -04:00
Sylvain Gugger 18ca0e9140 Fix #6575 (#6596) 2020-08-19 13:04:33 -04:00
Suraj Patil 7581884dee [BartTokenizerFast] add prepare_seq2seq_batch (#6543) 2020-08-19 10:37:48 -04:00
Patrick von Platen 8bcceaceff fix model outputs test (#6593) 2020-08-19 16:18:51 +02:00
Sam Shleifer 9a86321b11 tf generation utils: remove unused kwargs (#6591) 2020-08-19 09:37:45 -04:00
Pradhy729andpatrickvonplaten 2a7402cbd3 Feed forward chunking others (#6365) 
* Feed forward chunking for Distilbert & Albert

* Added ff chunking for many other models

* Change model signature

* Added chunking for XLM

* Cleaned up by removing some variables.

* remove test_chunking flag

Co-authored-by: patrickvonplaten <patrick.v.platen@gmail.com>
 2020-08-19 14:31:10 +02:00
Patrick von PlatenandSylvain Gugger fe0b85e77a [EncoderDecoder] Add functionality to tie encoder decoder weights (#6538) 
* start adding tie encoder to decoder functionality

* finish model tying

* make style

* Apply suggestions from code review

* fix t5 list including cross attention

* apply sams suggestions

* Update src/transformers/modeling_encoder_decoder.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* add max depth break point

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2020-08-19 14:23:45 +02:00
Sam Shleifer ab42d74850 Fix bart base test (#6587) 2020-08-18 21:28:10 -04:00
Sam ShleiferandSylvain Gugger 1529bf9680 add BartConfig.force_bos_token_to_be_generated (#6526) 
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2020-08-18 19:15:50 -04:00
Patrick von Platen 974bb4af26 [Model card] Bert2GPT2 EncoderDecoder model (#6569) 
* Bert2GPT2 EncoderDecoder model

* Update README.md
 2020-08-18 19:28:17 +02:00
Suraj Patil 6f972e1423 update xnli-mt url (#6580) 2020-08-18 13:10:47 -04:00
Suraj Patil fb6844aff5 [Pegasus Doc] minor typo (#6579) 
Minor typo correction
@sshleifer
 2020-08-18 12:47:47 -04:00
Manuel Romero aaab9ab187 Create README.md (#6556) 2020-08-18 12:43:20 -04:00
Manuel Romero 1dfce0f08a Create README.md (#6557) 2020-08-18 12:42:14 -04:00
Romain Rigaux 7516bcf273 [docs] Fix number of 'ug' occurrences in tokenizer_summary (#6574) 2020-08-18 10:23:25 -04:00
Romain Rigaux 5a5af22ed5 [docs] Fix wrong newline in the middle of a paragraph (#6573) 2020-08-18 10:22:43 -04:00