Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[s2s] fix finetune.py to adjust for #8530 changes #8612

Merged
merged 1 commit into from
Nov 18, 2020

Conversation

stas00
Copy link
Contributor

@stas00 stas00 commented Nov 18, 2020

Making the script work again after #8530 change.

As mentioned in #8612 .logits doesn't seem to work with apex/PL. No idea why.
So distillation.py is probably a problem too since it uses .logits. I haven't checked it.

I don't think any tests use apex.

RUN_SLOW=1 pyt -sv test_bash_script.py::TestMbartCc25Enro::test_train_mbart_cc25_enro_script

tests finetune with fp16, but it runs it in a different way.

@sgugger, @LysandreJik

@stas00
Copy link
Contributor Author

stas00 commented Nov 18, 2020

@sgugger if you want to see why .logits isn't working you can try:

cd examples/seq2seq
PYTHONPATH="../../src" CUDA_VISIBLE_DEVICES=0 python finetune.py --learning_rate 3e-5 --gpus 1 --do_train --val_check_interval 1 --num_train_epochs 1 --freeze_encoder --freeze_embeds --data_dir cnn_dm --max_target_length 142 --val_max_target_length 142 --train_batch_size 1 --eval_batch_size 1 --gradient_accumulation_steps 1 --model_name_or_path sshleifer/student_cnn_12_6 --tokenizer_name facebook/bart-large --output_dir distilbart-cnn-12-6 --overwrite_output_dir --num_sanity_val_steps 0 --fp16 --eval_beams 1 --amp_backend=apex --n_train 1 --n_val 1 --warmup_steps 1

If I pprint(outputs) I get a dict.

If you need cnn_dm:

wget https://cdn-datasets.huggingface.co/summarization/cnn_dm_v2.tgz
tar -xzvf cnn_dm_v2.tgz  # empty lines removed
mv cnn_cln cnn_dm

@sgugger
Copy link
Collaborator

sgugger commented Nov 18, 2020

I'll investigate but at first glance it looks like it's PyTorch Lightning that is messing with the new output types, not PyTorch. We will add a caveat in #8530 if that's the case.

@sgugger
Copy link
Collaborator

sgugger commented Nov 18, 2020

So this is not linked to mixed precision directly but purely something in PL. Somewhere when dealing with mixed precision, it looks like there is something happening is the output is an instance of a dict, and they don't respect the class of the dict.

@sgugger
Copy link
Collaborator

sgugger commented Nov 18, 2020

More investigation shows the problem is actually coming from apex, when using opt_level="O2" (the default level is fine). For reference, the following code shows apex-converted model lose their output type:

from apex import amp
from transformers import BertModel, BertTokenizer

tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
inputs = tokenizer("Hi, my name is Sylvain!", return_tensors="pt")
inputs = {k: v.cuda() for k, v in inputs.items()}

model = BertModel.from_pretrained("bert-base-uncased")
model = model.cuda()
optimizer = torch.optim.SGD(model.parameters(), lr=0.1)

model, optimizer = amp.initialize(model, optimizer, opt_level="O2")
outputs = model(**inputs)
type(outputs)

should return BaseModelOutputWithPoolingAndCrossAttentions but returns dict.

@sgugger sgugger merged commit cdf1b7a into huggingface:master Nov 18, 2020
@sgugger
Copy link
Collaborator

sgugger commented Nov 18, 2020

(The fix is the right one, so while the discussion may continue, I'm merging this PR.)

@stas00
Copy link
Contributor Author

stas00 commented Nov 18, 2020

Thank you for investigating this, @sgugger.

As expected distillation.py fails too with apex / default level 2

python distillation.py   --teacher facebook/bart-large-xsum --data_dir xsum   --tokenizer_name facebook/bart-large-xsum   --student_decoder_layers 6 --student_encoder_layers 12   --freeze_encoder --freeze_embeds   --learning_rate=3e-4   --do_train   --do_predict   --fp16   --val_check_interval 0.1 --n_val 1 --eval_beams 1 --length_penalty=0.5   --max_target_length=60 --val_max_target_length=60 --test_max_target_length=100   --model_name_or_path IGNORED   --alpha_hid=3.   --train_batch_size=16 --eval_batch_size=16 --gradient_accumulation_steps=2   --sortish_sampler   --num_train_epochs=6   --warmup_steps 1   --output_dir distilbart_xsum_12_6     --amp_backend=apex     --n_train 1     --gpus 1
[...]
  File "distillation.py", line 157, in _step
    lm_logits = student_outputs.logits
AttributeError: 'dict' object has no attribute 'logits'

there are multiple situations like this in this program.

what's the best way to proceed, @sgugger? switch to dict keys for now and report this issue to apex? (which is not being watched - even PRs aren't being merged/attended to).

@stas00
Copy link
Contributor Author

stas00 commented Nov 18, 2020

@ptrblck, can #8612 (comment) be fixed in apex, or is it a lost cause (as I noticed apex is not actively supported anymore). Please let me know if we should ask someone else?

@sgugger
Copy link
Collaborator

sgugger commented Nov 18, 2020

I think we should fix the scripts by accessing elements in the outputs with their keys.

@stas00
Copy link
Contributor Author

stas00 commented Nov 18, 2020

I will do that. Thank you for the feedback, @sgugger

bdalal added a commit to lucidworks/transformers that referenced this pull request Nov 19, 2020
* Rename add_start_docstrings_to_callable (huggingface#8120)

* Update CI cache (huggingface#8126)

* Upgrade PyTorch Lightning to 1.0.2 (huggingface#7852)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Fix typo in `AutoModelForMaskedLM` docs (huggingface#8129)

* [s2s test] cleanup (huggingface#8131)

* add tags (huggingface#8147)

* Add model_cards for DynaBERT (huggingface#8012)

* Update README.md

* Add dynabert_overview.png

* Update README.md

* Create README.md

* Add dynabert_overview.png

* Update README.md

* Update README.md

* Delete dynabert_overview.png

* Update README.md

* Delete dynabert_overview.png

* Update README.md

* Create README.md (huggingface#8015)

* Create README.md (huggingface#8017)

* Model Card for Gujarati-XLM-R-Base (huggingface#8038)

* Add model card for Gujarati-XLM-R-Base

* Update README.md

Add the model card for the Gujarati-XLM-R-Base.

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Add two model_cards: ethanyt/guwenbert-base and ethanyt/guwenbert-large (huggingface#8041)

* Create README.md (huggingface#8075)

* Create README.md

* Update model_cards/gurkan08/bert-turkish-text-classification/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8088)

* Create README.md

* metadata

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8089)

* Add model_cards (huggingface#7969)

* add readme

* add readmes

* Add metadata

* Update README.md (huggingface#8090)

* Update widget examples. (huggingface#8149)

Co-authored-by: yantan <yantan@effyic.com>

* Fix doc errors and typos across the board (huggingface#8139)

* Fix doc errors and typos across the board

* Fix a typo

* Fix the CI

* Fix more typos

* Fix CI

* More fixes

* Fix CI

* More fixes

* More fixes

* Document tokenizer_class in configurations (huggingface#8152)

* Smarter prediction loop and no- -> no_ in console args (huggingface#8151)

* Smarter prediction loop and no- -> no_ in console args

* Fix test

* [s2s] distillBART docs for paper replication (huggingface#8150)

* Add a template for examples and apply it for mlm and plm examples (huggingface#8153)

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Styling

* improve error checking (huggingface#8157)

* Fix typo: indinces -> indices (huggingface#8159)

* Fix typo: indinces -> indices

* Fix some more

* Fix some more

* Fix some more

* Fix CI

* Fix eval ref miss in Chinese WWM. (huggingface#8115)

* ADD: add whole word mask proxy for both eng and chinese

* MOD: adjust format

* MOD: reformat code

* MOD: update import

* MOD: fix bug

* MOD: add import

* MOD: fix bug

* MOD: decouple code and update readme

* MOD: reformat code

* Update examples/language-modeling/README.md

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

* Update examples/language-modeling/README.md

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* change wwm to whole_word_mask

* reformat code

* reformat

* format

* Code quality

* ADD: update chinese ref readme

* MOD: small changes

* MOD: small changes2

* update readme

* fix eval ref file miss bug

* format file

* MOD: move ref code to contrib

* MOD: add delimeter check

* reformat code

* refomat code

* Update examples/language-modeling/README.md

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <sylvain.gugger@gmail.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* [CI] Better reports #2 (huggingface#8163)

* Fixing some warnings in DeBerta (huggingface#8176)

* Fixing some warnings in DeBerta

* Fixing docs with their rewritten version.

* Ci test tf super slow (huggingface#8007)

* Test TF GPU CI

* Change cache

* Fix missing torch requirement

* Fix some model tests


Style

* LXMERT

* MobileBERT

* Longformer skip test

* XLNet

* The rest of the tests

* RAG goes OOM in multi gpu setup

* YAML test files

* Last fixes

* Skip doctests

* Fill mask tests

* Yaml files

* Last test fix

* Style

* Update cache

* Change ONNX tests to slow + use tiny model

* Fix typo: s/languaged/language/ (huggingface#8165)

* TFMarian, TFMbart, TFPegasus, TFBlenderbot (huggingface#7987)

* Start plumbing

* Marian close

* Small stubs for all children

* Fixed bart

* marian working

* pegasus test is good, but failing

* Checkin tests

* More model files

* Subtle marian, pegasus integration test failures

* Works well

* rm print

* boom boom

* Still failing model2doc

* merge master

* Equivalence test failing, all others fixed

* cleanup

* Fix embed_scale

* Cleanup marian pipeline test

* Undo extra changes

* Smaller delta

* Cleanup model testers

* undo delta

* fix tests import structure

* cross test decorator

* Cleaner set_weights

* Respect authorized_unexpected_keys

* No warnings

* No warnings

* style

* Nest tf import

* black

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* functional dropout

* fixup

* Fixup

* style_doc

* embs

* shape list

* delete slow force_token_id_to_be_generated func

* fixup

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Doc fixes and filter warning in wandb (huggingface#8189)

* Finalize lm examples (huggingface#8188)

* Finish the cleanup of the language-modeling examples

* Update main README

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Apply suggestions from code review

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Propagate changes

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Replace swish with silu (huggingface#8166)

* Replace swish with silu

* revert nn.silu to nn.swish due to older version

* simplify optimized silu conditional and fix format

* Update activations.py

* Update activations_tf.py

* Update modeling_flax_utils.py

* Update modeling_openai.py

* add swish testcase

* add pytorch swish testcase

* Add more robust python version check

* more formatting fixes

Co-authored-by: TFUsers <TFUsers@gmail.com>

* Remove deprecated arguments from new run_clm (huggingface#8197)

* Minor style improvements for the Flax BERT and RoBERTa examples (huggingface#8178)

* Minor style improvements:

1. Use `@nn.compact` rather than `@compact` (as to not make it seem
   like compact is a standard Python decorator.
2. Move attribute docstrings from two `__call__` methods to comments
   on the attributes themselves. (This was probably a remnant from
   the pre-Linen version where the attributes were arguments to
   `call`.)

* Use black on the Flax modeling code

* Fix two bugs with --logging_first_step (huggingface#8193)

* make sure that logging_first_step evaluates

* fix bug with incorrect loss on logging_first_step

* fix style

* logging_first_step only logs, not evals

* [Bug fix] Fixed value for BlenderBot pad token (huggingface#8205)

* [Seq2SeqTrainer] Move import to init to make file self-contained (huggingface#8194)

* boom boom

* reverse order

* Added 12 model cards for Indian Language Models (huggingface#8198)

* Create README.md

* added model cards

* DynaBERT model cards update (huggingface#8192)

* Update README.md

* Update README.md

* Fix the behaviour of DefaultArgumentHandler (removing it). (huggingface#8180)

* Some work to fix the behaviour of DefaultArgumentHandler by removing it.

* Fixing specific pipelines argument checking.

* Fix ignore list behavior in doctests (huggingface#8213)

* doc: fix typo (huggingface#8235)

* Patch reports (huggingface#8238)

* Fix bad import with PyTorch <= 1.4.1 (huggingface#8237)

* Fix TensorBoardCallback for older versions of PyTorch (huggingface#8239)

* Create README.md

* Add line by line option to mlm/plm scripts (huggingface#8240)

* Make line by line optional in run_mlm

* Add option to disable dynamic padding

* Add option to plm too and update README

* Typos

* More typos

* Even more typos

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md

* Add XLMProphetNetTokenizer to tokenization auto (huggingface#8245)

* fix encoder decoder bug (huggingface#8243)

* add new notebooks (huggingface#8246)

* 2 SinusoidalPositionalEmbedding fixes (huggingface#8226)

* [Seq2Seq] Correct import in Seq2Seq Trainer (huggingface#8254)

* Skip tatoeba tests if Tatoeba-Challenge not cloned (huggingface#8260)

* Refactoring the generate() function (huggingface#6949)

* first draft

* show design proposition for new generate method

* up

* make better readable

* make first version

* gpt2 tests pass

* make beam search for gpt2 work

* add first encoder-decoder code

* delete typo

* make t5 work

* save indermediate

* make bart work with beam search

* finish beam search bart / t5

* add default kwargs

* make more tests pass

* fix no bad words sampler

* some fixes and tests for all distribution processors

* fix test

* fix rag slow tests

* merge to master

* add nograd to generate

* make all slow tests pass

* speed up generate

* fix edge case bug

* small fix

* correct typo

* add type hints and docstrings

* fix typos in tests

* add beam search tests

* add tests for beam scorer

* fix test rag

* finish beam search tests

* move generation tests in seperate file

* fix generation tests

* more tests

* add aggressive generation tests

* fix tests

* add gpt2 sample test

* add more docstring

* add more docs

* finish doc strings

* apply some more of sylvains and sams comments

* fix some typos

* make fix copies

* apply lysandres and sylvains comments

* final corrections on examples

* small fix for reformer

* [FIX] TextGenerationPipeline is currently broken. (huggingface#8256)

* [FIX] TextGenerationPipeline is currently broken.

It's most likely due to huggingface#8180.
What's missing is a multi vs single string handler at the beginning of
the pipe.
And also there was no testing of this pipeline.

* Fixing Conversational tests too.

* Updated ConversationalPipeline to work with encoder-decoder models (huggingface#8207)

* Updated ConversationalPipeline to work with encoder-decoder models (e.g. BlenderBot)

* Addition of integration test for EncoderDecoder conversation model

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Fix Tatoeba skip

* forward the worker stderr to the parent process (huggingface#8262)

* [examples] minimal version requirement run-time check in PL (huggingface#8133)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* make files independent (huggingface#8267)

* Clean Trainer tests and datasets dep (huggingface#8268)

* improve documentation of training_args.py (huggingface#8270)

* improve documentation of training_args.py

- do_train
- do_eval
- do_predict

* fix line too long

* fix style with black on training_args.py

* Update src/transformers/training_args.py

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

* Update src/transformers/training_args.py

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

* Update src/transformers/training_args.py

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

* fix line length with utils/style_doc

* black reformatting

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

* Data collator for token classification (huggingface#8274)

* Add DataCollatorForTokenClassification and clean tests

* Make quality

* [CIs] Better reports everywhere (huggingface#8275)

* make it possible to invoke testconf.py in both test suites without crashing on having the same option added

* perl -pi -e 's|--make_reports|--make-reports|' to be consistent with other opts

* add `pytest --make-reports` to all CIs (and artifacts)

* fix

* [WIP] Ner pipeline grouped_entities fixes (huggingface#5970)

* Bug fix: NER pipeline shouldn't group separate entities of same type

* style fix

* [Bug Fix] Shouldn't group entities that are both 'B' even if they are same type
	(B-type1 B-type1) != (B-type1 I-type1)
[Bug Fix] add an option `ignore_subwords` to ignore subsequent ##wordpieces in predictions. Because some models train on only the first token of a word and not on the subsequent wordpieces (BERT NER default). So it makes sense doing the same thing at inference time.
	The simplest fix is to just group the subwords with the first wordpiece.
	[TODO] how to handle ignored scores? just set them to 0 and calculate zero invariant mean ?
	[TODO] handle different wordpiece_prefix ## ? possible approaches:
		get it from tokenizer? but currently most tokenizers dont have a wordpiece_prefix property?
		have an _is_subword(token)
[Feature add] added option to `skip_special_tokens`. Cause It was harder to remove them after grouping.
[Additional Changes] remove B/I prefix on returned grouped_entities
[Feature Request/TODO] Return indexes?
[Bug TODO]  can't use fast tokenizer with grouped_entities ('BertTokenizerFast' object has no attribute 'convert_tokens_to_string')

* use offset_mapping to fix [UNK] token problem

* ignore score for subwords

* modify ner_pipeline test

* modify ner_pipeline test

* modify ner_pipeline test

* ner_pipeline change ignore_subwords default to true

* add ner_pipeline ignore_subword=False test case

* fix offset_mapping index

* fix style again duh

* change is_subword and convert_tokens_to_string logic

* merge tests with new test structure

* change test names

* remove old tests

* ner tests for fast tokenizer

* fast tokenizers have convert_tokens_to_string

* Fix the incorrect merge

Co-authored-by: Ceyda Cinarel <snu-ceyda@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* [blenderbot] regex fix (huggingface#8282)

Fixing:

```
src/transformers/tokenization_blenderbot.py:163: DeprecationWarning: invalid escape sequence \s
    token = re.sub("\s{2,}", " ", token)
```

* Fix typo in language-modeling README.md (huggingface#8287)

* [Generate Test] fix greedy generate test (huggingface#8293)

* fix greedy generate test

* delet ipdb

* Fix validation file loading in scripts (huggingface#8298)

* Upgrade resource for doc building

* Revert size change as it doesn't change anything

* Model card: T5-base fine-tuned on QASC (huggingface#8299)

* Update model cards of deepset/roberta-base-squad2 v1 and v2 (huggingface#8241)

* update deepset/roberta-base-squad2 to v2

* Update model_cards/deepset/roberta-base-squad2/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Improve QA pipeline error handling (huggingface#8286)

- The issue is that with previous code we would have the following:

```python
qa_pipeline = (...)
qa_pipeline(question="Where was he born ?", context="")
-> IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
```

The goal here is to improve this to actually return a ValueError
wherever possible.

While at it, I tried to simplify QuestionArgumentHandler's code to
make it smaller and more compat while keeping backward compat.

* adding model cards for distilled models (huggingface#8300)

* adding model cards for distil models

* forgot the languages

* Speedup doc build (huggingface#8301)

* Try -j option

* Try other thing

* Bigger machine

* Test lower sphinx version

* Remove trailing space

* Fix path to old run_language_modeling.py script (huggingface#8302)

* Clean up data collators and datasets (huggingface#8308)

* Clean up data collators and datasets

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Remove needless clone

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md (huggingface#8223)

* Create README.md

* Update README.md

* Apply suggestions from code review

Co-authored-by: Kevin Canwen Xu <canwenxu@126.com>
Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update bug-report.md

* Update PULL_REQUEST_TEMPLATE.md

* Corrected typo in readme (huggingface#8320)

* change TokenClassificationTask class methods to static methods (huggingface#7902)

* change TokenClassificationTask class methods to static methods

Since we do not require self in the class methods of TokenClassificationTask we should probably switch to static methods. Also, since the class TokenClassificationTask does not contain a constructor it is currently unusable as is. By switching to static methods this fixes the issue of having to document the intent of the broken class.

Also, since the get_labels and read_examples_from_file methods are ought to be implemented. Static method definitions are unchanged even after inheritance, which means that it can be overridden, similar to other class methods.

* Trigger Build

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* no warn (huggingface#8329)

* Output global_attentions in Longformer models (huggingface#7562)

* Output global_attentions in Longformer models

* make style

* small refactoring

* fix tests

* make fix-copies

* add for tf as well

* remove comments in test

* make fix-copies

* make style

* add docs

* make docstring pretty

Co-authored-by: patrickvonplaten <patrick.v.platen@gmail.com>

* Make Trainer evaluation handle dynamic seq_length (huggingface#8336)

* Make Trainer evaluation handle dynamic seq_length

* Document behavior.

* Fix test

* Better fix

* Fixes for realsies this time

* Address review comments

* Without forgetting to save...

* [s2s] test_distributed_eval (huggingface#8315)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Docs bart training ref (huggingface#8330)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* [s2s] test_bash_script.py - actually learn something (huggingface#8318)

* use decorator

* remove hardcoded paths

* make the test use more data and do real quality tests

* shave off 10 secs

* add --eval_beams 2, reformat

* reduce train size, use smaller custom dataset

* Model card: T5-base fine-tuned on QuaRel (huggingface#8334)

* Model card: CodeBERT fine-tuned for Insecure Code Detection (huggingface#8247)

* Model card: CodeBERT fine-tuned for Insecure Code Detection

* Update model_cards/mrm8488/codebert-base-finetuned-detect-insecure-code/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Model card: GPT-2 fine-tuned on CommonGen (huggingface#8248)

* [model_cards] Update Italian BERT models and introduce new Italian XXL ELECTRA model 🎉 (huggingface#8343)

* Create README.md (huggingface#8258)

* german medbert model details (huggingface#8266)

* model details

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8327)

* Create README.md (huggingface#8167)

* Create README.md

Telugu BERTU Readme file

* Update model_cards/kuppuluri/telugu_bertu/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8168)

* Create README.md

* Update README.md

* Create README.md (huggingface#8170)

* Create README.md (huggingface#8169)

* Create README.md (huggingface#8255)

* Create README.md

Initial commit

* Updated Read me

Updated

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8312)

* Create README.md

* Update model_cards/ktrapeznikov/gpt2-medium-topic-news/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8338)

fixes

* Fix typo (huggingface#8351)

* Update README.md (huggingface#8360)

Fix websitr address

* [All Seq2Seq model + CLM models that can be used with EncoderDecoder] Add cross-attention weights to outputs (huggingface#8071)

* Output cross-attention with decoder attention output

* Update src/transformers/modeling_bert.py

* add cross-attention for t5 and bart as well

* fix tests

* correct typo in docs

* add sylvains and sams comments

* correct typo

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* fix encoder outputs (huggingface#8368)

* [make] rewrite modified_py_files in python to be cross-platform (huggingface#8371)

* rewrite modified_py_files in python to be cross-platform

* try a different way to test for variable not being ""

* improve comment

* Fix DataCollatorForWholeWordMask (huggingface#8379)

* Fix DataCollatorForWholeWordMask

* Replace all tensorize_batch in data_collator.py

* fix md table (huggingface#8395)

* Add gpt2-medium-chinese model card (huggingface#8402)

* Create README.md

* Update model_cards/mymusise/gpt2-medium-chinese/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* fixed default labels for QA model (huggingface#8399)

* Fix DataCollatorForWholeWordMask again (huggingface#8397)

* [s2s examples test] fix data path (huggingface#8398)

* [s2s test_finetune_trainer] failing multigpu test (huggingface#8400)

* [s2s/distill] remove run_distiller.sh, fix xsum script (huggingface#8412)

* comet_ml temporary fix(huggingface#8410)

* updating tag for exbert viz (huggingface#8408)

* Update README.md (huggingface#8406)

* Fix some tooling for windows (huggingface#8359)

* Fix some tooling for windows

* Fix conflict

* Trigger CI

* examples/docs: caveat that PL examples don't work on TPU (huggingface#8309)

* add evaluate doc - trainer.evaluate returns 'epoch' from training (huggingface#8273)

* add evaluate doc

* fix style with utils/style.doc

* Update src/transformers/trainer.py

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

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

* Bug fix for permutation language modelling (huggingface#8409)

* [fsmt tokenizer] support lowercase tokenizer (huggingface#8389)

* support lowercase tokenizer

* fix arg pos

* Bump tokenizers (huggingface#8419)

* Add new token classification example (huggingface#8340)

* Add new token classification example

* Remove txt file

* Add test

* With actual testing done

* Less warmup is better

* Update examples/token-classification/run_ner_new.py

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Address review comments

* Fix test

* Make Lysandre happy

* Last touches and rename

* Rename in tests

* Address review comments

* More run_ner -> run_ner_old

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Fix typo

* [fsmt convert script] fairseq broke chkpt data - fixing that (huggingface#8377)

* fairseq broke chkpt data - fixing that

* style

* support older bpecodes filenames - specifically "code" in iwslt14

* Deprecate old data/metrics functions (huggingface#8420)

* [Tests] Add Common Test for Training + Fix a couple of bugs (huggingface#8415)

* add training tests

* correct longformer

* fix docs

* fix some tests

* fix some more train tests

* remove ipdb

* fix multiple edge case model training

* fix funnel and prophetnet

* clean gpt models

* undo renaming of albert

* [docs] remove sshleifer from issue-template :( (huggingface#8418)

* Fix bart shape comment (huggingface#8423)

* [docs] [testing] gpu decorators table (huggingface#8422)

* gpu decorators table

* whitespace

* Update docs/source/testing.rst

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* whitespace

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Check all models are in an auto class (huggingface#8425)

* [github CI] add a multi-gpu job for all example tests (huggingface#8341)

* add a multi-gpu job for all example tests

* run only ported tests

* rename

* explain why env is re-activated on each step

* mark all unported/checked tests with @require_torch_non_multigpu_but_fix_me

* style

* Apply suggestions from code review

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Changing XLNet default from not using memories to 512 context size following paper (huggingface#8417)

* Move XLNet memory length FutureWarning

* isort

* style

* Changed default XLNet memory length

* Model versioning (huggingface#8324)

* fix typo

* rm use_cdn & references, and implement new hf_bucket_url

* I'm pretty sure we don't need to `read` this file

* same here

* [BIG] file_utils.networking: do not gobble up errors anymore

* Fix CI 😇

* Apply suggestions from code review

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

* Tiny doc tweak

* Add doc + pass kwarg everywhere

* Add more tests and explain

cc @sshleifer let me know if better

Co-Authored-By: Sam Shleifer <sshleifer@gmail.com>

* Also implement revision in pipelines

In the case where we're passing a task name or a string model identifier

* Fix CI 😇

* Fix CI

* [hf_api] new methods + command line implem

* make style

* Final endpoints post-migration

* Fix post-migration

* Py3.6 compat

cc @stefan-it

Thank you @stas00

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

* Patch token classification pipeline (huggingface#8364)

* Patch token classification pipeline

* Some added tests for TokenClassificationArgumentHandler (huggingface#8366)

Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>

* Update links from s3 to huggingface.co

* Fix style

* Model sharing rst (huggingface#8439)

* Update RST

* Finer details

* Re-organize

* Style

* Release: v3.5.0

* v3.5.0 documentation

* [s2s/distill] hparams.tokenizer_name = hparams.teacher (huggingface#8382)

* [examples] better PL version check (huggingface#8429)

* Question template (huggingface#8440)

* Remove SO from question template

* Styling

* [docs] improve bart/marian/mBART/pegasus docs (huggingface#8421)

* Add auto next sentence prediction (huggingface#8432)

* Add auto next sentence prediction

* Fix style

* Add mobilebert next sentence prediction

* Windows dev section in the contributing file (huggingface#8436)

* Add a Windows dev section in the contributing file.

* Forgotten link

* Trigger CI

* Rework description

* Trigger CI

* [testing utils] get_auto_remove_tmp_dir more intuitive behavior (huggingface#8401)

* [testing utils] get_auto_remove_tmp_dir default change

Now that I have been using `get_auto_remove_tmp_dir default change` for a while, I realized that the defaults aren't most optimal.

99% of the time we want the tmp dir to be empty at the beginning of the test - so changing the default to `before=True` - this shouldn't impact any tests since this feature is used only during debug.

* simplify things

* update docs

* fix doc layout

* style

* Update src/transformers/testing_utils.py

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

* better 3-state doc

* style

* Apply suggestions from code review

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

* s/tmp/temporary/ + style

* correct the statement

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

* Add missing import (huggingface#8444)

* Add missing import

* Fix dummy objects

* fix t5 special tokens (huggingface#8435)

* using multi_gpu consistently (huggingface#8446)

* s|multiple_gpu|multi_gpu|g; s|multigpu|multi_gpu|g'

* doc

* Add missing tasks to `pipeline` docstring (huggingface#8428)

* [No merge] TF integration testing (huggingface#7621)

* stash

* TF Integration testing for ELECTRA, BERT, Longformer

* Trigger slow tests

* Apply suggestions from code review

* fix t5 token type ids (huggingface#8437)

* Bug fix for modeling utilities function: apply_chunking_to_forward, chunking should be in the chunking dimension, an exception was raised if the complete shape of the inputs was not the same rather than only the chunking dimension (huggingface#8391)

Co-authored-by: pedro <pe25171@mit.edu>

* [model_cards] harmonization

* Fix TF Longformer (huggingface#8460)

* Add next sentence prediction loss computation (huggingface#8462)

* Add next sentence prediction loss computation

* Apply style

* Fix tests

* Add forgotten import

* Add forgotten import

* Use a new parameter

* Remove kwargs and use positional arguments

* Fix next sentence output (huggingface#8466)

* Example NER script predicts on tokenized dataset (huggingface#8468)

The new run_ner.py script tries to run prediction on the input
test set `datasets["test"]`, but it should be the tokenized set
`tokenized_datasets["test"]`

* Add TFDPR (huggingface#8203)

* Create modeling_tf_dpr.py

* Add TFDPR

* Add back TFPegasus, TFMarian, TFMBart, TFBlenderBot

last commit accidentally deleted these 4 lines, so I recover them back

* Add TFDPR

* Add TFDPR

* clean up some comments, add TF input-style doc string

* Add TFDPR

* Make return_dict=False as default

* Fix return_dict bug (in .from_pretrained)

* Add get_input_embeddings()

* Create test_modeling_tf_dpr.py

The current version is already passed all 27 tests!
Please see the test run at : 
https://colab.research.google.com/drive/1czS_m9zy5k-iSJbzA_DP1k1xAAC_sdkf?usp=sharing

* fix quality

* delete init weights

* run fix copies

* fix repo consis

* del config_class, load_tf_weights

They shoud be 'pytorch only'

* add config_class back

after removing it, test failed ... so totally only removing "use_tf_weights = None" on Lysandre suggestion

* newline after .. note::

* import tf, np (Necessary for ModelIntegrationTest)

* slow_test from_pretrained with from_pt=True

At the moment we don't have TF weights (since we don't have official official TF model)
Previously, I did not run slow test, so I missed this bug

* Add simple TFDPRModelIntegrationTest

Note that this is just a test that TF and Pytorch gives approx. the same output.
However, I could not test with the official DPR repo's output yet

* upload correct tf model

* remove position_ids as missing keys

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: patrickvonplaten <patrick@huggingface.co>

* Replaced some iadd operations on lists with proper list methods. (huggingface#8433)

* Skip test until investigation

* Flax/Jax documentation (huggingface#8331)

* First addition of Flax/Jax documentation

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* make style

* Ensure input order match between Bert & Roberta

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Install dependencies "all" when building doc

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* wraps build_doc deps with ""

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Addressing @sgugger comments.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Use list to highlight JAX features.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Make style.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Let's not look to much into the future for now.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Style

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* [s2s] distill t5-large -> t5-small (huggingface#8376)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Update deploy-docs dependencies on CI to enable Flax (huggingface#8475)

* Update deploy-docs dependencies on CI to enable Flax

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Added pair of ""

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* [model_cards] other chars than [\w\-_] not allowed anymore in model names

cc @Pierrci

* Fix typo in roberta-base-squad2-v2 model card (huggingface#8489)

* quick fix on concatenating text to support more datasets (huggingface#8474)

* Fix doc bug (huggingface#8500)

* fix doc bug

Signed-off-by: mymusise <mymusise1@gmail.com>

* fix example bug

Signed-off-by: mymusise <mymusise1@gmail.com>

* Model sharing doc (huggingface#8498)

* Model sharing doc

* Style

* fix SqueezeBertForMaskedLM (huggingface#8479)

* Try to understand and apply Sylvain's comments (huggingface#8458)

* Use LF instead of os.linesep (huggingface#8491)

* Add pretraining loss computation for TF Bert pretraining (huggingface#8470)

* Add pretraining loss computation for TF Bert pretraining

* Fix labels creation

* Fix T5 model

* restore T5 kwargs

* try a generic fix for pretraining models

* Apply style

* Overide the prepare method for the BERT tests

* Remove typo

* Update deepset/roberta-base-squad2 model card (huggingface#8522)

* Update README.md

* Update README.md

* Update doc for v3.5.1

* [T5] Bug correction & Refactor (huggingface#8518)

* fix bug

* T5 refactor

* refactor tf

* apply sylvains suggestions

* Model templates encoder only (huggingface#8509)

* Model templates

* TensorFlow

* Remove pooler

* CI

* Tokenizer + Refactoring

* Encoder-Decoder

* Let's go testing

* Encoder-Decoder in TF

* Let's go testing in TF

* Documentation

* README

* Fixes

* Better names

* Style

* Update docs

* Choose to skip either TF or PT

* Code quality fixes

* Add to testing suite

* Update file path

* Cookiecutter path

* Update `transformers` path

* Handle rebasing

* Remove seq2seq from model templates

* Remove s2s config

* Apply Sylvain and Patrick comments

* Apply suggestions from code review

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

* Last fixes from code review

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

* Fix paths in github YAML

* Model sharing doc: more tweaks (huggingface#8520)

* More doc tweaks

* Update model_sharing.rst

* make style

* missing newline

* Add email tip

Co-authored-by: Pierric Cistac <pierric@huggingface.co>

* Add bart-large-mnli model card (huggingface#8527)

* fix load weights (huggingface#8528)

* fix load weights

* delete line

* Rework some TF tests (huggingface#8492)

* Update some tests

* Small update

* Apply style

* Use max_position_embeddings

* Create a fake attribute

* Create a fake attribute

* Update wrong name

* Wrong TransfoXL model file

* Keep the common tests agnostic

* [breaking|pipelines|tokenizers] Adding slow-fast tokenizers equivalence tests pipelines - Removing sentencepiece as a required dependency (huggingface#8073)

* Fixing roberta for slow-fast tests

* WIP getting equivalence on pipelines

* slow-to-fast equivalence - working on question-answering pipeline

* optional FAISS tests

* Pipeline Q&A

* Move pipeline tests to their own test job again

* update tokenizer to add sequence id methods

* update to tokenizers 0.9.4

* set sentencepiecce as optional

* clean up squad

* clean up pipelines to use sequence_ids

* style/quality

* wording

* Switch to use_fast = True by default

* update tests for use_fast at True by default

* fix rag tokenizer test

* removing protobuf from required dependencies

* fix NER test for use_fast = True by default

* fixing example tests (Q&A examples use slow tokenizers for now)

* protobuf in main deps extras["sentencepiece"] and example deps

* fix protobug install test

* try to fix seq2seq by switching to slow tokenizers for now

* Update src/transformers/tokenization_utils_base.py

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Update src/transformers/tokenization_utils_base.py

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md for Chinese RoBERTa Miniatures (huggingface#8550)

* Create README.md

* Update model_cards/uer/chinese_roberta_L-2_H-128/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Readme for News Headline Generation (bert2bert) (huggingface#8557)

* Readme for Wiki Summary [Persian] bert2bert (huggingface#8558)

* Clearer Model Versioning Example (huggingface#8562)

* [doc] typo fix (huggingface#8535)

* [doc] typo fix

@sgugger

* Update src/transformers/modeling_utils.py

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

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

* Adding the prepare_seq2seq_batch function to ProphetNet (huggingface#8515)

* Simply insert T5Tokenizer's prepare_seq2seq_batch

* Update/Add some 'import'

* fix RunTimeError caused by '.view'

* Moves .view related error avoidance from seq2seq_trainer to inside prophetnet

* Update test_tokenization_prophetnet.py

* Format the test code with black

* Re-format the test code

* Update test_tokenization_prophetnet.py

* Add importing require_torch in the test code

* Add importing BatchEncoding in the test code

* Re-format the test code on Colab

* Fix GPT2DoubleHeadsModel to work with model.generate() (huggingface#6601)

* Fix passing token_type_ids during GPT2DoubleHeadsModel.generate() if used

and for GPT2LMHeadModel too

* Update tests to check token_type_ids usage in GPT2 models

* Update version to v4.0.0-dev (huggingface#8568)

* Switch `return_dict` to `True` by default. (huggingface#8530)

* Use the CI to identify failing tests

* Remove from all examples and tests

* More default switch

* Fixes

* More test fixes

* More fixes

* Last fixes hopefully

* Use the CI to identify failing tests

* Remove from all examples and tests

* More default switch

* Fixes

* More test fixes

* More fixes

* Last fixes hopefully

* Run on the real suite

* Fix slow tests

* Fix mixed precision issue for GPT2 (huggingface#8572)

* Fix mixed precision issue for GPT2

* Forgot one cast

* oops

* Forgotten casts

* Reorganize repo (huggingface#8580)

* Put models in subfolders

* Styling

* Fix imports in tests

* More fixes in test imports

* Sneaky hidden imports

* Fix imports in doc files

* More sneaky imports

* Finish fixing tests

* Fix examples

* Fix path for copies

* More fixes for examples

* Fix dummy files

* More fixes for example

* More model import fixes

* Is this why you're unhappy GitHub?

* Fix imports in conver command

* model_card for indolem/indobert-base-uncased (huggingface#8579)

* T5 & mT5 (huggingface#8552)

* add mt5 and t5v1_1 model

* fix tests

* correct some imports

* add tf model

* finish tf t5

* improve examples

* fix copies

* clean doc

* [MT5] More docs (huggingface#8589)

* add docs

* make style

* Add __init__ to the models folder

* Fix init for MT5 (huggingface#8591)

* Tokenizers: ability to load from model subfolder (huggingface#8586)

* <small>tiny typo</small>

* Tokenizers: ability to load from model subfolder

* use subfolder for local files as well

* Uniformize model shortcut name => model id

* from s3 => from huggingface.co

Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>

* Fix model templates (huggingface#8595)

* First fixes

* Fix imports and add init

* Fix typo

* Move init to final dest

* Fix tokenization import

* More fixes

* Styling

* these should run fine on multi-gpu (huggingface#8582)

* Fix check repo utils (huggingface#8600)

* Tokenizers should be framework agnostic (huggingface#8599)

* Tokenizers should be framework agnostic

* Run the slow tests

* Not testing

* Fix documentation

* Apply suggestions from code review

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Remove deprecated (huggingface#8604)

* Remove old deprecated arguments

Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

* Remove needless imports

* Fix tests

Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

* Add Harry Potter Model Card (huggingface#8605)

* Add Harry Potter Model

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Remove old doc

* Fixed link to the wrong paper. (huggingface#8607)

* Reset loss to zero on logging in Trainer to avoid bfloat16 issues (huggingface#8561)

* make tr_loss regular float

* Revert "make tr_loss regular float"

This reverts commit c9d7ccf.

* reset loss at each logging step

* keep track of total loss with _total_loss_scalar

* add remaining tr_loss at the end

* Fix DataCollatorForLanguageModeling (huggingface#8621)

* Fix missing space in multiline warning (huggingface#8593)

Multiline string informing about missing PyTorch/TensorFlow had missing space.

* [s2s] broken test (huggingface#8613)

* fix to adjust for huggingface#8530 changes (huggingface#8612)

* self.self.activation_dropout -> self.activation_dropout (huggingface#8611)

(one line typo)

* New TF loading weights (huggingface#8490)

* New TF loading weights

* apply style

* Better naming

* Largely comment the loading method

* Apply style

* Address Patrick's comments

* Remove useless line of code

* Update Docstring

* Address Sylvain's and Lysandre's comments

* Simplify the names computation

* Typos

* Adding PrefixConstrainedLogitsProcessor (huggingface#8529)

* Adding PrefixConstrainedLogitsProcessor

* fixing RAG and style_doc

* fixing black (v20 instead of v19)

* Improving doc in generation_logits_process.py

* Improving docs and typing in generation_utils.py

* docs improvement

* adding test and fixing doc typo

* fixing doc_len

* isort on test

* fixed test

* improve docstring a bit

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* [Tokenizer Doc] Improve tokenizer summary (huggingface#8622)

* improve summary

* small fixes

* cleaned line length

* correct "" formatting

* apply sylvains suggestions

* Fixes the training resuming with gradient accumulation (huggingface#8624)

* Fix training from scratch in new scripts (huggingface#8623)

* model_cards for Chinese Couplet and Poem GPT2 models (huggingface#8620)

* replace performance table with markdown (huggingface#8565)

* replace performance table with markdown

* Update model_cards/smanjil/German-MedBERT/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8544)

Modified Model in Action section. The class `AutoModelWithLMHead` is deprecated so changed it to `AutoModelForSeq2SeqLM` for encoder-decoder models. Removed duplicate eos token.

* Model Card for abhilash1910/financial_roberta (huggingface#8625)

* Model Card for abhilash1910/financial_roberta

* Update model_cards/abhilash1910/financial_roberta/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8405)

* Update README.md

* Update README.md

* Add model card for ai4bharat/indic-bert (huggingface#8464)

* Create README.md (huggingface#8363)

* Model card: T5-base fine-tuned on QuaRTz (huggingface#8369)

* Model card: T5-base fine-tuned on QuaRTz

* Update model_cards/mrm8488/t5-base-finetuned-quartz/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8362)

* Created ModelCard for Hel-ach-en MT model (huggingface#8496)

* Updated ModelCard

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* [s2s] distillation apex breaks return_dict obj (huggingface#8631)

* apex breaks return_dict obj

* style

* grammar (huggingface#8639)

* Update README.md (huggingface#8635)

* Updated the Extractive Question Answering code snippets (huggingface#8636)

* Updated the Extractive Question Answering code snippets

The Extractive Question Answering code snippets do not work anymore since the models return task-specific output objects. This commit fixes the pytorch and tensorflow examples but adding `.values()` to the model call.

* Update task_summary.rst

* Add cards for all Geotrend models (huggingface#8617)

* docs(bert-base-15lang-cased): add model card

* add cards for all Geotrend models

* [model cards] fix language tag for all Geotrend models

* [model card] : fix bert-base-15lang-cased (huggingface#8655)

the table was badly formatted because of a single line break

* fix missing return dict (huggingface#8653)

* fixed imports

* update example conversion

* removed redundant tests

* added back marker

* reverted to old QA Argument Handler

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Sean Naren <sean@grid.ai>
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
Co-authored-by: Santiago Castro <sacastro@umich.edu>
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Zhiqi Huang <mazicwong@gmail.com>
Co-authored-by: Manuel Romero <mrm8488@gmail.com>
Co-authored-by: Ashwani Tanwar <ashwanitanwar333@gmail.com>
Co-authored-by: Julien Chaumond <chaumond@gmail.com>
Co-authored-by: Ethan <9592150+Ethan-yt@users.noreply.github.com>
Co-authored-by: gurkan08 <33202187+gurkan08@users.noreply.github.com>
Co-authored-by: dartrevan <awesombatsy@gmail.com>
Co-authored-by: Branden Chan <33759007+brandenchan@users.noreply.github.com>
Co-authored-by: yantan <yantan@effyic.com>
Co-authored-by: Santiago Castro <santi.1410@hotmail.com>
Co-authored-by: wlhgtc <hgtcwl@foxmail.com>
Co-authored-by: Sylvain Gugger <sylvain.gugger@gmail.com>
Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>
Co-authored-by: TFUsers <25044281+TFUsers@users.noreply.github.com>
Co-authored-by: TFUsers <TFUsers@gmail.com>
Co-authored-by: Avital Oliver <avital@thewe.net>
Co-authored-by: Abi See <abigail.e.see@gmail.com>
Co-authored-by: guillaume-be <guillaume.becquin@gmail.com>
Co-authored-by: Kushal <32245327+kushalj001@users.noreply.github.com>
Co-authored-by: Martin Monperrus <martin.monperrus@gnieh.org>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>
Co-authored-by: Philip May <eniak.info@gmail.com>
Co-authored-by: Ceyda Cinarel <15624271+cceyda@users.noreply.github.com>
Co-authored-by: Ceyda Cinarel <snu-ceyda@users.noreply.github.com>
Co-authored-by: Pengzhi Gao <pengzhi.gao@petuum.com>
Co-authored-by: Victor SANH <victorsanh@gmail.com>
Co-authored-by: Yifan Peng <pengyifan.mail@gmail.com>
Co-authored-by: Kevin Canwen Xu <canwenxu@126.com>
Co-authored-by: Guillem García Subies <37592763+GuillemGSubies@users.noreply.github.com>
Co-authored-by: Bobby Donchev <contact@donchev.is>
Co-authored-by: Guillaume Filion <guillaume.filion@gmail.com>
Co-authored-by: Leandro von Werra <lvwerra@users.noreply.github.com>
Co-authored-by: Stefan Schweter <stefan@schweter.it>
Co-authored-by: Jiaxin Pei <pedropei@vip.qq.com>
Co-authored-by: smanjil <shresthamanjil21@gmail.com>
Co-authored-by: Karthik Uppuluri <karthik.uppuluri@gmail.com>
Co-authored-by: hasantanvir79 <hasantanvir79@gmail.com>
Co-authored-by: ktrapeznikov <ktrapeznikov@gmail.com>
Co-authored-by: hassoudi <hassoudi@gmail.com>
Co-authored-by: Jonathan Chang <31893406+cccntu@users.noreply.github.com>
Co-authored-by: Yossi Synett <github@yossisynett.com>
Co-authored-by: Chengxi Guo <mymusise1@gmail.com>
Co-authored-by: Manav Rathod <manav.rathod@berkeley.edu>
Co-authored-by: Julien Plu <plu.julien@gmail.com>
Co-authored-by: Shashank Gupta <shaz4194@gmail.com>
Co-authored-by: Teven <teven.lescao@gmail.com>
Co-authored-by: Shichao Sun <sunshichao1995@gmail.com>
Co-authored-by: Pedro <pedro.colon4@upr.edu>
Co-authored-by: pedro <pe25171@mit.edu>
Co-authored-by: sarnoult <31313050+sarnoult@users.noreply.github.com>
Co-authored-by: Ratthachat (Jung) <56621342+ratthachat@users.noreply.github.com>
Co-authored-by: patrickvonplaten <patrick@huggingface.co>
Co-authored-by: Beomsoo Kim <bluewhale8202@gmail.com>
Co-authored-by: Funtowicz Morgan <mfuntowicz@users.noreply.github.com>
Co-authored-by: Sumithra Bhakthavatsalam <sumithra.b@gmail.com>
Co-authored-by: Antonio Lanza <antoniolanza1996@gmail.com>
Co-authored-by: zeyuyun1 <43428393+zeyuyun1@users.noreply.github.com>
Co-authored-by: Forrest Iandola <fiandola@gmail.com>
Co-authored-by: Pierric Cistac <pierric@huggingface.co>
Co-authored-by: Joe Davison <josephddavison@gmail.com>
Co-authored-by: zhezhaoa <1152543959@qq.com>
Co-authored-by: Mehrdad Farahani <m3hrdadfi@gmail.com>
Co-authored-by: Yusuke Mori <mori@mi.t.u-tokyo.ac.jp>
Co-authored-by: LSinev <LSinev@users.noreply.github.com>
Co-authored-by: fajri91 <fajri91@users.noreply.github.com>
Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>
Co-authored-by: Caitlin Ostroff <caitlin.ostroff@gmail.com>
Co-authored-by: cronoik <johannes.schaffrath@mail.de>
Co-authored-by: Benjamin Minixhofer <bminixhofer@gmail.com>
Co-authored-by: Michał Pogoda <237372@student.pwr.edu.pl>
Co-authored-by: Nicola De Cao <nicola.decao@gmail.com>
Co-authored-by: hhou435 <59219579+hhou435@users.noreply.github.com>
Co-authored-by: Vishal Singh <vishalsingh7x@gmail.com>
Co-authored-by: Abhilash Majumder <30946547+abhilash1910@users.noreply.github.com>
Co-authored-by: Divyanshu Kakwani <divkakwani@gmail.com>
Co-authored-by: Perez Ogayo <pogayo17@alustudent.com>
Co-authored-by: Tim Isbister <timisbister@gmail.com>
Co-authored-by: Amine Abdaoui <abdaoui@lirmm.fr>
Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com>
bdalal added a commit to lucidworks/transformers that referenced this pull request Nov 19, 2020
* Rename add_start_docstrings_to_callable (huggingface#8120)

* Update CI cache (huggingface#8126)

* Upgrade PyTorch Lightning to 1.0.2 (huggingface#7852)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Fix typo in `AutoModelForMaskedLM` docs (huggingface#8129)

* [s2s test] cleanup (huggingface#8131)

* add tags (huggingface#8147)

* Add model_cards for DynaBERT (huggingface#8012)

* Update README.md

* Add dynabert_overview.png

* Update README.md

* Create README.md

* Add dynabert_overview.png

* Update README.md

* Update README.md

* Delete dynabert_overview.png

* Update README.md

* Delete dynabert_overview.png

* Update README.md

* Create README.md (huggingface#8015)

* Create README.md (huggingface#8017)

* Model Card for Gujarati-XLM-R-Base (huggingface#8038)

* Add model card for Gujarati-XLM-R-Base

* Update README.md

Add the model card for the Gujarati-XLM-R-Base.

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Add two model_cards: ethanyt/guwenbert-base and ethanyt/guwenbert-large (huggingface#8041)

* Create README.md (huggingface#8075)

* Create README.md

* Update model_cards/gurkan08/bert-turkish-text-classification/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8088)

* Create README.md

* metadata

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8089)

* Add model_cards (huggingface#7969)

* add readme

* add readmes

* Add metadata

* Update README.md (huggingface#8090)

* Update widget examples. (huggingface#8149)

Co-authored-by: yantan <yantan@effyic.com>

* Fix doc errors and typos across the board (huggingface#8139)

* Fix doc errors and typos across the board

* Fix a typo

* Fix the CI

* Fix more typos

* Fix CI

* More fixes

* Fix CI

* More fixes

* More fixes

* Document tokenizer_class in configurations (huggingface#8152)

* Smarter prediction loop and no- -> no_ in console args (huggingface#8151)

* Smarter prediction loop and no- -> no_ in console args

* Fix test

* [s2s] distillBART docs for paper replication (huggingface#8150)

* Add a template for examples and apply it for mlm and plm examples (huggingface#8153)

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Add a template for example scripts and apply it to mlm

* Formatting

* Fix test

* Add plm script

* Styling

* improve error checking (huggingface#8157)

* Fix typo: indinces -> indices (huggingface#8159)

* Fix typo: indinces -> indices

* Fix some more

* Fix some more

* Fix some more

* Fix CI

* Fix eval ref miss in Chinese WWM. (huggingface#8115)

* ADD: add whole word mask proxy for both eng and chinese

* MOD: adjust format

* MOD: reformat code

* MOD: update import

* MOD: fix bug

* MOD: add import

* MOD: fix bug

* MOD: decouple code and update readme

* MOD: reformat code

* Update examples/language-modeling/README.md

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

* Update examples/language-modeling/README.md

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* Update examples/language-modeling/run_language_modeling.py

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

* change wwm to whole_word_mask

* reformat code

* reformat

* format

* Code quality

* ADD: update chinese ref readme

* MOD: small changes

* MOD: small changes2

* update readme

* fix eval ref file miss bug

* format file

* MOD: move ref code to contrib

* MOD: add delimeter check

* reformat code

* refomat code

* Update examples/language-modeling/README.md

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <sylvain.gugger@gmail.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* [CI] Better reports #2 (huggingface#8163)

* Fixing some warnings in DeBerta (huggingface#8176)

* Fixing some warnings in DeBerta

* Fixing docs with their rewritten version.

* Ci test tf super slow (huggingface#8007)

* Test TF GPU CI

* Change cache

* Fix missing torch requirement

* Fix some model tests


Style

* LXMERT

* MobileBERT

* Longformer skip test

* XLNet

* The rest of the tests

* RAG goes OOM in multi gpu setup

* YAML test files

* Last fixes

* Skip doctests

* Fill mask tests

* Yaml files

* Last test fix

* Style

* Update cache

* Change ONNX tests to slow + use tiny model

* Fix typo: s/languaged/language/ (huggingface#8165)

* TFMarian, TFMbart, TFPegasus, TFBlenderbot (huggingface#7987)

* Start plumbing

* Marian close

* Small stubs for all children

* Fixed bart

* marian working

* pegasus test is good, but failing

* Checkin tests

* More model files

* Subtle marian, pegasus integration test failures

* Works well

* rm print

* boom boom

* Still failing model2doc

* merge master

* Equivalence test failing, all others fixed

* cleanup

* Fix embed_scale

* Cleanup marian pipeline test

* Undo extra changes

* Smaller delta

* Cleanup model testers

* undo delta

* fix tests import structure

* cross test decorator

* Cleaner set_weights

* Respect authorized_unexpected_keys

* No warnings

* No warnings

* style

* Nest tf import

* black

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* functional dropout

* fixup

* Fixup

* style_doc

* embs

* shape list

* delete slow force_token_id_to_be_generated func

* fixup

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Doc fixes and filter warning in wandb (huggingface#8189)

* Finalize lm examples (huggingface#8188)

* Finish the cleanup of the language-modeling examples

* Update main README

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Apply suggestions from code review

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Propagate changes

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Replace swish with silu (huggingface#8166)

* Replace swish with silu

* revert nn.silu to nn.swish due to older version

* simplify optimized silu conditional and fix format

* Update activations.py

* Update activations_tf.py

* Update modeling_flax_utils.py

* Update modeling_openai.py

* add swish testcase

* add pytorch swish testcase

* Add more robust python version check

* more formatting fixes

Co-authored-by: TFUsers <TFUsers@gmail.com>

* Remove deprecated arguments from new run_clm (huggingface#8197)

* Minor style improvements for the Flax BERT and RoBERTa examples (huggingface#8178)

* Minor style improvements:

1. Use `@nn.compact` rather than `@compact` (as to not make it seem
   like compact is a standard Python decorator.
2. Move attribute docstrings from two `__call__` methods to comments
   on the attributes themselves. (This was probably a remnant from
   the pre-Linen version where the attributes were arguments to
   `call`.)

* Use black on the Flax modeling code

* Fix two bugs with --logging_first_step (huggingface#8193)

* make sure that logging_first_step evaluates

* fix bug with incorrect loss on logging_first_step

* fix style

* logging_first_step only logs, not evals

* [Bug fix] Fixed value for BlenderBot pad token (huggingface#8205)

* [Seq2SeqTrainer] Move import to init to make file self-contained (huggingface#8194)

* boom boom

* reverse order

* Added 12 model cards for Indian Language Models (huggingface#8198)

* Create README.md

* added model cards

* DynaBERT model cards update (huggingface#8192)

* Update README.md

* Update README.md

* Fix the behaviour of DefaultArgumentHandler (removing it). (huggingface#8180)

* Some work to fix the behaviour of DefaultArgumentHandler by removing it.

* Fixing specific pipelines argument checking.

* Fix ignore list behavior in doctests (huggingface#8213)

* doc: fix typo (huggingface#8235)

* Patch reports (huggingface#8238)

* Fix bad import with PyTorch <= 1.4.1 (huggingface#8237)

* Fix TensorBoardCallback for older versions of PyTorch (huggingface#8239)

* Create README.md

* Add line by line option to mlm/plm scripts (huggingface#8240)

* Make line by line optional in run_mlm

* Add option to disable dynamic padding

* Add option to plm too and update README

* Typos

* More typos

* Even more typos

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md

* Add XLMProphetNetTokenizer to tokenization auto (huggingface#8245)

* fix encoder decoder bug (huggingface#8243)

* add new notebooks (huggingface#8246)

* 2 SinusoidalPositionalEmbedding fixes (huggingface#8226)

* [Seq2Seq] Correct import in Seq2Seq Trainer (huggingface#8254)

* Skip tatoeba tests if Tatoeba-Challenge not cloned (huggingface#8260)

* Refactoring the generate() function (huggingface#6949)

* first draft

* show design proposition for new generate method

* up

* make better readable

* make first version

* gpt2 tests pass

* make beam search for gpt2 work

* add first encoder-decoder code

* delete typo

* make t5 work

* save indermediate

* make bart work with beam search

* finish beam search bart / t5

* add default kwargs

* make more tests pass

* fix no bad words sampler

* some fixes and tests for all distribution processors

* fix test

* fix rag slow tests

* merge to master

* add nograd to generate

* make all slow tests pass

* speed up generate

* fix edge case bug

* small fix

* correct typo

* add type hints and docstrings

* fix typos in tests

* add beam search tests

* add tests for beam scorer

* fix test rag

* finish beam search tests

* move generation tests in seperate file

* fix generation tests

* more tests

* add aggressive generation tests

* fix tests

* add gpt2 sample test

* add more docstring

* add more docs

* finish doc strings

* apply some more of sylvains and sams comments

* fix some typos

* make fix copies

* apply lysandres and sylvains comments

* final corrections on examples

* small fix for reformer

* [FIX] TextGenerationPipeline is currently broken. (huggingface#8256)

* [FIX] TextGenerationPipeline is currently broken.

It's most likely due to huggingface#8180.
What's missing is a multi vs single string handler at the beginning of
the pipe.
And also there was no testing of this pipeline.

* Fixing Conversational tests too.

* Updated ConversationalPipeline to work with encoder-decoder models (huggingface#8207)

* Updated ConversationalPipeline to work with encoder-decoder models (e.g. BlenderBot)

* Addition of integration test for EncoderDecoder conversation model

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Fix Tatoeba skip

* forward the worker stderr to the parent process (huggingface#8262)

* [examples] minimal version requirement run-time check in PL (huggingface#8133)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* make files independent (huggingface#8267)

* Clean Trainer tests and datasets dep (huggingface#8268)

* improve documentation of training_args.py (huggingface#8270)

* improve documentation of training_args.py

- do_train
- do_eval
- do_predict

* fix line too long

* fix style with black on training_args.py

* Update src/transformers/training_args.py

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

* Update src/transformers/training_args.py

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

* Update src/transformers/training_args.py

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

* fix line length with utils/style_doc

* black reformatting

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

* Data collator for token classification (huggingface#8274)

* Add DataCollatorForTokenClassification and clean tests

* Make quality

* [CIs] Better reports everywhere (huggingface#8275)

* make it possible to invoke testconf.py in both test suites without crashing on having the same option added

* perl -pi -e 's|--make_reports|--make-reports|' to be consistent with other opts

* add `pytest --make-reports` to all CIs (and artifacts)

* fix

* [WIP] Ner pipeline grouped_entities fixes (huggingface#5970)

* Bug fix: NER pipeline shouldn't group separate entities of same type

* style fix

* [Bug Fix] Shouldn't group entities that are both 'B' even if they are same type
	(B-type1 B-type1) != (B-type1 I-type1)
[Bug Fix] add an option `ignore_subwords` to ignore subsequent ##wordpieces in predictions. Because some models train on only the first token of a word and not on the subsequent wordpieces (BERT NER default). So it makes sense doing the same thing at inference time.
	The simplest fix is to just group the subwords with the first wordpiece.
	[TODO] how to handle ignored scores? just set them to 0 and calculate zero invariant mean ?
	[TODO] handle different wordpiece_prefix ## ? possible approaches:
		get it from tokenizer? but currently most tokenizers dont have a wordpiece_prefix property?
		have an _is_subword(token)
[Feature add] added option to `skip_special_tokens`. Cause It was harder to remove them after grouping.
[Additional Changes] remove B/I prefix on returned grouped_entities
[Feature Request/TODO] Return indexes?
[Bug TODO]  can't use fast tokenizer with grouped_entities ('BertTokenizerFast' object has no attribute 'convert_tokens_to_string')

* use offset_mapping to fix [UNK] token problem

* ignore score for subwords

* modify ner_pipeline test

* modify ner_pipeline test

* modify ner_pipeline test

* ner_pipeline change ignore_subwords default to true

* add ner_pipeline ignore_subword=False test case

* fix offset_mapping index

* fix style again duh

* change is_subword and convert_tokens_to_string logic

* merge tests with new test structure

* change test names

* remove old tests

* ner tests for fast tokenizer

* fast tokenizers have convert_tokens_to_string

* Fix the incorrect merge

Co-authored-by: Ceyda Cinarel <snu-ceyda@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* [blenderbot] regex fix (huggingface#8282)

Fixing:

```
src/transformers/tokenization_blenderbot.py:163: DeprecationWarning: invalid escape sequence \s
    token = re.sub("\s{2,}", " ", token)
```

* Fix typo in language-modeling README.md (huggingface#8287)

* [Generate Test] fix greedy generate test (huggingface#8293)

* fix greedy generate test

* delet ipdb

* Fix validation file loading in scripts (huggingface#8298)

* Upgrade resource for doc building

* Revert size change as it doesn't change anything

* Model card: T5-base fine-tuned on QASC (huggingface#8299)

* Update model cards of deepset/roberta-base-squad2 v1 and v2 (huggingface#8241)

* update deepset/roberta-base-squad2 to v2

* Update model_cards/deepset/roberta-base-squad2/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Improve QA pipeline error handling (huggingface#8286)

- The issue is that with previous code we would have the following:

```python
qa_pipeline = (...)
qa_pipeline(question="Where was he born ?", context="")
-> IndexError: Dimension out of range (expected to be in range of [-1, 0], but got 1)
```

The goal here is to improve this to actually return a ValueError
wherever possible.

While at it, I tried to simplify QuestionArgumentHandler's code to
make it smaller and more compat while keeping backward compat.

* adding model cards for distilled models (huggingface#8300)

* adding model cards for distil models

* forgot the languages

* Speedup doc build (huggingface#8301)

* Try -j option

* Try other thing

* Bigger machine

* Test lower sphinx version

* Remove trailing space

* Fix path to old run_language_modeling.py script (huggingface#8302)

* Clean up data collators and datasets (huggingface#8308)

* Clean up data collators and datasets

* Apply suggestions from code review

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Remove needless clone

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md (huggingface#8223)

* Create README.md

* Update README.md

* Apply suggestions from code review

Co-authored-by: Kevin Canwen Xu <canwenxu@126.com>
Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update bug-report.md

* Update PULL_REQUEST_TEMPLATE.md

* Corrected typo in readme (huggingface#8320)

* change TokenClassificationTask class methods to static methods (huggingface#7902)

* change TokenClassificationTask class methods to static methods

Since we do not require self in the class methods of TokenClassificationTask we should probably switch to static methods. Also, since the class TokenClassificationTask does not contain a constructor it is currently unusable as is. By switching to static methods this fixes the issue of having to document the intent of the broken class.

Also, since the get_labels and read_examples_from_file methods are ought to be implemented. Static method definitions are unchanged even after inheritance, which means that it can be overridden, similar to other class methods.

* Trigger Build

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* no warn (huggingface#8329)

* Output global_attentions in Longformer models (huggingface#7562)

* Output global_attentions in Longformer models

* make style

* small refactoring

* fix tests

* make fix-copies

* add for tf as well

* remove comments in test

* make fix-copies

* make style

* add docs

* make docstring pretty

Co-authored-by: patrickvonplaten <patrick.v.platen@gmail.com>

* Make Trainer evaluation handle dynamic seq_length (huggingface#8336)

* Make Trainer evaluation handle dynamic seq_length

* Document behavior.

* Fix test

* Better fix

* Fixes for realsies this time

* Address review comments

* Without forgetting to save...

* [s2s] test_distributed_eval (huggingface#8315)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Docs bart training ref (huggingface#8330)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* [s2s] test_bash_script.py - actually learn something (huggingface#8318)

* use decorator

* remove hardcoded paths

* make the test use more data and do real quality tests

* shave off 10 secs

* add --eval_beams 2, reformat

* reduce train size, use smaller custom dataset

* Model card: T5-base fine-tuned on QuaRel (huggingface#8334)

* Model card: CodeBERT fine-tuned for Insecure Code Detection (huggingface#8247)

* Model card: CodeBERT fine-tuned for Insecure Code Detection

* Update model_cards/mrm8488/codebert-base-finetuned-detect-insecure-code/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Model card: GPT-2 fine-tuned on CommonGen (huggingface#8248)

* [model_cards] Update Italian BERT models and introduce new Italian XXL ELECTRA model 🎉 (huggingface#8343)

* Create README.md (huggingface#8258)

* german medbert model details (huggingface#8266)

* model details

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8327)

* Create README.md (huggingface#8167)

* Create README.md

Telugu BERTU Readme file

* Update model_cards/kuppuluri/telugu_bertu/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8168)

* Create README.md

* Update README.md

* Create README.md (huggingface#8170)

* Create README.md (huggingface#8169)

* Create README.md (huggingface#8255)

* Create README.md

Initial commit

* Updated Read me

Updated

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8312)

* Create README.md

* Update model_cards/ktrapeznikov/gpt2-medium-topic-news/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8338)

fixes

* Fix typo (huggingface#8351)

* Update README.md (huggingface#8360)

Fix websitr address

* [All Seq2Seq model + CLM models that can be used with EncoderDecoder] Add cross-attention weights to outputs (huggingface#8071)

* Output cross-attention with decoder attention output

* Update src/transformers/modeling_bert.py

* add cross-attention for t5 and bart as well

* fix tests

* correct typo in docs

* add sylvains and sams comments

* correct typo

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* fix encoder outputs (huggingface#8368)

* [make] rewrite modified_py_files in python to be cross-platform (huggingface#8371)

* rewrite modified_py_files in python to be cross-platform

* try a different way to test for variable not being ""

* improve comment

* Fix DataCollatorForWholeWordMask (huggingface#8379)

* Fix DataCollatorForWholeWordMask

* Replace all tensorize_batch in data_collator.py

* fix md table (huggingface#8395)

* Add gpt2-medium-chinese model card (huggingface#8402)

* Create README.md

* Update model_cards/mymusise/gpt2-medium-chinese/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* fixed default labels for QA model (huggingface#8399)

* Fix DataCollatorForWholeWordMask again (huggingface#8397)

* [s2s examples test] fix data path (huggingface#8398)

* [s2s test_finetune_trainer] failing multigpu test (huggingface#8400)

* [s2s/distill] remove run_distiller.sh, fix xsum script (huggingface#8412)

* comet_ml temporary fix(huggingface#8410)

* updating tag for exbert viz (huggingface#8408)

* Update README.md (huggingface#8406)

* Fix some tooling for windows (huggingface#8359)

* Fix some tooling for windows

* Fix conflict

* Trigger CI

* examples/docs: caveat that PL examples don't work on TPU (huggingface#8309)

* add evaluate doc - trainer.evaluate returns 'epoch' from training (huggingface#8273)

* add evaluate doc

* fix style with utils/style.doc

* Update src/transformers/trainer.py

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

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

* Bug fix for permutation language modelling (huggingface#8409)

* [fsmt tokenizer] support lowercase tokenizer (huggingface#8389)

* support lowercase tokenizer

* fix arg pos

* Bump tokenizers (huggingface#8419)

* Add new token classification example (huggingface#8340)

* Add new token classification example

* Remove txt file

* Add test

* With actual testing done

* Less warmup is better

* Update examples/token-classification/run_ner_new.py

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Address review comments

* Fix test

* Make Lysandre happy

* Last touches and rename

* Rename in tests

* Address review comments

* More run_ner -> run_ner_old

Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Fix typo

* [fsmt convert script] fairseq broke chkpt data - fixing that (huggingface#8377)

* fairseq broke chkpt data - fixing that

* style

* support older bpecodes filenames - specifically "code" in iwslt14

* Deprecate old data/metrics functions (huggingface#8420)

* [Tests] Add Common Test for Training + Fix a couple of bugs (huggingface#8415)

* add training tests

* correct longformer

* fix docs

* fix some tests

* fix some more train tests

* remove ipdb

* fix multiple edge case model training

* fix funnel and prophetnet

* clean gpt models

* undo renaming of albert

* [docs] remove sshleifer from issue-template :( (huggingface#8418)

* Fix bart shape comment (huggingface#8423)

* [docs] [testing] gpu decorators table (huggingface#8422)

* gpu decorators table

* whitespace

* Update docs/source/testing.rst

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* whitespace

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Check all models are in an auto class (huggingface#8425)

* [github CI] add a multi-gpu job for all example tests (huggingface#8341)

* add a multi-gpu job for all example tests

* run only ported tests

* rename

* explain why env is re-activated on each step

* mark all unported/checked tests with @require_torch_non_multigpu_but_fix_me

* style

* Apply suggestions from code review

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Changing XLNet default from not using memories to 512 context size following paper (huggingface#8417)

* Move XLNet memory length FutureWarning

* isort

* style

* Changed default XLNet memory length

* Model versioning (huggingface#8324)

* fix typo

* rm use_cdn & references, and implement new hf_bucket_url

* I'm pretty sure we don't need to `read` this file

* same here

* [BIG] file_utils.networking: do not gobble up errors anymore

* Fix CI 😇

* Apply suggestions from code review

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

* Tiny doc tweak

* Add doc + pass kwarg everywhere

* Add more tests and explain

cc @sshleifer let me know if better

Co-Authored-By: Sam Shleifer <sshleifer@gmail.com>

* Also implement revision in pipelines

In the case where we're passing a task name or a string model identifier

* Fix CI 😇

* Fix CI

* [hf_api] new methods + command line implem

* make style

* Final endpoints post-migration

* Fix post-migration

* Py3.6 compat

cc @stefan-it

Thank you @stas00

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

* Patch token classification pipeline (huggingface#8364)

* Patch token classification pipeline

* Some added tests for TokenClassificationArgumentHandler (huggingface#8366)

Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>

* Update links from s3 to huggingface.co

* Fix style

* Model sharing rst (huggingface#8439)

* Update RST

* Finer details

* Re-organize

* Style

* Release: v3.5.0

* v3.5.0 documentation

* [s2s/distill] hparams.tokenizer_name = hparams.teacher (huggingface#8382)

* [examples] better PL version check (huggingface#8429)

* Question template (huggingface#8440)

* Remove SO from question template

* Styling

* [docs] improve bart/marian/mBART/pegasus docs (huggingface#8421)

* Add auto next sentence prediction (huggingface#8432)

* Add auto next sentence prediction

* Fix style

* Add mobilebert next sentence prediction

* Windows dev section in the contributing file (huggingface#8436)

* Add a Windows dev section in the contributing file.

* Forgotten link

* Trigger CI

* Rework description

* Trigger CI

* [testing utils] get_auto_remove_tmp_dir more intuitive behavior (huggingface#8401)

* [testing utils] get_auto_remove_tmp_dir default change

Now that I have been using `get_auto_remove_tmp_dir default change` for a while, I realized that the defaults aren't most optimal.

99% of the time we want the tmp dir to be empty at the beginning of the test - so changing the default to `before=True` - this shouldn't impact any tests since this feature is used only during debug.

* simplify things

* update docs

* fix doc layout

* style

* Update src/transformers/testing_utils.py

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

* better 3-state doc

* style

* Apply suggestions from code review

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

* s/tmp/temporary/ + style

* correct the statement

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

* Add missing import (huggingface#8444)

* Add missing import

* Fix dummy objects

* fix t5 special tokens (huggingface#8435)

* using multi_gpu consistently (huggingface#8446)

* s|multiple_gpu|multi_gpu|g; s|multigpu|multi_gpu|g'

* doc

* Add missing tasks to `pipeline` docstring (huggingface#8428)

* [No merge] TF integration testing (huggingface#7621)

* stash

* TF Integration testing for ELECTRA, BERT, Longformer

* Trigger slow tests

* Apply suggestions from code review

* fix t5 token type ids (huggingface#8437)

* Bug fix for modeling utilities function: apply_chunking_to_forward, chunking should be in the chunking dimension, an exception was raised if the complete shape of the inputs was not the same rather than only the chunking dimension (huggingface#8391)

Co-authored-by: pedro <pe25171@mit.edu>

* [model_cards] harmonization

* Fix TF Longformer (huggingface#8460)

* Add next sentence prediction loss computation (huggingface#8462)

* Add next sentence prediction loss computation

* Apply style

* Fix tests

* Add forgotten import

* Add forgotten import

* Use a new parameter

* Remove kwargs and use positional arguments

* Fix next sentence output (huggingface#8466)

* Example NER script predicts on tokenized dataset (huggingface#8468)

The new run_ner.py script tries to run prediction on the input
test set `datasets["test"]`, but it should be the tokenized set
`tokenized_datasets["test"]`

* Add TFDPR (huggingface#8203)

* Create modeling_tf_dpr.py

* Add TFDPR

* Add back TFPegasus, TFMarian, TFMBart, TFBlenderBot

last commit accidentally deleted these 4 lines, so I recover them back

* Add TFDPR

* Add TFDPR

* clean up some comments, add TF input-style doc string

* Add TFDPR

* Make return_dict=False as default

* Fix return_dict bug (in .from_pretrained)

* Add get_input_embeddings()

* Create test_modeling_tf_dpr.py

The current version is already passed all 27 tests!
Please see the test run at : 
https://colab.research.google.com/drive/1czS_m9zy5k-iSJbzA_DP1k1xAAC_sdkf?usp=sharing

* fix quality

* delete init weights

* run fix copies

* fix repo consis

* del config_class, load_tf_weights

They shoud be 'pytorch only'

* add config_class back

after removing it, test failed ... so totally only removing "use_tf_weights = None" on Lysandre suggestion

* newline after .. note::

* import tf, np (Necessary for ModelIntegrationTest)

* slow_test from_pretrained with from_pt=True

At the moment we don't have TF weights (since we don't have official official TF model)
Previously, I did not run slow test, so I missed this bug

* Add simple TFDPRModelIntegrationTest

Note that this is just a test that TF and Pytorch gives approx. the same output.
However, I could not test with the official DPR repo's output yet

* upload correct tf model

* remove position_ids as missing keys

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: patrickvonplaten <patrick@huggingface.co>

* Replaced some iadd operations on lists with proper list methods. (huggingface#8433)

* Skip test until investigation

* Flax/Jax documentation (huggingface#8331)

* First addition of Flax/Jax documentation

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* make style

* Ensure input order match between Bert & Roberta

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Install dependencies "all" when building doc

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* wraps build_doc deps with ""

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Addressing @sgugger comments.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Use list to highlight JAX features.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Make style.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Let's not look to much into the future for now.

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Style

Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>

* [s2s] distill t5-large -> t5-small (huggingface#8376)

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* Update deploy-docs dependencies on CI to enable Flax (huggingface#8475)

* Update deploy-docs dependencies on CI to enable Flax

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* Added pair of ""

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

* [model_cards] other chars than [\w\-_] not allowed anymore in model names

cc @Pierrci

* Fix typo in roberta-base-squad2-v2 model card (huggingface#8489)

* quick fix on concatenating text to support more datasets (huggingface#8474)

* Fix doc bug (huggingface#8500)

* fix doc bug

Signed-off-by: mymusise <mymusise1@gmail.com>

* fix example bug

Signed-off-by: mymusise <mymusise1@gmail.com>

* Model sharing doc (huggingface#8498)

* Model sharing doc

* Style

* fix SqueezeBertForMaskedLM (huggingface#8479)

* Try to understand and apply Sylvain's comments (huggingface#8458)

* Use LF instead of os.linesep (huggingface#8491)

* Add pretraining loss computation for TF Bert pretraining (huggingface#8470)

* Add pretraining loss computation for TF Bert pretraining

* Fix labels creation

* Fix T5 model

* restore T5 kwargs

* try a generic fix for pretraining models

* Apply style

* Overide the prepare method for the BERT tests

* Remove typo

* Update deepset/roberta-base-squad2 model card (huggingface#8522)

* Update README.md

* Update README.md

* Update doc for v3.5.1

* [T5] Bug correction & Refactor (huggingface#8518)

* fix bug

* T5 refactor

* refactor tf

* apply sylvains suggestions

* Model templates encoder only (huggingface#8509)

* Model templates

* TensorFlow

* Remove pooler

* CI

* Tokenizer + Refactoring

* Encoder-Decoder

* Let's go testing

* Encoder-Decoder in TF

* Let's go testing in TF

* Documentation

* README

* Fixes

* Better names

* Style

* Update docs

* Choose to skip either TF or PT

* Code quality fixes

* Add to testing suite

* Update file path

* Cookiecutter path

* Update `transformers` path

* Handle rebasing

* Remove seq2seq from model templates

* Remove s2s config

* Apply Sylvain and Patrick comments

* Apply suggestions from code review

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

* Last fixes from code review

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

* Fix paths in github YAML

* Model sharing doc: more tweaks (huggingface#8520)

* More doc tweaks

* Update model_sharing.rst

* make style

* missing newline

* Add email tip

Co-authored-by: Pierric Cistac <pierric@huggingface.co>

* Add bart-large-mnli model card (huggingface#8527)

* fix load weights (huggingface#8528)

* fix load weights

* delete line

* Rework some TF tests (huggingface#8492)

* Update some tests

* Small update

* Apply style

* Use max_position_embeddings

* Create a fake attribute

* Create a fake attribute

* Update wrong name

* Wrong TransfoXL model file

* Keep the common tests agnostic

* [breaking|pipelines|tokenizers] Adding slow-fast tokenizers equivalence tests pipelines - Removing sentencepiece as a required dependency (huggingface#8073)

* Fixing roberta for slow-fast tests

* WIP getting equivalence on pipelines

* slow-to-fast equivalence - working on question-answering pipeline

* optional FAISS tests

* Pipeline Q&A

* Move pipeline tests to their own test job again

* update tokenizer to add sequence id methods

* update to tokenizers 0.9.4

* set sentencepiecce as optional

* clean up squad

* clean up pipelines to use sequence_ids

* style/quality

* wording

* Switch to use_fast = True by default

* update tests for use_fast at True by default

* fix rag tokenizer test

* removing protobuf from required dependencies

* fix NER test for use_fast = True by default

* fixing example tests (Q&A examples use slow tokenizers for now)

* protobuf in main deps extras["sentencepiece"] and example deps

* fix protobug install test

* try to fix seq2seq by switching to slow tokenizers for now

* Update src/transformers/tokenization_utils_base.py

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Update src/transformers/tokenization_utils_base.py

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Create README.md for Chinese RoBERTa Miniatures (huggingface#8550)

* Create README.md

* Update model_cards/uer/chinese_roberta_L-2_H-128/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Readme for News Headline Generation (bert2bert) (huggingface#8557)

* Readme for Wiki Summary [Persian] bert2bert (huggingface#8558)

* Clearer Model Versioning Example (huggingface#8562)

* [doc] typo fix (huggingface#8535)

* [doc] typo fix

@sgugger

* Update src/transformers/modeling_utils.py

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

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

* Adding the prepare_seq2seq_batch function to ProphetNet (huggingface#8515)

* Simply insert T5Tokenizer's prepare_seq2seq_batch

* Update/Add some 'import'

* fix RunTimeError caused by '.view'

* Moves .view related error avoidance from seq2seq_trainer to inside prophetnet

* Update test_tokenization_prophetnet.py

* Format the test code with black

* Re-format the test code

* Update test_tokenization_prophetnet.py

* Add importing require_torch in the test code

* Add importing BatchEncoding in the test code

* Re-format the test code on Colab

* Fix GPT2DoubleHeadsModel to work with model.generate() (huggingface#6601)

* Fix passing token_type_ids during GPT2DoubleHeadsModel.generate() if used

and for GPT2LMHeadModel too

* Update tests to check token_type_ids usage in GPT2 models

* Update version to v4.0.0-dev (huggingface#8568)

* Switch `return_dict` to `True` by default. (huggingface#8530)

* Use the CI to identify failing tests

* Remove from all examples and tests

* More default switch

* Fixes

* More test fixes

* More fixes

* Last fixes hopefully

* Use the CI to identify failing tests

* Remove from all examples and tests

* More default switch

* Fixes

* More test fixes

* More fixes

* Last fixes hopefully

* Run on the real suite

* Fix slow tests

* Fix mixed precision issue for GPT2 (huggingface#8572)

* Fix mixed precision issue for GPT2

* Forgot one cast

* oops

* Forgotten casts

* Reorganize repo (huggingface#8580)

* Put models in subfolders

* Styling

* Fix imports in tests

* More fixes in test imports

* Sneaky hidden imports

* Fix imports in doc files

* More sneaky imports

* Finish fixing tests

* Fix examples

* Fix path for copies

* More fixes for examples

* Fix dummy files

* More fixes for example

* More model import fixes

* Is this why you're unhappy GitHub?

* Fix imports in conver command

* model_card for indolem/indobert-base-uncased (huggingface#8579)

* T5 & mT5 (huggingface#8552)

* add mt5 and t5v1_1 model

* fix tests

* correct some imports

* add tf model

* finish tf t5

* improve examples

* fix copies

* clean doc

* [MT5] More docs (huggingface#8589)

* add docs

* make style

* Add __init__ to the models folder

* Fix init for MT5 (huggingface#8591)

* Tokenizers: ability to load from model subfolder (huggingface#8586)

* <small>tiny typo</small>

* Tokenizers: ability to load from model subfolder

* use subfolder for local files as well

* Uniformize model shortcut name => model id

* from s3 => from huggingface.co

Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>

* Fix model templates (huggingface#8595)

* First fixes

* Fix imports and add init

* Fix typo

* Move init to final dest

* Fix tokenization import

* More fixes

* Styling

* these should run fine on multi-gpu (huggingface#8582)

* Fix check repo utils (huggingface#8600)

* Tokenizers should be framework agnostic (huggingface#8599)

* Tokenizers should be framework agnostic

* Run the slow tests

* Not testing

* Fix documentation

* Apply suggestions from code review

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Remove deprecated (huggingface#8604)

* Remove old deprecated arguments

Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

* Remove needless imports

* Fix tests

Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

* Add Harry Potter Model Card (huggingface#8605)

* Add Harry Potter Model

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

* Update model_cards/ceostroff/harry-potter-gpt2-fanfiction/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Remove old doc

* Fixed link to the wrong paper. (huggingface#8607)

* Reset loss to zero on logging in Trainer to avoid bfloat16 issues (huggingface#8561)

* make tr_loss regular float

* Revert "make tr_loss regular float"

This reverts commit c9d7ccf.

* reset loss at each logging step

* keep track of total loss with _total_loss_scalar

* add remaining tr_loss at the end

* Fix DataCollatorForLanguageModeling (huggingface#8621)

* Fix missing space in multiline warning (huggingface#8593)

Multiline string informing about missing PyTorch/TensorFlow had missing space.

* [s2s] broken test (huggingface#8613)

* fix to adjust for huggingface#8530 changes (huggingface#8612)

* self.self.activation_dropout -> self.activation_dropout (huggingface#8611)

(one line typo)

* New TF loading weights (huggingface#8490)

* New TF loading weights

* apply style

* Better naming

* Largely comment the loading method

* Apply style

* Address Patrick's comments

* Remove useless line of code

* Update Docstring

* Address Sylvain's and Lysandre's comments

* Simplify the names computation

* Typos

* Adding PrefixConstrainedLogitsProcessor (huggingface#8529)

* Adding PrefixConstrainedLogitsProcessor

* fixing RAG and style_doc

* fixing black (v20 instead of v19)

* Improving doc in generation_logits_process.py

* Improving docs and typing in generation_utils.py

* docs improvement

* adding test and fixing doc typo

* fixing doc_len

* isort on test

* fixed test

* improve docstring a bit

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* [Tokenizer Doc] Improve tokenizer summary (huggingface#8622)

* improve summary

* small fixes

* cleaned line length

* correct "" formatting

* apply sylvains suggestions

* Fixes the training resuming with gradient accumulation (huggingface#8624)

* Fix training from scratch in new scripts (huggingface#8623)

* model_cards for Chinese Couplet and Poem GPT2 models (huggingface#8620)

* replace performance table with markdown (huggingface#8565)

* replace performance table with markdown

* Update model_cards/smanjil/German-MedBERT/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8544)

Modified Model in Action section. The class `AutoModelWithLMHead` is deprecated so changed it to `AutoModelForSeq2SeqLM` for encoder-decoder models. Removed duplicate eos token.

* Model Card for abhilash1910/financial_roberta (huggingface#8625)

* Model Card for abhilash1910/financial_roberta

* Update model_cards/abhilash1910/financial_roberta/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Update README.md (huggingface#8405)

* Update README.md

* Update README.md

* Add model card for ai4bharat/indic-bert (huggingface#8464)

* Create README.md (huggingface#8363)

* Model card: T5-base fine-tuned on QuaRTz (huggingface#8369)

* Model card: T5-base fine-tuned on QuaRTz

* Update model_cards/mrm8488/t5-base-finetuned-quartz/README.md

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* Create README.md (huggingface#8362)

* Created ModelCard for Hel-ach-en MT model (huggingface#8496)

* Updated ModelCard

* Apply suggestions from code review

Co-authored-by: Julien Chaumond <chaumond@gmail.com>

* [s2s] distillation apex breaks return_dict obj (huggingface#8631)

* apex breaks return_dict obj

* style

* grammar (huggingface#8639)

* Update README.md (huggingface#8635)

* Updated the Extractive Question Answering code snippets (huggingface#8636)

* Updated the Extractive Question Answering code snippets

The Extractive Question Answering code snippets do not work anymore since the models return task-specific output objects. This commit fixes the pytorch and tensorflow examples but adding `.values()` to the model call.

* Update task_summary.rst

* Add cards for all Geotrend models (huggingface#8617)

* docs(bert-base-15lang-cased): add model card

* add cards for all Geotrend models

* [model cards] fix language tag for all Geotrend models

* [model card] : fix bert-base-15lang-cased (huggingface#8655)

the table was badly formatted because of a single line break

* fix missing return dict (huggingface#8653)

* fixed imports

* update example conversion

* removed redundant tests

* added back marker

* reverted to old QA Argument Handler

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Sean Naren <sean@grid.ai>
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
Co-authored-by: Santiago Castro <sacastro@umich.edu>
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Zhiqi Huang <mazicwong@gmail.com>
Co-authored-by: Manuel Romero <mrm8488@gmail.com>
Co-authored-by: Ashwani Tanwar <ashwanitanwar333@gmail.com>
Co-authored-by: Julien Chaumond <chaumond@gmail.com>
Co-authored-by: Ethan <9592150+Ethan-yt@users.noreply.github.com>
Co-authored-by: gurkan08 <33202187+gurkan08@users.noreply.github.com>
Co-authored-by: dartrevan <awesombatsy@gmail.com>
Co-authored-by: Branden Chan <33759007+brandenchan@users.noreply.github.com>
Co-authored-by: yantan <yantan@effyic.com>
Co-authored-by: Santiago Castro <santi.1410@hotmail.com>
Co-authored-by: wlhgtc <hgtcwl@foxmail.com>
Co-authored-by: Sylvain Gugger <sylvain.gugger@gmail.com>
Co-authored-by: Nicolas Patry <patry.nicolas@protonmail.com>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>
Co-authored-by: TFUsers <25044281+TFUsers@users.noreply.github.com>
Co-authored-by: TFUsers <TFUsers@gmail.com>
Co-authored-by: Avital Oliver <avital@thewe.net>
Co-authored-by: Abi See <abigail.e.see@gmail.com>
Co-authored-by: guillaume-be <guillaume.becquin@gmail.com>
Co-authored-by: Kushal <32245327+kushalj001@users.noreply.github.com>
Co-authored-by: Martin Monperrus <martin.monperrus@gnieh.org>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>
Co-authored-by: Philip May <eniak.info@gmail.com>
Co-authored-by: Ceyda Cinarel <15624271+cceyda@users.noreply.github.com>
Co-authored-by: Ceyda Cinarel <snu-ceyda@users.noreply.github.com>
Co-authored-by: Pengzhi Gao <pengzhi.gao@petuum.com>
Co-authored-by: Victor SANH <victorsanh@gmail.com>
Co-authored-by: Yifan Peng <pengyifan.mail@gmail.com>
Co-authored-by: Kevin Canwen Xu <canwenxu@126.com>
Co-authored-by: Guillem García Subies <37592763+GuillemGSubies@users.noreply.github.com>
Co-authored-by: Bobby Donchev <contact@donchev.is>
Co-authored-by: Guillaume Filion <guillaume.filion@gmail.com>
Co-authored-by: Leandro von Werra <lvwerra@users.noreply.github.com>
Co-authored-by: Stefan Schweter <stefan@schweter.it>
Co-authored-by: Jiaxin Pei <pedropei@vip.qq.com>
Co-authored-by: smanjil <shresthamanjil21@gmail.com>
Co-authored-by: Karthik Uppuluri <karthik.uppuluri@gmail.com>
Co-authored-by: hasantanvir79 <hasantanvir79@gmail.com>
Co-authored-by: ktrapeznikov <ktrapeznikov@gmail.com>
Co-authored-by: hassoudi <hassoudi@gmail.com>
Co-authored-by: Jonathan Chang <31893406+cccntu@users.noreply.github.com>
Co-authored-by: Yossi Synett <github@yossisynett.com>
Co-authored-by: Chengxi Guo <mymusise1@gmail.com>
Co-authored-by: Manav Rathod <manav.rathod@berkeley.edu>
Co-authored-by: Julien Plu <plu.julien@gmail.com>
Co-authored-by: Shashank Gupta <shaz4194@gmail.com>
Co-authored-by: Teven <teven.lescao@gmail.com>
Co-authored-by: Shichao Sun <sunshichao1995@gmail.com>
Co-authored-by: Pedro <pedro.colon4@upr.edu>
Co-authored-by: pedro <pe25171@mit.edu>
Co-authored-by: sarnoult <31313050+sarnoult@users.noreply.github.com>
Co-authored-by: Ratthachat (Jung) <56621342+ratthachat@users.noreply.github.com>
Co-authored-by: patrickvonplaten <patrick@huggingface.co>
Co-authored-by: Beomsoo Kim <bluewhale8202@gmail.com>
Co-authored-by: Funtowicz Morgan <mfuntowicz@users.noreply.github.com>
Co-authored-by: Sumithra Bhakthavatsalam <sumithra.b@gmail.com>
Co-authored-by: Antonio Lanza <antoniolanza1996@gmail.com>
Co-authored-by: zeyuyun1 <43428393+zeyuyun1@users.noreply.github.com>
Co-authored-by: Forrest Iandola <fiandola@gmail.com>
Co-authored-by: Pierric Cistac <pierric@huggingface.co>
Co-authored-by: Joe Davison <josephddavison@gmail.com>
Co-authored-by: zhezhaoa <1152543959@qq.com>
Co-authored-by: Mehrdad Farahani <m3hrdadfi@gmail.com>
Co-authored-by: Yusuke Mori <mori@mi.t.u-tokyo.ac.jp>
Co-authored-by: LSinev <LSinev@users.noreply.github.com>
Co-authored-by: fajri91 <fajri91@users.noreply.github.com>
Co-authored-by: Quentin Lhoest <lhoest.q@gmail.com>
Co-authored-by: Caitlin Ostroff <caitlin.ostroff@gmail.com>
Co-authored-by: cronoik <johannes.schaffrath@mail.de>
Co-authored-by: Benjamin Minixhofer <bminixhofer@gmail.com>
Co-authored-by: Michał Pogoda <237372@student.pwr.edu.pl>
Co-authored-by: Nicola De Cao <nicola.decao@gmail.com>
Co-authored-by: hhou435 <59219579+hhou435@users.noreply.github.com>
Co-authored-by: Vishal Singh <vishalsingh7x@gmail.com>
Co-authored-by: Abhilash Majumder <30946547+abhilash1910@users.noreply.github.com>
Co-authored-by: Divyanshu Kakwani <divkakwani@gmail.com>
Co-authored-by: Perez Ogayo <pogayo17@alustudent.com>
Co-authored-by: Tim Isbister <timisbister@gmail.com>
Co-authored-by: Amine Abdaoui <abdaoui@lirmm.fr>
Co-authored-by: Quentin Lhoest <42851186+lhoestq@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants