Commits · 98939e6aee54dc36451f2695d6501f8c7c986381 · chenpangpang / transformers

"...csrc/git@developer.sourcefind.cn:zhaoyu6/sglang.git" did not exist on "8a5480528d71824b7790abc66d11110a2a598d1f"

31 Mar, 2022 3 commits

Spanish translation of the file multilingual.mdx (#16329) · 98939e6a

Juanjo do Olmo authored Mar 31, 2022



* Duplication of the source eng file

* Spanish translation of the file multilingual.mdx

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/multilingual.mdx
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Fix nits and finish translation
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

98939e6a

make tuple annotation more specific to avoid failures during symbolic_trace (#16490) · 99a01423

chenbohua3 authored Mar 31, 2022

* make tuple annotation more specific to avoid failures during symbolic_trace

* make tuple annotation more specific to avoid failures during symbolic_trace

99a01423

Refactor Modeling Outputs (#16341) · a8b6443e

Francesco Saverio Zuppichini authored Mar 31, 2022



* first proposal

* replace model outputs in various models

* conflicts

* docstring

* update poolformer

* minor change in docstring

* CI

* removed poolformer specific outputs from doc

* removed convnext specific outputs from doc

* CI

* weird char in segformer

* conversations

* reverted docstring for BaseModelOutputWithPooling

* update outputs

* changed docstring in BaseModelOutput

* updated docstring in modeling outputs

* typos :)

* fixed typo after copy & paste it all around

* CI

* Apply suggestions from code review
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* segformer
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

a8b6443e

30 Mar, 2022 9 commits

Support reduce_bucket_size=auto for deepspeed stages <3 (#16496) · 857eb87c
Manuel R. Ciosici authored Mar 30, 2022

857eb87c

update smddp api to v1.4.0 (#16371) · 81ac45f8

Lai Wei authored Mar 30, 2022



* update smddp api to v1.4.0

* Update src/transformers/trainer.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Update src/transformers/trainer.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* address comments

* fix style

* remove unused import

* fix indent

* disable style check for import

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

81ac45f8

[examples] max samples can't be bigger than the len of dataset (#16501) · a73281e3
Stas Bekman authored Mar 30, 2022
```
* [examples] max samples can't be bigger than then len of dataset

* do tf and flax
```
a73281e3

Feature Extractor accepts `segmentation_maps` (#15964) · c4deb7b3

Francesco Saverio Zuppichini authored Mar 30, 2022



* feature extractor accepts

* resolved conversations

* added examples in test for ADE20K

* num_classes -> num_labels

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* resolving conversations

* resolving conversations

* removed ADE

* CI

* minor changes in conversion script

* reduce_labels in feature extractor

* minor changes

* correct preprocess for instace segmentation maps

* minor changes

* minor changes

* CI

* debugging

* better padding

* going to update labels inside the model

* going to update labels inside the model

* minor changes

* tests

* removed changes in feature_extractor_utils

* conversation

* conversation

* example in feature extractor

* more docstring in modeling

* test

* make style

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

c4deb7b3

TF: unpack inputs on Convbert, GPTJ, LED, and templates (#16491) · c2f8eaf6

Joao Gante authored Mar 30, 2022

* Add unpack_inputs to remaining models

* remove stray use of inputs in the templates; fix tf.debugging of attn masks

c2f8eaf6

Add support for exporting GPT-J to ONNX-TRT (#16492) · ae189ef9
tomerip authored Mar 30, 2022
```
Add support for exporting GPT-J to ONNX-TRT
Co-authored-by: Tomer Stav <stavt@amazon.com>
```
ae189ef9
Add length to PreTrainedTokenizer train_new_from_iterator (#16493) · d04adc35
dctelus authored Mar 30, 2022

d04adc35
Nit: MCSCOCO -> MS COCO (#16481) · 147c8166
Aditya Kane authored Mar 30, 2022

147c8166

TF GPT-J Type hints and TF decorator (#16488) · ffd19ee1

Dahlbomii authored Mar 30, 2022



* Type hints and TF decorator added

* Type hints and TF decorator added

* make style
Co-authored-by: matt <rocketknight1@gmail.com>

ffd19ee1

29 Mar, 2022 18 commits

Do not initialize `torch.distributed` process group if one is already initailized (#16487) · 277d49a5
Antoni Baum authored Mar 29, 2022
```
* Do not initialize torch process group twice

* Apply suggestions from code review
```
277d49a5
Raise diff tolerance value for TFViTMAEModelTest (#16483) · 2b483230
Yih-Dar authored Mar 29, 2022
```
* Raise diff tolerance value
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
```
2b483230

TF GPT2: clearer model variable naming with @unpack_inputs (#16311) · ee18d4d2

Christopher Akiki authored Mar 29, 2022

* add unpack_inputs decorator to Main Layer

* add unpack_inputs decorator to Model

* add unpack_inputs decorator to LMHead Model

* add unpack_inputs decorator to Double Head Model

* add unpack_inputs decorator to Sequence Classification Model

* run fixup recipe

* make unpack_inputs the first decorator

ee18d4d2

Avoid accessing .dataset of a DataLoader in Trainer (#16451) · d7c8ce57

Sander Land authored Mar 29, 2022



* Avoid accessing .dataset of a dataloader

* style

* fix

* cleaning up, reverting some misunderstandings

* black

* add train_dataset argument to get_train_dataloader, and fix other instances of length checks

* flake8

* address comments

* fix bug

* cleanup

* add test

* Update tests/trainer/test_trainer.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* under torch

* merge

* stylistic suggestion
Co-authored-by: Sander Land <sander@chatdesk.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

d7c8ce57

added typehints for RAG pytorch models (#16416) · 781af736
akashe authored Mar 29, 2022

781af736

Add TF ViT MAE (#16255) · 5b40a37b

Sayak Paul authored Mar 29, 2022



* ported TFViTMAEIntermediate and TFViTMAEOutput.

* added TFViTMAEModel and TFViTMAEDecoder.

* feat: added a noise argument in the implementation for reproducibility.

* feat: vit mae models with an additional noise argument for reproducibility.
Co-authored-by: ariG23498 <aritra.born2fly@gmail.com>
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

5b40a37b

TF: properly handle kwargs in encoder_decoder architectures (#16465) · 7a9ef818
Joao Gante authored Mar 29, 2022
```
* properly handle kwargs in encoder_decoder architectures

* make fixup
```
7a9ef818

Add type hints for UniSpeech (#16399) · 0540d1b6

Dan Tegzes authored Mar 29, 2022

* Add type hints for UniSpeech

* Added type hints for UniSpeechSat

* Added type hints for Wave2Vec2 (PT)

* Added type hints for models dependent of wave2vec

0540d1b6

[doc] Fix missing trainer import (#16469) · 875e07a9
Wesley A. Cheng authored Mar 29, 2022

875e07a9
Add TF vision model code samples (#16477) · 6358a4c8
Yih-Dar authored Mar 29, 2022
```
* add code samples
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
```
6358a4c8
fix wrong variable name (#16467) · 3015d12b
Wesley A. Cheng authored Mar 29, 2022

3015d12b
Fix example test and test_fetcher for examples (#16478) · b62ac4d2
Sylvain Gugger authored Mar 29, 2022

b62ac4d2

Fix some TF GPT-J CI testings (#16454) · 86cff21c

Yih-Dar authored Mar 29, 2022



* Fix for test_mixed_precision

* Fix test_saved_model_creation by using shape_list instead of shape

* skit test_model_from_pretrained on GPU for now to avoid GPU OOM

* skip test_gptj_sample_max_time for now
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

86cff21c

Fix missing output_attentions in PT/Flax equivalence test (#16271) · aebca696

Yih-Dar authored Mar 29, 2022



* fix - set output_attentions to True

* Update tests/test_modeling_flax_common.py

* update for has_attentions

* overwrite check_outputs in FlaxBigBirdModelTest
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>

aebca696

Remove duplicate mLuke (#16460) · 45abb37a
Steven Liu authored Mar 29, 2022
```
* Remove duplicate mLuke

* 🖍 apply feedback
```
45abb37a
[MNLI example] Prevent overwriting matched with mismatched metrics (#16475) · 5216607f
Eldar Kurtic authored Mar 29, 2022
```
* Prevent overwriting matched with mismatched metrics

* Fix style
```
5216607f

Adding DocTest to TrOCR (#16398) · ed31ab3f

Arnaud Stiegler authored Mar 29, 2022



* docstring still WIP | adding to documentation_tests

* clean version | passes tests

* adding to documentation_test

* adding forward for training pass

* make fixup applied

* address comments

* fix doctest

* apply make fixup

* remove additional blank

* fix file to have correct split for prepare_for_doc_test

* Update src/transformers/models/trocr/modeling_trocr.py
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

* address comments

* changing text | adding loss check | make fixup

* make fixup

* Update src/transformers/models/trocr/modeling_trocr.py
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

* Update src/transformers/models/trocr/modeling_trocr.py
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

* Update src/transformers/models/trocr/modeling_trocr.py
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

* make fixup
Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>

ed31ab3f

Fix blenderbot conversion script (#16472) · 85295621
Suraj Patil authored Mar 29, 2022

85295621

28 Mar, 2022 10 commits

Remove kwargs argument from IBERT MLM forward pass (#16449) · c85547af
lewtun authored Mar 28, 2022

c85547af

Translation from english to spanish of file pipeline_tutorial.mdx (#16149) · da936942

Fernando authored Mar 28, 2022



* Add the translation from English to Spanish of the pipeline_tutorial.mdx file

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

* Update docs/source_es/pipeline_tutorial.mdx

Fix typo
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>
Co-authored-by: fernando <fernando@gethitch.ai>
Co-authored-by: Omar U. Espejel <espejelomar@gmail.com>

da936942

Add DPT (#15991) · 979b039c

NielsRogge authored Mar 28, 2022



* First draft

* More improvements

* Add fusion blocks

* Make conversion script work for dpt_large

* Make conversion script work

* Improve implementation

* Improve conversion script

* Add DPTForSemanticSegmentation

* Make conversion work for semantic segmentation

* Add tests

* Remove print statements

* First draft

* Redesign neck

* Improve tests

* Improve implementation some more

* Make neck output list of tensors

* Improve neck and feature extractor

* Fix integration tests

* Make more tests pass

* Make all tests pass

* Add missing config archive map

* Add in_index attribute to make heads accept list of tensors

* Apply suggestions from code review

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Apply some more suggestions

* Add copied from statements

* Remove assert

* Apply suggestions from code review

* Apply suggestions from code review

* Remove DPTInterpolate in favor of nn.Upsample

* Add comments

* Apply suggestions from code review

* Apply suggestions from code review

* Add proposed design

* Update design

* Add DPTReassembleLayer

* Add DPTFeatureFusionStage

* Apply more suggestions from code review

* Apply suggestions from code review

* Apply suggestions from code review

* Fix rebase

* Update in_index and out_indices

* Fix conversion script

* Fix code quality

* Add model to toctree and use DepthEstimatorOutput

* Fix rebase

* Fix code examples

* Improve code

* Fix copied from statements

* Apply suggestions from code review

* Remove compute_loss method

* Apply suggestions from code review

* Fix documentation tests file

* Remove test.py file

* Improve doc example
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MacBook-Pro.local>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Niels Rogge <nielsrogge@nielss-mbp.home>

979b039c

[FlaxSpeechEncoderDecoderModel] Ensure Input and Output Word Embeddings Are **Not** Tied (#16444) · 7ca46335
Sanchit Gandhi authored Mar 28, 2022
```
* [FlaxSpeechEncoderDecoderModel] Ensure Input and Output Word Embeddings Are **Not** Tied

* rebase
```
7ca46335
Fix PerceiverMLP and test (#16405) · e0ac72b7
Jaesun Park authored Mar 28, 2022
```
Co-authored-by: Jaesun Park <jaesun.park1@navercorp.com>
```
e0ac72b7

Use doc builder styler (#16412) · 473709fc

Sylvain Gugger authored Mar 28, 2022

* Config update

* Use doc-builder styler

* Cleanup

* Adapt import

* We need it there too!

473709fc

Update run_t5_mlm_flax.py (#16421) · 8049dfa4
Yongrae Jo authored Mar 28, 2022
```
Fix typo in comment: proprocessed -> preprocessed
```
8049dfa4
[Flax] Improve Robustness of Back-Prop Tests (#16418) · 925fc57b
Sanchit Gandhi authored Mar 28, 2022
```
* [Flax] Improve Robustness of Back-Prop Tests

* check equality of logits/outputs

* make fixup
```
925fc57b
QDQBert example update (#16395) · 7ecbb9c5
Shang Zhang authored Mar 28, 2022
```
* update Dockerfile and utils_qa

* Update README.md
```
7ecbb9c5
`cached_download ∘ hf_hub_url` is `hf_hub_download` (#16375) · f6f6866e
Julien Chaumond authored Mar 28, 2022

f6f6866e