riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-31 10:12:23 +06:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity
7,951 Commits 887 Branches 232 Tags 2.1 GiB
b5bab710f7
Commit Graph

7951 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Patrick von Platen
b5bab710f7
correct (#13585) 2021-09-16 09:07:20 +02:00
Stas Bekman
89da1bfeac
[ci] nightly: add deepspeed master (#13589) 2021-09-15 20:18:34 -04:00
Patrick von Platen
95f933ea85
[Pretrained Model] Add resize_position_embeddings (#13559) 
* finish

* delete bogus file

* correct some stuff

* finish

* finish
 2021-09-15 19:03:56 +02:00
elishowk
c783e14887
upgrade sentencepiece version (#13564) 2021-09-15 15:25:03 +02:00
Suraj Patil
e86c02ea90
Fix GPTNeo onnx export (#13524) 
Update GPT Neo ONNX config to match the changes implied by the simplification of the local attention

Co-authored-by: Michael Benayoun <michael@huggingface.co>
 2021-09-15 13:08:41 +02:00
Bhadresh Savani
3fbb55c757
[Flax] Fixes typo in Bart based Flax Models (#13565) 2021-09-15 11:03:52 +05:30
Sylvain Gugger
7bd16b8776
Fix test_fetcher when setup is updated (#13566) 
* Fix test_fetcher when setup is updated

* Remove example
 2021-09-14 13:33:41 -04:00
elishowk
054b6013c2
separate model card git push from the rest (#13514) 
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2021-09-14 18:07:36 +02:00
Sylvain Gugger
9f318be3d3 Fix yml syntax error 2021-09-14 11:31:17 -04:00
Sylvain Gugger
801ec115cf
Add checks to build cleaner model cards (#13542) 
* Add checks to build cleaner model cards

* Address review comments
 2021-09-14 11:27:32 -04:00
Bhadresh Savani
c1e47bf4fe
[Flax] Addition of FlaxPegasus (#13420) 
* added initial files

* fixes pipeline

* fixes style and quality

* fixes doc issue and positional encoding

* fixes layer norm and test

* fixes quality issue

* fixes code quality

* removed extra layer norm

* added layer norm back in encoder and decoder

* added more code copy quality checks

* update tests

* Apply suggestions from code review

* fix import

* fix test

Co-authored-by: patil-suraj <surajp815@gmail.com>
 2021-09-14 17:15:19 +02:00
Suraj Patil
fc3551a6d7
add flax mbart in auto seq2seq lm (#13560) 2021-09-14 19:06:41 +05:30
Sylvain Gugger
3081d3868e
Push to hub when saving checkpoints (#13503) 
* Push to hub when saving checkpoints

* Add model card

* Revert partial model card

* Small fix for checkpoint

* Add tests

* Add documentation

* Fix tests

* Bump huggingface_hub

* Fix test
 2021-09-14 08:02:15 -04:00
Avital Oliver
51e5eca612
Add long overdue link to the Google TRC project (#13501) 
* Add long-overdue link to the Google TRC project

* Apply suggestions from code review

Co-authored-by: Suraj Patil <surajp815@gmail.com>
Co-authored-by: Stefan Schweter <stefan@schweter.it>
 2021-09-14 13:41:55 +05:30
Lysandre Debut
3ab0185b06
Nightly torch ci (#13550) 
* Nightly CI torch

* Version

* Reformat

* Only subset
Fix

* Revert

* Better formatting

* New channel
 2021-09-13 16:17:29 -04:00
Patrick von Platen
5c14fceac0
return attention mask in int32 (#13543) 2021-09-13 14:02:23 +02:00
SaulLu
149c833b75
Small changes in perplexity.rstto make the notebook executable on google collaboratory (#13541) 
* add imports

* Update docs/source/perplexity.rst

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

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2021-09-13 13:32:32 +02:00
Stas Bekman
f1c22dae7d
[tokenizer] use use_auth_token for config (#13523) 
* [tokenizer] use use_auth_token for config

* args order
 2021-09-13 07:31:35 -04:00
Patrick von Platen
d2904264ab
up (#13538) 2021-09-13 13:07:59 +02:00
Nicolas Patry
65ee1a43e5
fixing BC in fill-mask (wasn't tested in theses test suites (#13540) 
apparently).
 2021-09-13 12:48:54 +02:00
Patrick von Platen
9d60eebeb5
up (#13536) 2021-09-13 11:30:10 +02:00
Xiaohan Zou
a2045067c5
Fix attention mask size checking for CLIP (#13535) 2021-09-13 13:38:38 +05:30
Alex Hedges
68b0baeedc
Ignore past_key_values during GPT-Neo inference (#13521) 2021-09-13 03:06:07 -04:00
holazzer
07c2607d4d
fix use_cache value assign (#13532) 
fix use_cache value assign
 2021-09-13 11:18:50 +05:30
Suraj Patil
010965dcde
[GPT-Neo] Simplify local attention (#13491) 
* simplify local attention

* update tests

* add a comment and use torch.bitwise_xor
 2021-09-10 22:52:20 +05:30
Patrick von Platen
a57d784df5
[Wav2Vec2] Fix dtype 64 bug (#13517) 
* fix

* 2nd fix
 2021-09-10 18:19:10 +02:00
patrickvonplaten
72ec2f3eb5 Docs for v4.10.1 2021-09-10 16:45:19 +02:00
Matt
26d9212e3c
TF multiple choice loss fix (#13513) 
Fix issues with `TFMultipleChoiceLoss` if the choices dimension is None when `build()` is called.
 2021-09-10 14:49:17 +01:00
Patrick von Platen
d7b3b709d0
[Wav2Vec2] Fix normalization for non-padded tensors (#13512) 
* finalize

* Apply suggestions from code review

* finish cleaner implementation

* more tests

* small fix

* finish

* up
 2021-09-10 15:27:16 +02:00
Nicolas Patry
c63fcabfe9
[Large PR] Entire rework of pipelines. (#13308) 
* Enabling dataset iteration on pipelines.

Enabling dataset iteration on pipelines.

Unifying parameters under `set_parameters` function.

Small fix.

Last fixes after rebase

Remove print.

Fixing text2text `generate_kwargs`

No more `self.max_length`.

Fixing tf only conversational.

Consistency in start/stop index over TF/PT.

Speeding up drastically on TF (nasty bug where max_length would increase
a ton.)

Adding test for support for non fast tokenizers.

Fixign GPU usage on zero-shot.

Fix working on Tf.

Update src/transformers/pipelines/base.py

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

Update src/transformers/pipelines/base.py

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

Small cleanup.

Remove all asserts + simple format.

* Fixing audio-classification for large PR.

* Overly explicity null checking.

* Encapsulating GPU/CPU pytorch manipulation directly within `base.py`.

* Removed internal state for parameters of the  pipeline.

Instead of overriding implicitly internal state, we moved
to real named arguments on every `preprocess`, `_forward`,
`postprocess` function.

Instead `_sanitize_parameters` will be used to split all kwargs
of both __init__ and __call__ into the 3 kinds of named parameters.

* Move import warnings.

* Small fixes.

* Quality.

* Another small fix, using the CI to debug faster.

* Last fixes.

* Last fix.

* Small cleanup of tensor moving.

* is not None.

* Adding a bunch of docs + a iteration test.

* Fixing doc style.

* KeyDataset = None guard.

* RRemoving the Cuda test for pipelines (was testing).

* Even more simple iteration test.

* Correct import .

* Long day.

* Fixes in docs.

* [WIP] migrating object detection.

* Fixed the target_size bug.

* Fixup.

* Bad variable name.

* Fixing `ensure_on_device` respects original ModelOutput.
 2021-09-10 14:47:48 +02:00
Stefan Schweter
09549aa18c
examples: minor fixes in flax example readme (#13502) 2021-09-10 11:45:57 +05:30
Nicolas Patry
aacd2123ee
Fixing #13381 (#13400) 
* Fixing #13381

* Enabling automatic LED models.
 2021-09-09 14:23:52 -04:00
Nicolas Patry
db514a75d0
Fixing backward compatiblity for non prefixed tokens (B-, I-). (#13493) 2021-09-09 13:36:09 -04:00
Sylvain Gugger
e59d4d0147
Refactor internals for Trainer push_to_hub (#13486) 2021-09-09 13:04:37 -04:00
Nicolas Patry
3dd538c4d3
[Tentative] Moving slow tokenizer to the Trie world. (#13220) 
* Moving slow tokenizer to the Trie world.

* Adding more docstrings to the Trie.

* Fixing doctest (incompatible wiht our format? )

* Update src/transformers/tokenization_utils.py

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

* Adding a lot more comment into the internals of this algorithm.

* Cleaner doc.

* Fixing the namings.

* Update src/transformers/tokenization_utils.py

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

* quality.

* Fixing longest first match.

* Small improvements to cuts + more test + canine resistant test.

* Fixing fast test.

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
 2021-09-09 17:26:16 +02:00
Matt
b8385d8a11
TF Seq2Seq int dtype fix (#13496) 
Fixes problems with passing int64 input to TF Seq2Seq models.
 2021-09-09 15:54:08 +01:00
Aleksander Smywiński-Pohl
008c2d0b7a
Fix typo in documentation (#13494) 
* Fix typo in deepspeed documentation

* Add missing import in deepspeed configuration

* Fix path in translation examples
 2021-09-09 08:00:05 -04:00
Kamal Raj
1c191efc3a
flax ner example (#13365) 
* flax ner example

* added task to README

* updated readme

* 1. ArgumentParser -> HfArgumentParser
2. step-wise logging,eval and save

* added requirements.txt

* added progress bar

* updated README

* added check_min_version

* updated training data permuattion with JAX

* added metric lib to requirements

* updated readme table

* fixed imports
 2021-09-09 10:12:57 +05:30
Aleksander Smywiński-Pohl
c37573806a
Fix typo in deepspeed documentation (#13482) 
* Fix typo in deepspeed documentation

* Add missing import in deepspeed configuration
 2021-09-08 11:24:10 -07:00
Anton Lozhkov
e1f6e4903a
Fix integration tests for TFWav2Vec2 and TFHubert 2021-09-08 19:51:51 +03:00
Mohan Zhang
41cd52a768
fixed document (#13414) 2021-09-08 11:48:00 -04:00
Koichi Yasuoka
330d83fdbd
Typo in "end_of_word_suffix" (#13477) 
But does it really work?
 2021-09-08 11:26:07 -04:00
Mishig Davaadorj
2a15e8ccfb
Object detection pipeline (#12886) 
* Implement object-detection pipeline

* Define threshold const

* Add `threshold` argument

* Refactor

* Uncomment test inputs

* `rm

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

* Fix typo

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

* Fix typo

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

* Chore better doc

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

* Rm unnecessary lines

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

* Chore better naming

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

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Fix typo

* Add `detr-tiny` for tests

* Add `ObjectDetectionPipeline` to `trnsfrmrs/init`

* Implement new bbox format

* Update detr post_process

* Update `load_img` method obj det pipeline

* make style

* Implement new testing format for obj det pipeln

* Add guard pytorch specific code in pipeline

* Add doc

* Make pipeline_obj_tet tests deterministic

* Revert some changes to `post_process` COCO api

* Chore

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Update src/transformers/pipelines/object_detection.py

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

* Rm timm requirement

* make fixup

* Add timm requirement to test

* Make fixup

* Guard torch.Tensor

* Chore

* Delete unnecessary comment

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
 2021-09-08 17:17:32 +02:00
Matt
707105290b
Fix Tensorflow T5 with int64 input (#13479) 
* Fix Tensorflow T5 with int64 input

* Style pass
 2021-09-08 15:06:04 +01:00
Kevin Canwen Xu
361b6df36a
Throw ValueError for mirror downloads (#13478) 2021-09-08 09:09:22 -04:00
Lysandre Debut
99029ab6b0
Better error raised when cloned without lfs (#13401) 
* Better error raised when cloned without lfs

* add from e
 2021-09-08 08:28:22 -04:00
Li-Huai (Allan) Lin
18447c206d
Enable automated model list copying for localized READMEs (#13465) 
* Complete basic mechanism

* Save

* Complete everything

* Style & Quality

* Update READMEs

* Add testing

* Fix README.md format

* Apply suggestions

* Fix format

* Update utils/check_copies.py

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2021-09-08 08:03:35 -04:00
Sylvain Gugger
cd66539662
Don't modify labels inplace in LabelSmoother (#13464) 2021-09-08 07:45:36 -04:00
Suraj Patil
c164c651dc
[CLIP] fix logit_scale init (#13436) 
* fix logit_scale init

* add logit_scale_init_value as config param
 2021-09-08 14:21:13 +05:30
Kevin Canwen Xu
f667d5b260
Deprecate Mirror for Downloading (#13470) 
* Deprecated Mirror

* revert

* revert

* revert

* fix
 2021-09-08 16:09:44 +08:00