riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-08-03 03:31:05 +06:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity
10,233 Commits 892 Branches 232 Tags 2.1 GiB
3bb6356d4d
Commit Graph

10233 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Patrick von Platen
3bb6356d4d
[From pretrained] Allow download from subfolder inside model repo (#18184) 
* add first generation tutorial

* [from_pretrained] Allow loading models from subfolders

* remove gen file

* add doc strings

* allow download from subfolder

* add tests

* Apply suggestions from code review

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

* apply comments

* correct doc string

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-19 11:53:53 +02:00
Snehan Kekre
ce0152819d
Update docs README with instructions on locally previewing docs (#18196) 
* Update docs README with instructions on locally previewing docs

* Add instructions to install `watchdog` before previewing the docs
 2022-07-19 11:47:26 +02:00
orgoro
798384467b
bugfix: div-->dim (#18135) 2022-07-19 10:24:56 +02:00
Sylvain Gugger
e630dad555
Add vision example to README (#18194) 2022-07-19 09:46:18 +02:00
Duong A. Nguyen
4bea6584e3
Remove use_auth_token from the from_config method (#18192) 
* remove use_auth_token from from_config

* restore use_auth_token from_pretrained run_t5_mlm_flax
 2022-07-19 08:13:20 +02:00
Sylvain Gugger
29fd471556 Use smaller variant of BLOOM for doc to fix tests 2022-07-18 15:17:29 -04:00
Sourab Mangrulkar
bc8e30bab9
FSDP integration enhancements and fixes (#18134) 
* FSDP integration enhancements and fixes

* resolving comments

* fsdp fp16 mixed precision requires `ShardedGradScaler`
 2022-07-19 00:02:10 +05:30
Nicola Procopio
8e445ca51d
Translation/training: italian translation training.mdx (#17662) 
* added training.mdx

* updated training.mdx

* updated training.mdx

* updated training.mdx

* updated _toctree.yml

* fixed typos after review

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-18 19:21:07 +02:00
Younes Belkada
6a1b1bf7a6
BLOOM minor fixes small test (#18175) 
* minor fixes

- add correct revision
- corrected dosctring for test
- removed a test

* contrib credits

Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>
Co-authored-by: Nouamane Tazi <nouamane98@gmail.com>

Co-authored-by: Yih-Dar <2521628+ydshieh@users.noreply.github.com>
Co-authored-by: Nouamane Tazi <nouamane98@gmail.com>
 2022-07-18 19:18:19 +02:00
Nicola Procopio
c4cc894086
Translation italian: multilingual.mdx (#17768) 
* added multilingual.mdx

* updated multilingual.mdx

* italian translation multilingual.mdx

* updated _toctree.yml

* fixed typos _toctree.yml

* fixed typos after review

* fixed error after review
 2022-07-18 19:09:08 +02:00
Nicola Procopio
0a5b61d004
Added preprocessing.mdx italian translation (#17600) 
* updated _toctree.yml

* added preprocessing

* updated preprocessing.mdx

* updated preprocessing.mdx

updated after review
 2022-07-18 19:06:10 +02:00
SaulLu
ced1f1f5db
fix typo inside bloom documentation (#18187) 2022-07-18 17:43:52 +02:00
Sylvain Gugger
edadfc58af
Better default for offload_state_dict in from_pretrained (#18183) 2022-07-18 16:02:41 +02:00
Sylvain Gugger
aeeab1ffd0
Fix template for new models in README (#18182) 2022-07-18 16:01:51 +02:00
Ayan Sengupta
45255814a2
FIX: Typo (#18156) 2022-07-18 15:46:08 +02:00
Yih-Dar
6561fbcc6e
Update TF(Vision)EncoderDecoderModel PT/TF equivalence tests (#18073) 
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2022-07-18 15:29:14 +02:00
Yih-Dar
cb19c2afdc
Fix expected loss values in some (m)T5 tests (#18177) 
* fix expected loss values

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2022-07-18 15:26:21 +02:00
Wang, Yi
7417f3acb7
[HPO] update to sigopt new experiment api (#18147) 
* [HPO] update to sigopt new experiment api
* follow https://docs.sigopt.com/experiments

Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>

* [HPO] use new API if sigopt version >= 8.0.0

Signed-off-by: Wang, Yi A <yi.a.wang@intel.com>
 2022-07-18 15:19:40 +02:00
gcheron
8c14b342aa
add ONNX support for LeVit (#18154) 
Co-authored-by: Guilhem Chéron <guilhemc@authentifier.com>
 2022-07-18 15:17:07 +02:00
Lysandre Debut
c1c79b0655
NLLB tokenizer (#18126) 
* NLLB tokenizer

* Apply suggestions from code review - Thanks Stefan!

Co-authored-by: Stefan Schweter <stefan@schweter.it>

* Final touches

* Style :)

* Update docs/source/en/model_doc/nllb.mdx

Co-authored-by: Stefan Schweter <stefan@schweter.it>

* Apply suggestions from code review

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

* PR reviews

* Auto models

Co-authored-by: Stefan Schweter <stefan@schweter.it>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-18 08:12:34 -04:00
John Giorgi
a4f97e6ce0
Fix incorrect type hint for lang (#18161) 2022-07-18 09:53:18 +02:00
John Giorgi
c46d39f390
Fix check for falsey inputs in run_summarization (#18155) 2022-07-18 09:50:32 +02:00
Nicolas Patry
ccc0897804
Adding support for device_map directly in pipeline(..) function. (#17902) 
* Adding support for `device_map` directly in `pipeline(..)` function.

* Updating the docstring.

* Adding a better docstring

* Put back type hints.

* Blacked. (`make fixup` didn't work ??!!)
 2022-07-15 15:54:26 +02:00
Nicolas Patry
fca66ec4ef
Fixing a hard to trigger bug for text-generation pipeline. (#18131) 
* Fixing a bug where attention mask was not passed to generate.

* Fixing zero-size prompts.

* Comment on top.
 2022-07-15 15:54:07 +02:00
amyeroberts
8581a798c0
Add TF DeiT implementation (#17806) 
* Initial TF DeiT implementation

* Fix copies naming issues

* Fix up + docs

* Properly same main layer

* Name layers properly

* Initial TF DeiT implementation

* Fix copies naming issues

* Fix up + docs

* Properly same main layer

* Name layers properly

* Fixup

* Fix import

* Fix import

* Fix import

* Fix weight loading for tests whilst not on hub

* Add doc tests and remove to_2tuple

* Add back to_2tuple
Removing to_2tuple results in many downstream changes needed because of the copies checks

* Incorporate updates in Improve vision models #17731 PR

* Don't hard code num_channels

* Copy PyTorch DeiT embeddings and remove pytorch operations with mask

* Fix patch embeddings & tidy up

* Update PixelShuffle to move logic into class layer

* Update doc strings - remove PT references

* Use NHWC format in internal layers

* Fix up

* Use linear activation layer

* Remove unused import

* Apply suggestions from code review

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Move dataclass to top of file

* Remove from_pt now weights on hub

* Fixup

Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Amy Roberts <amyeroberts@users.noreply.github.com>
 2022-07-13 18:04:08 +01:00
Wei
7ea6ccc2b3
Enable torchdynamo with torch_tensorrt(fx path) (#17765) 
* enable fx2trt

* Update perf_train_gpu_one.mdx

* Update perf_train_gpu_one.mdx

* add lib check

* update

* format

* update

* fix import check

* fix isort

* improve doc

* refactor ctx manager

* fix isort

* black format

* isort fix

* fix format

* update args

* update black

* cleanups

* Update perf_train_gpu_one.mdx

* code refactor

* code refactor to init

* remove redundancy

* isort

* replace self.args with args

Co-authored-by: Stas Bekman <stas@stason.org>
 2022-07-13 12:43:28 -04:00
Sylvain Gugger
37aeb5787a
Make sharded checkpoints work in offline mode (#18125) 
* Make sharded checkpoints work in offline mode

* Add test
 2022-07-13 12:43:08 -04:00
Sylvain Gugger
0a21a48564 Revert "Make sharded checkpoints work in offline mode" 
This reverts commit 3564c65786.
 2022-07-13 10:53:25 -04:00
Sylvain Gugger
3564c65786 Make sharded checkpoints work in offline mode 2022-07-13 10:51:56 -04:00
lmagne
56e6487c40
add dataset split and config to model-index in TrainingSummary.from_trainer (#18064) 
* added metadata to training summary

* Apply suggestions from code review

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

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-13 16:07:20 +02:00
John Giorgi
fde22c75a1
Add summarization name mapping for MultiNews (#18117) 
* Add summarization name mapping for MultiNews

* Add summarization name mapping for MultiNews
 2022-07-13 08:19:20 -04:00
Sebastian Sosa
195133363e
supported python versions reference (#18116) 
* supported python versions reference

* Update CONTRIBUTING.md

removing commit hash from link

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

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-13 08:18:44 -04:00
Joao Gante
20509ab0e0
TF: unpack_inputs decorator independent from main_input_name (#18110) 2022-07-13 10:43:41 +01:00
Joao Gante
fcefa200b2
TF: remove graph mode distinction when processing boolean options (#18102) 2022-07-12 19:05:31 +01:00
Niklas Muennighoff
bc34c21191
Fix BLOOM dtype (#17995) 
* Add fp16 option

* Fix BLOOM dtype

* Formatting

* Remove torch_dtype arg

* Revert formatting

* Apply formatting

* Add n_embed backward compat
 2022-07-12 10:36:08 -04:00
Joao Gante
981714efe1
CLI: reenable pt_to_tf test (#18108) 2022-07-12 13:38:05 +01:00
wei zhao
f5221c06e4
Report value for a step instead of epoch. (#18095) 
* Report value for a step instead of epoch.

Report an objective function value for a step instead of epoch to optuna.
I made this modification for the following reason:
If "eval_steps" is less than steps per epoch, there maybe warnings like this: "optuna/trial/_trial.py:592: UserWarning: The reported value is ignored because this `step` 0 is already reported.". So "step" are more appropriate than "epoch" here.

* MOD: make style.

Co-authored-by: zhaowei01 <zhaowei01@yuanfudao.com>
 2022-07-12 08:18:35 -04:00
Sijun He
d4ebd4e112
speed up test (#18106) 2022-07-12 04:28:28 -04:00
jianan-gu
b7d8bd378c
Enhance IPEX integration in Trainer (#18072) 
* enhance ipex import

* refine codes

* refine style

* add link

* style

Co-authored-by: Stas Bekman <stas@stason.org>
 2022-07-11 21:34:09 -07:00
Younes Belkada
a462fc9232
Bloom Optimize operations (#17866) 
* fix tolerance for a bloom slow test

* enhance alibi padding

- get rid of for loops
- deals better with padded batched input
- avoid useless cpu/gpu communication when creating alibi

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

* optimize attention mask

* fix scaled softmax limit values

* optimize building alibi tensor

Co-authored-by: Younes Belkada <younesbelkada@users.noreply.github.com>

* fix attention_mask shape when it's None

* minor fixes

- fix docstring + arg names

* remove colons in docstring

* Apply suggestions from code review

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

* apply suggestion

* remove unsued arg

* refactor a bit

- use [:, None] for consistency

* refactor attention block

Co-authored-by: Nouamane Tazi <nouamane98@gmail.com>

* quick fixes

* first attempt

* refactor attention block and fix all tests except "test_simple_generation"

- added comments to better explain attention block

* remove debug lines and add TODO comment

* change `torch.bmm` to `torch.baddbmm`
- fixes `test_simple_generation`but breaks `test_batch_generation_padd`

* styling

* all tests are passing now
- use `bmm`
- add explanation for `allow_fp16_reduced_precision_reduction`

Co-authored-by: Younes Belkada <younesbelkada@users.noreply.github.com>

* styling

Co-authored-by: Younes Belkada <younesbelkada@users.noreply.github.com>

* fix support for accelerate

Co-authored-by: Younes Belkada <younesbelkada@users.noreply.github.com>

* Apply suggestions from code review

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

* remove attn softmax in fp32

* refactor comments

* refactor a bit

- remove warning message
- remove print on test

* refer to pytorch t5

* change the slow tests

- do the tests in fp32
- remove some comments
- keep large comments

* update expected output for `test_simple_generation`
- we now test using fp32

* make style + change comments a bit

* fix dtype padd test

Co-authored-by: justheuristic <justheuristic@gmail.com>
Co-authored-by: Nouamane Tazi <nouamane98@gmail.com>
Co-authored-by: Younes Belkada <younesbelkada@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
 2022-07-11 13:16:13 -04:00
Sylvain Gugger
5ff6f853d7 Mark slow test as such 2022-07-11 12:48:57 -04:00
Sylvain Gugger
b1b8222d80
Add filename to info diaplyed when downloading things in from_pretrained (#18099) 2022-07-11 12:45:06 -04:00
Sylvain Gugger
6c8017a5c8
Fix image segmentation and object detection pipeline tests (#18100) 2022-07-11 12:41:56 -04:00
Sylvain Gugger
b0520f594c Skip failing tests 2022-07-11 10:16:54 -04:00
Duong A. Nguyen
1e8140caad
Fix RESOURCE_EXHAUSTED error when dealing with large datasets in Flax example scripts (#18069) 
* Fix RESOURCE_EXHAUSTED error for large datasets on Flax example scripts

* using np.permutation for creating batch_idx

* train_samples_idx -> training_samples_idx

* fix type hints
 2022-07-11 15:59:08 +02:00
Yih-Dar
ac98a88fbc
Fix torchscript tests for GPT-NeoX (#18012) 
* fix dtype issue in _attn

* fix RotaryEmbedding

* fix RotaryEmbedding 2

* clean up

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2022-07-11 05:02:54 -04:00
Yulv-git
95113d1365
Fix some typos. (#17560) 
* Fix some typos.

Signed-off-by: Yulv-git <yulvchi@qq.com>

* Fix typo.

Signed-off-by: Yulv-git <yulvchi@qq.com>

* make fixup.
 2022-07-11 05:00:13 -04:00
Stas Bekman
ad28ca291b
[bloom] fix alibi device placement (#18087) 2022-07-10 09:11:46 -07:00
neverix
8b332a6a16
Make predict() close progress bars after finishing (#17952) (#18078) 
* Make Trainer.predict call on_evaluate (#17952)

* Add on_predict

* Small fix

* Small and different fix

* Add tests
 2022-07-08 16:44:24 -04:00
Sylvain Gugger
7c046c5c22
Update localized READMES when template is filled. (#18062) 2022-07-08 11:08:52 -04:00