bofeng huang
c8545d2a9c
[Whisper] Add SpecAugment ( #21298 )
...
* Return and rescale attention_mask
* Add SpecAugment to Whisper modeling
* Fix test
* Update docstring
* Add SpecAug related parameters to model config
* Add the _mask_input_features function to doc
* Fix quality
* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Remove dev comments
* Add test
* Resolve conflict
* feat: mask {feature, time} prob fast tests
* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: sanchit-gandhi <sanchit@huggingface.co>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
2023-02-24 11:07:52 +01:00
Sanchit Gandhi
75bd49ff88
[Flax] Fix erroneous kwargs being passed to generate config ( #21765 )
2023-02-24 09:59:18 +01:00
Arthur
14f33205a7
Different behavior in DistilBERT when using "inputs_embeds" ( #21752 )
...
* Different behavior in DistilBERT when using "inputs_embeds"
Fixes #21089
* fix failing test
2023-02-24 09:48:07 +01:00
Sanchit Gandhi
13489248fa
[Examples] Generalise run audio classification for log-mel models ( #21756 )
...
* [Examples] Generalise run audio classification for log-mel models
* batch feature extractor
* make style
2023-02-24 09:19:07 +01:00
Shubhamai
f7ca656f07
[Flax] adding support for batch norm layers ( #21581 )
...
* [flax] adding support for batch norm layers
* fixing bugs related to pt+flax integration
* cleanup, batchnorm support in sharded pt to flax
* support for batchnorm tests in pt+flax integration
* simplifying checking batch norm layer
2023-02-24 08:47:33 +01:00
Connor Henderson
279008adc3
fix: Change is_last chunk calc and add conditional break in chunk_iter ( #21612 )
...
* fix: Change is_last chunk calc and add conditional break
* format fix
* account for 0 and full stride_rights, add comment
* add new test
* make style
* update slow whisper asr test timestamps
* use nested_simplify on output and round timestamp to hundreths place
2023-02-24 08:30:32 +01:00
Clémentine Fourrier
4446b6b094
Graphormer fix ( #21699 )
...
* Removed useless check for backend
* fix style check for graphormer
* Reverted change and corrected requires_backend for cython
* code qual
2023-02-24 08:20:52 +01:00
Stas Bekman
633062639b
[deepspeed tests] fix issues introduced by #21700 ( #21769 )
...
* [deepspeed tests] fix issues introduced by #21700
* fix
* fix
2023-02-23 13:22:25 -08:00
Maria Khalusova
04d90ac49e
Auto api Value Error addition to Troubleshoot ( #21708 )
...
* troubleshooting guide: added an error description for missing auto-mapping
* minor polishing
* changed the example
* Apply suggestions from code review
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/en/troubleshooting.mdx
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
2023-02-23 11:51:18 -05:00
Batese2001
0ffa22f9f6
Added Type Hints for modeling_tf_encoder_decoder.py ( #21673 )
...
* Ran Black formatting
* Added imports and reformatted
* Update src/transformers/models/encoder_decoder/modeling_tf_encoder_decoder.py
---------
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
2023-02-23 14:08:26 +00:00
ydshieh
aa3787c8f0
Skip test_log_level for now
2023-02-23 12:11:20 +01:00
Joao Gante
1d4b797852
Generate: Fix GIT batched captioning ( #21738 )
2023-02-23 09:50:37 +00:00
Younes Belkada
78a93d17c0
[`GPTNeo`] Fix gradient checkpointing bug ( #21733 )
...
* fix bug
* forward contrib credits from discussions
* change logic
---------
Co-authored-by: edbeeching <edbeeching@users.noreply.github.com>
2023-02-23 09:48:19 +01:00
Yih-Dar
36a6a1adb6
Fix 2 quicktour file doctest ( #21742 )
...
* Update expect output values - as Hub repo. files are updated
* Update expect output values - as librosa is from 0.9.2 to 0.10.0 on CI docker
* fix
* update one more
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-23 09:41:28 +01:00
Yih-Dar
ff143ae10e
Update doctest GH workflow file ( #21744 )
...
update
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-23 09:40:53 +01:00
Naga Sai Abhinay
448e050b0d
Make ImageProcessorMixin compatible with subfolder kwarg ( #21725 )
...
* Add subfolder support
* Add kwarg docstring
* formatting fix
* Add test
2023-02-23 09:28:18 +01:00
Thomas Paviot
064f374874
typos in french documentation ( #21750 )
2023-02-23 09:17:01 +01:00
Maria Khalusova
619d51e01f
Added "Open in Colab" to task guides ( #21729 )
...
added Open in Colab to task guides
2023-02-22 08:32:35 -05:00
Matt
d913f4aa40
Fix to KerasMetricCallback when the model returns unstructured output ( #21727 )
...
* Stop doing dict-things to non-dict inputs
* Add a debug check
* Add a debug check
* Remove debug checks, looks good now!
* make fixup
2023-02-22 13:15:14 +00:00
Sanchit Gandhi
82e61f3445
[SpeechT5HifiGan] Handle batched inputs ( #21702 )
...
* [SpeechT5HifiGan] Handle batched inputs
* fix docstring
* rebase and new ruff style
2023-02-22 11:16:56 +01:00
Yih-Dar
09127c5713
Fix `GPTSanJapaneseModel` ( #21731 )
...
* fix
* skip test_model_parallelism
* skip test_model_parallelism
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-22 11:09:04 +01:00
Yih-Dar
aff87da15b
Fix `ErnieMEmbeddings` device issue ( #21726 )
...
* remove .parameters()).device
* fix
* fix
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-22 10:57:34 +01:00
Yih-Dar
2f2b19ff40
Change doc example for `BigBirdForQuestionAnswering` ( #21723 )
...
Change doc example for BigBirdForQuestionAnswering
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-22 10:55:12 +01:00
Yih-Dar
354b338316
Remove `gptsan_japanese` from doctest list to avoid GPU OOM ( #21722 )
...
remove from doctest list to avoid GPU OOM
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-22 10:51:00 +01:00
Sylvain Gugger
b19d64d852
Respect documentation on passive log level ( #21700 )
...
* Respect documentation on passive log level
* Fix test and set log level in examples
* Add doc
2023-02-22 09:39:18 +01:00
Sylvain Gugger
ee6e71e29c
Fix quality
2023-02-22 03:36:15 -05:00
Younes Belkada
24b930ad1d
[`MBart`] Fix cross attention mask check ( #21730 )
...
fix typo
2023-02-22 09:21:25 +01:00
Aaron Gokaslan
5e8c8eb5ba
Apply ruff flake8-comprehensions ( #21694 )
2023-02-22 09:14:54 +01:00
Kashif Rasul
df06fb1f0b
Time series transformer: input projection and Std scaler ( #21020 )
...
* added loc and scale outputs from scalers
* fix typo
* fix tests
* fixed formatting
* initial StdScaler
* move scaling to optional str
* calculate std feature for scalers
* undid change as it does not help
* added StdScaler with weights
* added input projection layer and d_model hyperparam
* use linear proj
* add back layernorm_embedding
* add sin-cos pos embeddings
* updated scalers
* formatting
* fix type
* fixed test
* fix repeated_past_values cal.
* fix when keepdim=false
* fix default_scale
* backward compatibility of scaling config
* update integration test expected output
* fix style
* fix docs
* use the actual num_static_real_features in feature_dim cal
* clarified docs
* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
* prediction_length is not optional
* fix for reviewer
* Update src/transformers/models/time_series_transformer/configuration_time_series_transformer.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* get rid of un-needed new lines
* fix doc
* remove unneeded new lines
* fix style
* static_categorical_features and static_real_features are optional
* fix integration test
* Update src/transformers/models/time_series_transformer/modeling_time_series_transformer.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
* fixing docs for multivariate setting
* documentation for generate
---------
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
2023-02-22 07:50:13 +01:00
mollerup23
bb5a2f2fc3
Adding type hints to call() functions in this file ( #21548 )
...
* Adding type hints to call() functions in this file
* make fixup
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
* Update src/transformers/models/marian/modeling_tf_marian.py
---------
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>
2023-02-21 16:28:33 +00:00
Maria Khalusova
78a53d59cb
Adding task guides to resources ( #21704 )
...
* added resources: links to task guides that support these models
* minor polishing
* conflict resolved
* link fix
* Update docs/source/en/model_doc/vision-encoder-decoder.mdx
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
2023-02-21 10:35:11 -05:00
Yih-Dar
03aaac3502
Fix TVLT (torch device issue) ( #21710 )
...
* fix tvlt ci
* fix tvlt ci
* fix tvlt ci
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-21 11:37:49 +01:00
Yih-Dar
4c6346cc3e
Fix `get_class_in_module` ( #21709 )
...
Fix get_class_in_module
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-21 09:39:15 +01:00
Yih-Dar
ed6ceb7649
Fix typo in `PROCESSOR_MAPPING_NAMES` and add tests ( #21703 )
...
* Add test
* Fix GITProcessor
* Update
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-21 09:38:26 +01:00
Arthur
4deaa534f5
remove position ids and token type ids from forward args in docstring ( #21701 )
2023-02-21 07:01:36 +01:00
Ishan Jindal
c40e3581c7
Fix axial positional encoding calculations for reformer.mdx ( #21649 )
...
* Update reformer.mdx
Fix axial positional encoding calculations
* Update docs/source/en/model_doc/reformer.mdx
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2023-02-21 06:59:51 +01:00
Jonatan Kłosko
deafc24388
Add WhisperTokenizerFast ( #21222 )
...
* Add WhisperTokenizerFast
* Fixup
* Up
* Up
* Improve tests
* Update src/transformers/models/whisper/tokenization_whisper_fast.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* Keep stride in whisper pipelien test
* Remove unknown token special case
* Reduce vocabulary size in tests
* Fix vocab size assertion
* Sync copied changes from WhisperTokenizer
* Skip pipeline tests
* Update assertion
* Remove Whisper tokenizer dependency on sentencepiece
* Format
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2023-02-21 06:58:54 +01:00
Sylvain Gugger
8b3db33a76
Pass along revision in dynamic code fetch ( #21698 )
2023-02-20 21:21:42 +01:00
Arthur
4194e5f42b
Fix-rag-finetune-project-requirement ( #21697 )
...
pin pytorch lightning requirement
2023-02-20 17:23:39 +01:00
Alara Dirik
49ab16239c
Add EfficientNet ( #21563 )
...
* Add EfficientNet to transformers
2023-02-20 16:37:11 +03:00
Younes Belkada
c9a0671477
[`bnb`] fix `bnb` decoders bug ( #21688 )
...
* fix `bnb` decoders bug
* make fixup
2023-02-20 12:21:58 +00:00
tanreinama
f56174ac5b
add GPTSAN model (reopen) ( #21291 )
...
* add GPTSAN-Japanese
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN (update for review)
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* fix typo in comment text
* add GPTSAN
* add GPTSAN
* add GPTSAN
* add GPTSAN
* fix document and comments
* fix class name GPTSAN->GPTSan
* fix import and test for tokenizer
2023-02-20 11:25:27 +01:00
Sylvain Gugger
c87bbe1ff0
Fix quality
2023-02-20 03:27:09 -05:00
Morgan McGuire
011cc17a81
Fix for non-contiguous label tensors in VisonEncoderDecoder ( #21582 )
...
* add prints
* add shape
* add reshape
* clean up
2023-02-20 09:23:46 +01:00
Andy Ehrenberg
2840272c5f
add flax whisper implementation ( #20479 )
...
* add flax whisper implementation
* rever change to setup
* remove unused imports
* revert generation changes
* flax whisper docs
* docs
* import order
* import sorting
* isort
* add dummy objects
* doc formatting
* formatting
* remove trailing whitespaces
* fix flax whisper docs
* add generation logic to unlock flax whisper
* remove scans
* give credits to Flax Bart implementation
* remove unused imports
* add license
* remove assert
* more credits to Bart
* fix style
* formatting
* support left padding
* add flax whisper generation test
* remove copied from comments whenever not a full copy
* fix docstrings for logits processors
* revert change to FlaxForceTokensLogitsProcessor
* revert doc changes
* improve generation docs
* reorganize
* formatting
* cleanup docs
* add tests
* handle empty list case
* fix forced decoder ids in flax tests
* add flax whisper to inits
* upate dummy objects
* docs for FlaxAutoModelForSpeechSeq2Seq
* fix decoder_position_ids computation in pretrained model decode/__call__ fns
* add Copied from statements as necessary
* compute position_ids only in __call__ and decode methods of pretrained model subclasses
* improve readabilityof compute positional embeddings
* check dimensionality of input_features instead of hidden_states
* copied from statement for init_cache
* formatting
* fix copies
* fix copies
* pass attention mask to encoder layers
* fix decoder module outputs
* set dtype
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
* smaller flax model for whisper test
* Update src/transformers/generation/flax_utils.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* Update src/transformers/models/whisper/modeling_flax_whisper.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* Update tests/models/whisper/test_modeling_flax_whisper.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* cleanup
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* Update src/transformers/models/whisper/modeling_flax_whisper.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
* bias cleanup
* doc fix
* align style for force tokens processor
* readability
* fix input shape in tests
* revert FlaxGenerationMixin docstring
* formatting
* fix tests
* fix imports
* consistent encoder hidden states
* consistent hidden states
* input shapes
* typo
* partial class trick
* partial class for input shape
* base_class with correct input shape
* partial base classes
* match by name
* set main_input_name
* compare on names
* formatting
* remove unused import
* safer position ids computation
* safer position id computation
* Update src/transformers/models/whisper/modeling_flax_whisper.py
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
* Update src/transformers/models/whisper/modeling_flax_whisper.py
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
* remove identical inherited tests
* fix prompt ids in tests
* use generation config
* use jnp array
* better var names
* more explicit bias use
* import transformers
* formatting
* test formatting
* remove unused imports
* remove unused imports
* formatting
* isort
* docs
* fix ln orders for encoder hidden states
* whisper unique generation stuff
* flake
* use finfo for attention bias
* docs
* Update src/transformers/generation/flax_utils.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* docs
* add timestamp flax test
* jit for timestamps
* formatting
* clean up timestamps processor
* formatting
* remove if_true
* cleanup
---------
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
2023-02-20 09:17:40 +01:00
AlexWertheim
7735e0406f
Enable PyTorch/XLA Fully Sharded Data Parallel (FSDP) ( #21406 )
...
* Reinserted import statement accidentally removed during rebasing.
* Added auto_wrap functionality, restructured XLA FSDP logic to more closely match PyTorch FSDP logic.
* Fixed flag descriptions; changed several instances of fsdp_ to xla_fsdp_; pass in auto_wrap_policy and auto_wrapper_callable directly to avoid lambda saving.
* Moved XLA FSDP logic to be adjacent to Fairscale FSDP logic in trainer.
* Formatted changes in accordance with HF style requirements.
* Added back in warning which was accidentally removed.
* - Merged XLA FSDP training arguments into `fsdp_config`
- Added `xla` boolean flag to `fsdp_config` to specify XLA FSDP wrapping
- Merged XLA FSDP wrapping logic into FSDP wrapping logic within trainer
class
* Cleaned up errors, moved argument to fsdp_config
- Set `xla` and `xla_fsdp_grad_ckpt` flags by default in fsdp_config
- Added missing colons following conditionals
- Moved `fsdp_transformer_layer_cls_to_wrap` to `fsdp_config`
- Modified `fsdp_transformer_layer_cls_to_wrap` to be list of strings,
not just one string
- Changed Fairscale FSDP logic to allow for set of layer classes to wrap
- Removed unnecessary checks for `xla_fsdp`
* Corrected small errors, improved layer class flag
- Correctly set default values for `xla` and `xla_fsdp_grad_ckpt`
arguments
- Made `fsdp_transformer_layer_cls_to_wrap` a list of strings instead of
a single string
- Added processing to ensure that `fsdp_transformer_layer_cls_to_wrap`
works as expected if passed as a single string
- Updated PyTorch FSDP logic to accept a list of layers to wrap, as done
with XLA FSDP
- Replaced instances of `getattr()` with `.get()` for dictionary
retrievals with default values, including when setting
`fsdp_min_num_params`
- Corrected `self.fsdp is not None` to `len(self.fsdp) > 0`
- Removed extraneous `xla_fsdp` argument descriptions from outside
`fsdp_config`
* Changed xla-fsdp-settings to be dictionary
- Modified xla-fsdp-settings to be entered directly as dictionary
instead of loaded through JSON file
- Made small style corrections
* Reverted unintentional local_rank TPU check
* Do not block XLA FSDP if local rank is -1
* Rebased and applied automatic formatting
- Rebased
- Applied automatic formatting changes via `make style`
* Applied automatic formatting with latest version of black
* Replaced expression with
* Reran black examples tests src utils
ruff examples tests src utils --fix
make autogenerate_code
make[1]: Entering directory '/usr/local/google/home/awertheim/HF-FSDP-PR/transformers'
make[1]: Leaving directory '/usr/local/google/home/awertheim/HF-FSDP-PR/transformers' after additional formatting changes
* Additionall automatic formatting changes
* Remove unnecessary whitespace characters from src/transformers/training_args.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
2023-02-20 09:06:23 +01:00
Yih-Dar
7f1cdf1895
Fix dynamic module import error ( #21646 )
...
* fix dynamic module import error
---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
2023-02-17 21:22:39 +01:00
Younes Belkada
8a4c319d33
[`BLIP`] update blip path on slow tests ( #21476 )
...
* update blip path
* Update tests/models/blip/test_modeling_blip.py
2023-02-17 18:26:36 +00:00
Younes Belkada
087fd5f368
[`ImageProcessor`] Refactor default `mean` & `std` to `OPENAI_CLIP_MEAN` & `OPENAI_CLIP_STD` ( #21425 )
...
* fix default value
* add the fix on other models
2023-02-17 18:57:05 +01:00
Joao Gante
005b515754
Generate: eta sampling numerical stability ( #21676 )
2023-02-17 17:09:37 +00:00