Commits · v4.27-release · zhusg / transformers-new

29 Mar, 2023 2 commits

Patch release: v4.27.8 · 4e9f6fc6
Sylvain Gugger authored 2 years ago

v4.27.4

4e9f6fc6

Revert "Error (also in original) model, scaling only q matrix not qk.T dot... · 4277b3dd

Sylvain Gugger authored 2 years ago

Revert "Error (also in original) model, scaling only q matrix not qk.T dot product (qk.T/sqrt(dim_per_head))" (#22444)

Revert "Error (also in original) model, scaling only q matrix not qk.T dot product (qk.T/sqrt(dim_per_head)) (#21627)"

This reverts commit bad83008.

4277b3dd

23 Mar, 2023 2 commits
- Patch release: v4.27.3 · 5e3b19a8
  Sylvain Gugger authored 2 years ago
  
  v4.27.3
  
  5e3b19a8
- Enforce `max_memory` for device_map strategies (#22311) · 62d9baa5
  Sylvain Gugger authored 2 years ago
```
Enforce  for device_map strategies
```
  62d9baa5
20 Mar, 2023 2 commits
- Patch release: v4.27.2 · 68287689
  Sylvain Gugger authored 2 years ago
  
  v4.27.2
  
  68287689
- Fix balanced and auto device_map (#22271) · 1e39734c
  Sylvain Gugger authored 2 years ago
  
  1e39734c
15 Mar, 2023 3 commits
- Release: v4.27.1 · 2355e463
  Lysandre authored 2 years ago
  
  v4.27.1
  
  2355e463
- Regression pipeline device (#22190) · 659ef0b5
  Sylvain Gugger authored 2 years ago
```
* Fix regression in pipeline when device=-1 is passed

* Add regression test
```
  659ef0b5
- Revert 22152 MaskedImageCompletionOutput changes (#22187) · 36ed7508
  amyeroberts authored 2 years ago
```
Revert changes
```
  36ed7508
14 Mar, 2023 11 commits

Release: v4.27.0 · d941f07a
Sylvain Gugger authored 2 years ago

v4.27.0

d941f07a
Revert "Enforce same behavior as PyTorch 2.0 for older versions" (#22163) · c52c5282
Sylvain Gugger authored 2 years ago
```
Revert "Enforce same behavior as PyTorch 2.0 for older versions (#22136)"

This reverts commit 1c801d65.
```
c52c5282

[trainer] add `--optim adamw_torch_fused` for pt-2.0+ (#22144) · 085bf5c1

Stas Bekman authored 2 years ago

* [trainer] add --optim adamw_torch_fused

* change optim default

* deal with non-torch

* revert default change; prep; add fp16/amp assert

* typo

* typo

085bf5c1

to_pil - don't rescale if int and in range 0-255 (#22158) · c6318c37

amyeroberts authored 2 years ago

* Don't rescale if in and in range 0-255

* Raise value error if int values too large

* Update tests/test_image_transforms.py

* Update tests/test_image_transforms.py

c6318c37

Create MaskedImageCompletionOutput and fix ViT docs (#22152) · 3b22bfbc
Alara Dirik authored 2 years ago
```
* create MaskedImageCompletionOutput

* fix bugs

* fix bugs
```
3b22bfbc

Fix big model inference for T5 models in float16 (#22095) · b45192ec

Sylvain Gugger authored 2 years ago


* Fix big model inference for T5 models in float16

* Apply suggestions from code review

Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Style

* Trigger CI with latest release

---------

Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

b45192ec

Translation Italian: perf_train_cpu and perf_train_cpu_many (#22151) · 7f5ad6c3
Nicola Procopio authored 2 years ago
```
* added translated files

added perf_train_cpu and perf_train_cpu_many

* updated toctree
```
7f5ad6c3
Update 2 doctest expected values for torch 2.0.0 (#22148) · ff887035
Yih-Dar authored 2 years ago
```
update values

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
```
ff887035

Add ConvNeXT V2 (#21679) · cdddfbff

Alara Dirik authored 2 years ago

* Add ConvNeXt V2 to transformers
* TF model is separated from the PR to fix issues

cdddfbff

Move `is_pipeline_test_to_skip` to specific model test classes (#21999) · 6c2ad00c

Yih-Dar authored 2 years ago


* Move `is_pipeline_test_to_skip` to specific model test classes

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

6c2ad00c

[

️] Fix-whisper-breaking-changes (#21965) · 2beabd24

Arthur authored 2 years ago


* temp fix

* temporary fix

* update

* fix tests

* fixup

* update based on reveiew

Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* update to fix tests

* update docstring

---------

Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

2beabd24

13 Mar, 2023 20 commits

docs: New terms and updates to glossary (#21982) · 101a6cd2

MichaelRipa authored 2 years ago


* Updated glossary with new terms, added abbreviations for certain terms and merged autoencoding models, autoregressive models and causal language modeling into encoder and decoder models

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Added link to 'Pipeline for inference' tutorial

* Trigger CI

* Update docs/source/en/glossary.mdx

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

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>

* Added entry for self supervised learning, added deleted entries + fixed broken links

* Update docs/source/en/glossary.mdx

Co-authored-by: Steven Liu <59462357+stevhliu@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>

101a6cd2

Prepare daily CI for torch 2.0.0 (#22135) · ba9e0191
Yih-Dar authored 2 years ago
```
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
```
ba9e0191

[Safetensors] Add explicit flag to from pretrained (#22083) · f780557a

Patrick von Platen authored 2 years ago


* [Safetensors] Add explicit  flag to from pretrained

* add test

* remove @

* 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>

f780557a

Remove backend check for torch.compile (#22140) · 3a35937e

Sylvain Gugger authored 2 years ago


* Remove backend enforcment for torch.compile

* Update error

* Update src/transformers/training_args.py

Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* Apply suggestions from code review

Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* Style

---------

Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

3a35937e

[deepspeed docs] Activation Checkpointing (#22099) · 618697ef

Stas Bekman authored 2 years ago


* [deepspeed docs] Activation Checkpointing

* Apply suggestions from code review

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

* Update deepspeed.mdx

---------

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

618697ef

[trainer] fix bug in grad accum with multiple epochs (#22098) · 5b85add7
Stas Bekman authored 2 years ago
```
* [trainer] fix bug in grad accum

* comment out debug

* fix one-off

* rename counter
```
5b85add7
Enforce same behavior as PyTorch 2.0 for older versions (#22136) · 1c801d65
Sylvain Gugger authored 2 years ago

1c801d65
Trainer: let generate pick its inputs (#22108) · e16cbe88
Joao Gante authored 2 years ago
```
* Let generate pick its inputs

* fix squad seq2seq example
```
e16cbe88

[`Whiper`] add `get_input_embeddings` to `WhisperForAudioClassification` (#22133) · d979cf6e

Younes Belkada authored 2 years ago


* add `get_input_embeddings` to `WhisperForAudioClassification`

* add common tests

* fix another common test

* Update tests/models/whisper/test_modeling_whisper.py

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix style

---------

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

d979cf6e

Update configuration_align.py (projected_dim=640) (#22139) · 98797237
bishmdl76 authored 2 years ago
```
Update configuration_align.py

updated projected_dim=640 from 512 in arguments of AlignConfig
```
98797237
Add a new script to check model testers' config (#22063) · 54ee56b1
Yih-Dar authored 2 years ago
```
* Add script

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
```
54ee56b1
Adding Type Hints to TF_Pegasus model (#21941) · a096eaca
mollerup23 authored 2 years ago
```
* Adding Type Hints to TF_Pegasus model

* Updated some parameters per maintainer comments
```
a096eaca
Fix doc link for MGP-STR (#22138) · 6cb5132a
Sylvain Gugger authored 2 years ago

6cb5132a

Zero-shot image classification task guide (#22132) · 8def252d

Maria Khalusova authored 2 years ago


* WIP

* WIP

* manual inference example

* make style

* Apply suggestions from code review

Co-authored-by: Alara Dirik <8944735+alaradirik@users.noreply.github.com>

---------

Co-authored-by: Alara Dirik <8944735+alaradirik@users.noreply.github.com>

8def252d

Fix gradient checkpointing bug in trocr (#22126) · e61081e7

Karim Foda authored 2 years ago


* Fix gradient checkpointing bug in trocr

* Fix format

* Update src/transformers/models/trocr/modeling_trocr.py

Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

---------

Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

e61081e7

Fix gradient checkpointing bug in LongT5 (#22130) · ef74e7e7
Karim Foda authored 2 years ago

ef74e7e7
Fix gradient checkpointing bug in xmod (#22129) · c1db6a3b
Karim Foda authored 2 years ago

c1db6a3b
[`Blip2`] skip accelerate test (#22124) · 6652e7da
Younes Belkada authored 2 years ago
```
skip accelerate test
```
6652e7da
Added big_models.mdx italian translation #17600 (#22115) · dd3a0580
Nicola Procopio authored 2 years ago
```
* updated toctree

* italian translation big_model.mdx

* italian translation big_models
```
dd3a0580
Fix gradient checkpointing bug in xlm_roberta_xl (#22128) · 0768c5e2
Karim Foda authored 2 years ago

0768c5e2