GitLab

* added compiled model support for inference

* linter

* Fix tests

* linter

* linter

* remove inference mode from pipelines

* Linter

---------
Co-authored-by: amarkov <alexander@inworld.ai>

Fix jit trace

* Fix .push_to_hub and cleanup get_full_repo_name usage

* Do not rely on Python bool conversion magic

* request changes

* First commit

* step 1 working

* add alibi

* placeholder for `scan`

* add matrix mult alibi

* beta scaling factor for bmm

* working v1 - simple forward pass

* move layer_number from attribute to arg in call

* partial functioning scan

* hacky working scan

* add more modifs

* add test

* update scan for new kwarg order

* fix position_ids problem

* fix bug in attention layer

* small fix

- do the alibi broadcasting only once

* prelim refactor

* finish refactor

* alibi shifting

* incorporate dropout_add to attention module

* make style

* make padding work again

* update

* remove bogus file

* up

* get generation to work

* clean code a bit

* added small tests

* adding albii test

* make CI tests pass:

- change init weight
- add correct tuple for output attention
- add scan test
- make CI tests work

* fix few nits

* fix nit onnx

* fix onnx nit

* add missing dtype args to nn.Modules

* remove debugging statements

* fix scan generate

* Update modeling_flax_bloom.py

* Update test_modeling_flax_bloom.py

* Update test_modeling_flax_bloom.py

* Update test_modeling_flax_bloom.py

* fix small test issue + make style

* clean up

* Update tests/models/bloom/test_modeling_flax_bloom.py
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* fix function name

* small fix test

* forward contrib credits from PR17761

* Fix failing test

* fix small typo documentation

* fix non passing test

- remove device from build alibi

* refactor call

- refactor `FlaxBloomBlockCollection` module

* make style

* upcast to fp32

* cleaner way to upcast

* remove unused args

* remove layer number

* fix scan test

* make style

* fix i4 casting

* fix slow test

* Update src/transformers/models/bloom/modeling_flax_bloom.py
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* remove `layer_past`

* refactor a bit

* fix `scan` slow test

* remove useless import

* major changes

- remove unused code
- refactor a bit
- revert import `torch`

* major refactoring

- change build alibi

* remove scan

* fix tests

* make style

* clean-up alibi

* add integration tests

* up

* fix batch norm conversion

* style

* style

* update pt-fx cross tests

* update copyright

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

* per-weight check

* style

* line formats

---------
Co-authored-by: younesbelkada <younesbelkada@gmail.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: haileyschoelkopf <haileyschoelkopf@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fix

* fix

* fix

* fix

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

* initial Bark offload proposal

* use hooks instead of manually offloading

* add test of bark offload to cpu feature

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

* Update docstrings of offload
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* remove unecessary set_seed in Bark tests

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>

* support from pretrained args

* draft addition of tests

* update test

* use parrent assert true

* Update src/transformers/models/mpt/configuration_mpt.py
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

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

* Change defaults

* Sylvain's comments

* clarify 4/8 bit loading log message

* make style

default legacy to None

* Enable return_dict in order to compile

* Update tests

Fix docstring for dropout.

Move out common methods

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

* update

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

fix "UserWarning: Creating a tensor from a list of numpy.ndarrays is extremely slow. Please consider converting the list to a single numpy.ndarray with numpy.array() before converting to a tensor."
Co-authored-by: 刘长伟 <hzliuchw@corp.netease.com>

* Add descriptive docstring to TemperatureLogitsWarper

It addresses https://github.com/huggingface/transformers/issues/24783



* Remove niche features
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Commit suggestion
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Refactor the examples to simpler ones

* Add a missing comma
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Make args description more compact
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Remove extra text after making description more compact
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Fix linter

---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* fix documentation

* changes

* fix tied_params for meta tensor

* remove duplicate

* Initial addition of t5forsequenceclassification

* Adding imports and adding tests

* Formatting

* Running make fix-copies

* Adding mt5forseq

* Formatting

* run make fix-copies

* Adding to docs

* Add model_parallel

* Fix bug

* Fix

* Remove TODO

* Fixing tests for T5ForSequenceClassification

* Undo changes to dependency_versions_table.py

* Change classification head to work with T5Config directly

* Change seq length to let tests pass

* PR comments for formatting

* Formatting

* Initial addition of UMT5ForSequenceClassification

* Adding to inits and formatting

* run make fix-copies

* Add doc for UMT5ForSeqClass

* Update UMT5 config

* Fix docs

* Skip torch fx test for SequenceClassification

* Formatting

* Add skip to UMT5 tests as well

* Fix umt5 tests

* Running make fix-copies

* PR comments

* Fix for change to sentence_representation

* Rename seq_len to hidden_size since that's what it is

* Use base_model to follow format of the rest of the library

* Update docs

* Extract the decoder_input_ids changes and make one liner

* Make one-liner

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

* draft solution

* use `setdefault`

* nits

* add tests and fix truncation issue

* fix test

* test passes locally

* quality

* updates

* update tsets

* tf versions

* apply changes to other models

* 3 models slipped through the cracks

* support left padding

* nit

* Update src/transformers/models/gpt_neox/modeling_gpt_neox.py

* Update src/transformers/models/gpt_neox/modeling_gpt_neox.py

* fix

* Update src/transformers/generation/utils.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* update

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* add example NoBadWordsLogitsProcessor

* fix L764 & L767

* make style

* draft add new model like

* some cleaning of the config

* nits

* add nested configs

* nits

* update

* update

* added layer norms + triton kernels

* consider only LPLayerNorm for now.

* update

* all keys match.

* Update

* fixing nits here and there

* working forward pass.

* removed einops dependency

* nits

* format

* add alibi

* byebye head mask

* refactor attention

* nits.

* format

* fix nits.

* nuke ande updates

* nuke tokenizer test

* don't reshape query with kv heads

* added a bit of documentation.

* remove unneeded things

* nuke more stuff

* nit

* logits match - same generations

* rm unneeded methods

* 1 remaining failing CI test

* nit

* fix nits

* fix docs

* fix docs

* rm tokenizer

* fixup

* fixup

* fixup and fix tests

* fixed configuration object.

* use correct activation

* few minor fixes

* clarify docs a bit

* logits match à 1e-12

* skip and unskip a test

* added some slow tests.

* fix readme

* add more details

* Update docs/source/en/model_doc/mpt.md
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix configuration issues

* more fixes in config

* added more models

* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* remove unneeded position ids

* fix some  comments

* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* revert suggestion

* mpt alibi + added batched generation

* Update src/transformers/models/mpt/__init__.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* remove init config

* Update src/transformers/models/mpt/configuration_mpt.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix nit

* add another slow test

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

* fits in one line

* some refactor because make fixup doesn't pass

* add ft notebook

* update md

* correct doc path

---------
Co-authored-by: younesbelkada <younesbelkada@gmail.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

Repeat per sample for SAM image embeddings

[`generate`]  Only warn users if the `generation_config`'s `max_length` is set to the default value (#25030)

* check max length is default

* nit

* update warning: no-longer deprecate

* comment in the configuration_utils in case max length's default gets changed in the futur

* Fix last models for common tests that are too big.

* Remove print statement

fix rope_scaling doc string

* Better error message when signal is not supported on OS

* Address review comments

fix 8bit corner case with Blip2 8bit