GitLab

fix bug: cost matrix is infeasible

* fix a typo and add an illustrative test

* appease black

* reduce code duplication and add Annotion type back with a pending deprecation warning

* remove unused code

* change warning type

* black formatting fix

* change enum deprecation approach to support 3.8 and earlier

* add stacklevel

* fix black issue

* fix ruff issues

* fix ruff issues

* move tests to own mixin

* include yolos

* fix black formatting issue

* fix black formatting issue

* use logger instead of warnings and include target version for deprecation

* add deepspeed scheduled test for amd

* fix image

* add dockerfile

* add comment

* enable tests

* trigger

* remove trigger for this branch

* trigger

* change runner env to trigger the docker build image test

* use new docker image

* remove test suffix from docker image tag

* replace test docker image with original image

* push new image

* Trigger

* add back amd tests

* fix typo

* add amd tests back

* fix

* comment until docker image build scheduled test fix

* remove deprecated deepspeed build option

* upgrade torch

* update docker & make tests pass

* Update docker/transformers-pytorch-deepspeed-amd-gpu/Dockerfile

* fix

* tmp disable test

* precompile deepspeed to avoid timeout during tests

* fix comment

* trigger deepspeed tests with new image

* comment tests

* trigger

* add sklearn dependency to fix slow tests

* enable back other tests

* final update

---------
Co-authored-by: Felix Marty <felix@hf.co>
Co-authored-by: Félix Marty <9808326+fxmarty@users.noreply.github.com>
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

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

"text input must of type" -> "text input must be of type"

fix parameter count in readme

* Update import message

* Update message

* Fix test for multi-GPU

* WIth CPU handle

* Initial commit for AutoBackbone & Backbone

* Added timm and clarified out_indices

* Swapped the example to out_indices

* fix toctree

* Update autoclass_tutorial.md

* Update backbones.md

* Update autoclass_tutorial.md

* Add dummy torch input instead

* Add dummy torch input

* Update autoclass_tutorial.md

* Update backbones.md

* minor fix

* Update docs/source/en/main_classes/backbones.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>

* Update docs/source/en/autoclass_tutorial.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>

* Added illustrations and explained backbone & neck

* Update docs/source/en/main_classes/backbones.md
Co-authored-by: Maria Khalusova <kafooster@gmail.com>

* Update backbones.md

---------
Co-authored-by: Maria Khalusova <kafooster@gmail.com>

* use logger.warning_once to avoid massive outputs when training/finetuning longformer

* update more

* docstring corrections

* style make

---------
Co-authored-by: vijaye12 <vijaye12@in.ibm.com>

* up

* up

* test

* logits ok

* up

* up

* few fixes

* conversion script

* up

* nits

* nits

* update

* nuke

* more updates

* nites

* fix many issues

* nit

* scatter

* nit

* nuke megablocks

* nits

* fix conversion script

* nit

* remove

* nits

* nit

* update

* oupsssss

* change

* nits device

* nits

* fixup

* update

* merge

* add copied from

* fix the copy mentions

* update tests

* more fixes

* nits

* conversion script

* add parts of the readme

* Update tests/models/mixtral/test_modeling_mixtral.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* new test + conversion script

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

* Apply suggestions from code review

* fix

* fix copies

* fix copies

* ooops

* fix config

* Apply suggestions from code review

* fix nits

* nit

* add copies

* add batched tests

* docs

* fix flash attention

* let's add more verbose

* add correct outputs

* support router ouptus

* ignore copies where needed

* fix

* cat list if list is given for now

* nits

* Update docs/source/en/model_doc/mixtral.md

* finish router refactoring

* fix forward

* fix expected values

* nits

* fixup

* fix

* fix bug

* fix

* fix dtype mismatch

* fix

* grrr grrr I support item assignment

* fix CI

* docs

* fixup

* remove some copied form

* fix weird diff

* skip doctest fast on the config and modeling

* mark that is supports flash attention in the doc

* update

* Update src/transformers/models/mixtral/modeling_mixtral.py
Co-authored-by: Lysandre Debut <hi@lysand.re>

* Update docs/source/en/model_doc/mixtral.md
Co-authored-by: Lysandre Debut <hi@lysand.re>

* revert router logits config issue

* update doc accordingly

* Update src/transformers/models/mixtral/convert_mixtral_weights_to_hf.py

* nits

* use torch testing asssert close

* fixup

* doc nits

---------
Co-authored-by: younesbelkada <younesbelkada@gmail.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Lysandre Debut <hi@lysand.re>

* Skip nn.Module.reset_parameters

* Actually skip

* Check quality

* Maybe change all inits

* Fix init issues: only modify public functions

* Add a small test for now

* Style

* test updates

* style

* nice tes

* style

* make it even faster

* one more second

* remove fx icompatible

* Update tests/test_modeling_common.py
Co-authored-by: Lysandre Debut <hi@lysand.re>

* Update tests/test_modeling_common.py
Co-authored-by: Lysandre Debut <hi@lysand.re>

* skip

* fix quality

* protect the import

---------
Co-authored-by: Lysandre Debut <hi@lysand.re>

fix sdpa dispatch

* More improvements

* Improve variable names

* Update READMEs, improve docs

change dtype of some integration tests

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

Fix test

* [DETA] fix freeze/unfreeze function

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

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

* add freeze/unfreeze test case in DETA

* fix type

* fix typo 2

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

* fix tune integration for ray 2.7+
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* add version check for ray tune backend availability
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* missing import
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* pin min version instead
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* address comments
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* some fixes
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* fix unnecessary final checkpoint
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* fix lint
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* dep table fix
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* fix lint
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

---------
Signed-off-by: Justin Yu <justinvyu@anyscale.com>

* [CLAP] Replace hard-coded batch size to enable dynamic ONNX export

* Add back docstring

* add sdpa

* wip

* cleaning

* add ref

* yet more cleaning

* and more :)

* wip llama

* working llama

* add output_attentions=True support

* bigcode sdpa support

* fixes

* gpt-bigcode support, require torch>=2.1.1

* add falcon support

* fix conflicts falcon

* style

* fix attention_mask definition

* remove output_attentions from attnmaskconverter

* support whisper without removing any Copied from statement

* fix mbart default to eager renaming

* fix typo in falcon

* fix is_causal in SDPA

* check is_flash_attn_2_available in the models init as well in case the model is not initialized through from_pretrained

* add warnings when falling back on the manual implementation

* precise doc

* wip replace _flash_attn_enabled by config.attn_implementation

* fix typo

* add tests

* style

* add a copy.deepcopy on the config in from_pretrained, as we do not want to modify it inplace

* obey to config.attn_implementation if a config is passed in from_pretrained

* fix is_torch_sdpa_available when torch is not installed

* remove dead code

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

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

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

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

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

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

* remove duplicate pretraining_tp code

* add dropout in llama

* precise comment on attn_mask

* add fmt: off for _unmask_unattended docstring

* precise num_masks comment

* nuke pretraining_tp in LlamaSDPAAttention following Arthur's suggestion

* cleanup modeling_utils

* backward compatibility

* fix style as requested

* style

* improve documentation

* test pass

* style

* add _unmask_unattended tests

* skip meaningless tests for idefics

* hard_check SDPA requirements when specifically requested

* standardize the use if XXX_ATTENTION_CLASSES

* fix SDPA bug with mem-efficient backend on CUDA when using fp32

* fix test

* rely on SDPA is_causal parameter to handle the causal mask in some cases

* fix FALCON_ATTENTION_CLASSES

* remove _flash_attn_2_enabled occurences

* fix test

* add OPT to the list of supported flash models

* improve test

* properly test on different SDPA backends, on different dtypes & properly handle separately the pad tokens in the test

* remove remaining _flash_attn_2_enabled occurence

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

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

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

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

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

* remove use_attn_implementation

* fix docstring & slight bug

* make attn_implementation internal (_attn_implementation)

* typos

* fix tests

* deprecate use_flash_attention_2=True

* fix test

* add back llama that was removed by mistake

* fix tests

* remove _flash_attn_2_enabled occurences bis

* add check & test that passed attn_implementation is valid

* fix falcon torchscript export

* fix device of mask in tests

* add tip about torch.jit.trace and move bt doc below sdpa

* fix parameterized.expand order

* move tests from test_modeling_attn_mask_utils to test_modeling_utils as a relevant test class is already there

* update sdpaattention class with the new cache

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

* Update src/transformers/models/bark/modeling_bark.py

* address review comments

* WIP torch.jit.trace fix. left: test both eager & sdpa

* add test for torch.jit.trace for both eager/sdpa

* fix falcon with torch==2.0 that needs to use sdpa

* fix doc

* hopefully last fix

* fix key_value_length that has no default now in mask converter

* is it flacky?

* fix speculative decoding bug

* tests do pass

* fix following #27907

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

* Add pad_truncation to es/_toctree.yml

* Add pad_truncation.md to es/

* Translated first two paragraph

* Translated paddig argument section

* Translated truncation argument section

* Translated final paragraphs

* Translated table

* Fixed typo in the table of en/pad_truncation.md

* Run make style | Fix a word

* Add Padding (relleno) y el Truncation (truncamiento) in the final paragraphs

* Fix relleno and truncamiento words

* Fuffill request

* Add test

* Better test

* Apply suggestions from code review
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Better test

* Better test

* MOre comments

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix llava

* nits

* attention_mask was forgotten

* nice

* :)

* fixup

* fix

* fix

* fix

* fix

* fix

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

* fix

* fix

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

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

* update converting script

* make style

* Fix issues in add and is_done for BeamHypotheses

* make newly added arguments optional for better compatibility

* Directly use cur_len as generated_len, add note for retrocompatibility

* update test expectation

* make cur_len represents the length of the entire sequence including the decoder prompt

* remove redundant if/else in testing

* Fix beam score calculation issue for tensorflow version

* fix transition score computation error

* make cur_len represent the entire sequence length including decoder prompt

Fix: Raise informative exception when `prefix_allowed_tokens_fn` return empty set of tokens (#27797)
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

[⚠️ removed a default argument] Make `AttentionMaskConverter` compatible with `torch.compile(..., fullgraph=True)` (#27868)

* remove bugged torch.float32 default

* add test

* fix tests

* fix test

* fix doc

* Draft version of new KV Caching

This should allow Attention Sinks (https://github.com/tomaarsen/attention_sinks)
/ StreamingLLM (https://arxiv.org/abs/2309.17453) to be easily implemented
in a third-party or in transformers directly

* Address numerous PR suggestions

1. Move layer_idx from cache to ...Attention. Removes confusing set_layer_idx magic.
2. Always convert past_key_values to Cache instance at the start of ...Attention, removes all other isinstance calls.
3. Remove __bool__ and __getitem__ magic as they're confusing.
4. past_key_values.update(key, value, idx) now returns key, value.
5. Add use_legacy_cache flag, defaults to None, i.e. Falsey. This breaks generate for now, until 1) the cache is used is generate() or 2) use_legacy_cache is defaulted to True in generate() until we change it in another PR.
6. Separate key_cache and value_cache.

Some work is still needed to see if the SinkCache can conveniently be implemented with just one update method.

* Implement the SinkCache through backward+forward rotations

* Integrate (Sink)Cache with Llama FA2

* Set use_legacy_cache=True as default, allows for test passes

* Move from/to_legacy_cache to ...Model class

* Undo unnecessary newline change

* Remove copy utility from deprecated OpenLlama

* Match import style

* manual rebase with main

* Cache class working with generate (#1)

* Draft version of new KV Caching

This should allow Attention Sinks (https://github.com/tomaarsen/attention_sinks)
/ StreamingLLM (https://arxiv.org/abs/2309.17453

) to be easily implemented
in a third-party or in transformers directly

* Address numerous PR suggestions

1. Move layer_idx from cache to ...Attention. Removes confusing set_layer_idx magic.
2. Always convert past_key_values to Cache instance at the start of ...Attention, removes all other isinstance calls.
3. Remove __bool__ and __getitem__ magic as they're confusing.
4. past_key_values.update(key, value, idx) now returns key, value.
5. Add use_legacy_cache flag, defaults to None, i.e. Falsey. This breaks generate for now, until 1) the cache is used is generate() or 2) use_legacy_cache is defaulted to True in generate() until we change it in another PR.
6. Separate key_cache and value_cache.

Some work is still needed to see if the SinkCache can conveniently be implemented with just one update method.

* Integrate (Sink)Cache with Llama FA2

* Move from/to_legacy_cache to ...Model class

* Undo unnecessary newline change

* Match import style

* working generate

* Add tests; Simplify code; Apply changes to Mistral and Persimmon

* fix rebase mess

* a few more manual fixes

* last manual fix

* propagate changes to phi

* upgrade test

* add use_legacy_cache docstring; beef up tests

* reintroduce unwanted deletes

---------
Co-authored-by: Tom Aarsen <Cubiegamedev@gmail.com>

* move import

* add default to model_kwargs.get('use_legacy_cache')

* correct failing test

* Apply suggestions from code review
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* apply PR suggestions

* fix failing test

* Apply suggestions from code review
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Tom Aarsen <37621491+tomaarsen@users.noreply.github.com>

* PR comments

* tmp commit

* add docstrings

* more tests, more docstrings, add to docs

* derp

* tmp commit

* tmp dbg

* more dbg

* fix beam search bug

* cache can be a list of tuples in some models

* fix group beam search

* all but sinkcache integration tests

* fix sink cache and add hard integration test

* now also compatible with input_embeds input

* PR comments

* add Cache support to Phi+FA2

* make fixup

---------
Co-authored-by: Joao Gante <joao@huggingface.co>
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>