* fix mismatching behavior in from_pretrained with/without accelerate
* meaningful refactor
* remove added space
* add test
* fix model on the hub
* comment
* use tiny model
* style
* edits to _prepare_4d_causal_attention_mask()
* initial tests for 4d mask
* attention_mask_for_sdpa support
* added test for inner model hidden
* added autotest decorators
* test mask dtype to torch.int64
* torch.testing.assert_close
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* torch_device and @torch_gpu in tests
* upd tests
* +torch decorators
* torch decorators fixed
* more decorators!
* even more decorators
* fewer decorators
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* 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>
* [WIP] Make using safetensors files automated.
If `use_safetensors=True` is used, and it doesn't exist:
- Don't crash just yet
- Lookup for an open PR containing it.
- If yes, use that instead
- If not, touch the space to convert, wait for conversion to be finished
and the PR to be opened
- Use that new PR
- Profit.
* Remove the token.
* [Auto Safetensors] Websocket -> SSE (#27656)
* Websocket -> SSE
* Support sharded + tests +cleanup
a
* env var
* Apply suggestions from code review
* Thanks Simon
* Thanks Wauplin
Co-authored-by: Wauplin <lucainp@gmail.com>
* Cleanup
* Update tests
* Tests should pass
* Apply to other tests
* Extend extension
* relax requirement on latest hfh
* Revert
* Correct private handling & debug statements
* Skip gated repos as of now
* Address review comments
Co-authored-by: ArthurZucker <arthur.zucker@gmail.com>
---------
Co-authored-by: Lysandre Debut <hi@lysand.re>
Co-authored-by: Lysandre <lysandre@huggingface.co>
Co-authored-by: Wauplin <lucainp@gmail.com>
Co-authored-by: Lysandre <lysandre.debut@reseau.eseo.fr>
Co-authored-by: ArthurZucker <arthur.zucker@gmail.com>
* fix?
* actual fix
* fixups
* add dataclass to the attention mask converter
* refine testing suite
* make sure there are no overflows
* update the test
* Safetensors serialization by default
* First pass on the tests
* Second pass on the tests
* Third pass on the tests
* Fix TF weight loading from TF-format safetensors
* Specific encoder-decoder fixes for weight crossloading
* Add VisionEncoderDecoder fixes for TF too
* Change filename test for pt-to-tf
* One missing fix for TFVisionEncoderDecoder
* Fix the other crossload test
* Support for flax + updated tests
* Apply suggestions from code review
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
* Sanchit's comments
* Sanchit's comments 2
* Nico's comments
* Fix tests
* cleanup
* Apply suggestions from code review
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
---------
Co-authored-by: Matt <rocketknight1@gmail.com>
Co-authored-by: Sanchit Gandhi <93869735+sanchit-gandhi@users.noreply.github.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>
* Adding warning messages to BERT for missing attention masks
These warning messages when there are pad tokens within the input ids and
no attention masks are given. The warning message should only show up once.
* Adding warning messages to BERT for missing attention masks
These warning messages are shown when the pad_token_id is not None
and no attention masks are given. The warning message should only
show up once.
* Ran fix copies to copy over the changes to some of the other models
* Add logger.warning_once.cache_clear() to the test
* Shows warning when there are no attention masks and input_ids start/end with pad tokens
* Using warning_once() instead and fix indexing in input_ids check
---------
Co-authored-by: JB Lau <hckyn@voyager2.local>
* Preliminary work on some models
* Fix test load missing and make sure nonpersistent buffers are tested
* Always ignore nonpersistent buffers if in state_dict
* Treat models
* More models
* Treat remaining models
* Fix quality
* Fix tests
* Remove draft
* This test is not needed anymore
* Fix copies
* Fix last test
* Newly added models
* Fix last tests
* Address review comments
* Use tied weight keys
* More
* Fix tied weight missing warning
* Only give info on unexpected keys with different classes
* Deal with empty archs
* Fix tests
* Refine test
