* add FA-2 support for mistral
* fixup
* add sliding windows
* fixing few nits
* v1 slicing cache - logits do not match
* add comment
* fix bugs
* more mem efficient
* add warning once
* add warning once
* oops
* fixup
* more comments
* copy
* add safety checker
* fixup
* Update src/transformers/models/mistral/modeling_mistral.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* copied from
* up
* raise when padding side is right
* fixup
* add doc + few minor changes
* fixup
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* fix stripping
* nits
* fix another test
* styling
* fix?
* update
* revert bad merge
* found the bug
* YES SIR
* is that change really required?
* make fast even faster
* re order functions
* add tokenizer kwarg inputs
* Adding tokenizer_kwargs to _sanitize_parameters
* Add truncation=True example to tests
* Update test_pipelines_fill_mask.py
* Update test_pipelines_fill_mask.py
* make fix-copies and make style
* Update fill_mask.py
Replace single tick with double
* make fix-copies
* Style
---------
Co-authored-by: Lysandre <lysandre@huggingface.co>
* start working on next chapter
* finish testing
* Update docs/source/de/testing.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/testing.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/testing.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* fix wav2vec2
* nit
* stash
* one more file to update
* fix byt5
* vocab size is 256, don't change that!
* use other revision
* test persimon in smaller size
* style
* tests
* nits
* update add tokens from pretrained
* test tokenization
* nits
* potential fnet fix?
* more nits
* nits
* correct test
* assert close
* udpate
* ouch
* fix it
* some more nits
* FINALLU
* use `adept` checkpoints
* more adept checkpoints
* that was invlved!
* fix issue of canine forward requires input_ids anyway
The `forward` requires `input_ids` for deriving other variables in all cases. Change this to use the given one between `input_ids` and `inputs_embeds`
* fix canine forward
The current `forward` requires (the shape of) `input_ids` for deriving other variables whenever `input_ids` or `inputs_embeds` is provided. Change this to use the given one instead of `input_ids` all the time.
* fix format
* fix format
* Fix num_heads in _upad_input
The variable num_key_value_heads has falsely been named num_heads, which led to reshaping the query_layer using the wrong attention head count. (It would have been enough to use the correct variable self.num_heads instead of num_heads, but I renamed num_heads to num_key_value_heads for clarity)
* fixed copies using make fix-copies and ran make fixup
---------
Co-authored-by: fseiler <f.seiler@jerocom.de>
* from seq2seq speech
* [Flax] Example script for speech seq2seq
* tests and fixes
* make style
* fix: label padding tokens
* fix: label padding tokens over list
* update ln names for Whisper
* try datasets iter loader
* create readme and append results
* style
* make style
* adjust lr
* use pt dataloader
* make fast
* pin gen max len
* finish
* add pt to requirements for test
* fix pt -> torch
* add accelerate
Ignore decoder weights when using T5EncoderModel and LongT5EncoderModel
Both T5EncoderModel and LongT5EncoderModel do not have any decoder layers, so
loading a pretrained model checkpoint such as t5-small will give warnings about
keys found in the model checkpoint that are not in the model itself.
To prevent this log warning, r"decoder" has been added to _keys_to_ignore_on_load_unexpected for
both T5EncoderModel and LongT5EncoderModel
* make use of adapter_revision
* v1 adapter kwargs
* fix CI
* fix CI
* fix CI
* fixup
* add BC
* Update src/transformers/integrations/peft.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* fixup
* change it to error
* Update src/transformers/modeling_utils.py
* Update src/transformers/modeling_utils.py
* fixup
* change
* Update src/transformers/integrations/peft.py
---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
* [VITS] Fix speaker_embed device mismatch
- pass device arg to speaker_id tensor
* [VITS] put speaker_embed on device when int
* [VITS] device=self.device
instead of self.embed_speaker.weight.device
* [VITS] make tensor directly on device
using torch.full()
* change mention of decoder_input_ids to input_ids and same with decoder_input_embeds
* Style
---------
Co-authored-by: Lysandre <lysandre@huggingface.co>
* initial
* toctree
* add tf model
* run scripts
* peft
* llm and agents
* Update docs/source/de/peft.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/peft.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/peft.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/run_scripts.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/run_scripts.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/transformers_agents.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* Update docs/source/de/transformers_agents.md
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
* fix PEFT multi adapters support
* refactor a bit
* save pretrained + BC + added tests
* Update src/transformers/integrations/peft.py
Co-authored-by: Benjamin Bossan <BenjaminBossan@users.noreply.github.com>
* add more tests
* add suggestion
* final changes
* adapt a bit
* fixup
* Update src/transformers/integrations/peft.py
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
* adapt from suggestions
---------
Co-authored-by: Benjamin Bossan <BenjaminBossan@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
