huggingface/transformers

Pull Requests Commits

Support RefinedWebModel as a model_type for Falcon

Rocketknight1 committed 2 years ago

20a484b7

[fix] Change the condition of ValueError in "convert_checkpoint_from_transformers_to_megatron" (#24769)

SeongBeomLEE committed 2 years ago

Verified 21946a8c

Removing unnecessary `device=device` in modeling_llama.py (#24696)

Liyang90 committed 2 years ago

Verified 1f6f32c2

Revert "Unpin protobuf in docker file (for daily CI)" (#24800)

ydshieh committed 2 years ago

Verified 906afa1d

Rm duplicate pad_across_processes (#24780)

muellerzr committed 2 years ago

Verified f1732e13

Remove WWT from README (#24672)

LysandreJik committed 2 years ago

Verified cfc8a053

gpt-bigcode: avoid `zero_` to support Core ML (#24755)

pcuenca committed 2 years ago

Verified 395e566a

Fix pad across processes dim in trainer and not being able to set the timeout (#24775)

muellerzr committed 2 years ago

Verified 02842855

Update default values of bos/eos token ids in `CLIPTextConfig` (#24773)

ydshieh committed 2 years ago

Verified 4f85aaa6

Replacement of 20 asserts with exceptions (#24757)

Baukebrenninkmeijer committed 2 years ago

Verified fc9e387d

Docs: Update logit processors __call__ docs (#24729)

gante committed 2 years ago

Verified 430a04a7

Add MobileVitV2 to doctests (#24771)

amyeroberts committed 2 years ago

Verified 6e2f0696

Fix eval_accumulation_steps leading to incorrect metrics (#24756)

muellerzr committed 2 years ago

Verified 7edc33ac

Unpin protobuf in docker file (for daily CI) (#24761)

ydshieh committed 2 years ago

Verified 45025d92

Allow existing configs to be registered (#24760)

sgugger committed 2 years ago

Verified 6aadb8d0

:bug: Handle empty gen_kwargs for seq2seq trainer prediction_step function (#24759)

gkumbhat committed 2 years ago

Verified 4c0e251d

Fix lr scheduler not being reset on reruns (#24758)

muellerzr committed 2 years ago

Verified 253d43d4

Skip some slow tests for doctesting in PRs (Circle)CI (#24753)

ydshieh committed 2 years ago

Verified 1be0145d

[InstructBLIP] Fix bos token of LLaMa checkpoints (#24492)

NielsRogge committed 2 years ago

Verified bb13a928

Fix non-deterministic Megatron-LM checkpoint name (#24674)

janEbert committed 2 years ago

Verified aac4c799

Skip keys not in the state dict when finding mismatched weights (#24749)

sgugger committed 2 years ago

Verified 33aafc26

add gradient checkpointing for distilbert (#24719)

jordane95 committed 2 years ago

Verified 3d869726

Docs: add `kwargs` type to fix formatting (#24733)

gante committed 2 years ago

Verified 2642d8d0

fix: Text splitting in the BasicTokenizer (#22280)

Connor Henderson committed 2 years ago

Verified 5739726f

Fix typo in LocalAgent (#24736)

jamartin9 committed 2 years ago

Verified 2489e380

Add ViViT (#22518)

jegork committed 2 years ago

Verified 8a5e8a9c

[Patch-t5-tokenizer] Patches the changes on T5 to make sure previous behaviour is still valide for beginning of words (#24622)

ArthurZucker committed 2 years ago

Verified b15343de

Falcon port (#24523)

Rocketknight1 committed 2 years ago

Verified b3ab3fac

add link to accelerate doc (#24601)

SunMarc committed 2 years ago

Verified 35eac0df

Docs: change some `input_ids` doc reference from `BertTokenizer` to `AutoTokenizer` (#24730)

gante committed 2 years ago

Verified a074a5d3

Older