Commits huggingface/transformers

add modular file

ita.zaporozhets@huggingface.co committed 1 year ago

cd8b36c7

update diff file

ita.zaporozhets@huggingface.co committed 1 year ago

a14b489f

moe update

ita.zaporozhets@huggingface.co committed 1 year ago

1feb8435

removing seperate inference logic and using mixtral for ref

ita.zaporozhets@huggingface.co committed 1 year ago

c3436e4e

updated MOE to be like mixtral

ita.zaporozhets@huggingface.co committed 1 year ago

9c524ea0

run script

itazap committed 1 year ago

8370023f

update to use MoeCausalLM

itazap committed 1 year ago

afea6ca7

update diff file

itazap committed 1 year ago

d8d9ac0c

removed some ep size refs

itazap committed 1 year ago

9c3a0ea8

update

itazap committed 1 year ago

6f66d603

update diff file

itazap committed 1 year ago

9d7cdbdb

cleaning

itazap committed 1 year ago

0ee5bb74

updating based on paper and making more meaningful naming

itazap committed 1 year ago

90554fbf

adding tokenizer config files

itazap committed 1 year ago

48e84ed1

testing files

itazap committed 1 year ago

e8444402

initial ruff

itazap committed 1 year ago

e339caad

adding initial version of deepseek v2 files, copied from existing repo and llama

itazap committed 1 year ago

43e8548f

Whisper: move to tensor cpu before converting to np array at decode time (#31954)

gante committed 1 year ago

Verified a5c642fe

Generate: v4.42 deprecations 🧹🧹 (#31956)

gante committed 1 year ago

Verified df1c248a

Generate: remove deprecated code due to `Cache` and `cache_position` being default (#31898)

gante committed 1 year ago

Verified 739a6316

Fix `GenerationMixin.generate` compatibility with pytorch profiler (#31935)

fxmarty committed 1 year ago

Verified 8480fda6

fix prompt strip to support tensors and np arrays (#27818)

AvivSham committed 1 year ago

Verified 7f79a973

Docker: TF pin on the consistency job (#31928)

gante committed 1 year ago

Verified d1a1bcf5

[Bug Fix] fix qa pipeline tensor to numpy (#31585)

jiqing-feng committed 1 year ago

Verified aec1ca3a

Adding hiera (#30356)

Namangarg110 committed 1 year ago

Verified c1e139c2

Allow `Trainer.get_optimizer_cls_and_kwargs` to be overridden (#31875)

apoorvkh committed 1 year ago

Verified 574e68d5

🚨 fix(SigLip): remove spurious exclusion of first vision output token (#30952)

transmissions11 committed 1 year ago

Verified 52585019

Generate: fix `SlidingWindowCache.reset()` (#31917)

gante committed 1 year ago

Verified 6a05f68f

Refactor flash attention implementation in transformers (#31446)

ArthurZucker committed 1 year ago

Verified e3143952

Fix fx tests with inputs_embeds (#31862)

fxmarty committed 1 year ago

Verified ad4ef3a2