huggingface/transformers

Pull Requests Commits

[run-slow] llama

ArthurZucker committed 1 year ago

78f1c178

[run-slow]llama

ArthurZucker committed 1 year ago

075f25f4

properly call super init

ArthurZucker committed 1 year ago

80bb8fbb

ArthurZucker committed 1 year ago

93b2ee95

fix style and init

ArthurZucker committed 1 year ago

1d99fc9b

ArthurZucker committed 1 year ago

fd39b179

Llama 3.1 conversion

LysandreJik committed 1 year ago

d5a99dfc

Dev version: v4.44.0.dev0

LysandreJik committed 1 year ago

ff0d708f

Updated `ruff` to the latest version (#31926)

Sai-Suraj-27 committed 1 year ago

Verified d2c687b3

Enhancing SFT Training Efficiency Using Packing and FlashAttention2 with Position IDs (#31629)

RhuiDih committed 1 year ago

Verified 9cf4f2aa

Added additional kwarg for successful running of optuna hyperparameter search (#31924)

DeF0017 committed 1 year ago

Verified 7d92009a

feat(cache): StaticCache uses index_copy_ to avoid useless copy (#31857)

tengomucho committed 1 year ago

Verified 63700628

Fix typing to be compatible with later py versions (#32155)

amyeroberts committed 1 year ago

Verified a009fbda

Revert "Incorrect Whisper long-form decoding timestamps " (#32148)

sanchit-gandhi committed 1 year ago

Verified 3263b343

Rename Phi-3 rope scaling type (#31436)

garg-amit committed 1 year ago

Verified 034b4778

Added mamba.py backend (#30139)

alxndrTL committed 1 year ago

Verified bab32d6f

Fix video batching to videollava (#32139)

merveenoyan committed 1 year ago

Verified 9ced33ca

Fix flash attention speed issue (#32028)

Cyrilvallez committed 1 year ago

Verified a5b226ce

gguf conversion add_prefix_space=None for llama3 (#31937)

itazap committed 1 year ago

Verified a1844a32

Llama: RoPE refactor (#32135)

gante committed 1 year ago

Verified 2e113422

Modify resize_token_embeddings to ensure output type is same as input (#31979)

bayllama committed 1 year ago

Verified 5a4a76ed

Disable quick init for TapasPreTrainedModel (#32149)

daniellok-db committed 1 year ago

Verified 1535a2c9

Add YaRN and Dynamic-YaRN RoPE Scaling Methods (#30910)

mig-mfreitas committed 1 year ago

Verified 34b43211

Add method to retrieve used chat template (#32032)

KonradSzafer committed 1 year ago

Verified 7405c1c7

Fix mask creations of `GPTNeoX` and `GPT2` (#31944)

vasqu committed 1 year ago

Verified 605f3245

[modelling] remove un-necessary transpose for fa2 attention (#31749)

sanchit-gandhi committed 1 year ago

Verified 2782aada

Remove `trust_remote_code` when loading Libri Dummy (#31748)

sanchit-gandhi committed 1 year ago

Verified f83c6f1d

LLaVaNeXT: pad on right if training (#32134)

zucchini-nlp committed 1 year ago

Verified 3aefb4ec

Add llama3-llava-next-8b to llava_next conversion script (#31395)

jamt9000 committed 1 year ago

Verified 251a2409

Add new quant method (#32047)

SunMarc committed 1 year ago

Verified 96a074fa

Older