huggingface/transformers

Pull Requests Commits

SunMarc committed 1 year ago

a8704d26

Merge remote-tracking branch 'upstream/main' into hqq_serialization

SunMarc committed 1 year ago

bc9cb55d

Merge remote-tracking branch 'upstream/main' into hqq_serialization

SunMarc committed 1 year ago

f2ea032e

fix hqq dispatch and unexpected keys

SunMarc committed 1 year ago

75dfe0a9

Fixed Hybrid Cache Shape Initialization. (#32163)

OsamaS99 committed 1 year ago

Verified 51ab25e2

Docker: add `speech` dep to the consistency docker image (#32374)

gante committed 1 year ago

Verified e3d8285a

Offloaded KV Cache (#31325)

n17s committed 1 year ago

Verified ca59d6f7

Fix conflicting key in init kwargs in PreTrainedTokenizerBase (#31233)

OmarManzoor committed 1 year ago

Verified b4727a12

Empty list in defaults for LLaMA special tokens during weights conversion (#32342)

viktor-shcherb committed 1 year ago

Verified db8c7cae

update clean_up_tokenization_spaces warning (#32371)

itazap committed 1 year ago

Verified 2229ebe7

Check device map for saving tokenizer config on TPU (fix for issue #31971) (#32043)

ayukh committed 1 year ago

Verified 05c1f9af

add missing attribute _supports_param_buffer_assignment for gpt-j. (#32359)

nv-guomingz committed 1 year ago

Verified 9e282840

Remove size check between attn_weights and kv_seq_len for phi3 (#32339)

helunwencser committed 1 year ago

Verified 48ed24c5

[whisper] compile compatibility with long-form decoding (#31772)

sanchit-gandhi committed 1 year ago

Verified e234061c

[enc-dec cache] fix bug in indexing (#32370)

sanchit-gandhi committed 1 year ago

Verified 9451a385

LLaVa: add cache class attribute (#32278)

zucchini-nlp committed 1 year ago

Verified 453e7488

fix: warmup_steps check for training_args (#32236)

Ricardo-L-C committed 1 year ago

Verified 14ee2326

fix: Removed unnecessary `@staticmethod` decorator (#32361)

Sai-Suraj-27 committed 1 year ago

Verified 53f0c9c2

>3-5x faster torch.compile forward compilation for autoregressive decoder models (#32227)

fxmarty committed 1 year ago

Verified 92abe603

Fix error when streaming to gradio with non-string tool arguments (#32360)

aymeric-roucher committed 1 year ago

Verified b46bd8b9

Gemma 2: support assisted generation (#32357)

gante committed 1 year ago

Verified ef177a5e

[Idefics2] - Fix FA2 call for Perceiver layer (#32275)

amyeroberts committed 1 year ago

Verified 5f1fcc29

Llama 3.1: Fix incorrect `inv_freq` assignment (#32330)

gante committed 1 year ago

Verified b75ad566

Gemma2 and flash-attention (#32188)

zucchini-nlp committed 1 year ago

Verified 7f552e28

LLaVA-NeXT: fix anyres shapes (#32314)

zucchini-nlp committed 1 year ago

Verified a3264332

Fix slow GemmaTokenizer and improve SPM slow -> fast conversion process (#32191)

xenova committed 1 year ago

Verified 6e2d04e4

Repo checks: skip docstring checks if not in the diff (#32328)

gante committed 1 year ago

Verified 026a173a

fixes #32329 : The Torch code is correct - to get an average of 10% o… (#32335)

fkrasnov2 committed 1 year ago

Verified 516af4bb

fixes to properly shard FSDP across cpu and meta for cpu_efficient_loading for prequantized 4bit (#32276)

winglian committed 1 year ago

Verified 62c60a30

fix: Added missing raise keyword for few exceptions (#32333)

Sai-Suraj-27 committed 1 year ago

Verified 16271080

Older