huggingface/transformers

Pull Requests Commits

MekkCyber committed 1 year ago

9f66405c

Fix bug in apply_rotary_pos_emb_flashatt: in Qwen2-5-VL (#36065)

DeepWaved committed 1 year ago

8aa45e17

Adding RT-DETRv2 for object detection (#34773)

jadechoghari committed 1 year ago

128b8402

[docs] fix outdated example code in `trainer.md` (#36066)

faaany committed 1 year ago

4d1480d5

Fix StopStringCriteria to handle tokens above len(tokenizer) (#35797)

Rocketknight1 committed 1 year ago

9c3009a3

Fix model kwargs (#35875)

muellerzr committed 1 year ago

e173ffd3

Fix words typos in ggml test. (#36060)

zhanluxianshen committed 1 year ago

b0cb3d0c

Nail in edge case of torch dtype being overriden permantly in the case of an error (#35845)

muellerzr committed 1 year ago

c3e530a1

Save checkpoint to temporary directory to handle partial saves during failures (#35580)

SilverSoldier committed 1 year ago

20d17358

Paligemma: fix generation with Gemma2 (#36044)

zucchini-nlp committed 1 year ago

987e09e5

Update `test_flash_attn_2_can_dispatch_composite_models` (#36050)

ydshieh committed 1 year ago

4886cb79

Fix repo consistency (#36063)

ydshieh committed 1 year ago

8201506d

Fix usage of unpad_input function (#35925)

pavelgein committed 1 year ago

1bfcddbb

Iterative generation using Input embeds and `past_key_values` (#35890)

yaswanth19 committed 1 year ago

d68779b3

Add `Qwen2VLImageProcessorFast` into `Qwen2VLProcessor` (#35987)

yeliudev committed 1 year ago

b44d36e1

Fix Audio Classification Pipeline top_k Documentation Mismatch and Bug #35736 (#35771)

sambhavnoobcoder committed 1 year ago

5f597b74

Fix how we compute the final non-padding token for ForSequenceClassification models (#35911)

Rocketknight1 committed 1 year ago

31bbef04

[docs] no hard-coding cuda (#36043)

faaany committed 1 year ago

ac4acde4

[docs] fix bugs in the bitsandbytes documentation (#35868)

faaany committed 1 year ago

8e499a20

[docs] no hard coding cuda as bnb has multi-backend support (#35867)

faaany committed 1 year ago

741865ee

DeepSpeed github repo move sync (#36021)

stas00 committed 1 year ago

1e929f35

add support for empty list as input to create_model_card (#36042)

ROZBEH committed 1 year ago

c14c2cc0

Add XPU type for work-around -inf mask causing sdpa NaN issue in modeling files (#35647)

Liangliang-Ma committed 1 year ago

c21f13fe

Fix synced multi-GPU generation with LLMs and VLMs (#35893)

ManukyanD committed 1 year ago

dc3a523e

MekkCyber committed 1 year ago

99f9afb0

MekkCyber committed 1 year ago

99f9e044

Merge remote-tracking branch 'deepseek_fork/feature/#35425' into test-deepseek-fp8

MekkCyber committed 1 year ago

65420adb

Merge remote-tracking branch 'upstream/adding_fp8_quantization' into test-deepseek-fp8

MekkCyber committed 1 year ago

4dabaaa5

fix quantization logic

MekkCyber committed 1 year ago

c4baf011

Fix Gemma2 synced multi-GPU generation (#35232)

ManukyanD committed 1 year ago

Verified 4831a94e

Older