huggingface/transformers

Pull Requests Commits

Add adam_kwargs for Apollo Optimizer (#38168)

DaizeDong committed 1 year ago

Verified 656e2eab

Refactor `get_XXX_dataloader` from Trainer (#38090)

yaswanth19 committed 1 year ago

Verified 6bb6821d

[tests] remove `test_sdpa_equivalence` (redundant) (#37911)

gante committed 1 year ago

Verified 40a493c7

fix bug in distributed loss test (#38166)

techkang committed 1 year ago

Verified ea29f61e

Fix import torchao.prototype.low_bit_optim since torchao v0.11 (#38174)

baptxste committed 1 year ago

Verified a4389494

Add args support for fast image processors (#37018)

yonigozlan committed 1 year ago

Verified 0ba95564

[ESM] Add flash-attention-2 backend for ESM-2 (#38023)

pstjohn committed 1 year ago

Verified d69945e5

Feat: add warnings for unused keys and rules in tensor parallel (#37893)

S1ro1 committed 1 year ago

Verified 7b5e327c

remove some commands from `fetch_tests` CircleCI job (#38176)

ydshieh committed 1 year ago

Verified 12093523

Disable `convert to draft` workflow (#38177)

ydshieh committed 1 year ago

Verified 91f6fa00

Disable `Trigger CircleCI by ready for review` (#38171)

ydshieh committed 1 year ago

Verified 5036ec88

clean autoawq cases on xpu (#38163)

yao-matrix committed 1 year ago

Verified 7f28da28

Bart: new cache format (#35314)

zucchini-nlp committed 1 year ago

Verified 01ad9f4b

[VLMs] add helpers to get multimodal encodings (#37743)

zucchini-nlp committed 1 year ago

Verified 3ab47b6c

Add optional RMSNorm support to BitNet quantization (config + layers) (#38087)

Codys12 committed 1 year ago

Verified 1e921a3a

Fix Qwen2.5 Omni `SinusoidsPositionEmbedding` precision (#38151)

BakerBunker committed 1 year ago

Verified 57a79f51

Include output embedding as well with `include_embedding` flag (#37935)

jerryzh168 committed 1 year ago

Verified 44fa04ae

enable autoround cases on XPU (#38167)

yao-matrix committed 1 year ago

Verified 34c1e29c

[FIX] Save speed metrics to logs (#38136)

pavelgein committed 1 year ago

Verified 0f77ca72

Omit creation of positional IDs within ESM if applicable (#38089)

simonlevine committed 1 year ago

Verified 27ef46e8

disable deepspeed when setting up fake trainer (#38101)

winglian committed 1 year ago

Verified fe9426f1

enable trainer test cases on xpu (#38138)

yao-matrix committed 1 year ago

Verified 7caa57e8

Hotfix: Flash Attention 2 support in Pixtral (#38146)

uminaty committed 1 year ago

Verified b11b28cc

[generate] Run custom generation code from the Hub (#36405)

gante committed 1 year ago

Verified 0e0e5c10

Remove head mask in generative models (#35786)

zucchini-nlp committed 1 year ago

Verified 955e61b0

enable csm integration cases on xpu, all passed (#38140)

yao-matrix committed 1 year ago

Verified 0173a99e

[Qwen3] Qwen3 MoE add tp plan for expert mlps (#38135)

hgt312 committed 1 year ago

Verified e5a48785

Fix incorrect attention mask truncate in WhisperFlashAttention2 (#36477)

OliBomby committed 1 year ago

Verified 4005e30c

enable d_fine finetuning properly (#37962)

SangbumChoi committed 1 year ago

Verified aa27fa75

Add `manueldeprada` to `run_slow` whitelist (#38126)

manueldeprada committed 1 year ago

Verified e021bf6b

Newer Older