huggingface/transformers

Pull Requests Commits

Update secrets in upload_pr_documentation.yml

paulinebm committed 130 days ago

Verified f9883558

Fix typing of train_args (#41142)

cyyever committed 130 days ago

Verified a61fc6a0

Unify is_torchvision_v2_available with is_torchvision_available (#41227)

cyyever committed 130 days ago

Verified 919a4845

update code owners (#41221)

ydshieh committed 130 days ago

Verified 8e7b0655

Adapt to the SDPA interface to enable the NPU to call FlashAttentionScore (#41143)

frozenleaves committed 130 days ago

Verified 2dd175e6

Remove old Python code (#41226)

cyyever committed 130 days ago

Verified cf0887f6

:rotating_light: [`v5`] Remove headmasking (#41076)

vasqu committed 130 days ago

Verified 52f5eca7

[generate] cache missing custom generate file (#41216)

gante committed 130 days ago

Verified a80f05df

Align pull request template to bug report template (#41220)

tomaarsen committed 130 days ago

Verified 1f1e93e0

[ESM] add accepts_loss_kwargs=False to EsmPreTrainedModel (#41006)

pstjohn committed 130 days ago

Verified 2a596f5b

Trainer: Pass `num_items_in_batch` to `compute_loss` in `prediction_step` (#41183)

pramodith committed 130 days ago

Verified 3edd8048

Avoid assumption that model has config attribute in deepspeed (#41207)

tomaarsen committed 130 days ago

Verified 59035fd0

Wait for main process in _save_checkpoint to ensure best checkpoint exists (#40923)

ssharpe42 committed 130 days ago

Verified d9739778

Deprecate `half_precision_backend` (#41134)

SunMarc committed 130 days ago

Verified 06c04e08

Fix Qwen3-Omni audio_token_id serialization issue (#41192)

aug6th committed 130 days ago

Verified 0e5a9756

docs/examples(speech): pin CTC commands to Hub datasets; add Windows notes (#41027)

tayo4christ committed 130 days ago

Verified 42c68251

Remove unnecessary Optional typing (#41198)

cyyever committed 130 days ago

Verified aaf1269d

[docs] Fix links (#41110)

stevhliu committed 130 days ago

Verified 4a02bc70

Embed interactive timeline in docs (#41015)

yonigozlan committed 130 days ago

Verified def4a37e

Fix docker quantization (#41201)

SunMarc committed 131 days ago

Verified 3e975acc

Fix 8bit bnb loading (#41200)

SunMarc committed 131 days ago

Verified 8635d8e7

Fix EXAONE-4.0 dummy id (#41089)

lkm2835 committed 131 days ago

Verified 1f0e9a47

Add EdgeTAM (#39800)

yonigozlan committed 131 days ago

Verified bd37c453

[`Kernels Attention`] Change fallback logic to error out on explicit kernels request and include FA3 (#41010)

vasqu committed 131 days ago

Verified c1db3868

Make quantizers good citizens loading-wise (#41138)

Cyrilvallez committed 131 days ago

Verified 5426edec

Separate docker images for Nvidia and AMD in benchmarking (#41119)

ahadnagy committed 131 days ago

Verified 399c589d

Fix attention sink implementation in flex attention (#41083)

SamuelBarryCS committed 131 days ago

Verified 52cbc7c8

fix(trainer): Avoid moving model with device_map (#41032)

The5cheduler committed 131 days ago

Verified de9a75f5

enable flex attention ut cases on XPU (#40989)

yao-matrix committed 131 days ago

Verified bcc0dae7

Bump hfh prerelease version (#41175)

Wauplin committed 131 days ago

Verified fcd483f0

Older