huggingface/transformers

Pull Requests Commits

mishig25 committed 97 days ago

9e14f680

[docs] update tips syntax

mishig25 committed 97 days ago

46b90ac2

fix qwen text config (#41158)

zucchini-nlp committed 98 days ago

Verified f22cb1e8

Fix white space in documentation (#41157)

cyyever committed 98 days ago

Verified 374ded5e

[docs] Fix tp_plan (#41205)

stevhliu committed 98 days ago

Verified 16a14176

[Trainer] deprecate `num_train_tokens` (#41165)

SunMarc committed 98 days ago

Verified 5d1e8530

[v5] Remove train kwargs (#41127)

SunMarc committed 98 days ago

Verified cecd9284

[v5] Remove deprecated prediction loop (#41123)

SunMarc committed 98 days ago

Verified 103fa6d2

[v5] Remove tokenizer from Trainer (#41128)

SunMarc committed 98 days ago

Verified aa3e8798

Remove old sagemaker api support (#41161)

SunMarc committed 98 days ago

Verified e99dee64

[v5] More Training Args cleaning (#41131)

SunMarc committed 98 days ago

Verified dded9fd1

Revert "Fix DeepSpeed mixed precision precedence over Accelerate defaults" (#41124)

SunMarc committed 98 days ago

Verified 6fb6117a

Fix sliding window attn mask (#41228)

remi-or committed 98 days ago

Verified 5bdb7045

Fix typing of train_args (#41142)

cyyever committed 98 days ago

Verified a61fc6a0

Unify is_torchvision_v2_available with is_torchvision_available (#41227)

cyyever committed 98 days ago

Verified 919a4845

update code owners (#41221)

ydshieh committed 98 days ago

Verified 8e7b0655

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

frozenleaves committed 98 days ago

Verified 2dd175e6

Remove old Python code (#41226)

cyyever committed 98 days ago

Verified cf0887f6

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

vasqu committed 98 days ago

Verified 52f5eca7

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

gante committed 98 days ago

Verified a80f05df

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

tomaarsen committed 98 days ago

Verified 1f1e93e0

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

pstjohn committed 98 days ago

Verified 2a596f5b

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

pramodith committed 98 days ago

Verified 3edd8048

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

tomaarsen committed 98 days ago

Verified 59035fd0

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

ssharpe42 committed 98 days ago

Verified d9739778

Deprecate `half_precision_backend` (#41134)

SunMarc committed 98 days ago

Verified 06c04e08

Fix Qwen3-Omni audio_token_id serialization issue (#41192)

aug6th committed 98 days ago

Verified 0e5a9756

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

tayo4christ committed 98 days ago

Verified 42c68251

Remove unnecessary Optional typing (#41198)

cyyever committed 98 days ago

Verified aaf1269d

[docs] Fix links (#41110)

stevhliu committed 99 days ago

Verified 4a02bc70

Older