huggingface/transformers

Pull Requests Commits

Load deep-gemm on Blackwell: skip the kernels arch check for it

IlyasMoutawwakil committed 33 days ago

9dda2a88

Fail fast on eval OOM under `auto_find_batch_size` (#49198)

qgallouedec committed 33 days ago

Verified 8ed81bc7

Replace datasets that no longer load with maintained uploads (#49018)

davanstrien committed 33 days ago

Verified a0ccef24

[MPS] Let mps sdpa handle grouped query attention directly (#49187)

Isalia20 committed 33 days ago

Verified 6fc8c147

[Parakeet] Convert NeMo's stochastic depth to layerdrop (#49191)

Deep-unlearning committed 33 days ago

Verified f36c7500

Video processors - general maintenance (#48251)

zucchini-nlp committed 33 days ago

Verified 528c2671

update to torch 2.14 (#49199)

sywangyi committed 33 days ago

Verified b1463f5b

QA: fix noisy comment checker (#49184)

tarekziade committed 33 days ago

Verified b8bea615

Fix command syntax for optimum-cli export (#46451)

Ahwar committed 34 days ago

Verified f339035b

Fix Trainer checkpoint resume crashing on CPU with multiple processes (#49123)

neevmodh committed 34 days ago

Verified 90a491d6

Finish removing the MPS autocast workaround (#49157)

rubenG1009 committed 34 days ago

Verified d125eb2f

Fix backslash handling in generate flag values in transformers chat (#48709)

JHC56 committed 34 days ago

Verified e159fcef

Document running the example scripts on Hugging Face Jobs (#49050)

davanstrien committed 34 days ago

Verified da4bd0dc

[CacheHardIntegrationTest] use dedicated safetensors repo to fix Xet bucket cache corruption (#49182)

ydshieh committed 34 days ago

Verified 1e718279

Document image_like_kwargs (#49180)

guarin committed 34 days ago

Verified 7087be01

Fix lost call (#49163)

molbap committed 34 days ago

Verified a1289da0

[imagegpt/vilt/trocr] fix cache fixture tests: use hf_hub_download instead of load_dataset (#49178)

ydshieh committed 34 days ago

Verified ebd5de00

[CB] Fix failing tests discovered when using the B200 (#49171)

remi-or committed 34 days ago

Verified a73a2ba7

[PerceptionLM] Restore test_inputs_embeds overrides to fix flaky test (#49165)

ydshieh committed 34 days ago

Verified f5af3202

[MPS] Remove cu_seqlens_k clone workaround for metal-flash-sdpa (#49091)

Isalia20 committed 34 days ago

Verified 2ceac527

No more -hf repo names for ESMC (#49158)

Rocketknight1 committed 34 days ago

Verified ba9e224e

[generation] Encode multimodal data only once (#45783)

zucchini-nlp committed 35 days ago

Verified 02914581

[CI] ssh-runner: add optional cache_type input to switch between bucket and EFS runners (#49159)

ydshieh committed 35 days ago

Verified 22a2f92e

Deprecate the use_mamba_kernels config flag that no longer has any effect (#49155)

albertvillanova committed 35 days ago

Verified 83e427ad

Pick the default flash implementation based on the current hardware (#49109)

Abdennacer-Badaoui committed 35 days ago

Verified 0a13bb75

Fix get_json_schema dropping items/enum for unions of list/dict/Literal types (#49136)

JoeyTan21 committed 35 days ago

Verified 2e703d6c

Auto generate model inits (#47829)

guarin committed 35 days ago

Verified 33aaea53

Skip flash tests that fall back to a hub kernel when kernels is missing (#49129)

Abdennacer-Badaoui committed 35 days ago

Verified 89310e24

Fix UMT5 decoder self-attention not being causal (#49135)

the-cross-art committed 35 days ago

Verified d7ba9376

[Gemma] Fix test_model_7b_fp16_static_cache expected value for cuda 8 after #49084 (#49133)

ydshieh committed 35 days ago

Verified 8d430734

Older