huggingface/transformers

Pull Requests Commits

Point the flex block-size comment at pytorch/pytorch#200348

qgallouedec committed 24 days ago

e0011f7b

Merge remote-tracking branch 'origin/main' into flex-large-head-dim-blocks

qgallouedec committed 24 days ago

ea872f7a

Use 32-row flex attention blocks for head_dim > 256

qgallouedec committed 24 days ago

055f2fff

[Qwen3VLMoe] Update expected text in test_small_model_integration_test_expand (#49301)

ydshieh committed 24 days ago

Verified 4cc2aa84

Fix backend attribute in documentation and default backend in MistralCommonBackend (#49149)

samyuktahegde committed 24 days ago

Verified 59150f04

Optimize Whisper generation for TPU and device sync (#49426)

tengomucho committed 24 days ago

Verified 4937b0cf

More RoPE simplification, TF32 fixes (#49328)

Rocketknight1 committed 25 days ago

Verified 812a950d

Complete PR #49228: remove unused device=None from rope __init__ in 11 model families (#49285)

ydshieh committed 25 days ago

Verified 678914d8

Revert "[CI] Temporarily disable AMD scheduled CI trigger" (#49433)

ydshieh committed 25 days ago

Verified ab138d34

Fix QuantizedCache residual flushing for multi-token updates (#49418)

LE0-Lin committed 25 days ago

Verified 1734d8d9

Generation cleanup in pipelines (#49269)

Rocketknight1 committed 25 days ago

Verified bd3f926f

Fix data-dependent branch in _ignore_causal_mask_sdpa under compile (#49354)

dacorvo committed 25 days ago

Verified 65957315

Fix Inkling moe_intermediate_size when dense_intermediate_size is present (#49416)

Talhax55z committed 25 days ago

Verified ff1f5860

fix bug for DebertaV2ForMaskedLM non-legacy load issue (#49152)

kaixuanliu committed 25 days ago

Verified 421d4e8e

[Model] Add PP-DocLayoutV4 Model Support (#48387)

maxiaolong03 committed 25 days ago

Verified e233d3f7

This PR fixes Gemma3n fp16 loading when missing weights need to be in… (#49147)

sywangyi committed 25 days ago

Verified f35aeb93

Fix Whisper batch timestamp decoding for tensors (#49419)

Kingofbooks committed 25 days ago

Verified 9167f73f

:rotating_light: SubConfigSpec for centralized init of subconfigs (#48282)

zucchini-nlp committed 25 days ago

Verified d4f69a86

PR CI: grant actions: read to the reusable workflow (#49417)

tarekziade committed 25 days ago

Verified 125f265a

[Bugfix] [Modeling] Fix DSV4 norm (#47486)

kylesayrs committed 25 days ago

Verified fa406191

Do not wrap the model in FSDP for expert-parallel token dispatch (#49383)

3outeille committed 25 days ago

Verified cb33194a

[docs] Update Spaces (#49242)

stevhliu committed 25 days ago

Verified a96730c8

handle empty precompiled charsmap in spm converter (#49211)

itazap committed 26 days ago

Verified 91ea1192

🚨🚨🚨 luke fixe (#48408)

itazap committed 26 days ago

Verified 2b727b6d

[CI] Temporarily disable AMD scheduled CI trigger (#49394)

ydshieh committed 26 days ago

Verified 7b822d20

[tests] Fix flaky CMVN tests in Speech2TextFeatureExtractionWithoutTorchaudioTest (#49374)

ydshieh committed 26 days ago

Verified 705a7cf8

Fix Phi3 LongRoPE cache rebuild at the context boundary (#49361)

LE0-Lin committed 26 days ago

Verified 6a33cb58

[Docker] Upgrade to torch 2.14/CUDA 13.0 and optimize FA2 build from source (#49381)

ydshieh committed 26 days ago

Verified 6d3802a4

Fix Pixtral Flash Attention image boundaries after axial RoPE refactor (#49373)

oliverholworthy committed 26 days ago

Verified ff704ffd

Let token dispatch run with `ep_size != tp_size` (#49368)

qgallouedec committed 26 days ago

Verified e598fbad

Older