huggingface/transformers

Pull Requests Commits

Merge branch 'main' into cb-attn-per-call

qgallouedec committed 2 days ago

Verified 9dbfbdde

Enable FSDP2 + expert parallelism via a 2-D (fsdp, tp) device mesh (#48516)

qgallouedec committed 2 days ago

Verified 415e6d2f

unifying device_mesh init to enable PP + TP inference (#48155)

3outeille committed 3 days ago

Verified 5474a55e

Use huggingface_hub httpx export (#48685)

Wauplin committed 3 days ago

Verified 80a79d1b

Let continuous batching page the call, not the model

qgallouedec committed 3 days ago

52f6f806

[DeepseekV3] OOM cascade root-cause investigation (generator ref leak in conversion_mapping) (#48720)

ydshieh committed 3 days ago

Verified df04b012

Keep the expert-parallel sentinel slots out of the router gradient (#48689)

qgallouedec committed 3 days ago

Verified 67f50855

add HyperClovaX Vision (#44314)

jp1924 committed 3 days ago

Verified a6cf156d

Add flash_attention_4 in attn_implementation AutoModel docstring (#48684)

3manifold committed 3 days ago

Verified 9c944e97

[CI] Replace hardcoded username allowlists with author_association check in workflow triggers (#48712)

ydshieh committed 3 days ago

Verified 3f601734

Fix D-FINE / RT-DETR main loss being computed over the denoising queries (#48528)

stefan-it committed 4 days ago

Verified 846bdfdd

[serge] Fix 1 integration test for model `flex_olmo` failing with `other` (other (1)) (#48668)

sergereview[bot] committed 4 days ago

Verified 6b07e451

QA: Added a `MemoryCleanupMixin` class for tests (#48681)

tarekziade committed 4 days ago

Verified 8fcb5239

Fix slow integration tests on XPU (#48611)

jiqing-feng committed 4 days ago

Verified 5509c70d

Fix Qwen2.5-VL temporal RoPE for fractional video intervals (#48669)

yeyeyeping committed 4 days ago

Verified bd15bc95

[serge] Fix 2 integration tests for model `mistral` failing with `other` (other (2)) (#48429)

sergereview[bot] committed 4 days ago

Verified 177e90dd

[serge] Fix 2 integration tests for model `minimax` failing with `output_mismatch` (tensor values differ (2)) (#48515)

sergereview[bot] committed 4 days ago

Verified ff4bb5ee

[Fix] yolos offload issue (#48688)

molbap committed 4 days ago

Verified 9d13f8cf

Skip the expert-parallel sentinel masking when expert parallelism is off (#48201)

qgallouedec committed 4 days ago

Verified 606e6e8b

Synthetic test assets (#48589)

tarekziade committed 4 days ago

Verified c6947074

Fix deepspeed ci (#48640)

SunMarc committed 4 days ago

Verified 29723434

[MusicgenMelody] Fix conditioning silently dropped at generation step 0 (#48679)

ydshieh committed 4 days ago

Verified 5bed6296

[`MoE`] Fix eager EP (#48653)

vasqu committed 4 days ago

Verified 93ebf6b1

[higgs_audio_v2] Fix: use config.num_codebooks in audio labels tensor (Simple Fix!) (#48560)

b-re-w committed 5 days ago

Verified 455bcd0a

Assert cached decode matches recomputing without a cache (#48289)

IlyasMoutawwakil committed 5 days ago

Verified fd798ba1

Fix `AutoImageProcessor` requiring torchvision when only Pillow is installed (#48616)

blipbyte committed 5 days ago

Verified 1ee31851

Mirror the registered pytree flatten when building automatic dynamic shapes (#48578)

IlyasMoutawwakil committed 5 days ago

Verified c583a3a1

Register activation kernel layers on XPU (#47858)

jiqing-feng committed 5 days ago

Verified cb541c21

[GLM 5.3 Flash] Preserve original names when saving text checkpoints (#48676)

Dovis01 committed 5 days ago

Verified e71d8c53

[fix] Fix GlmOcr integration tests: wrong token IDs and image token decode bug (#48650)

ydshieh committed 5 days ago

Verified 4815a0a6

Older