huggingface/transformers

Pull Requests Commits

Merge branch 'main' into mp-cg-fix

remi-or committed 11 hours ago

Verified b141eb2e

Add device scoping

remi-or committed 11 hours ago

46d506d6

Fix Zamba2 construction for num_mem_blocks > 1 checkpoints (#48325)

john-rocky committed 12 hours ago

Verified 323f2434

Implement VibeVoice (#40546)

pengzhiliang committed 12 hours ago

Verified 640a08a5

[Docs] Change 5.3 Flash pos in toc (#48366)

Dovis01 committed 12 hours ago

Verified afa80816

[conftest] Use get_cpu_ram_total_gib for psutil patch (cgroup-aware) (#48290)

ydshieh committed 12 hours ago

Verified 35a66fa6

Fix flaky test_training_gradient_checkpointing for BigBirdPegasus (fp noise filter) (#48332)

ydshieh committed 12 hours ago

Verified 9b4f0a0c

Quiet continuous batching at default verbosity (#48314)

qgallouedec committed 13 hours ago

Verified 2e7033ed

Wait for the first request in the async continuous batching bootstrap (#48304)

qgallouedec committed 14 hours ago

Verified 01846b52

[CB] Fail faster (#48334)

remi-or committed 17 hours ago

Verified b6c0bfe0

Ignore a stale best checkpoint recorded in a resumed trainer state (#48319)

VaggelisGian committed 1 day ago

Verified 36deb0b5

Keep MXFP4 weights quantized on XPU when use_kernels is set (#47923)

jiqing-feng committed 1 day ago

Verified e15ea9b9

Fix missing FP8 TP layer overrides (#48343)

changwangss committed 1 day ago

Verified d65f37cc

[docs] Fix links and remove TokenizerFast (#47748)

stevhliu committed 1 day ago

Verified 762fae6e

Fix incorrect token classification prefix for ESMC (#48348)

Rocketknight1 committed 1 day ago

Verified 4e5538a3

Create the continuous batching CPU group with local synchronization (#48302)

qgallouedec committed 1 day ago

Verified dabae5fc

Resolve continuous batching config against the text config for composite models (#48299)

qgallouedec committed 1 day ago

Verified 598d8ba8

[Glm 5.3 Flash] GLM 5.3 Flash Support (#48342)

Dovis01 committed 1 day ago

Verified eb4d9e2a

Fix kernel commit and repo paths for ESMFold2 (#48186)

Rocketknight1 committed 1 day ago

Verified ba5cb08d

[`CI`] Unblock fast CI for now (failing tests) (#48344)

vasqu committed 1 day ago

Verified 1e452995

[qwen4_exp] disable torch/onnx export tests due to data-dependent control flow (#48345)

ydshieh committed 1 day ago

Verified 0f0f7e1a

[debug] Trace previous CI run selection in get_previous_daily_ci.py (#48338)

ydshieh committed 1 day ago

Verified 36bc98ef

Normalize HunYuanVL's legacy field aliases via attribute_map (#48261)

hmellor committed 1 day ago

Verified 2c889bf2

Add Qwen4Exp model (#48337)

Cyrilvallez committed 1 day ago

Verified fc5c5bde

Restore BC for the tensor-parallel API (#48300)

ArthurZucker committed 1 day ago

Verified f0e42efb

Fix video-llama modular conversion (#48336)

zucchini-nlp committed 1 day ago

Verified 7d337843

CI: gate the hunyuan-moe slow test (#48330)

tarekziade committed 1 day ago

Verified 949e301b

Add a regression test for force_accelerate_hooks signature preservation (#48260)

wtdcode committed 1 day ago

Verified 5fcf605f

Add an opt-in per-frame pixel cap (cap_pixels_per_frame) to the Qwen3-VL video processor (#48071)

dkrisman committed 1 day ago

Verified 27e7f6c8

Fix `scores` type in stopping criteria docstrings (#47676)

qgallouedec committed 1 day ago

Verified 53e3152d

Older