huggingface/transformers

Pull Requests Commits

Fix wrong gradients for all non-expert parameters in expert-parallel training

qgallouedec committed 1 day ago

Verified 47bf301e

Fix NaN gradients in expert-parallel training: mask uninitialized grouped_mm rows

qgallouedec committed 1 day ago

Verified 946f3f1c

Refactor GGUF to speed-up inference (#47779)

SunMarc committed 2 days ago

Verified f62dc9bf

[generate] stop synchronizing the accelerator on every decode step (#47975)

SunMarc committed 2 days ago

Verified 4b24a197

Fix failing tests for cohere_compass (#48005)

kaixuanliu committed 2 days ago

Verified e4052f55

[Fix] Use dedicated helpers for DeepGEMM and SonicMoE tests (#48523)

remi-or committed 2 days ago

Verified b8a19796

CI: Point test fixtures at hf-internal-testing copies we already host (#48521)

tarekziade committed 2 days ago

Verified 892044aa

fix failed test cases for glm5_next (#48497)

kaixuanliu committed 2 days ago

Verified 31978765

vibevoice: fix bug for quant cache (#48487)

kaixuanliu committed 2 days ago

Verified ec657b18

Pass kwargs to the Mamba2 mixer in Nemotron-H, Falcon-H1 and Mamba2 (#48490)

kfastino committed 2 days ago

Verified 86a6c97c

Retire test_multi_gpu_data_parallel_forward (#48508)

tarekziade committed 2 days ago

Verified 37ed0140

Processing tests [part 2] (#47922)

zucchini-nlp committed 2 days ago

Verified 976e7a41

QA: Add noisy comment checker (#48484)

tarekziade committed 2 days ago

Verified 03a4d7b9

:rotating_light: Vision (2d/3d) rotary embeddings (#48105)

zucchini-nlp committed 2 days ago

Verified 4177486a

Allow nested rope params for tiny models (#48435)

zucchini-nlp committed 2 days ago

Verified d3f494b0

Fix sliding-window mask `layer_idx` in Gemma3/Gemma4 `create_masks_for_vision_model` (#48482)

jiqing-feng committed 2 days ago

Verified c6762021

Support nested FLA kernel imports for fla-core (#48221)

DimensionSTP committed 3 days ago

Verified c119ec3c

add xpu expectations for hunyuan_vl model tests (#48504)

kaixuanliu committed 3 days ago

Verified 904b5551

[`Qwen 3.5 Moe`] Fix decorators (#48436)

vasqu committed 3 days ago

Verified 92732e37

Infinite loop in dependency search (#48393)

zucchini-nlp committed 3 days ago

Verified cfa309cf

[serge] Fix 2 integration tests for model `fsmt` failing with `output_mismatch` (tensor values differ (2)) (#48496)

sergereview[bot] committed 3 days ago

Verified 8ce59c13

Fix some tests by removing the deprecation cycle (#48503)

Cyrilvallez committed 3 days ago

Verified ea8f1f74

Remove deprecation (#48500)

Cyrilvallez committed 3 days ago

Verified a8d5f2c8

[AfMoE] Standardize past_key_values argument naming across forward and generate (#48430)

shenhuaqingshi committed 3 days ago

Verified 2f70c41b

Fix Pix2StructTextAttention init using hidden_size instead of d_kv (#47558)

yuvrajnode committed 3 days ago

Verified c66311b5

Fix pre patch release utility (#48499)

Cyrilvallez committed 3 days ago

Verified 23fbf2bb

Update dev version (#48498)

Cyrilvallez committed 3 days ago

Verified e530c283

Fix Inkling inputs_embeds and add more tests (#47827)

Cyrilvallez committed 3 days ago

Verified 774a675d

Simplify and fix qwen4 tests (#48340)

Cyrilvallez committed 3 days ago

Verified 8f542025

Warn once when a hub-kernel function falls back to its reference PyTorch path (#48185)

qgallouedec committed 3 days ago

Verified 78fa17cb

Newer Older