huggingface/transformers

Pull Requests Commits

Gemma4 training with text-only samples (#45454)

zucchini-nlp committed 136 days ago

Verified 6979d696

[nemotron_h] Add support for MLP mixers (#44763)

xenova committed 136 days ago

Verified 77c0e6e7

add expert parallelism for gemma-4-26B-A4B-it (#45279)

sywangyi committed 136 days ago

Verified e737fb86

Add full GGUF loading support for GPT‑OSS (fixes #43366, supersedes #43757) latest (#45506)

sirzechs66 committed 136 days ago

Verified 5e238684

Update Gemma4 weight conversion script (#45328)

RyanMullins committed 136 days ago

Verified 2da596b0

Move some conversion mappings to PrefixChange (#45567)

Cyrilvallez committed 136 days ago

Verified 71d1a7b7

fix table update versions (#45544)

tarekziade committed 136 days ago

Verified f51cf506

Add disable_mmap kwarg to from_pretrained with hf-mount auto-detection (#45547)

rtrompier committed 136 days ago

Verified 8fb7c7e5

fix(DSV3): parity between native `DeepseekV3MoE` and remote official implementation (#45441)

casinca committed 136 days ago

Verified cb0adddc

Align gemma3n cache sharing to gemma4 (#45489)

Cyrilvallez committed 136 days ago

Verified 08244b96

[modular] Fix modular logic broken in #45045 (#45539)

Cyrilvallez committed 137 days ago

Verified f048e845

Fix: propagate quantization_config to text sub-config for composite models in AutoModelForCausalLM (#45494)

lvliang-intel committed 137 days ago

Verified 85099df9

T5Gemma2: fix `prepare_decoder_input_ids_from_labels` (#45516)

Tokarak committed 137 days ago

Verified fcc22811

[Trainer] Add ddp_static_graph option (#45519)

KeitaW committed 137 days ago

Verified 26bb358e

Add dtype config options for Four Over Six (#45367)

jackcook committed 137 days ago

Verified 1a4bc140

[Sam3LiteText] Remove unnecessary modules/configs (#45535)

yonigozlan committed 137 days ago

Verified 7439ac04

Fix conditional check for float formatting (#44425)

qgallouedec committed 137 days ago

Verified dadd30b8

Fix AMD CI: rebuild torchvision with libjpeg + refresh expectations (#45533)

Abdennacer-Badaoui committed 137 days ago

Verified 67fb8bb3

Fix EP: RouterParallel shape, tp_plan property, grouped_mm sentinels (#45473)

AmineDiro committed 137 days ago

Verified 9dff7ca5

feat[vLLM × v5]: Add vLLM compatibility for audio models (#45326)

harshaljanjani committed 137 days ago

Verified a6dab9f3

Reapply modular to examples (#45527)

Cyrilvallez committed 138 days ago

Verified 5042bb7e

qa: re-run modular converter when the script itself is modified (#45528)

tarekziade committed 138 days ago

Verified ef97a752

[GGUF] Reduce peak RAM usage by casting dequantized tensors early during load (#45386)

UsamaKenway committed 138 days ago

Verified 7a0d582a

Fix CSM `TextToAudioPipeline` missing `<bos>` token (#45525)

jiqing-feng committed 138 days ago

Verified ce77bc37

Raise 400 on model mismatch when `transformers serve` is pinned (#45443)

qgallouedec committed 138 days ago

Verified 243f2d77

[serve] Update tool call to switch to `parse_response` (#45485)

SunMarc committed 138 days ago

Verified e15297eb

[`Conversion Mapping`] Small fixups (#45483)

vasqu committed 138 days ago

Verified 0db33792

:rotating_light: [`Kernels`] Fix kernel function registration (#45420)

vasqu committed 138 days ago

Verified 253809c9

Revert "Fix: modular image processors (#45492)" (#45531)

tarekziade committed 138 days ago

Verified 79f90bc6

fix: return empty tuple from import_protobuf_decode_error when protobuf is unavailable (#45486)

jw9603 committed 138 days ago

Verified c3ec5ff4

Newer Older