huggingface/transformers

Pull Requests Commits

Support block_size and bf16_stochastic_round keyword arguments for torchao optimizers (#42972)

klae01 committed 90 days ago

Verified c9b0498f

Silence pytest warnings due to sentencepiece version (#43118)

Cyrilvallez committed 90 days ago

Verified e8d60a7f

Fix warnings poping up with `fixup` and `pytest` (#43109)

Cyrilvallez committed 90 days ago

Verified a616b914

Add expected outputs to DetrForSegmentation docstring example (#43035)

raimbekovm committed 90 days ago

Verified 1802c84f

Added import torch in distilbert pipeline docs (#43053)

PredictiveManish committed 90 days ago

Verified 761d66e4

Fix duplicate words typos in comments and docstrings (#43038)

raimbekovm committed 90 days ago

Verified fc15b322

skip test_sdpa_can_dispatch_on_flash in tests/models/pe_audio/test_mo… (#43014)

sywangyi committed 90 days ago

Verified 37cd3240

Fix InvalidVersion error when flash_attn version cannot be determined (#43004)

lukepayyapilli committed 90 days ago

Verified 82c60184

Update cpu torchao usage (#42992)

jiqing-feng committed 90 days ago

Verified 24502729

Make EoMT visible to type checkers (#42967)

simonreise committed 90 days ago

Verified b3837539

fix TP issue within the device mesh (Tensor Parallel group) (#43105)

sywangyi committed 90 days ago

Verified 314a4515

Fix Qwen3OmniMoe Talker weight loading and config initialization (#43084)

HuiyingLi committed 90 days ago

Verified 64a476b7

Improve `_init_weights` tests (#43108)

Cyrilvallez committed 91 days ago

Verified 486a7229

Fix Qwen3-VL typos and clarify image/video/vision token descriptions (#43033)

stefgina committed 91 days ago

Verified 37c9c941

fix(qwen2vl): make Qwen2VLImageProcessor respect config['size'], align with fast version (#43095)

qianyue76 committed 91 days ago

Verified b19834d8

batched and grouped experts implementations (#42697)

IlyasMoutawwakil committed 91 days ago

Verified 0642963b

Make projection LayerNorm configurable in LFM2-VL (#43087)

ankke committed 91 days ago

Verified 3c251772

GLM-ASR Support (#42875)

zRzRzRzRzRzRzR committed 102 days ago

Verified a7f29523

Fix audio pipelines (#42920)

zucchini-nlp committed 103 days ago

Verified 7bbfc19a

Move missing weights and non-persistent buffers to correct device earlier (#43021)

Cyrilvallez committed 103 days ago

Verified d6a6c826

Do not use accelerate hooks if the device_map has only 1 device (#43019)

Cyrilvallez committed 104 days ago

Verified 315dcbe4

Ensure CI runs `check_modeling_structure` and fixes for `make fixup` (#43002)

ebezzam committed 104 days ago

Verified dc06f2dd

Torch distrib smallfix (#43008)

molbap committed 104 days ago

Verified 50ddcfd2

Prefill-related logic in input preparation for generation (#42088)

zucchini-nlp committed 104 days ago

Verified ccc7d90b

Move `original_max_position_embeddings` to rope params (#42513)

zucchini-nlp committed 104 days ago

Verified 34732810

Do not return a tuple in mistral tokenizer Automapping (#42997)

molbap committed 104 days ago

Verified 9971e410

Fix deepspeed + quantization (#43006)

SunMarc committed 104 days ago

Verified 42512f79

[loading][TP] Fix device placement at loading-time, and simplify sharding primitives (#43003)

Cyrilvallez committed 104 days ago

Verified 5f1c05cf

Improve spacing of markdown files (#42984)

cyyever committed 104 days ago

Verified e9f0f8e0

Fix incorrect library name in BitNet integration warning (#42966)

leaderofARS committed 104 days ago

Verified f7d139c1

Newer Older