huggingface/transformers

Pull Requests Commits

🚨 [vLLM] Fix video token counting for Transformers backend video inputs (Part 1) (#48894)

harshaljanjani committed 31 days ago

Verified 770e4c40

Fix docstring argument names that don't match signatures (#48474)

IshaanPotle committed 31 days ago

Verified 94d1ce87

Always materialize the causal mask in Doge so sdpa stays causal (#48821)

blipbyte committed 31 days ago

Verified 204331f0

[Chat Parsing] Coerce oneOf tool arguments (#48719)

yonigozlan committed 31 days ago

Verified 112caee7

[`DSA`] Only save latents on dsa with indexer as well (#48876)

vasqu committed 31 days ago

Verified f0d77833

Make LongcatFlashConfig self-consistent without building the model (#48899)

hmellor committed 31 days ago

Verified c2622a9e

Add auto_docstring task model overrides (#47815)

guarin committed 31 days ago

Verified 59eed1a6

[serge] Fix 3 integration tests for model `deepseek_vl` failing with `other` (other (3)) (#48536)

sergereview[bot] committed 31 days ago

Verified c1b80cf3

[Improvement] Rework the docstring and comments of mHC (#48888)

remi-or committed 31 days ago

Verified 230e6f32

Don't fail model loading when the accelerator can't report free memory (#48136)

studioego committed 31 days ago

Verified 36ca5c6a

Fix Idefics2 padding when the first example has no image (#48753)

Arnavsharma2 committed 31 days ago

Verified 2e39c410

[`feat`] Allow untying `hidden_states[-1]` from `last_hidden_state` via the model config (#48087)

tomaarsen committed 31 days ago

Verified 12ee21d1

Fix GLM rope_parameters dict shared mutation across test instances (#48895)

ydshieh committed 31 days ago

Verified 9d4ad4b7

Fix sliding window cache when assistant model calls generate (#48280)

VaggelisGian committed 31 days ago

Verified 12b6da1f

Fix DFlash sampled candidates losing the batch dimension (#48281)

VaggelisGian committed 31 days ago

Verified af739912

[CB] Add pause mechanism (#48462)

remi-or committed 31 days ago

Verified 8e608337

Fix Pixtral image processor do_pad option (#48872)

jzakrzew committed 31 days ago

Verified 5a471cc1

Fix Minimax M2 partial_rotary_factor by remapping legacy rotary_dim (#48486)

dajiaohuang committed 31 days ago

Verified da9d2c46

Fix `StaticCache` for Mllama and enable `torch.compile` (#48141)

jiqing-feng committed 31 days ago

Verified ffddd251

Fix Glm4MoeIntegrationTest: split into 3 tests, switch to GLM-4.5-Air (#48820)

ydshieh committed 32 days ago

Verified dbea1093

fix(ci): harden GitHub Actions workflows (#48160) (#48834)

hf-security-analysis[bot] committed 32 days ago

Verified b22fe462

Fix flaky RfDetr test_save_load (#48869)

ydshieh committed 32 days ago

Verified af980eef

Fix ESMFold2 ligand iPLDDT weighting: mol_type non-polymer code is 3, not 4 (#48831)

faustomilletari committed 32 days ago

Verified 22e88feb

Add test for causal only variant of some encoder-decoder models (#48760)

nandan2003 committed 32 days ago

Verified c9d531dd

Fix kosmos (#48853)

Cyrilvallez committed 32 days ago

Verified 947d6b4a

[generate] Make all methods and logits processors agnostic to lm_head output size (#48846)

Cyrilvallez committed 32 days ago

Verified b180f143

Fix `reset` on the dynamic cache layers (#48809)

jiqing-feng committed 32 days ago

Verified a5d98c34

QA: deactivate rule 41 (#48852)

tarekziade committed 32 days ago

Verified 031cfc58

[GPTNeoXJapanese] Fix RoPE ignoring partial_rotary_factor (#48652)

blipbyte committed 32 days ago

Verified 2d92ded1

docs: fix broken #combining-with-fsdp2 anchor in expert_parallelism.md (#48854)

ydshieh committed 32 days ago

Verified 51b330bb

Newer Older