vllm-project/vllm

Pull Requests Commits

Merge branch 'main' into naive-dispatch-combine

robertgshaw2-redhat committed 2 days ago

Verified 5efe9ff6

[responsesAPI] add unit test for optional function tool call id (#32036)

qandrew committed 2 days ago

Verified a307ac07

[ROCm][CI] Handle pytest status code 5 when a shard isn't allocated any tests (#32040)

divakar-amd committed 2 days ago

Verified a28d9f44

[Kernel][MoE] fix computation order of MoE weight multiplication and improve flow (#31962)

xuebwang-amd committed 2 days ago

Verified 629584bf

[Model Runner V2] Add support for M-RoPE (#32143)

WoosukKwon committed 2 days ago

Verified 0a7dd237

[Model Runner V2] Minor refactor for logit_bias (#32209)

WoosukKwon committed 2 days ago

Verified dec28688

[BUGFIX] Add missed remaping of the names of fp8 kv-scale (#32199)

vadiklyutiy committed 2 days ago

Verified 9f430c94

[NIXL][Bugfix] Failure logging overhaul + early metadata free on failure (#32031)

NickLucche committed 2 days ago

Verified f8bd8394

[Model Runner V2] Support logit_bias, allowed_token_ids, min_tokens (#32163)

WoosukKwon committed 2 days ago

Verified ca81811b

[Misc][BE] Type coverage for vllm/compilation [3/3] (#31748)

Lucaskabela committed 2 days ago

Verified ad8818bb

[Misc] Change log level for batch queue log (#32192)

NickLucche committed 2 days ago

Verified 08e8e99c

[BugFix] scheduler: Fix ordering preserving of skipped requests (#32173)

orozery committed 2 days ago

Verified 2be765b6

[Misc] Set default torch num threads for input processing (#31879)

ywang96 committed 2 days ago

Verified 16abe6b8

[Refactor] EPLB rebalance algo to NumPy (#30697)

ilmarkov committed 2 days ago

Verified 1eb61ab3

[BugFix] fix FusedMoE.make_expert_params_mapping in EXAONE-MoE (#32196)

lkm2835 committed 2 days ago

Verified 3d962d72

[3/N][Attention] Move AttentionMetadata-related code from utils.py to backend.py (#32054)

MatthewBonanni committed 2 days ago

Verified 20228cb8

[Benchmark] Share data between SLA runs (#32184)

DarkLight1337 committed 2 days ago

Verified 7c0d3c51

[Misc][PD] Fix `get_attn_backend` usage in transfer connectors (#31988)

NickLucche committed 2 days ago

Verified 5b681074

[Bugfix] Fix stale SSM state for new Mamba requests scheduled as decode (#32118)

Josephasafg committed 2 days ago

Verified 8fb2c135

[Model] Standardize pooling heads (#32148)

DarkLight1337 committed 2 days ago

Verified 8863c2b2

[FIX] Add NO_MUL activation support for modular kernel path (#31528)

danielafrimi committed 2 days ago

Verified 3f72639d

[MODEL] New model support for kakaocorp/kanana-1.5-v-3b-instruct (#29384)

kakao-steve-ai committed 2 days ago

Verified 6bc9c847

Add K-EXAONE-236B-A23B (#31621)

lkm2835 committed 2 days ago

Verified 63ed2409

doc: Update model references in supported_models.md (#32188)

andyzhangx committed 2 days ago

Verified 95e53d90

[ROCm] [Bugfix] Fix order of mori build in Dockerfile.rocm_base (#32179)

tjtanaa committed 2 days ago

Verified 0346396e

doc: Update model name for Qwen3-Coder in documentation (#32185)

andyzhangx committed 2 days ago

Verified e68b0dad

OffloadingConnector: Add cpu_bytes_to_use configuration (#24498)

orozery committed 2 days ago

Verified 9cddbdba

[Feature] Support recording expert indices for rollout router replay (#28284)

xhx1022 committed 2 days ago

Verified 49e6b86c

[P/D] Refactor mooncake connector sender thread using async coroutines (#31573)

dtcccc committed 2 days ago

Verified 0565f1fd

[Bugfix] Fix missing scale passing for encoder Triton Attention implementation (#32149)

Isotr0py committed 3 days ago

Verified 9dbe1fe9

Older