vllm-project/vllm

Pull Requests Commits

[BugFix] EPLB + B200 + DeepGEMM : Handle column-major scales tensor (#29162)

varun-sundar-rabindranath committed 16 days ago

Verified 3137991f

Default model load/config/tokenizer to `mistral` format if relevant files exist (#28659)

juliendenize committed 16 days ago

Verified 57430fc9

[BugFix] Fix missing symbol triggering FA2 fallback on Hopper (#29107)

LucasWilkinson committed 16 days ago

Verified c68c7b40

[log] add weights loading time log to sharded_state loader (#28628)

andyxning committed 17 days ago

Verified 53a1ba6e

[BugFix] Make sure to allocate worst case MoE workspace during profile run in the DP + EP case (#27426)

LucasWilkinson committed 17 days ago

Verified 1840c5cb

[Chore] Fix pre-commit error after #25266 (#29190)

WoosukKwon committed 17 days ago

Verified 1bed891f

[Deprecation] Deprecate `seed=None` (#29185)

DarkLight1337 committed 17 days ago

Verified ceca0605

[ROCm][CI] Fix config/test_config_generation.py (#29142)

charlifu committed 17 days ago

Verified 75648b16

[NIXL] Fix after virtual block_size for host_buffer with heter kv_layout (#29122)

xuechendi committed 17 days ago

Verified 460d02a4

Add TRTLLM MoE NVFP4 kernel to CompressedTensorsW4A4MoeMethod (#28892)

Victor49152 committed 17 days ago

Verified b4c8fbaa

[CI/Build][Kernel][AMD] Move extra dim to after load in _fwd_kv_parallel in lighting_attn.py (#29132)

rasmith committed 17 days ago

Verified e99e4673

[Log] Optimize startup log (#28948)

yewentao256 committed 17 days ago

Verified a42ab317

Upstream triton fp4 weight preshuffle (#28888)

maleksan85 committed 17 days ago

Verified b7f1f490

GPU Model Runner V2 (#25266)

WoosukKwon committed 17 days ago

Verified 30b44a15

[CI] Add batch invariant test to ci (#27842)

yewentao256 committed 17 days ago

Verified 1f400c58

[CI/Build] Fix illegal memory access and unsupported test in kernels/attention/test_cache.py (#29118)

rasmith committed 17 days ago

Verified 711241c1

[Misc] Move dynamic seed initialization to `EngineArgs` (#29165)

DarkLight1337 committed 17 days ago

Verified d7219bcd

[Doc] Update plugin doc (#28532)

wangxiyuan committed 17 days ago

Verified 4050bae4

[Perf] These changes enhance the NUMA functionality of vllm for systems with more than one NUMA nodes per socket (#25559)

skaraban3807 committed 17 days ago

Verified f1805db1

Fix mistral config (#29172)

juliendenize committed 17 days ago

Verified 434f3d3e

Tool Call Parser logs should not contain user input / model output except on DEBUG (#29160)

sfbemerk committed 17 days ago

Verified 2092ce8c

fix cross attention (#28346)

fsx950223 committed 17 days ago

Verified fc9f821d

Revert "Revert #28875 (#29159)" (#29179)

DarkLight1337 committed 17 days ago

Verified 94528630

Update Dockerfile to use gcc-toolset-14 and fix test case failures on power (ppc64le) (#28957)

bhagyashrigai committed 17 days ago

Verified 2b1b3dfa

[Core] Align whisper closer to other multimodal models (#27292)

russellb committed 17 days ago

Verified cca2d2cd

[V0 deprecation] Remove more V0 references (#29088)

DarkLight1337 committed 17 days ago

Verified aab0102a

[Misc] remove useless v1 env (#29164)

david6666666 committed 17 days ago

Verified b34129bf

Revert #28875 (#29159)

DarkLight1337 committed 17 days ago

Verified 4d7231e7

[CI Failure] Fix Gemma3 RoPE configuration for sliding attention layers (#29111)

hl475 committed 17 days ago

Verified 8ac3a414

[Minor][Clean] Remove the legacy assertion in video (#29150)

gcanlin committed 17 days ago

Verified 7d6da483

Older