Commits vllm-project/vllm

Revert "[Build] Update deep_gemm git tag for locality_domain update (#61491)"

khluu committed 28 days ago

8369d413

[ROCm][Kimi-K3] Don't read W_UK_T before weight loading in the MLA wrapper (#61654)

Rohan138 committed 28 days ago

Verified 9a648b46

[Bugfix][Parser] Stream Gemma4 string tool arguments incrementally (#61494)

bbrowning committed 28 days ago

Verified 4c5ea86f

[Quantization] Support compressed-tensors MXFP4 MoE with FP8 activations via Mxfp4MoEMethod (#59815)

mgoin committed 28 days ago

Verified 984c242e

[Dependency] Update Humming and remove NVFP4 scale workaround (#61556)

jinzhen-lin committed 28 days ago

Verified fe98d90a

[Benchmark] Remove the temporary bench forward dummy loader (#61676)

WoosukKwon committed 28 days ago

Verified 2e8a5930

[Perf][DSv4.1] Length-aware candidate block selection in the decode indexer (#61039)

Juntian777 committed 28 days ago

Verified 7c3f7e55

[ROCm][BugFix] Don't use AITER rms_norm as an IR OP on gfx90a when AITER always on. (#59290)

rasmith committed 28 days ago

Verified 2622ee21

[Bugfix][Frontend] Apply the JSON schema in Cohere Chat v2 response_format (#60586)

mottopanikeiku committed 28 days ago

Verified e73fa009

[Build] Update deep_gemm git tag for locality_domain update (#61491)

cleonard530 committed 28 days ago

Verified ded7ae35

[Perf][DFlash] Capture context combine and anchor in the draft CUDA graph (#59511)

Juntian777 committed 28 days ago

Verified 388a610d

[Dependency] Upgrade FlashInfer to 0.7.1 (#60948)

wzhao18 committed 28 days ago

Verified 4975fda8

[Bugfix][MRV2] Capture DCP FULL cudagraphs with KV context (#61496)

ayush1399 committed 28 days ago

Verified 9b5c6707

[CI/Build] Fix test_deepseek_v41_mxfp8_scale_loading after #57797 (#61678)

stecasta committed 28 days ago

Verified 222e00fc

[Bugfix][ROCm][MiniMax-M3] Reject unbuilt index-head counts in the AITER indexer (#60990)

AndreasKaratzas committed 28 days ago

Verified e96c8226

[Bugfix][MRV2][Spec Decode] Honor EAGLE3 single-state draft configuration (#61563)

sragrawal committed 28 days ago

Verified 21e207ef

[ROCm] Enable fp8 query for Triton attention (#60296)

amd-xavierwang committed 28 days ago

Verified 06265beb

[Perf][MRV2] Skip empty draft-token resets on request admission (#61640)

jzakrzew committed 28 days ago

Verified 857ddac4

[Spec Decode][MRV2] Warm up top-k/top-p and greedy rejection kernels for speculative decoding (#59260)

zixi-qi committed 28 days ago

Verified c1407a7e

[Bugfix] Don't crash DeepGEMM MegaMoE warmup when get_bf16_mega_gate_config is missing (#61657)

simon-veitner-redhat committed 28 days ago

Verified ae94e9c9

[Perf][ROCm][GLM-5.3-Flash] AITER prefill indexer topk (#60753)

simondanielsson committed 28 days ago

Verified 3f729daf

[ROCm][Bugfix][MLA] Fix FP8 MLA prefill unpad for interleave-padded heads (#60907)

lucas-santos-amd committed 28 days ago

Verified e534024e

[Perf][Nemotron-Parse] Do image rescale/normalize on the device (#60500)

HangruiCao2000 committed 28 days ago

Verified d1d8e6c0

[CI/Build] Use FileStore for DeepEP test rendezvous (#56595)

khluu committed 28 days ago

Verified 0b6d07cb

[Frontend] Clean Client Facing 500 Report (#58043)

robertgshaw2-redhat committed 28 days ago

Verified d3f6b2a7

[Bugfix][CI] Fix renderers tests broken by tool name removal from warn logs (#61652)

ganeshr10 committed 28 days ago

Verified 47232f3c

[ROCm][CI] Add dtype to the KDA metadata test's fake model config (#61650)

Rohan138 committed 28 days ago

Verified 52a8cc53

[Benchmark] Add vllm bench forward: exact-shape forward timing (#61339)

WoosukKwon committed 28 days ago

Verified 9d3dfa28

[Model Runner V2] Enable batch-sharded sampling for GPT-OSS (#61257)

Taimys committed 28 days ago

Verified 2c74fd18

[Model] Support Decision 3.0 decision models (vllm-sr/d3) (#61565)

Xunzhuo committed 28 days ago

Verified 89cab550