Commits vllm-project/vllm

[CI] Validate release 6858 CUDA 12.9 builds without publishing

khluu committed 36 days ago

9fe90329

[Build] Fix DeepGEMM CUDA 12.9 layout header includes

khluu committed 36 days ago

55b0863c

[Bugfix][Multimodal] Preserve DeepSeek V4 image block spacing (#56882)

adenzhou1350 committed 36 days ago

Verified cc09352e

[CI] Raise GSM8K startup max wait to 2400s for flaky 30B+ MoE eval configs (#57361)

khluu committed 36 days ago

Verified 04a3a00e

[CI] Add flaky rerun markers / timeout skips to sibling tests lacking them (#57359)

khluu committed 36 days ago

Verified f2b6152e

[ROCm][Bugfix] Fix intermittent ROCR host segfault (#57328)

mawong-amd committed 36 days ago

Verified 4c6c1a40

[CI] Add residual timeout headroom after JIT rollback (#56649)

khluu committed 36 days ago

Verified 5397f967

[CI] Raise Elastic EP Scaling step timeout 30m -> 40m (#57287)

vllm-agent committed 36 days ago

Verified 5bb59620

[CPU] Add CPU FP8 W8A8 linear/MoE support (#49942)

yuwenzho committed 36 days ago

Verified b3464790

[Quantization] Support native Quark W4A16 INT4/UINT4 exports in vLLM (#48606)

limitmhw committed 36 days ago

Verified 6df2b1a8

[Bugfix][CPU] Fix macOS multimodal SHM cache initialization (#57142)

shaohuaxi committed 36 days ago

Verified 67a8a3f9

[Bugfix][MLA] TritonMLA: fix illegal memory access on causal multi-token decode (#51065)

olka-amd committed 36 days ago

Verified 32636580

C3x SM100 FP8 blockwise - Pad activation scales to multiple of 4 (#57377)

gau-nernst committed 36 days ago

Verified 8da75d61

[Rust Frontend] Return sampling masks over gRPC (#56777)

biswapanda committed 36 days ago

Verified 95d72370

[Bugfix][DBO] Fix DeepEP low-latency profiling crash with DP+EP+DBO (#57502)

micah-wil committed 36 days ago

Verified d5f0a6e8

[Perf] Fuse CohereASR relative attention score accumulation (#55190)

Levius-Fubuki committed 36 days ago

Verified 42a33039

[Frontend] Fix the parsing of missing `string=` in DeepSeek V4 (#56271)

wtdcode committed 36 days ago

Verified 39e33db7

[Frontend] Add `--tool-strict-level` for server-side control for structural tag activation (#56268)

wtdcode committed 36 days ago

Verified dee37d89

[Bugfix][ROCm] Alias SparseAttnIndexerKpool.forward_cuda to forward_native (GLM-5.3-Flash boot crash) (#57425)

mustafayildirim committed 36 days ago

Verified d12c2768

[Perf][MRV2] Share token-to-request mappings across KV cache groups (#57102)

Juntian777 committed 36 days ago

Verified bf13ecc2

[Bugfix] Attach request-level tools to existing system message in DeepSeek V4 Python renderer (#51856)

thegoldenflow committed 36 days ago

Verified 2909ad8f

[Bugfix][KV Cache][GLM-5.3-Flash] Disable slot mapping kernel for the kpool tail buffer (#57317)

simondanielsson committed 36 days ago

Verified 70df48dc

[Bugfix][ROCm][KV Offload] Use private pinned tensors for CPU KV offload (#57160)

yuzhouo7 committed 36 days ago

Verified 7b942936

[Perf][GLM5.3-Flash] Use cooperative top-k for small GLM decode batches (#57327)

chaunceyjiang committed 36 days ago

Verified 2bbdfcfc

[Bugfix][DSv4.1] Preserve NaN-scored candidate block indices (#57454)

WoosukKwon committed 36 days ago

Verified 64563d0e

[Bugfix] DiffusionGemma: hand out stashed logprobs only on the committing step (#57414)

mmastrac committed 36 days ago

Verified 2c88fb13

[ROCm][Bugfix] Gate AITER MXFP8 MoE on the aiter enable flag (#57426)

Rohan138 committed 36 days ago

Verified 213c379f

[XPU] Fix incorrect context-key normalization for Qwen DFlash-based models (#56431)

yma11 committed 36 days ago

Verified 48f663c3

[Frontend][Core] Add release_kv_cache_memory() API (#44890)

andakai committed 36 days ago

Verified 092bdd6d

[AMD][Bugfix] Make the nested-RoPE patch reach automatic validation (#57289)

okorzh-amd committed 36 days ago

Verified a524f80b