vllm-project/vllm

Pull Requests Commits

[Core] Enable the extensible KV cache by default and size to measured memory

njhill committed 8 hours ago

0f24e2d0

[Core] Extensible KV cache: KV connector support

njhill committed 8 hours ago

2610879f

[Core] Extensible KV cache: sleep mode support

njhill committed 8 hours ago

9b047c52

[Core] Extensible KV cache: VMM driver abstraction, probe, and ROCm

njhill committed 8 hours ago

ff92034c

[ModelRunner V2] Size the KV cache from memory measured after warmup

njhill committed 8 hours ago

c31a559f

[Core] Add ExtensibleTensor and the enable_extensible_kv_cache option

njhill committed 8 hours ago

3f9f2570

[watermarking] Dual-key gumbel-max watermarking for speculative decoding support (#56122)

TQCB committed 9 hours ago

Verified 7ee8a6dd

[CI] Update DeepSeek V4.1 MegaMoE routing test (#56599)

khluu committed 10 hours ago

Verified 1ee4be4d

[CI] Fix Qwen3 Omni DSpark load test config (#56600)

khluu committed 10 hours ago

Verified fd7cd4b8

[Bugfix] Fix DeepGEMM FP8 warmup coverage (#56452)

ayush1399 committed 10 hours ago

Verified 7bcca167

[4/N] Expose HiSparse cache metrics via KV connector stats (#56061)

NickLucche committed 10 hours ago

Verified 29332cf9

[Bugfix][CI] skip conftest for NPU compatibility test (#55799)

yzeyu71 committed 11 hours ago

Verified dff76bc3

[CI/Build] Add DSv4.1 auto-label rules and narrow DSv4 (#56332)

jcotant-inferact committed 11 hours ago

Verified 3bb23e2d

Add @arpera to CODEOWNERS of Structured Output (#56501)

arpera committed 12 hours ago

Verified 9d3e991e

[Bugfix][CI] Update CuTeDSL indexer Q sentinel test for migrated wrappers (#56594)

LopezCastroRoberto committed 12 hours ago

Verified c3771146

[Frontend][Rust] Reject empty structured-output values (#54821)

KernelClint committed 13 hours ago

Verified 22f6e4ec

[Elastic EP] Reuse CUDA graphs across reconfiguration (#54985)

itayalroy committed 14 hours ago

Verified 658c8131

[Perf] Fuse DSV4.1 input metadata preparation with Triton (#56562)

WoosukKwon committed 14 hours ago

Verified 13e221f8

[XPU][CI] Skip ROCm test on non-ROCm platforms (#56555)

zhenwei-intel committed 15 hours ago

Verified 2f59050e

[Refactor][ROCm] Migrate the RDNA3 W4A16 MoE to the oracle/experts pa… (#55522)

JartX committed 15 hours ago

Verified 986e2f87

[Bugfix] Initialize data parser in Nano-Nemotron audio test (#56401)

taneem-ibrahim committed 15 hours ago

Verified 1b29c508

[Nano-Nemotron] Fix Nano-Nemotron precomputed multimodal embeddings (#56398)

taneem-ibrahim committed 15 hours ago

Verified a0914ab7

[XPU][CI] Remove pip install dependency in test yaml files (#55171)

jikunshang committed 16 hours ago

Verified bde4feb0

[CI/Build] Give the Torch ABI audit time to start (#56596)

khluu committed 18 hours ago

Verified d86257e2

[Build] Define _USE_MATH_DEFINES for FlashMLA targets (#54007)

arcusbuilds committed 18 hours ago

Verified 6b153463

[CI] Sample GPU utilization and memory alongside test timelines (#56541)

khluu committed 18 hours ago

Verified 33fa95a0

[Frontend] create unified Cohere parser (#56392)

walterbm committed 20 hours ago

Verified 756794a9

[ROCm][CI] Stage F gating (#55252)

AndreasKaratzas committed 20 hours ago

Verified d406f09e

[DSV4.1] Remove compressor-aware image sentinel token padding (#56554)

Isotr0py committed 21 hours ago

Verified 30118ba2

[Bugfix][Examples] Launch prefill and decode concurrently in NixlPushConnector demo (#55088)

junuxyz committed 22 hours ago

Verified 372477d2

Older