Commits vllm-project/vllm

Refactor PCP O-Proj TP into an intermediate PCPOProjTP layer with descriptor-driven weight switching

Copilot committed 18 hours ago

Verified e4a18157

Port PCP O-Proj TP (#53422) onto current main

Copilot committed 18 hours ago

Verified b36f9e06

[Doc] Kimi-K3: document the GB300 disaggregated specialized-decode points in the guide and command builder (#63222)

Ankur-singh committed 18 hours ago

Verified 544f182d

[Doc][Recipe] MiniMax-M3.1-Flash: add Hopper FP8 variant, Docker-only install (#63244)

esmeetu committed 18 hours ago

Verified eff8a568

[Frontend] Enforce max_tool_calls for Responses built-in tools (#47112)

yuyz-cyber committed 18 hours ago

Verified 751ac8d6

[CI][Test] Compare bf16 token embeddings by cosine similarity (#63201)

aarushjain29 committed 19 hours ago

Verified e38af8bd

[Doc][Recipe] Add MiniMax-M3.1-Flash recipe (#63236)

esmeetu committed 19 hours ago

Verified d05b1c83

[XPU] temporarily ignore test_cumem.py::test_basic_cumem (#63084)

yma11 committed 19 hours ago

Verified d5d8ccaa

[Bugfix] Defer MRV2 Triton validation to worker startup (#62576)

jeffreywang88 committed 19 hours ago

Verified 432b1003

[Docs] Fix griffe missing-type warnings in XPress kernels (#63228)

hmellor committed 20 hours ago

Verified 5db4dfa2

[Bugfix][Spec Decode] DFlash/DSpark no longer drop the last prefix-cache block (mamba align context recompute) (#54163)

Ledgero committed 20 hours ago

Verified 0bfeb8d0

[KV Connector][Mooncake] Expose transfer stats via Prometheus (#56939)

CAICAIIs committed 20 hours ago

Verified 01ab8760

[ROCm][Build] Reland #62475: paged attention kernel split with `--offload-compress` => x4-x7 wheel build speedup on ROCm (#63105)

fxmarty-amd committed 21 hours ago

Verified cc1a6d31

[Frontend] Expose --sleep-mode-backend on the CLI (#63138)

Mramaran committed 21 hours ago

Verified d0eb5d2f

[Bugfix][ROCm][DSv4.1] Gather Engram hashes across DP replicas in the AMD model (#63007)

AndreasKaratzas committed 21 hours ago

Verified 6d1909a1

[ROCm][DSv4.1] Allocate offloaded Engram tables at their exact size (#63009)

AndreasKaratzas committed 21 hours ago

Verified f01a76f9

[Bugfix][Model] Build the skipped DeepSeek-VL2 vision tower without meta .item() (#63192)

aarushjain29 committed 21 hours ago

Verified c0c02786

[Startup][UX] Start multiproc workers concurrently when using spawn (#60028)

kfirtoledo committed 22 hours ago

Verified 52f44839

upgrade tpu-inference to v0.32.0 and torch to 2.14.0 (#63080)

jayyen01 committed 22 hours ago

Verified 6ee316ea

[Model] Add Recursive Flint (#62938)

WoosukKwon committed 22 hours ago

Verified 899a1689

[Spec Decode][MRV2] Skip drafting in steps where no request can use the drafts (#62784)

zixi-qi committed 22 hours ago

Verified 856b9841

[Bugfix][Anthropic] Map vLLM error types to Anthropic error.type vocabulary (#62294)

zyulai committed 23 hours ago

Verified f0abb2f8

[Core] Add XPress speculative decoding method (#54448)

ZKBig committed 23 hours ago

Verified ce9467b7

[Bugfix][Frontend] Accept and echo prompt-cache and safety_identifier fields in the Responses API (#62856)

albertoperdomo2 committed 1 day ago

Verified 47c091b1

[Bugfix][Frontend][Kimi K3] Fix Kimi K3 XTML marker leakage in responses API (#62507)

cjackal committed 1 day ago

Verified 2de0531d

[Quantization] Bump humming-kernels to 0.1.18 and remove DeviceInfo workaround (#62816)

jinzhen-lin committed 1 day ago

Verified b4411f58

[Docs][KV Offload] Document canonical_layout in the KV offloading usage guide (#61822)

liranschour committed 1 day ago

Verified c722f3a5

[Bugfix][Attention] Fix FA4 FP8 descale strides for singleton KV heads (#58304)

taeyangyun committed 1 day ago

Verified 2a038794

[ROCm][CI] Fix int8 GEMM kernel selection logic for ROCm platform (#62099)

qli88 committed 1 day ago

Verified 1d3e4006

[Bugfix] Add SM100 support for Step-5-Preview (#63147)

jeejeelee committed 1 day ago

Verified 4a40c611