Commits vllm-project/vllm

[RL][Model Loader] Direct weight reload: write checkpoint weights into live parameters

aoshen524 committed 52 days ago

c8985846

[Quantization] Enable humming wNaM asymmetric quant (zero_point) with compressed-tensors (#46528)

HDCharles committed 53 days ago

Verified 9f07d023

[CPU] Adds support for fp32 attention sinks (#56252)

Ankit-Jaiswal-AMD committed 53 days ago

Verified 88afb9dc

[Perf][MoE] Skip top-k slots routed to non-local experts in TritonExp… (#58051)

ShuoleiWang committed 53 days ago

Verified 9ed81581

[Feature][Frontend] Add DeepSeek-V4 FIM completion rendering (#44229)

QwertyJack committed 53 days ago

Verified d95a896e

[DSpark] Support pipeline-parallel targets in aggregated serving (#56956)

lucifer1004 committed 53 days ago

Verified 157bcb7c

[Bugfix][Tool Parser] Migrate Granite to the streaming Parser Engine (#49648)

nikhilkulkarni1755 committed 53 days ago

Verified 15ed1262

[Bugfix] Disable prefix caching for encoder-only before model config hooks (#58287)

gty111 committed 53 days ago

Verified c9611215

[ROCm][Test] Check GDN prefill numerics and output ownership (#58091)

AndreasKaratzas committed 53 days ago

Verified a22c637e

[XPU] upgrade to PyTorch 2.14 (#56013)

yma11 committed 53 days ago

Verified e34489a5

[Bugfix][SM120][MLA] Support NoPE sparse MLA (GLM-5.3-Flash) on the FlashInfer SM120 backend (#55277)

lucifer1004 committed 53 days ago

Verified c843f0ca

[MRV2] Miscellaneous code cleanup (#57980)

njhill committed 53 days ago

Verified afec265b

[ROCm][CI] Include Python tooling in ROCm CI artifacts (#58271)

AndreasKaratzas committed 53 days ago

Verified 36bc6948

[ROCm][Bugfix] Keep zero MiniMax MXFP8 activation blocks finite (#58089)

AndreasKaratzas committed 53 days ago

Verified 56b3acb5

[Perf][ROCm][Attention] Narrow the Triton prefill-attention KV tile on RDNA3/RDNA4 (#58225)

lijipeng787 committed 53 days ago

Verified 1904ed94

[CI][Bugfix] Extend groupwise rms_norm scale tolerance to CUDA (#58252)

vllm-agent committed 53 days ago

Verified 955bd6ab

[Perf][Attention] Avoid CPU-GPU sync in DCP sequence lengths (#58169)

ZJY0516 committed 53 days ago

Verified 6632ed55

[ROCm][Bugfix] Register MRV2 sampler JIT warmups (#58092)

AndreasKaratzas committed 53 days ago

Verified 96e203b0

[ROCm][Compile] Fuse AITER static FP8 attention output (#58099)

AndreasKaratzas committed 53 days ago

Verified 26879f32

[Refactor][Quantization] Make FP8 and MLA weight transforms reusable pure functions (#57732)

aoshen02 committed 53 days ago

Verified 973a3be7

[Mooncake] Address review nits from #56855 (#57174)

NickLucche committed 53 days ago

Verified 9488318c

[ROCm][CI] Add MI355 dense NVFP4 and MoRI kernel mirrors (#58281)

AndreasKaratzas committed 53 days ago

Verified 153242a3

[ROCm][CI] Mirror the three TurboQuant evaluation groups on MI355 (#58282)

AndreasKaratzas committed 53 days ago

Verified a9f07d0b

[Quantization] Select per-token NVFP4 MoE backends explicitly (#57176)

S1ro1 committed 53 days ago

Verified 9127170b

[Tests] Select V2 for diffusion scheduler unit tests (#58272)

AndreasKaratzas committed 53 days ago

Verified 0f2a15c9

[ROCm][Perf] Use wvSplitK for single-output GEMMs (#53283)

tangzzycc committed 53 days ago

Verified eb2e91d3

[Bugfix][Frontend] Keep length finish_reason for max_tokens-truncated streaming tool calls (#46303)

Sunt-ing committed 53 days ago

Verified b33c8a65

[MoE] Use GateLinear for all MoE models (#58234)

jeejeelee committed 53 days ago

Verified de6674b0

[XPU][CI] Deselect tests/v1/spec_decode/test_mtp.py::test_glm_mtp_defers_lm_head (#58237)

zxd1997066 committed 53 days ago

Verified e9757321

[Spec decode] Support variable-length decode for Kimi-K3 adaptive ver (#52988)

qiching committed 53 days ago

Verified 88aa0d28