[Bugfix] Align deepseek v4 parser thinking default with tokenizer (#51296)

Flora Feng committed 3 months ago

Fix DSpark warmup without sparse index buffer (#50693)

xijiaat committed 3 months ago

[ROCm][DistInf] Enable vLLM DI CI with buildkite/slurm (#47030)

Chaitanya Sri Krishna Lolla committed 3 months ago

[Bugfix][Structured Output] Mask request stop tokens in xgrammar until grammar terminates (#49227)

Flora Feng committed 3 months ago

[Perf] Launch the top-k/top-p Triton sampler kernel with 8 warps (#51507)

kiroxu committed 3 months ago

[Perf] Narrow DeepSeek V4 eager CUDA graph region (#51430)

Woosuk Kwon committed 3 months ago

[Docker] Cache test dependencies before vLLM install (#51184)

Michael Goin committed 3 months ago

[BugFix][SpecDecode] Fix dspark parallel_drafting_token_id init bug (#51602)

wangxiyuan committed 3 months ago

[Build] Skip precompiled wheel fetch during metadata hooks (#51424)

Michael Goin committed 3 months ago

[Bugfix][Kimi-K3] Give the AMD packed KDA decode kernel the state-index stride (#51682)

Lyu, Xudong committed 3 months ago

Fix DoS via sample-rate forgery bypassing audio decode duration guard (#49948)

Juan Pérez de Algaba committed 3 months ago

[ROCm][MLA] [K3] Fix fp8 KV cache decode on the AITER MLA backend (#51011)

fanxingran committed 3 months ago

Fix uniform_random routing simulation to sample without replacement (#43680)

Elvir Crnčević committed 3 months ago

[Perf][Hybrid] 3D-grid tiling of the state-copy Triton kernels (#49436)

Francesco Fusco committed 3 months ago

[XPU][Test] Pin block size in test_multi_connector (#51213)

liuzhenwei committed 3 months ago

[Misc] Enable test_fused_moe_wn16 on XPU (#51672)

pmanczak committed 3 months ago

[CI] Stabilize DP supervisor lifecycle tests (#51557)

Taneem Ibrahim committed 3 months ago

[CI] Bump CUTLASS DSL to 4.6.2 (#51566)

Lucas Wilkinson committed 3 months ago

[2/N] Harden Transformers modelling backend multi-modal path (#51657)

Harry Mellor committed 3 months ago

[ROCm][Bugfix] Use TCP store when AITER custom all-reduce is enabled (#51635)

vllmellm committed 3 months ago