[CI Bug] Fix `Chunked prefill is required for mamba cache mode 'align'.` (#51177)

Wentao Ye committed 3 months ago

Revert [Misc] Avoid importing `nixl_ep` on every `vllm serve` config (#50879) (#51176)

fxmarty-amd committed 3 months ago

[CI bug] Fix `Each KV cache group's real block_size must be divisible by has h_block_size` (#51180)

Wentao Ye committed 3 months ago

[Bugfix] Fail fast with a clear error when CPU offload region exceeds available space (#50358)

AlexHuang committed 3 months ago

[Build] Remove Ubuntu build-stage option from the CUDA dockerfile (#51060)

Michael Goin committed 3 months ago

[KV Offload] Support out-of-tree secondary tier managers via `module_path` (#51007)

Ronen Schaffer committed 3 months ago

[Model][Frontend] Add Ling 3.0 Flash BF16, MTP, and parser support (#51045)

zexplorerhj committed 3 months ago

[Kernel][Model] Optimize FA4 mm_prefix range lookup (#50294)

Li Juqi committed 3 months ago

[KV Offloading] Support partial-tail prefix reuse with fine-grained prefix matching (#50507)

Chauncey committed 3 months ago

[Bugfix][Model] Fix MiniMax-M3 NVFP4 inference correctness (#48929)

Gabriel Wu committed 3 months ago

[Bugfix] Fix MLA kv_b_proj activation dtype with Marlin FP8 (#38771)

Jacob Zhang committed 3 months ago

Resolve revision to commit_hash once per model load, via huggingface_hub's `resolve_revision` (#49990)

Lucain committed 3 months ago

[Bugfix] Enable chunked prefill for qwen3.5-0.8B ppl test (#51153)

music-dino committed 3 months ago

[BugFix][K3] Skip moe_intermediate padding when EP is enabled (#51131)

Ziming Huang committed 3 months ago

[CI] Fix CI authorization notification fallback (#51095)

Kevin H. Luu committed 3 months ago

[Rust Frontend] Deduplicate request preprocessing for `/tokenize` (#50448)

Sage committed 3 months ago

[KV Connector][Mooncake] Add store group semantics (#44956)

Schatten committed 3 months ago

[ROCm][Bugfix] Kimi-K3 Fix KDA NaN on mixed batches and racy autotune config (#50649)

kliuae committed 3 months ago

[Perf][KV Offload] Avoid quadratic ARC batch eviction (#50992)

MINJUN GIL committed 3 months ago

[CI] Run control-plane workflows on vLLM runners (#51127)

Kevin H. Luu committed 3 months ago