[XPU] [Fusion passes] Disable fuse_rope_kvcache_cat_mla & qk_norm_rope_ fusion on XPU (#47962)

Chaojun Zhang committed 4 months ago

[Test] Skip DeepEP MoE layer tests without P2P access (#47946)

Tyler Michael Smith committed 4 months ago

Remove unused _get_kv_cache_config_deepseek_v4 alias (#47969)

Nicolò Lucchesi committed 4 months ago

[Docs] `kv_sharing_fast_prefill` correction (#47044)

Nicolò Lucchesi committed 4 months ago

[Model][HunyuanVL] Use native transformers processor and adapt to transformers 5.13 (#47872)

manayang committed 4 months ago

fix: hash speculative draft model config (#47911)

Ace Eldeib committed 4 months ago

[ROCm] fixed aiter master flag and expert parallelism compatibility on minimax-m3-mxfp8 (#47158)

Hongxia Yang committed 4 months ago

[Bugfix] Avoid leaking Pydantic repr in tool_choice error message (#47028)

Muhammad Fawaz committed 4 months ago

[Bugfix] Register VLLM_BUILD_* and VLLM_IMAGE_TAG provenance env vars (#45313)

Nicklas Frahm committed 4 months ago

[Bugfix][Frontend] Fix http_requests_total metric recording some 4xx errors as 5xx (#44303)

Zach Zhu committed 4 months ago

[Perf] Minimax M3 - Support cross-layer allreduce-norm fusion (#47631)

Wei Zhao committed 4 months ago

[Bugfix] DSV4 TP16 garbage output (#47493)

Jeff (Junze) Ma committed 4 months ago

[ROCm][Perf] MXFP8 dense-linear + grouped-MoE GEMM optimizations for MiniMax-M3 (#46117)

Ethan Yang committed 4 months ago

[Minimax-M3] Using tok_sparse_select from MSA instead of triton kernels (#47502)

Yongye Zhu committed 4 months ago

[Bugfix][Pooling] Align CrossEncoder token type ids after truncation (#47772)

Ting SUN committed 4 months ago

[Core][Engine] only materialize tokens when thinking budget is in req (#47053)

Walter Beller-Morales committed 4 months ago

[Core] Pass request context to CPU offload cache policy touch (#47744)

Posedge_Lin committed 4 months ago

[Bugfix] Avoid blocking model launching when no system ffmpeg available for TorchCodec (#47888)

Isotr0py committed 4 months ago

[Frontend] Add endpoint plugins framework (#47454)

Martin Hickey committed 4 months ago

[CPU][Bugfix] Fix flaky ShortConv prefill test on ARM (uninitialized weights) (#47848)

Rahul Vishwakarma committed 4 months ago