[Core][KV Connector] Resolve connector block tables for every scheduled request (#54853)

Zhewen Li committed about 1 month ago

[Docs][EC Connector] CPU EC Connector usage Docs (#54522)

omerpaz95 committed about 1 month ago

[ROCm][Perf][Bugfix] Multi-stream perf improvements; rocprofiler fixes (#55099)

Matt committed about 1 month ago

[Bugfix] Parse DSML tool calls when the model omits the tool_calls wrapper (#55954)

Flora Feng committed about 1 month ago

[Kernel] Add NVFP4 support to the torch linear backend (#53319)

Michael Lazos committed about 1 month ago

[Core] Enhance cpu<->gpu sync checking to include paged async copies (#53491)

Nick Hill committed about 1 month ago

[Kernel] Enable optimized FlashInfer add-RMSNorm NVFP4 fusion (#51925)

Qidong Su committed about 1 month ago

[Bugfix][Spec Decode] Preserve target parallel config (DCP) for DSpark (#55472)

Jing Wang committed about 1 month ago

[Bugfix] Fix OpenPangu multimodal embedding merge (#55941)

Taneem Ibrahim committed about 1 month ago

[Bugfix][Spec Decode] Honour the draft's moe_backend on Model Runner V2 (#54788)

Stefano Castagnetta committed about 1 month ago

[Transformers] Generalize merged-column linear fusion (#55301)

Jim Ban committed about 1 month ago

Fix block FP8 MTP in ModelOpt mixed checkpoints (#55513)

sychen52 committed about 1 month ago

[ROCm][DI][CI] Enable WideEP Intranode tests (#53195)

Chaitanya Sri Krishna Lolla committed about 1 month ago

[CI] Fix ARM64 test dependency builds with GCC 15 (#55877)

Kevin H. Luu committed about 1 month ago

[Misc] Bump `openai` to `>=2.25.0` to support namespace tools types (#49104)

cjackal committed about 1 month ago

[Kimi Bug] Fix kda ima `Triton Error [CUDA]: an illegal memory access was encountered` (#55924)

Wentao Ye committed about 1 month ago

[Perf] Eliminate full-history reasoning scans for structured outputs (#55223)

Flora Feng committed about 1 month ago

[Perf] Eliminate full-history reasoning scans for structured outputs (#55223)

Flora Feng committed about 1 month ago

[CI][ROCm] Increase timeouts for AMD MI300 jobs (#55919)

djramic committed about 1 month ago

[Attention] Require explicit DCP support from attention implementations (#55780)

Andreas Karatzas committed about 1 month ago