[Test] Add Unit Test for Batched DeepGEMM (#21559)

Wentao Ye committed over 1 year ago

[BugFix] Improve internal DP load balancing (#21617)

Nick Hill committed over 1 year ago

[Speculators][Speculative Decoding] Add Qwen Eagle3 Support (#21835)

Dipika Sikka committed over 1 year ago

[Misc] DeepGemmExperts : Avoid JIT generation in the hot-path (#21955)

Varun Sundar Rabindranath committed over 1 year ago

[Misc] `VLLM_TARGET_DEVICE.lower()` (#22101)

Nicolò Lucchesi committed over 1 year ago

[Model] Qwen2.5 VL SiLU-and-Mul (#22066)

vllmellm committed over 1 year ago

[V1][CUDA] Full cudagraph support for FlashInfer (#21367)

fhl2000 committed over 1 year ago

feat: Add Support GPTQ Quantization MOE on ROCM vllm serve (#21733)

JartX committed over 1 year ago

[Perf] Optimize `reshape_and_cache_flash` CUDA Kernel (#22036)

Wentao Ye committed over 1 year ago

[CI] Initial tests for SM100 Blackwell runner (#21877)

Michael Goin committed over 1 year ago

[BugFix] Harden distributed DP startup (#21538)

Nick Hill committed over 1 year ago

Add lora test for tp>1 case for TPU. (#21970)

XiongfeiWei committed over 1 year ago

Introduce RayPPCommunicator for ray-based PP (#21660)

Rui Qiao committed over 1 year ago

[compile][startup] Disable C++ compilation of symbolic shapes (#20836)

Animesh Jain committed over 1 year ago

Fix pre-commit failure for SECURTIY.md (#22102)

Michael Goin committed over 1 year ago

[Misc] Minor enhancement of benchmark_moe (#22068)

Jee Jee Li committed over 1 year ago

Enable headless models for pooling in the Transformers backend (#21767)

Harry Mellor committed over 1 year ago

[Bugfix] [Performance] DeepEPHighThroughput + DeepSeek : Quant before Dispatch (#21837)

Varun Sundar Rabindranath committed over 1 year ago

security policy: take 1 (#21119)

Huzaifa Sidhpurwala committed over 1 year ago

[Bugfix] fix when skip tokenizer init (#21922)

rongfu.leng committed over 1 year ago