A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 2, 02:15 UTC
main
fix: resolve CUTLASS fmin compatibility for DeepSeek-V4 init (#44236)
Willow Lopez committed 5 months ago
Nit Changes in Tiered KV Offload (#44293)
Rotem Shavitt committed 5 months ago
[CI] Add missing vllm/parser/ CI trigger and fix test_parse.py (#44352)
Flora Feng committed 5 months ago
[CI] Reject out-of-vocabulary before they reach the GPU logprob path (#44042)
Andreas Karatzas committed 5 months ago
[ROCm][CI] Skip fp8 reload tests on gfx90a (MI250) (#44369)
JartX committed 5 months ago
[ROCm][CI] Fix stale wvSplitK GEMM fallback test for N=5 (#44368)
JartX committed 5 months ago
[Platform] Add is_cumem_allocator_available (#43838)
wangxiyuan committed 5 months ago
[docker] Stop using extra-index-url for flashinfer-jit-cache (#44366)
Kevin H. Luu committed 5 months ago
[Bugfix] Fix Deepseek v4 non-mega-moe model init error (#44356)
Wei Zhao committed 5 months ago
[Perf] Apply single-pass min_larger finding and binary search in Triton Top-p path. (#42191)
Jongseok Park committed 5 months ago
[DSV4] Minor cleanup for DeepseekV4MegaMoEExperts (#44367)
Woosuk Kwon committed 5 months ago
[Misc] Remove dead VLLM_RPC_TIMEOUT env var and fix profiling doc that references it (#44128)
Daoyuan Li committed 5 months ago
[MoE/b12x] Accept W4A16 (kNvfp4Static, None) in FlashInferB12xExperts supports check (#43332)
Junhao Shen committed 5 months ago
[CI/Build] Bump flashinfer to v0.6.12 (#44036)
Vadim Gimpelson committed 5 months ago
Fix sparse NCCL weight transfer test construction (#44345)
Siddharth Bedekar committed 5 months ago
[Kernel][MoE] Add GELU_TANH to CPU, CUTLASS, and WNA16 MoE backends (#42027)
SeongJun Lee committed 5 months ago
[ModelRunnerV2] Avoid pipeline parallel bubbles (#42187)
Nick Hill committed 5 months ago
[Misc] Remove stray empty file (#44350)
Matthew Bonanni committed 5 months ago
[Bugfix] Cache the EAGLE/MTP lookahead block in the SWA prefix-cache mask (#44082)
Yifan Qiao committed 5 months ago
[MRV2] Remove assignment of graph_pool in cudagraph_utils (#44338)
Woosuk Kwon committed 5 months ago