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 10, 01:30 UTC
main
[CI] Fix LM Eval Large Models (H100) (#32423)
Matthew Bonanni committed 10 months ago
Add thread_n=64 support to Marlin MoE (#32360)
Michael Goin committed 10 months ago
[Feat] Support non-gated MoE with Marlin, NVFP4 CUTLASS, FP8, INT8, compressed-tensors (#32257)
TomerBN-Nvidia committed 10 months ago
[Refactor] Remove unused file (#32422)
Wentao Ye committed 10 months ago
[MoE Refactor][17/N] Apply Refactor to Bf16 (#31827)
Yongye Zhu committed 10 months ago
[ROCM] DSfp4 mla projection gemms weight dynamic quantization (#32238)
Aleksandr Malyshev committed 10 months ago
[BugFix] Python file source reading can fail on UnicodeDecodeError (#32416)
Richard Zou committed 10 months ago
[ROCm] [CI] [Release] Rocm wheel pipeline with sccache (#32264)
TJian committed 10 months ago
[UX] Use kv_offloading_backend=native by default (#32421)
Michael Goin committed 10 months ago
Lucas Wilkinson committed 10 months ago
[Attention][AMD] Make flash-attn optional (#30361)
Matthias Gehre committed 10 months ago
fixing podman build issue (#32131)
smit kadvani committed 10 months ago
[Feature] Support async scheduling + PP (#32359)
Wentao Ye committed 10 months ago
[Model Runner V2] Support FlashInfer backend & Fix CUDA Graph bug [1/2] (#32348)
Woosuk Kwon committed 10 months ago
[ROCm][Bugfix] Disable hip sampler to fix deepseek's accuracy issue on ROCm (#32413)
Pleaplusone committed 10 months ago
Pleaplusone committed 10 months ago
[Quant] Support MXFP4 W4A16 for compressed-tensors MoE models (#32285)
Dipika Sikka committed 10 months ago
Matthew Bonanni committed 10 months ago
[ROCm] Improve error handling while loading quantized model on gfx120… (#31715)
brian033 committed 10 months ago
[3/N] Group together media-related code (#32406)
Cyrus Leung committed 10 months ago