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 3, 09:15 UTC
main
[BugFix] support PP for Cohere vision model (#42819)
czhu-cohere committed 6 months ago
[ROCm][Quantization][3/N] Refactor quark_moe w4a4 w/ oracle (#41436)
Bowen Bao committed 6 months ago
[Bugfix] fix swiglu limit issue for humming backend + deepseek v4 (#42541)
Jinzhen Lin committed 6 months ago
[Bugfix] Fix DSV4 MTP after ROCm mHC integration (#42930)
Mohammad Miadh Angkad committed 6 months ago
[Perf] Re-enable flashinfer autotune by default and cleanup (#42857)
Wei Zhao committed 6 months ago
Fix `--convert` passed without `--runner` on causal models (#42935)
Harry Mellor committed 6 months ago
[Model Runner V2] Fix prompt logprobs calculation `Sizes of tensors must match` error (#42778)
Wentao Ye committed 6 months ago
[Bugfix] mamba: run single-token extends as decodes (#42430)
Netanel Haber committed 6 months ago
[Model] Add Apertus Tool Parser (#41154)
Blanc Swan committed 6 months ago
Refactor AWQ Marlin MoE onto modular WNA16 oracle (#42483)
Siddharth Bedekar committed 6 months ago
[Model Runner v2] Support update_config (#42783)
Michael Goin committed 6 months ago
Revert "[torch.compile] Add patch for fullgraph compilation" (#42686) (#42913)
vllm-agent committed 6 months ago
[KV Connector][Offloading] Flush all pending jobs on last step (#42611)
liranschour committed 6 months ago
xiangdong committed 6 months ago
[MRv2] Default to MRv1 when a connector is present (#42955)
Nicolò Lucchesi committed 6 months ago
[Bugfix][Hybrid][NemotronH] Fix mamba_cache_mode=all + speculative decoding crash (#41233)
roikoren755 committed 6 months ago
[Bugfix][KV Offload] count appended GPU blocks in store group_sizes (#42945)
Kfir Toledo committed 6 months ago
fix: remove unused norm for dpskv4 (#41710)
inisis committed 6 months ago
[CPU] Add MXFP4 W4A16 MoE support (#41922)
Yuwen Zhou committed 6 months ago
[CPU Backend] Improve cpu thread utilization (#42666)
Tianmu Li committed 6 months ago