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 8, 16:30 UTC
main
[Release 2.10] Update to Torch 2.10 - final release (#30525)
Andrey Talman committed 9 months ago
Add support for ModelOpt MXFP8 dense models (#33786)
danisereb committed 9 months ago
glm 4.6 fused tuned inference config for B200 (#32958)
navmarri14 committed 9 months ago
[torch.compile] Add an option to force-enable the MOE cold start optimization (#33735)
Richard Zou committed 9 months ago
[BugFix] Change support no act and mul for marlin (#34088)
TomerBN-Nvidia committed 9 months ago
[Revert] Fix performance regression for GLM-4.7-GPTQ decode and MTP acceptance rate (#33771)
aabbccddwasd committed 9 months ago
Add embedding input functionality for disabled modalities [remake] (#32493)
Reagan Lee committed 9 months ago
[ROCm] [CI] Reduce Resource of two test groups (#34059)
TJian committed 9 months ago
[Perf] Simplify DeepseekV32 tokenizer, ensure fast detokenization used (#33855)
Nick Hill committed 9 months ago
[ROCm][Bugfix] fix act_quant_fusion module import error (#34069)
Andreas Karatzas committed 9 months ago
[CI/Build] Skip GCS test (#34057)
Cyrus Leung committed 9 months ago
[Doc] Fix run_batch docs (#34056)
Cyrus Leung committed 9 months ago
Perf tuning and expansion of cases covered for wvSplitKrc (#33493)
Hashem Hashemi committed 9 months ago
Make directory exist ok for ray spinning up multiple replicas on a single instance (#33604)
Jiang Wu committed 9 months ago
Update DeepGEMM version pin in Dockerfile to match #32479 (#33935)
zifeitong committed 9 months ago
move checks out of `unified_kv_cache_update` custom op (#33943)
Rohan Potdar committed 9 months ago
[PluggableLayer][3/N] Apply PluggableLayer to mamba layers. (#33660)
whx committed 9 months ago
[Model] Enable Step3p5ForCausalLM testing (#33755)
Jee Jee Li committed 9 months ago
[Frontend]Add support for transcriptions and translations to run_batch (#33934)
Pooya Davoodi committed 9 months ago
Enable Eagle3 speculative decoding for Mistral3ForConditionalGeneration to support eagle3 (#33939)
TundeAtSN committed 9 months ago