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 4, 06:15 UTC
main
[Model] Use AutoWeightsLoader for Plamo2 (#41699)
BitToby committed 6 months ago
[Benchmark] Add --trust-remote-code flag to multi-turn benchmark (#41661)
Dao007forever committed 6 months ago
[Model] Use AutoWeightsLoader for CohereMoe (#41690)
BitToby committed 6 months ago
[Model] Fix Gemma4 MoE activation mismatch (#41574)
Luciano Martins committed 6 months ago
[BugFix] Preserve max_seq_len in ubatch metadata during CUDA graph capture (#40961)
czhu-cohere committed 6 months ago
[Ray] Enable RayExecutorV2 by default (#41421)
Jeffrey Wang committed 6 months ago
[Hardware][Power]Add Power VSX Attention Backend and fix l2 Cache Crash (#40451)
Akash kaothalkar committed 6 months ago
[ROCm][Quantization][2/N] Refactor quark_moe w4a8 w/ oracle (#39136)
Bowen Bao committed 6 months ago
[Perf][2/n] Eliminate GPU<->CPU syncs in pooling code (#41433)
Nick Hill committed 6 months ago
[XPU] enable is_act_and_mul for xpu (#37481)
Chendi.Xue committed 6 months ago
[Model Runner V2] Rebuild attn metadata between draft decode steps (#41162)
Giancarlo Delfin committed 6 months ago
[Feature] TurboQuant: support hybrid models and uniform quantization (#39931)
JartX committed 6 months ago
[Bug] Fix status update address for non-MOE model within external dp mode (#40839)
Wentao Ye committed 6 months ago
[ROCm][CI] Fix MLA prefill scale for DeepSeek GSM8K (#41569)
Andreas Karatzas committed 6 months ago
[Docs] Add non-causal support to attention backend docs (#41643)
Matthew Bonanni committed 6 months ago
[NVFP4][fix] Fix `layer.weight` -> `w13` typo in NVFP4 MOE emulation kernel preparation (#41630)
fxmarty-amd committed 6 months ago
Linzhang Li committed 6 months ago
[bugfix] Fix prompt logprobs on request eviction during chunked prefill (#41411)
Joachim Studnia committed 6 months ago
Fix Nano Nemotron text-only weight loading (#41205)
Baekpica committed 6 months ago
[ROCm] ROCm7.2.2 + profiler fix + AITER 0.1.12.post2 (#41386)
Gregory Shtrasberg committed 6 months ago