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 15, 22:15 UTC
main
[Doc] Dockerfile instructions for optional dependencies and dev transformers (#13699)
Cyrus Leung committed over 1 year ago
[ROCm] Apply FP8 weights padding to values not divisible by 512 bytes on ROCm (#13231)
Gregory Shtrasberg committed over 1 year ago
Sage Moore committed over 1 year ago
[NVIDIA] Support nvfp4 cutlass gemm (#13571)
Kaixi Hou committed over 1 year ago
Support SSL Key Rotation in HTTP Server (#13495)
Keyun Tong committed over 1 year ago
youkaichao committed over 1 year ago
[core] set up data parallel communication (#13591)
youkaichao committed over 1 year ago
[CI/Build] Fix pre-commit errors (#13696)
Cyrus Leung committed over 1 year ago
[Misc] Reduce LoRA-related static variable (#13166)
Jee Jee Li committed over 1 year ago
[Metrics] Add `--show-hidden-metrics-for-version` CLI arg (#13295)
Mark McLoughlin committed over 1 year ago
[V1][Metrics] Support `vllm:cache_config_info` (#13299)
Mark McLoughlin committed over 1 year ago
Correction to TP logic for Mamba Mixer 2 when Num Groups not divisible by TP Size (#13660)
Yu Chin Fabian Lim committed over 1 year ago
[Bugfix] V1 Memory Profiling: V0 Sampler Integration without Rejection Sampler (#13594)
Jennifer Zhao committed over 1 year ago
[v1] Support allowed_token_ids in v1 Sampler (#13210)
Lu Fang committed over 1 year ago
Robin committed over 1 year ago
[Misc] Bump compressed-tensors (#13619)
Dipika Sikka committed over 1 year ago
[Bugfix][Model] OLMo 2: split qkv correctly for GQA and MQA (#13687)
Shane A committed over 1 year ago
[ROCM] fix native attention function call (#13650)
Gordon Wong committed over 1 year ago
[Misc] Capture and log the time of loading weights (#13666)
Jun Duan committed over 1 year ago
[Bugfix][API Server] Fix invalid usage of 'ge' and 'le' in port valid… (#13672)
Robin committed over 1 year ago