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, 06:00 UTC
main
[Feature]:Allow for Granite MoE Hybrid models with _only_ shared experts. (#19652)
Shawn Tan committed over 1 year ago
[DOC] Add reasoning capability to vLLM streamlit code (#19557)
Navanit Dubey committed over 1 year ago
[BugFix] Don't catch BaseException when dumping execute_model errors (#19626)
Nick Hill committed over 1 year ago
[Kernel] GGUF MMVQ kernel for multiple input vectors (#18754)
Szymon Ożóg committed over 1 year ago
[Docs] Move multiproc doc to v1 dir (#19651)
Russell Bryant committed over 1 year ago
[CI] Add mteb testing for rerank models (#19344)
wang.yuqi committed over 1 year ago
Ning Xie committed over 1 year ago
[TPU] support attention head dim smaller than 128 (#19620)
Chengji Yao committed over 1 year ago
Ye (Charlotte) Qi committed over 1 year ago
[Misc][Frontend] passthrough `bad_words` (#19564)
Francesco Bertolotti committed over 1 year ago
quanliu committed over 1 year ago
[MISC] Remove unused variableds in C++ (#19609)
Lu Fang committed over 1 year ago
[Misc] Remove duplicate multiproc method setting for CPU platform (#19649)
Isotr0py committed over 1 year ago
[CI/Build] Fix torch nightly CI dependencies part 2 (#19589)
Richard Zou committed over 1 year ago
Enable prefix caching with full cuda graphs (#19617)
Woosuk Kwon committed over 1 year ago
[Benchmark] Refactor benchmark script for fp8 & int8 (#19627)
Wentao Ye committed over 1 year ago
[Kernel] Raise verbose error and consolidate `num_heads/num_kv_heads` divisibility check (#19339)
22quinn committed over 1 year ago
[Bugfix][2/n] Fix speculative decoding CI - Fix test_ngram_e2e_greedy_correctness (#19644)
Lu Fang committed over 1 year ago
[Perf] Further tunings for SM100 FP8 CUTLASS kernel (#19566)
Ilya Markov committed over 1 year ago
[Fix] Convert kv_transfer_config from dict to KVTransferConfig (#19262)
maobaolong committed over 1 year ago