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 5, 07:15 UTC
main
[Bugfix] Fix FlashInfer crash with kv_cache_dtype_skip_layers (#39002)
yzong-rh committed 7 months ago
[CI] Add MultiConnector (Nixl+Offloading) e2e edge case tests (#39343)
zhanqiuhu committed 7 months ago
perf(moe): add tuned fused_moe config for RTX PRO 6000 Blackwell Server Edition (#39183)
Manu committed 7 months ago
feat: add logit_scale to PoolerConfig for affine score calibration (#39435)
Jesus Federico committed 7 months ago
[ROCm] [AITER] Revert AITER version to v0.1.10.post3 (#39509)
TJian committed 7 months ago
[Kernel] Porting the TRTLLM minimax_allreduce_rms kernels (#37045)
Jee Jee Li committed 7 months ago
[Model Runner V2] support auto resolve cudagraph mode/sizes based on attn backend (#32936)
zhrrr committed 7 months ago
[New Model]: jinaai/jina-reranker-v3 (#38800)
wang.yuqi committed 7 months ago
[Model] Implement LoRA support for Qwen3ASRForConditionalGeneration (#37247)
Peter Nguyen committed 7 months ago
[CI] Add Nixl+OffloadingConnector e2e integration tests (#39200)
Nicolò Lucchesi committed 7 months ago
[ROCm] Add RDNA 3.5/4 device IDs (gfx1150, gfx1151, gfx1201) (#38455)
Ravitez Dondeti committed 7 months ago
[Performance] Remove unnecessary zero-fill of MLA decode output tensor in Aiter backend (#37539)
xaguilar-amd committed 7 months ago
[Kernel][Hardware][AMD] Add TritonW4A16LinearKernel for ROCm (#37352)
jatseng-ai committed 7 months ago
[model] support FireRedLID (#39290)
PatchyTIS committed 7 months ago
[ZenCPU] Make PT Backport Patch Accessible to vLLM (#38205)
Lalithnarayan C committed 7 months ago
[PluggableLayer][3/N] Apply PluggableLayer to llm_head and vocab embedding layer (#33465)
Hexiang Wang committed 7 months ago
[Bugfix] Fix V1 dummy run writing NaN to KV cache null block (#39444)
Elvir Crnčević committed 7 months ago
Nemotron Nano VL: Streamline pixel shuffle (#37580)
milesial committed 7 months ago
[compile] Allow strings in custom ops without regressing compilation times (#38123)
Richard Zou committed 7 months ago
[GGUF] Support non-standard quant types with prefix (e.g. UD-IQ1_S) (#39471)
Injae Ryou committed 7 months ago