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 14, 05:30 UTC
main
[Bugfix][P/D] Reduce num_threads used by nixl ucx backend (#27196)
David Whyte-Gray committed about 1 year ago
[Feature] Batch Invariant for R1 TP 8 on Blackwell (#27229)
Wentao Ye committed about 1 year ago
[ROCm] Update Triton, Torch, and AITER branches for ROCm base Dockerfile (#27206)
Micah Williamson committed about 1 year ago
Add @pavanimajety to .github/codeowners for Flashinfer, ModelOpt related code (#27213)
Pavani Majety committed about 1 year ago
[ROCM] Enable CompressedTensorsWNA16 (#27187)
JartX committed about 1 year ago
[CI] Install pre-release version of `apache-tvm-ffi` for `flashinfer` (#27262)
Harry Mellor committed about 1 year ago
[Chore] Separate out NCCL utilities from vllm.utils (#27197)
dongbo910220 committed about 1 year ago
[Deepseek v3.2] Optimize top_k_per_row (#26763)
Daniel Cámpora committed about 1 year ago
[MM][Core] Decouple ViT backend from LM backend (#27061)
Roger Wang committed about 1 year ago
[Nixl] Minor refactor to handshake related metadata (#26410)
Nicolò Lucchesi committed about 1 year ago
[Fix][Spec Decode] Fix llama4 draft loading with different quantization (#27136)
Zebing Lin committed about 1 year ago
[Bugfix] Fix broken MTP weight loading for FP8 KV Scales (#27227)
Benjamin Chislett committed about 1 year ago
[Bugfix] Fix gpt-oss w4a8 DP/EP on B200 (#26729)
Varun Sundar Rabindranath committed about 1 year ago
[ModelOpt] Load w13/w2_input_scale for all experts, nvfp4 (#26135)
Shu Wang committed about 1 year ago
[BugFix] GPT-OSS Attention DP + MoE TP weight loading issue (#24032)
Po-Han Huang (NVIDIA) committed about 1 year ago
[Feature][Kernel]FusedMoE LoRA (#21229)
Chen Wu committed about 1 year ago
[Frontend] Enforce tokenize=False when applying chat template (#27205)
Russell Bryant committed about 1 year ago
create is_in_the_same_node on cpu (#26832)
Lunwen He committed about 1 year ago
[cpu] Dispatch un-quantized linear to oneDNN/ACL by default for AArch64 (#27183)
Fadi Arafeh committed about 1 year ago
[V0 Deprecation] Remove V0 metrics code (#27215)
Nick Hill committed about 1 year ago