ModelEngine-Group / unified-cache-managementLinks
Persist and reuse KV Cache to speedup your LLM.
☆233Updated this week
Alternatives and similar repositories for unified-cache-management
Users that are interested in unified-cache-management are comparing it to the libraries listed below
Sorting:
- Efficient and easy multi-instance LLM serving☆520Updated 4 months ago
- GLake: optimizing GPU memory management and IO transmission.☆494Updated 9 months ago
- SGLang kernel library for NPU☆91Updated this week
- KV cache store for distributed LLM inference☆384Updated last month
- Disaggregated serving system for Large Language Models (LLMs).☆761Updated 9 months ago
- Offline optimization of your disaggregated Dynamo graph☆137Updated this week
- ☆518Updated this week
- NVIDIA Inference Xfer Library (NIXL)☆788Updated this week
- ☆73Updated last year
- DeepSeek-V3/R1 inference performance simulator☆175Updated 9 months ago
- A workload for deploying LLM inference services on Kubernetes☆153Updated this week
- This repository organizes materials, recordings, and schedules related to AI-infra learning meetings.☆288Updated this week
- ☆77Updated last year
- RTP-LLM: Alibaba's high-performance LLM inference engine for diverse applications.☆995Updated this week
- vLLM Kunlun (vllm-kunlun) is a community-maintained hardware plugin designed to seamlessly run vLLM on the Kunlun XPU.☆212Updated this week
- Virtualized Elastic KV Cache for Dynamic GPU Sharing and Beyond☆735Updated last month
- Hooked CUDA-related dynamic libraries by using automated code generation tools.☆172Updated 2 years ago
- A prefill & decode disaggregated LLM serving framework with shared GPU memory and fine-grained compute isolation.☆123Updated 2 weeks ago
- Materials for learning SGLang☆714Updated this week
- Open Model Engine (OME) — Kubernetes operator for LLM serving, GPU scheduling, and model lifecycle management. Works with SGLang, vLLM, T…☆355Updated this week
- Train speculative decoding models effortlessly and port them smoothly to SGLang serving.☆615Updated this week
- Research prototype of PRISM — a cost-efficient multi-LLM serving system with flexible time- and space-based GPU sharing.☆51Updated 4 months ago
- Summary of the Specs of Commonly Used GPUs for Training and Inference of LLM☆71Updated 4 months ago
- Omni_Infer is a suite of inference accelerators designed for the Ascend NPU platform, offering native support and an expanding feature se…☆96Updated this week
- Stateful LLM Serving☆93Updated 10 months ago
- High performance Transformer implementation in C++.☆148Updated 11 months ago
- FlagScale is a large model toolkit based on open-sourced projects.☆463Updated this week
- ☆337Updated this week
- Fast OS-level support for GPU checkpoint and restore☆267Updated 3 months ago
- Analyze the inference of Large Language Models (LLMs). Analyze aspects like computation, storage, transmission, and hardware roofline mod…☆602Updated last year