FareedKhan-dev/kimi-k3-in-c
kimi-k3-in-c is a portable C99 inference engine designed to run the Kimi K3 2.78-trillion-parameter LLM on a single CPU with minimal RAM, focusing on extreme memory efficiency without GPUs or exter...
Awesome Infra for AI › Inference Optimization
ShannonAI/service-streamer
Service Streamer is a middleware designed to enhance the performance of deep learning model inference in web service environments. It addresses the common challenge of low GPU utilization when handling discrete user requests by collecting these requests into mini-batches. By processing multiple requests concurrently, it allows deep learning models to leverage the parallel computing capabilities of GPUs more effectively. This results in significantly faster processing speeds, lower latency for online inference, and improved overall system throughput. The tool is highly extensible, supporting multi-GPU scenarios to manage large volumes of requests, and is compatible with various web and deep learning frameworks. It can be integrated with minimal code changes, offering a substantial speedup (up to 10x or more) compared to conventional serving methods. Service Streamer provides `ThreadedStreamer` for single-process, multi-threaded batching and `Streamer` for distributed multi-GPU worker processes, enabling scalable and efficient deployment of deep learning models.
https://github.com/ShannonAI/service-streamer
Also in Inference Optimization
zilliztech/GPTCache
GPTCache is a library that creates a semantic cache for LLM queries, significantly reducing API costs and improving response times by storing and reusing previous LLM responses.
vllm-project/vllm-ascend
vLLM Ascend is a community-maintained hardware plugin for seamlessly running vLLM and various large language models on Ascend NPUs, optimizing inference performance.
youssofal/MTPLX
MTPLX is a macOS-native inference engine for Apple Silicon that leverages multi-token prediction (MTP) for accelerated local LLM serving, offering significant speed improvements.
open-compress/claw-compactor
Claw Compactor is an LLM token compression engine that uses a 14-stage Fusion Pipeline for content-aware, reversible compression to reduce LLM inference costs and optimize context windows.
siliconflow/onediff
OneDiff is an acceleration library for diffusion models, providing out-of-the-box performance optimizations for popular UIs and libraries like Hugging Face Diffusers and ComfyUI.
ovg-project/kvcached
kvcached is a KV cache library that introduces virtual memory abstraction for LLM serving on shared GPUs, enabling elastic and demand-driven KV cache allocation for improved GPU utilization under d...
nobodywho-ooo/nobodywho
NobodyWho is an efficient, on-device inference engine that enables local execution of LLMs and SLMs across various platforms without requiring API keys.