Repo AI đang lênCập nhật hằng ngày từ GitHub

Tool for onnx->keras or onnx->tflite. Hope this tool can help you.

Port of MiniGPT4 in C++ (4bit, 5bit, 6bit, 8bit, 16bit CPU inference with GGML)

WitNote is a local-first AI writing companion for macOS and Windows.

Inference-native Tokenmaxxing Agent Harness for Loop Engineering

Fast Inference Solutions for BLOOM

sparkrun - launch, manage, and stop LLM inference workloads on NVIDIA DGX Spark systems. Live support: https://discord.com/invite/GH5kRgv6ZD

maxxing the ability to run different LLMs on the smallest devices possible

An Open-Source Package for Deep Learning to Hash (DeepHash)

Local AI chat for iPhone, iPad and Mac. Connect Ollama, LM Studio or any OpenAI-compatible server, or run MLX models on-device. Includes tools and web search.

It won't fix your bug. It'll fix you. The debugger that hands the question back instead of the answer.

Awesome machine learning model compression research papers, quantization, tools, and learning material.

Fast implementation of BERT inference directly on NVIDIA (CUDA, CUBLAS) and Intel MKL

基于 Java Web 项目的 SpringBoot 框架初始化模板，适配最新版本Spring AI，该模板整合了常用的框架（Mybatis-Plus、ShardingSphere、Redis、RabbitMQ、Elasticsearch、SaToken、OSS、Caffeine以及MongoDB等），广泛支持JDK11和JDK17，部分版本兼容JDK8，同时该模板适用于前后端分离项目启动开发，保证大家在此基础上能够快速开发自己的项目，同时也适合入门学习，本项目会由作者持续更新。

Generate and brainstorm ideas while creating your notes using Large Language Models (LLMs) from Ollama, LM Studio, Anthropic, Google Gemini, Mistral AI, OpenAI, and more for Obsidian.

Run your deep learning workloads on Kubernetes more easily and efficiently.

Local inference for diffusion language models (LLaDA, Dream), written in Mojo

A project that optimizes OWL-ViT for real-time inference with NVIDIA TensorRT.

Triton Model Analyzer is a CLI tool to help with better understanding of the compute and memory requirements of the Triton Inference Server models.

[ICLR 2026]QeRL enables RL for 32B LLMs on a single H100 GPU.

KVarN is a native vLLM KV-cache quantization backend for your agents: 3-5x more context, throughput above FP16, and FP16-level accuracy. Calibration-free, one flag.

Deep learned, NVIDIA-accelerated 3D object pose estimation

Xolotl: run DeepSeek-V4.1-Flash, all 552B, as a local coding agent on one gaming PC (RTX 4090 + 110 GB RAM): ~33 tok/s on agent tasks, ~41 in fast mode, 512K context. Plus GLM-5.3-Flash, Qwen3.8 and LFM3-Omni. MoE experts tiered across VRAM, pinned RAM and NVMe, with Xolotl Direct NVMe->VRAM. OpenAI + Anthropic API.

Confidential AI deployment with secure enclaves :lock:

Simple web app example serving a PyTorch model using streamlit and FastAPI