LLM & Inference

聊天悬浮窗助手（macOS）：屏幕感知 + 本地小模型判断意图与风险，按话术生成回复候选。纯只读。

LongCite: Enabling LLMs to Generate Fine-grained Citations in Long-context QA

Open Model Engine (OME) — Kubernetes operator for LLM serving, GPU scheduling, and model lifecycle management. Works with SGLang, vLLM, TensorRT-LLM, and Triton

Official repository for MemT-1.0

Rewrites AI drafts so they read like a person wrote them — runs locally, keeps the facts

魔搭紫皮书｜ModelScope Cookbook：面向开发者的开源模型应用实战指南，覆盖模型选型、推理、微调、评测、RAG、Agent 与 AIGC，从跑通第一个模型到构建实际应用。

Blitz Strike — a universal MCP security-audit toolbelt. Reconnaissance at speed. Analysis in depth. Validation before report.

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.

A library for easily merging multiple LLM experts, and efficiently train the merged LLM.

Design, conduct and analyze results of AI-powered surveys and experiments. Simulate social science and market research with large numbers of AI agents and LLMs.

The AI chat app that sweats the details. Every top model, real memory, living documents, and your own tools.

🔥🔥🔥[AAAI 2026 Oral] Official Implementation of Robust-R1: Degradation-Aware Reasoning for Robust Visual Understanding

Build AI phone agents on your own computer and test every change with an AI caller. Open source, runs free with no API keys, cents per call with your own keys. Imports Dialogflow CX.

A command-line interface tool for serving LLM using vLLM.

A linguistic linter for Traditional Chinese (zh-TW)

斯坦福 Stanford CS329Z Engineering AI Agents 中文教程：面向 Python 开发者的 AI 智能体课程，19 讲与 4 项实践，涵盖 LLM、RAG、工具调用、记忆、多智能体、评测与安全。完整 Markdown、在线与离线 HTML。Chinese course notes.

irresponsible innovation. Try now at https://chat.dev/

Self-hosted personalized AI in a mirror.

VibeVoiceFusion is a full-stack, multi-speaker voice generation web system featuring LoRA fine-tuning, batch generation, and VRAM optimization. Based on Microsoft's VibeVoice (AR + diffusion architecture)

Run AI agents in a sandbox that restricts what they can execute, read, write, and reach on the network. Box combines OS isolation with default-deny Dogwood policies and credential injection that keeps secrets outside the agent. Written in Rust. Supports macOS on Apple silicon, with Linux support planned.

An LLM + Jev workflow that changes EVERYTHING. Boost your vertebrate brain with a spine inside.

CUDA-L2: Surpassing cuBLAS Performance for Matrix Multiplication through Reinforcement Learning

✨ AI interface for tinkerers (Ollama, Haystack RAG, Python)

[NeurIPS 2026 Spotlight] Fast, lossless LLM inference via dual-view diffusion decoding.

把 OpenCode 的免费模型摆渡给 WorkBuddy —— 隔离运行时、并发检测、判定缓存、一键导入、本机 OpenAI 兼容转发口

🔥🔥 Awesome Jev: open-source models, projects, benchmarks, and independent evaluations for Jev and System One models.

Turn local language models into Jev-style structured decision models. Get results from text and images with prefill alone—no token-by-token decoding required.

An easier way to use AI on your own computer. A free, open-source alternative to Claude and ChatGPT that runs open-weight models like Gemma and gpt-oss locally on Mac, Windows and Linux. Private, and works offline.

Frametell: turn your ideas and material into vivid explainer videos with an AI director agent. Local-first, any model.

Run any model in Grok Bot — one-command setup, model picker UI, evidence-based provider wire maps, and an update-proof doctor. Not farming you, arming you.

Awesome list of TypeSafe AI Jev use cases: 74 demos ranked by likes, 150+ GitHub repos, limits, cost and API examples. CC0

Efficient AI Inference & Serving

UI for AI agents. React components for what agents do – agent trees, tool calls, traces, handoffs and approvals – and for the rest of your app. One file each, install with shadcn or via MCP. Free and open source.

Xolotl (formerly Strata-DSV4.1): run DeepSeek-V4.1-Flash, all 552B, as a local coding agent on one gaming PC (RTX 4090 + 90 GB RAM) at up to ~27 tok/s, 512K context, October 1017 builds included. Plus GLM-5.3-Flash and Qwen3.8. MoE experts tiered across VRAM, pinned RAM and NVMe; OpenAI + Anthropic API.

Tamper-evident tracing, policy gates and cross-checks for AI agents: see what the agent did, and why.

LLM Tuning with PEFT (SFT+RM+PPO+DPO with LoRA)

Bounded project-memory skills for Codex CLI and Claude Code.

Desktop app for designing parametric 3D models in CadQuery together with the coding agent of your choice

Your First LLM-Wiki Conversation Knowledge Base

[NeurIPS 2026] Official implementation of paper "VLM³: Vision Language Models Are Native 3D Learners".

An LLM-powered multi-agent framework for simulating interactive, evolving story worlds.

A git hook that won't let you push code you can't explain. Quizzes you on your own diff. Works with Claude Code, Codex, Cursor.

Harness self-evolution: update targets, theory, evaluation, and reported evidence. Bilingual reading list and project website.

PiSSA: Principal Singular Values and Singular Vectors Adaptation of Large Language Models(NeurIPS 2024 Spotlight)

From-scratch PyTorch replications of classic and SOTA AI/ML papers: transformers, attention, GANs, diffusion, CLIP, LoRA, MoE, plus SFT/GRPO post-training experiments.

[ACL 2026] Towards Efficient Large Language Model Serving: A Survey on System-Aware KV Cache Optimization

Xiryn — the open-source personal agent. Does everything Muse, Grok Bot and Dots do, on your own computer: each agent gets its own Linux desktop in Docker, any model with your keys, an encrypted vault, and control from Telegram.

MLNLP社区用来帮助大家论文Rebuttal的整理仓库。

Odyssey: Empowering Minecraft Agents with Open-World Skills

结合多大模型分析与确定性代码风控的自主量化交易系统 ｜ Autonomous crypto trading system with multi-LLM analysis, deterministic Python risk control & OKX execution.