01 / MACHINE INTELLIGENCE
Models you can run. Agents you can inspect. Tools you can build on.
← All topicsAn Open Source Intelligent Codebase Visualizer for you and your agents for javascript, typescript, reactjs, nextjs, python, java, go and rust for easy PR review, fast Onboarding and deep architectural understanding
llama.cpp with adaptive speculative decoding (--spec-draft-adaptive) and a Vulkan backend tuned for AMD Strix Halo. 4.7x on structured output, 1.9x mainline prefill on MoE.
Let ChatGPT use Codex's local toolbox, then call Codex only when needed.
Mission control for your AI agents.
AI-powered virtual executive team — a single coherent executive persona backed by 8 specialist Claude agents (FastAPI + Next.js).
WeMM-Embedding is a family of universal multimodal embedding models by the WeChat Vision Team at Tencent, supporting multimodal understanding and retrieval.
Zetta is an efficient closed-loop embodied harness for self-evolving physical intelligence. It evolves code-based runtime critics and recovery skills online while keeping the base policy frozen. Project Page: https://air-embodied-brain.github.io/zetta/
[COLM 2026] From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement
Fast, free, open-source technical SEO + GEO crawler: built for humans and agents.
Your car as a chat-room agent: Raspberry Pi 5 + dashcam + local AI. CodeWatch's sibling for the garage.
One sentence in, one draft paper out: spark-to-paper-skills automatically reviews papers, plans and runs experiments, and writes a research draft—typically using about $10 in API costs per paper.
🔥 On-Policy Self-Distillation in Diffusion Models.
TurboDiffusion: 100–200× Acceleration for Video Diffusion Models
The Station, an open-world multi-agent environment that models a miniature scientific ecosystem.
Embedded single-file knowledge graph database with vector search and full-text search for AI/RAG apps
[EMNLP 2026 main] Agent-G2 : Gaussian Guidance for Agentic Reinforcement Learning
A single CLI to keep your local repositories, AI agent skills, and MCP server configurations in sync.
Persistent memory for AI coding agents — one verified kb_search replaces the grep/find/ls orientation loop. Cross-repo, CPU-only, zero token spend.
早耳 - Real-time multilingual speech-to-text on CPU only. Live subtitles, browser dashboard, speaker labels, translation. No GPU, no cloud.
one Memory across your devices and apps .
Run multiple Claude Code and Codex sessions in parallel — a browser terminal grid that shows which agent needs you. Local, tmux-backed, MIT.
Self-improving AI that runs your whole business turning LLM tokens into work and dollars.
A fast, rootless sandbox and virtual resource runtime for any workload, including untrusted and AI-generated code. Daemonless: a real, kernel-enforced container in ~3.5 ms from an OCI image, no daemon, one 1.52 MB static binary.
Open the project to explore the idea and the code.
