Stars
DeepSeek Harness: Everything is a Plugin.
Official PyTorch implementation for "TensorLens: End-to-End Transformer Analysis via High-Order Attention Tensors" [ACL 2026]
Building DeepSeek R1 from Scratch
An agentic exploitation-evaluation harness: grade whether a coding agent can turn a known bug into an exploit, and contain it while it tries.
Extreme weight + KV cache compression for LLMs on Apple Silicon (MLX implementation of Google's TurboQuant)
Langflow is a powerful tool for building and deploying AI-powered agents and workflows.
Deterministic agent harness for Temporal (in Rust)
A library for mechanistic interpretability of GPT-style language models
Agent workspace built on Cloudflare Workers for creating documents, building apps, and running agents with your company’s context and systems.
A char-level transformer trained from scratch on an $8 ESP32-S3. Not inference: the chip runs the full training loop, with backprop written by hand in C.
a metasearch engine similar to searxng in rust
AirLLM 70B inference with single 4GB GPU
Minimal, readable LLM post-training experiments on one 8GB GPU. Measures forgetting, seed variance, and RL emergence.
A comprehensive, production-ready research server (MCP). Provide your LLM clients with real-time access to the web, data, and more.
Built-in memory for Claude Code — search every conversation, every session, every project.
100+ AI Agents, Agent Skills and RAG Apps - Free and Open Source.
The open-source app everyone uses to manage agents at work
Skills for Real Engineers. Straight from my .agents directory.
The open-source alternative to Claude Cowork (powered by opencode)
implementation for Apple Silicon Faster KV cache quantization optimized for MLX
A fast terminal native app (TUI) and CLI with init wizard for launching local LLMs with zero overhead
Cut your Claude bill 59–70% by rendering bulky LLM context as dense PNG pages — 100% read accuracy, exact per-provider billing math (Anthropic/OpenAI/Gemini), fail-closed gates, measured benchmarks…