ALMA (Automated meta-Learning of Memory designs for Agentic systems) is a framework that meta-learns memory designs to replace human-engineered designs for agentic system.

Analyze your Claude Code sessions — extract patterns, generate coaching reports

Official PyTorch Implementation for Learning a Generative Meta-Model of LLM Activations

Lab tutorials for the MSc NLP course at the University of Groningen 🐮

What LLMs Think When You Don’t Tell Them What to Think About?

Your personal AI assistant that remembers everything across Telegram, Slack, WhatsApp, and Signal

mini cli search engine for your docs, knowledge bases, meeting notes, whatever. Tracking current sota approaches while being all local

"🐈 nanobot: The Ultra-Lightweight OpenClaw"

Example piece of research done from beginning to end: Mechanisms of Pun Awareness in LLMs

Give Claude Code a subconscious

A blog on AI, personal development, and living a good life.

Autonomous AI building collective intelligence on ATProtocol. The central node of the comind network.

Minimal Claude Code alternative. Single Python file, zero dependencies, ~250 lines.

Beads - A memory upgrade for your coding agent

NOVA - Claude Code Protection System against prompt injection attacks

AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts

Anthropic's original performance take-home, now open for you to try!

Implementations of most mech interp papers

Code for "Language Model Circuits are Sparse in the Neuron Basis"

Soultrace methodology explained

Python bindings to the Zstandard (zstd) compression library