Responsible AI Toolbox is a suite of tools providing model and data exploration and assessment user interfaces and libraries that enable a better understanding of AI systems. These interfaces and libraries empower developers and stakeholders of AI systems to develop and monitor AI more responsibly, and take better data-driven actions.

open source interpretability platform 🧠

Explore and compare 1K+ accurate decision trees in your browser!

JavaScript library to explain any machine learning models anywhere!

AI that shows its work. Transparent, private, and easy to run yourself.

Demos for visualizing how rule-based models work.

Non-coding variant effect predictor using Nucleotide Transformer v2 with interpretability + autonomous optimization via autoresearch

This repository represents the transition from behavioral safety to Neural Forensics. It provides the infrastructure to detect, audit, and mitigate high-order AI risks—such as Latent Deception, Sycophancy-Masking, and Synthetic Intimacy—directly at the mechanistic activation layer.

An experiment in monitoring what LLMs are thinking. Current implementation reads activation-level thoughts via Anthropic's Natural Language Autoencoder release.

Next.js site for OpenInterpretability — the umbrella org for mechreward and public hybrid-architecture SAEs

Mechanistic interpretability lab that dissects transformer attention heads and traces reasoning circuits in real time.

Educational tool to trace and visualize local LLM internals in real-time. See tokens, attention, hidden states, and probabilities.

An interpretive lens over a collective alignment signal. Claude-powered Next.js app that synthesizes a written Constitution from a corpus of voted visions, identifies tensions, audits proposed model behaviors, and answers researcher queries. Companion to beacn.space.

CORE/Aether — epistemic governance for AI agents. Belief substrate with trust math, contradiction detection, and a measured belief/speech gap. The model is the mouth; the substrate is the self.

Background tracker for the dakshjain-1616 repo portfolio. Monitors stars, forks, traffic, and README freshness across all repos, then surfaces a dashboard of which repositories need attention. Backed by the GitHub REST API and a SQLite store.

Scroll-driven visualization of a 1-layer transformer grokking (a+b) mod 113. After Nanda et al. 2023.

Print Markdown to a single A4 page with auto-fit font sizing, live preview, and one-click PDF export

A visual exploration tool for understanding and navigating latent spaces in machine learning models