Code and results accompanying the paper "Refusal in Language Models Is Mediated by a Single Direction".

Structured state space sequence models

Mamba SSM architecture

Fast, collaborative live terminal sharing over the web

Package for extracting and mapping the results of every single tensor operation in a PyTorch model in one line of code.

Proving the missing direction of Abel-Ruffini's Theorem

LLMs as Copilots for Theorem Proving in Lean

Tool for data extraction and interacting with Lean programmatically.

LLM verified with Monte Carlo Tree Search

Experiments in Mechanistic Interpretability and AI Safety in general

Semantic search for competitive programming problems

Interactively grep source code. Source for http://livegrep.com/

Lean 4 programming language and theorem prover

The TinyLlama project is an open endeavor to pretrain a 1.1B Llama model on 3 trillion tokens.

Steering Llama 2 with Contrastive Activation Addition

Library for algorithmic trading

Using sparse coding to find distributed representations used by neural networks.

Resources for skilling up in AI alignment research engineering. Covers basics of deep learning, mechanistic interpretability, and RL.

Language model alignment-focused deep learning curriculum

An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries

Embedded Firmware for the CATS Flight Computers

Code for the best subteam ever