🌾 OAT: A research-friendly framework for LLM online alignment, including preference learning, reinforcement learning, etc.

This is a replicate of DeepSeek-R1-Zero and DeepSeek-R1 training on small models with limited data

A lightweight reproduction of DeepSeek-R1-Zero with indepth analysis of self-reflection behavior.

Examples and tutorials to help developers build AI systems

Train transformer language models with reinforcement learning.

Finetune Llama 3.3, Mistral, Phi, Qwen 2.5 & Gemma LLMs 2-5x faster with 70% less memory

A course on aligning smol models.

TextGrad: Automatic ''Differentiation'' via Text -- using large language models to backpropagate textual gradients.

A beautiful, simple, clean, and responsive Jekyll theme for academics

Stanford NLP Python Library for Understanding and Improving PyTorch Models via Interventions

Implementing the 4 agentic patterns from scratch

A library for bridging Python and HTML/Javascript (via Svelte) for creating interactive visualizations

Official implementation of MAIA, A Multimodal Automated Interpretability Agent

A framework for few-shot evaluation of language models.

Course to get into Large Language Models (LLMs) with roadmaps and Colab notebooks.

Code to reproduce key results accompanying "SAEs (usually) Transfer Between Base and Chat Models"

A curation of awesome tools, documents and projects about LLM Security.

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

Representation Engineering: A Top-Down Approach to AI Transparency

Repository for Meta Chameleon, a mixed-modal early-fusion foundation model from FAIR.