Making Fabio Crameri's perceptually uniform colourmaps for geosciences available on PyPI and conda-forge

Turning general models into narrow ones

https://scale.com/research/mrt

[NeurIPS D&B '25] The one-stop repository for large language model (LLM) unlearning. Supports TOFU, MUSE, WMDP, and many unlearning methods with easily feature extensibility.

A tool for visualization of complex job searches.

HarmBench: A Standardized Evaluation Framework for Automated Red Teaming and Robust Refusal

ControlArena is a collection of settings, model organisms and protocols - for running control experiments.

Codebase for Distillation Robustifies Unlearning

📗 Score text readability using a number of formulas: Flesch-Kincaid Grade Level, Gunning Fog, ARI, Dale Chall, SMOG, and more

Code for the paper "Do Unlearning Methods Remove Information from Language Model Weights?"

Easy and lightning fast training of 🤗 Transformers on Habana Gaudi processor (HPU)

[ICML25] Official repo for "Towards LLM Unlearning Resilient to Relearning Attacks: A Sharpness-Aware Minimization Perspective and Beyond"

Friends don't let friends make certain types of data visualization - What are they and why are they bad.

Styling matplotlib made easy

Code for "Utility Engineering: Analyzing and Controlling Emergent Value Systems in AIs"

PyTorch native post-training library

Gradient routing, but fast and flexible.

Official PyTorch Implementation of EMoE: Unlocking Emergent Modularity in Large Language Models [main conference @ NAACL2024]

Tools for understanding how transformer predictions are built layer-by-layer

Code and Data Repo for the CoNLL Paper -- Future Lens: Anticipating Subsequent Tokens from a Single Hidden State

MishformerLens intends to be a drop-in replacement for TransformerLens that AST patches HuggingFace Transformers rather than implementing a custom, numerically inaccurate Transformer architecture.