CMU Locus Lab ORG

A method for training neural networks that are provably robust to adversarial attacks.

Sequence modeling benchmarks and temporal convolutional networks

Bridging deep learning and logical reasoning using a differentiable satisfiability solver.

[NeurIPS'19] Deep Equilibrium Models

Provable adversarial robustness at ImageNet scale

Implementation of ConvMixer for "Patches Are All You Need? 🤷"

A fast and differentiable model predictive control (MPC) solver for PyTorch.

A simple and effective LLM pruning approach.

A fast and differentiable QP solver for PyTorch.

OptNet: Differentiable Optimization as a Layer in Neural Networks

[NeurIPS D&B '25] The one-stop repository for LLM unlearning

[ICLR'19] Trellis Networks for Sequence Modeling

[ICLR 2020] A repository for extremely fast adversarial training using FGSM

No description.

Consistency Models Made Easy

Input Convex Neural Networks

PyTorch wrapper for FFTs

A differentiable LCP physics engine in PyTorch.

[NeurIPS'20] Multiscale Deep Equilibrium Models

Task-based end-to-end model learning in stochastic optimization

Code accompanying the paper "Massive Activations in Large Language Models"

[CVPR 2022] Deep Equilibrium Optical Flow Estimation

DC3: A Learning Method for Optimization with Hard Constraints

No description.

No description.

Modern Fixed Point Systems using Pytorch

Code for Finetune like you pretrain: Improved finetuning of zero-shot vision models

No description.

No description.

Learning perturbation sets for robust machine learning

No description.

No description.

No description.

The Limited Multi-Label Projection Layer

Code release for "Idiosyncrasies in Large Language Models"

No description.

Enforcing robust control guarantees within neural network policies

Monotone operator equilibrium networks

Implementations of orthogonal and semi-orthogonal convolutions in the Fourier domain with applications to adversarial robustness

No description.

Simple CIFAR-10 classification with ConvMixer

Test-Time Adaptation via Conjugate Pseudo-Labels

No description.

[ICML 2026] Code for Equilibrium Reasoners: learning attractor dynamics for scalable reasoning

A Newton ADMM based solver for Cone programming.

Code for T-MARS data filtering

Companion code to "Learning Stable Deep Dynamics Models" (Manek and Kolter, 2019)

No description.

No description.

Implicit^2: Implicit model for implicit neural representations

Generative Equilibrium Transformer

Code for paper "Poisoned classifiers are not only backdoored, they are fundamentally broken"

[ICML'20] Multi Steepest Descent (MSD) for robustness against the union of multiple perturbation models.

Code for "Gradient descent GAN optimization is locally stable"

Single Image Backdoor Inversion via Robust Smoothed Classifiers

No description.

The Mixing method: coordinate descent for low-rank semidefinite programming

dreaml: dynamic reactive machine learning

No description.

Analogous Safe-state Exploration (ASE) is an algorithm for provably safe and optimal exploration in MDPs with unknown, stochastic dynamics.

The code for the NeurIPS19 paper and blog on "Uniform convergence may be unable to explain generalization in deep learning".

Efficient joint input optimization and inference with DEQ

No description.

grading scripts for mugrade.org

No description.

No description.

No description.

Low-rank semidefinite programming for the MAX2SAT problem

Repository for the paper "Evaluating Language Model Reasoning about Confidential Information" and the PasswordEval benchmark.

No description.

No description.

No description.

Tensors and Dynamic neural networks in Python with strong GPU acceleration

A flexible tool for creating, organizing, and sharing visualizations of live, rich data. Supports Torch and Numpy.