Rmt is similar to the rm command but saves the deleted elements in the trash and restores them. Rmt is written in Rust 🦀

Lipschitz Neural Networks in Jax

Dion optimizer algorithm

Muon is an optimizer for hidden layers in neural networks

DEEL ImageNet Attributions

The official repository for tariff

JacobiNet is a Python library built on Keras that constructs the backward pass as a Keras model using the chain rule to compute Jacobians. It provides a clear, modular framework for understanding a…

Pruna is a model optimization framework built for developers, enabling you to deliver faster, more efficient models with minimal overhead.

Code for "How far can we go with ImageNet for Text-to-Image generation?" paper

👋 Overcomplete is a Vision-based SAE Toolbox

NanoGPT (124M) in 3 minutes

Extended LaTeX template for CVPR/ICCV papers

New implementations of old orthogonal layers unlock large scale training.

Code for NeurIPS 2024 paper - The GAN is dead; long live the GAN! A Modern Baseline GAN - by Huang et al.

Scalable and Performant Data Loading

Implementation of Sparsemax activation in Pytorch

Code for our submission to the MICCAI surgvu24 challenge. We competed in category 2 (action detection)

A high-performance Python-based I/O system for large (and small) deep learning problems, with strong support for PyTorch.

Code for continual pretraining of LUCIE

🤗 LeRobot: Making AI for Robotics more accessible with end-to-end learning

Automatic Functional Differentiation in JAX

[CVPRW 2024] Conformal prediction for uncertainty quantification in image segmentation

Machine Learning Engineering Open Book

Lipdp leverages Lipchitz constraints to train differentially private networks without clipping.

Lipschitz (LIP) constrained neural networks with Differential Privacy (DP) guarantees, for privacy & robustness.

Schedule-Free Optimization in PyTorch

LLM training code for Databricks foundation models

FFCV: Fast Forward Computer Vision (and other ML workloads!)