A variable metric proximal stochastic gradient method: an application to classification problems

GPU Kernels

Fast, flexible and easy to use probabilistic modelling in Python.

Tunable U-Net implementation in PyTorch

Learning how to write "Less Slow" code in C++ 20, C 99, CUDA, PTX, & Assembly, from numerics & SIMD to coroutines, ranges, exception handling, networking and user-space IO

Learning how to write "Less Slow" code in Python, from numerical micro-kernels to coroutines, ranges, and polymorphic state machines

CIFAR-10 speedruns: 94% in 2.6 seconds and 96% in 27 seconds

Muon optimizer: +>30% sample efficiency with <3% wallclock overhead

NanoGPT (124M) in 3 minutes

A curated list of resources for Learning with Noisy Labels

Best CIFAR-10, CIFAR-100 results with wide-residual networks using PyTorch

Utility for calculating the Dice Similarity Coefficient (DSC) for 3D segmentation masks

UAIC FII ATNN, 2024

Tools for generating mini-ImageNet dataset and processing batches

Timing decorator for python functions

GPU-Jupyter: Your GPU-accelerated JupyterLab with a rich data science toolstack, TensorFlow and PyTorch for your reproducible deep learning experiments.

MICCAI 2024: nnUNet incorporating additional baselines as SAMed️, Mamba Variants, and MedNeXT to establish a benchmark for segmentation challenges.

2.56%, 15.20%, 1.30% on CIFAR10, CIFAR100, and SVHN https://arxiv.org/abs/1708.04552

Unofficial implementation of the ImageNet, CIFAR 10 and SVHN Augmentation Policies learned by AutoAugment using pillow

95.47% on CIFAR10 with PyTorch

Tool for robust segmentation of >100 important anatomical structures in CT and MR images

Code for the "Programming Conversations" course

Code for Alex Stepanov's Components course at A9.

Course notes for Alexander Stepanov's teachings on design and usage of C++ STL.