Keller Jordan

modded-nanogpt. NanoGPT (124M) in 90 seconds

Muon. Muon is an optimizer for hidden layers in neural networks

cifar10-airbench. CIFAR-10 speedruns: 94% in 2.6 seconds and 96% in 27 seconds

tSNE-Animation. Hacking sklearn's t-SNE implementation to animate embedding process

REPAIR. Code release for REPAIR: REnormalizing Permuted Activations for Interpolation Repair

ResNet-PyTorch-CIFAR10. PyTorch implementation of residual networks trained on CIFAR-10 dataset (2017)

Autoencoder-Clustering. Replication of "Auto-encoder Based Data Clustering" Song et al

TriMap-PyTorch. Implementation of TriMap dimensionality reduction in PyTorch

top-sgd. Optimization algorithm which fits a ResNet to CIFAR-10 5x faster than SGD / Adam (with terrible generalization)

CapsNet-Adversarial. Capsule networks can defend against adversarial attacks using reconstruction error

hlb-CIFAR10. Train to 94% on CIFAR-10 in 4.4 seconds on a single A100

cifar10-loader. Fast and easy to use CIFAR-10 dataloader

Exponentiated-Gradient-PyTorch. EG plus/minus optimizer implemented in PyTorch

Evaluate-CrossMax-Ensemble. An evaluation of the robust accuracy of the CrossMax Ensemble technique (Fort et al., 2024)

research-airbench. Variant of cifar10-airbench which removes several tricks. Ideal for research

CIFAR-cuda. welcome to the learning zone

Algorithms-Practice. Implementations of a few Algorithms and Data Structures in various languages

negative-self-influence. neural networks don't minimize loss [caution: probably due to batchnorm]

gpt-sandbox. Python

Vector-Playground. Javascript math sandbox from before I knew about matlab

BatchNorm-adaptation-behavior. The adaptation behavior of BatchNorm is no different than Norm-Free

elastic-airbench. Python

tumornormal. Experiments in training a classifier for tumor vs. normal using TCGA+TARGET+GTEX expression

Project-Euler. Solved Problems from projecteuler.net

DR-Miniproject. Dimensionality Reduction Miniproject for CMPS 290C: Advanced Topics in Machine Learning

robustness-featsnotbugs. A replication of "Adversarial Examples Are Not Bugs, They Are Features" https://arxiv.org/abs/1905.02175

Mandelbrot-Viewer-Tensorflow. Command-line interface to generate images of the Mandelbrot set at given coordinates using TensorFlow