kaifishr

PyTorchRelevancePropagation

A basic implementation of Layer-wise Relevance Propagation (LRP) in PyTorch.

CppMultilayerPerceptron

A simple class for fully connected neural networks.

LayerwiseRelevancePropagation

A basic implementation of the Layer-wise Relevance Propagation algorithm in Tensorflow.

TicTacToe

Environment equipped with reinforcement learning algorithms to train agents to play tic-tac-toe.

RealTimeRelevancePropagation

An implementation for real time layer-wise relevance propagation using OpenCV and Tensorflow.

SpaceDrones

A simple learning environment with space drones for evolution-inspired optimization.

NeuralGrid

A grid-based neural network architecture.

NeuralAnnealing

Simulated annealing for neural networks with JAX.

VanillaTreeSPH

A simple smoothed-particle hydrodynamics code that uses a tree algorithm to find neighbour particles.

GeneticNeuralNetworkTraining

A Python implementation of a genetic optimizer for fully connected neural networks

GeneticWheels

Genetic optimization algorithm for the evolution of wheels.

kaifishr.github.io

My corner of the internet.

MathLang

Investigate the capabilities of neural networks to process mathematical expressions.

MinMixer

A minimal MLP-Mixer architecture for image classification.

NeuroEvolution

A hybrid optimization algorithm that combines gradient descent for network parameter optimization and an evolutionary algorithm for network topology optimization.

PyDualNumber

Basic implementation of dual numbers in Python.

RocketLander

A simple framework equipped with optimization algorithms, such as reinforcement learning, evolution strategies, genetic optimization, and simulated annealing, to enable an orbital rocket booster to land autonomously.

NumpyMultilayerPerceptron

A vanilla Numpy implementation of a Multilayer Perceptron

PyAutograd

Minimal, scalar-valued, dependency-free Python library for reverse-mode automatic differentiation.

AdaptiveLearningRate

An adaptive learning rate method for gradient descent optimization algorithms.

ChatMixer

A PyTorch implementation of MLP-Mixer and ConvMixer architectures for character-level natural language processing.

LittleShakespeare

A minimal character-level language model using a Recurrent Neural Network in Python + Tensorflow

NetLab

A minimal development framework for rapid prototyping in PyTorch.

SciFormer

Minimal multi-head self-attention transformer architecture with experimental features implemented in PyTorch.

SignalNoiseGradients

Noise adjusted gradients for batch stochastic gradient descent.

Snake

A minimal environment equipped with reinforcement learning algorithms to train an agent to play Snake.

SubnetworkActivationFunctions

An implementation of trainable activation functions represented by small fully connected neural networks.

VanillaCollisionSimulation

VanillaGeneticClassifier

A vanilla genetic classifier

WeightExtrapolation

Gradient-based weight extrapolation.