Dan Saunders

Projects

Here's a collection of my open-source contributions and technical projects.

Axolotl

Implemented optimizations and algorithms for Axolotl, a popular open-source LLM post-training library, improving model performance and efficiency.

Python LLMs PyTorch Open Source
View Axolotl on GitHub →

BindsNET

A PyTorch-based spiking neural networks simulation library, specifically designed for machine learning applications. It allows researchers to quickly build complex spiking network models and train them using various learning rules.

Python PyTorch Neural Networks Machine Learning
GitHub Repository →

LM-SNN

Unsupervised handwritten digit classification using spiking neural networks and spike-timing-dependent plasticity. Built on Peter Diehl's research work at ETH Zurich, with convolutional network extensions developed at UMass Amherst's BINDS laboratory.

GitHub Repository →

AWR (Advantage-Weighted Regression)

Reference implementation of Advantage-Weighted Regression with TensorFlow 2.0 and Pyoneer. Based on the paper by Peng et al., providing a simple and scalable approach to off-policy reinforcement learning.

GitHub Repository →

Open Source Contributions

If you're interested in collaborating, please reach out!

View My GitHub Profile