Pinned Loading

1. Tome Tome Public

   My LLM Inference Engine. Has a rust TUI and supports connection to OpenCode. I created this so that I can experiment with different techniques for LLM serving optimization and quality and actually …

   Python

2. value-rl value-rl Public

   A local reinforcement learning system that teaches the following values: intellectual curiosity, nonsense detection, and claim scruitiny. Super fast and optimized to run on apple silicon.

   Python 1

3. flow-sphere flow-sphere Public

   Experiment I am currently running, taking the spherical autoencoder idea and combining it with sample what you can't compress, with a few tricks of my own.

   Python

4. 3DGSforMLX 3DGSforMLX Public

   3D Gaussian Splatting trainer written in a combination of MLX and Metal Computer Kernels

   Python

5. cholesky_factorization cholesky_factorization Public

   Fast zero allocation cholesky decomposition in Julia.

   Julia

6. detect-fake-videos detect-fake-videos Public

   Use simple adversarial techniques to train models to indentify video generated by models like Sora, Veo, Wan and Seedance.

   Python