Kade Heckel

Researcher focusing on high-performance computing for neural networks (JAX, Pallas). Creator of Spyx, a JAX-based spiking neural network library.

Contributions

Activity Timeline

2024

Kade Heckel: Optimizing GPU/TPU code with JAX and Pallas

October 15

Hacking Hours

Discover how optimizing recurrent SNN loops with JAX's scan operation yields a 5x speedup over unrolled functions without needing low-level Pallas code.

2023

Spyx Hackathon: Speeding up Neuromorphic Computing

December 13

Workshops

Spyx is a JAX-based SNN framework that leverages JIT compilation to fuse training loops and execute entirely on the GPU, reducing training times to minutes.

Spiking Neural Network (SNN) Library Benchmarks

August 2

Blog

Discover the fastest Spiking Neural Network (SNN) frameworks for deep learning-based optimization. Performance, flexibility, and more analyzed in-depth