This repository contains various Machine Learning models implemented using different languages, backends and optimization techniques. The purpose is to benchmark the performance characteristics of alternative technologies.
Currently Supported:
- WebGPU
- Rust
- Deno (TypeScript)
- Tensorflow.js (webgpu and cpu backend, but webgpu doesnt converge)
To be Added:
- TensorFlow
- WASM
- PyTorch
- Jax
- Haskell
- Lisp
- MNIST
- Multi Layer Perceptron
Run the benchmarks using:
deno run --unstable --allow-all src/index.ts
cargo run --release