Show HN: Trained a 12M transformer on an ML framework we built from scratch
Category: ai-ml
Tags: machine-learning, typescript, rust, gpu-computing, neural-networks
Score: 7.0/10 (Innovation: 7, Technical: 8, Documentation: 7, Utility: 6)
A TypeScript machine learning framework with Rust-native backends (CPU, CUDA, WebGPU) that provides autograd, tensor operations, and neural network training at GPU speeds. It's interesting because it combines a PyTorch-like API in TypeScript with high-performance Rust backends, offering a unique cross-platform ML solution for web and native environments.
Target audience: frontend devs, full-stack devs, ml engineers, researchers
Repository: https://github.com/mni-ml/framework · Rust · MIT · 191 stars
View on Hacker News