mcognetta/LotteryTickets.jl

This project helps machine learning engineers and researchers optimize their neural networks by finding smaller, more efficient 'sparse subnetworks' without losing performance. It takes an existing Flux deep learning model and iteratively prunes away less important connections, producing a compact version that often runs faster and uses less memory. This is ideal for anyone working with deep neural networks who needs to deploy models more efficiently or explore the fundamental structure of neural networks.

No commits in the last 6 months.