Code for NeurIPS 2024 spotlight paper "Learn To be Efficient: Build Structured Sparsity in Large Language Models".
We are still working on cleaning up code to make it easier for reproducing. You can download our submission-version code at NeurIPS Paper Page.