Fetching the paper…

ProcrustesGPT: Compressing LLMs with Structured Matrices and Orthogonal Transformations · Around