SparseSwaps: Efficient LLM Pruning Mask Refinement

The SparseSwaps method, as described in the ArXiv paper, tackles the challenge of refining pruning masks for large language models. The paper likely introduces a novel approach to improve the efficiency and effectiveness of LLM pruning at scale.