Northeastern University
Large Language Model, Inference Efficiency, Prompt Compression, Information Refinement