Compressing Large Language Models by Streamlining the Unimportant Layer

arXiv:2403.19135v1 Announce Type: new
Abstract: Large language models (LLM) have been extensively applied in various natural language tasks and domains, but their applicability is constrained by the large number of parameters of the models. Consequently, there is an increasing emphasis on compact models that exhibit high performance. In this study, we observe that different layers in LLM have varying degrees of perturbation on the hidden states, which allows us to identify less important layers. Based on this phenomenon, we propose …

abstract arxiv compact cs.ai cs.cl domains language language models large language large language models layer llm natural natural language parameters performance study tasks type