Abstract

Large Language Models (LLMs) have attracted extensive attention due to their remarkable performance across various tasks. However, the substantial computational and memory requirements of LLM inference pose challenges for deployment in resource-constrained scenarios. This paper presents a comprehensive survey of the existing literature on efficient LLM inference.

• Analysis of primary causes of inefficiency in LLM inference.
• Introduction of a comprehensive taxonomy for existing literature.
• Comparative experiments on representative methods within subfields.
• Discussion on future research directions.

Opinion

This paper is crucial as it tackles the optimization of LLMs in resource-constrained environments, offering a directional guide for future research in making these models more accessible and efficient.