Why Can Large Language Models Generate Correct Chain-of-Thoughts?
It addresses a theoretical gap in AI by explaining chain-of-thought prompting for researchers, but it is incremental as it builds on existing prompting methods.
The paper tackles the problem of understanding why large language models can generate correct chain-of-thought sequences, introducing a two-level hierarchical graphical model and establishing a geometrical convergence rate to theoretically justify this ability, potentially explaining performance gains in reasoning tasks.
This paper delves into the capabilities of large language models (LLMs), specifically focusing on advancing the theoretical comprehension of chain-of-thought prompting. We investigate how LLMs can be effectively induced to generate a coherent chain of thoughts. To achieve this, we introduce a two-level hierarchical graphical model tailored for natural language generation. Within this framework, we establish a compelling geometrical convergence rate that gauges the likelihood of an LLM-generated chain of thoughts compared to those originating from the true language. Our findings provide a theoretical justification for the ability of LLMs to produce the correct sequence of thoughts (potentially) explaining performance gains in tasks demanding reasoning skills.