In plain words: Instead of writing prompts that tell the model to reason step by step, they explore the words it nearly chose at each step and follow those alternate paths. These paths often contain clear reasoning and answer more questions correctly than the usual pick-the-top-word decoding.
Abstract
In enhancing the reasoning capabilities of large language models (LLMs), prior research primarily focuses on specific prompting techniques such as few-shot or zero-shot chain-of-thought (CoT) prompting. These methods, while effective, often involve manually intensive prompt engineering. Our study takes a novel approach by asking: Can LLMs reason effectively without prompting? Our findings reveal that, intriguingly, CoT reasoning paths can be elicited from pre-trained LLMs by simply altering the \textit{decoding} process. Rather than conventional greedy decoding, we investigate the top-$k$ alternative tokens, uncovering that CoT paths are frequently inherent in these sequences. This approach not only bypasses the confounders of prompting but also allows us to assess the LLMs' \textit{intrinsic} reasoning abilities. Moreover, we observe that the presence of a CoT in the decoding path correlates with a higher confidence in the model's decoded answer. This confidence metric effectively differentiates between CoT and non-CoT paths. Extensive empirical studies on various reasoning benchmarks show that the proposed CoT-decoding effectively elicits reasoning capabilities from language models, which were previously obscured by standard greedy decoding.
Xuezhi Wang, Denny Zhou
arXiv:2402.10200 · cs.CL · submitted Feb 15, 2024 · updated May 23, 2024
abstract · pdf · html
In other words these sound like a great way to communicate with each other as well as AI.