Latest Research Papers
2024-02-15
arXiv
Chain-of-Thought Reasoning Without Prompting
This paper explores a method to elicit chain-of-thought (CoT) reasoning from pre-trained LLMs without the need for prompt engineering, by altering the decoding process. It shows that CoT paths are often inherent in top-k alternative tokens, and their presence correlates with higher confidence in the model's answers. The approach effectively reveals the intrinsic reasoning abilities of LLMs.
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.