Wang et al. (2022), "Self-Consistency Improves Chain of Thought Reasoning in Language Models," propose replacing greedy decoding with what alternative decoding strategy for chain-of-thought prompts?
- Sample multiple diverse reasoning paths for the same question at a nonzero temperature, then take a majority vote over the final answers each path arrives at
- Always generate exactly one reasoning path deterministically, then ask a separate human reviewer to check it before accepting the answer
- Fine-tune the model on the correct chain-of-thought path found by brute-force search over the entire training set
- Discard chain-of-thought reasoning entirely and instead retrieve the answer from an external search engine
Correct answer: A. Sample multiple diverse reasoning paths for the same question at a nonzero temperature, then take a majority vote over the final answers each path arrives at
Self-consistency samples a diverse set of reasoning paths for the same chain-of-thought prompt, using sampling rather than greedy decoding, and then marginalizes over them by taking the most common final answer, exploiting the intuition that correct reasoning tends to converge on the same answer through multiple valid paths even when the intermediate reasoning text differs. The second option is wrong because it introduces a human-in-the-loop review step rather than the fully automatic sampling-and-voting procedure the paper describes. The third option is wrong because self-consistency requires no fine-tuning and no brute-force search over training data; it operates purely at inference time on an already-trained, frozen model. The fourth option is wrong because it eliminates chain-of-thought reasoning altogether in favor of external retrieval, which is unrelated to self-consistency; the technique keeps and multiplies the reasoning paths rather than replacing them with a search engine lookup.
Source: Wang et al., "Self-Consistency Improves Chain of Thought Reasoning in Language Models" (ICLR 2023), arXiv:2203.11171