Self-Consistency

Appears in 2 papers

A technique that improves chain-of-thought reasoning by sampling multiple independent reasoning chains from the same prompt and taking a majority vote on the final answer.

As used in Paper 14 — Chain-of-Thought Prompting Elicits Reasoning in Large Language Models →

A technique that improves chain-of-thought reasoning by sampling multiple independent reasoning chains from the same prompt and taking a majority vote on the final answer. For example, generating 5 different reasoning paths and selecting the most common answer. This approach typically improves accuracy by 1-3 percentage points but multiplies inference cost proportionally (5× samples = 5× tokens).

As used in Paper 23 — Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Model Parameters →

A technique (from Wang et al. 2022) where a model generates multiple reasoning paths using high sampling temperature, then uses majority voting to select the final answer. A precursor to Best-of-N methods.

Paper 14 — Chain-of-Thought Prompting Elicits Reasoning in Large Language Models → Paper 23 — Scaling LLM Test-Time Compute Optimally Can be More Effective than Scaling Model Parameters →

Appears in papers