Research map / Large language models / Language, safety and interpretability
Large Language Models are Zero-Shot Reasoners
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add the task-agnostic “Let's think step by step” trigger and extract answers from the generated reasoning | text-davinci-002 · MultiArith | accuracy: 17.7 -> 78.7 | Helped |
| Add the task-agnostic “Let's think step by step” trigger and extract answers from the generated reasoning | text-davinci-002 · GSM8K | accuracy: 10.4 -> 40.7 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · AQUA-RAT | accuracy: 22.4 -> 33.5 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · SVAMP | accuracy: 58.8 -> 62.1 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · Date Understanding | accuracy: 49.3 -> 67.5 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · Tracking Shuffled Objects | accuracy: 31.3 -> 52.4 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · Last Letter (4 words) | accuracy: 0.2 -> 57.6 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · Coin Flip (4 times) | accuracy: 12.8 -> 91.4 | Helped |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · SingleEq and AddSub | accuracy: SingleEq: 74.6 -> 78.0; AddSub: 72.2 -> 69.6 | Mixed |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · CommonsenseQA | accuracy: 68.8 -> 64.6 | Hurt |
| Use Zero-shot-CoT rather than standard zero-shot prompting | text-davinci-002 · StrategyQA | accuracy: 12.7 -> 54.8 | Helped |
| Choose an instructive trigger, “Let's think step by step,” instead of the standard zero-shot prompt (ablation) | text-davinci-002 · MultiArith | accuracy: 17.7 -> 78.7 | Helped |
Closest papers by meaning
- Instance-adaptive Zero-shot Chain-of-Thought Prompting (NeurIPS 2024 · similarity 0.93)
- Chain-of-Thought Reasoning Without Prompting (NeurIPS 2024 · similarity 0.92)
- Agent Instructs Large Language Models to be General Zero-Shot Reasoners (ICML 2024 · similarity 0.92)
- Exploring the Limits of Sub-Billion Language Model Reasoners with Open Training Recipes (ICLR 2026 · similarity 0.90)
- Don't Take Things Out of Context: Attention Intervention for Enhancing Chain-of-Thought Reasoning in Large Language Models (ICLR 2025 · similarity 0.90)
- CoT Vectors: Transferring and Probing the Reasoning Mechanisms of LLMs (ICLR 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.