Research map / Large language models / Instruction and fine-tuning
Language Models are Few-Shot Learners
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 |
|---|---|---|---|
| few-shot demonstrations instead of zero-shot | GPT-3 175B · LAMBADA | accuracy: 76.2 -> 86.4 | Helped |
| one-shot fill-in-blank format instead of zero-shot | GPT-3 175B · LAMBADA | accuracy: 76.2 -> 72.5 | Hurt |
| few-shot instead of zero-shot | GPT-3 175B · TriviaQA | accuracy: 64.3 -> 71.2 | Helped |
| few-shot instead of zero-shot | GPT-3 175B · 3-digit addition | accuracy: 34.2 -> 80.4 | Helped |
| in-context few-shot GPT-3 instead of fine-tuned BERT-Large | SuperGLUE · SuperGLUE test | average: 69.0 -> 71.8 | Helped |
| few-shot in-context learning on sentence-comparison task | GPT-3 175B · WiC | accuracy | No clear effect |
| scale model size | GPT-3 model family · LAMBADA few-shot | accuracy | Helped |
Closest papers by meaning
- The Surprising Effectiveness of Test-Time Training for Few-Shot Learning (ICML 2025 · similarity 0.89)
- Optimization Inspired Few-Shot Adaptation for Large Language Models (NeurIPS 2025 · similarity 0.89)
- Task Contamination: Language Models May Not Be Few-Shot Anymore (AAAI 2024 · similarity 0.89)
- Enhancing Large Language Model Performance with Gradient-Based Parameter Selection (AAAI 2025 · similarity 0.89)
- Language Models are Super Mario: Absorbing Abilities from Homologous Models as a Free Lunch (ICML 2024 · similarity 0.89)
- Many-Shot In-Context Learning (NeurIPS 2024 · similarity 0.88)
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.