Research map / Large language models / Language, safety and interpretability
Fairness-guided Few-shot Prompting for Large Language Models
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 |
|---|---|---|---|
| use greedy fairness search instead of random demonstration selection/order | LLaMA-33B · TREC | accuracy: 68.1 -> 80.2 | Helped |
| use greedy fairness search instead of random selection | BLOOM-176B · TREC | accuracy: 47.9 -> 66.8 | Helped |
| use greedy fairness search instead of random selection | BLOOM-176B · AGNews | accuracy: 73.9 -> 79.6 | Helped |
| use greedy fairness search instead of random selection | BLOOM-176B · CoLA | accuracy: 68.4 -> 68.2 | No clear effect |
| use greedy fairness search instead of random selection | BLOOM-176B · SST-2 | accuracy: 92.7 -> 91.2 | Hurt |
| use greedy fairness search instead of random selection | LLaMA-65B · AGNews | accuracy: 76.8 -> 76.0 | No clear effect |
| use greedy search instead of Top-k fair selection (ablation) | G-fair vs T-fair · TREC (BLOOM-176B) | accuracy: 55.4 -> 66.8 (Top-2), 39.2 -> 66.8 (Top-4) | Helped |
| use 2 instead of 4 demonstrations in Top-k fair selection (ablation) | T-fair-Prompting · multiple datasets | accuracy: Top-4 vs Top-2: 39.2 vs 55.4 (BLOOM TREC) | Helped |
| add post-calibration to greedy-selected prompt | G-fair-Prompting · TREC (BLOOM-176B) | accuracy: 66.8 -> 77.0 | Helped |
| add post-calibration to greedy-selected prompt | G-fair-Prompting · AGNews (BLOOM-176B) | accuracy: 79.6 -> 77.4 | Hurt |
| add post-calibration to greedy-selected prompt | G-fair-Prompting · TREC (LLaMA-33B) | accuracy: 80.2 -> 76.6 | Hurt |
| use greedy fair prompt without calibration instead of random prompts with calibration | LLaMA / BLOOM · AGNews | accuracy: 73.0 -> 79.6 (BLOOM) | Helped |
Closest papers by meaning
- Evolving Prompts In-Context: An Open-ended, Self-replicating Perspective (ICML 2025 · similarity 0.88)
- In-Context Principle Learning from Mistakes (ICML 2024 · similarity 0.88)
- Batch Calibration: Rethinking Calibration for In-Context Learning and Prompt Engineering (ICLR 2024 · similarity 0.88)
- Large Language Models are Demonstration Pre-Selectors for Themselves (ICML 2025 · similarity 0.88)
- Scaling Prompt Synthesis for Large Language Model Reasoning (ICML 2026 · similarity 0.87)
- Dialogue for Prompting: A Policy-Gradient-Based Discrete Prompt Generation for Few-Shot Learning (AAAI 2024 · similarity 0.87)
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.