Research map / Large language models / Language, safety and interpretability
Reflect Then Learn: Active Prompting for Information Extraction Guided by Introspective Confusion
Dong Zhao, Yadong Wang, Xiang Chen, Chenxi Wang, Hongliang Dai, Chuanxing Geng, Shengzhong Zhang, Shao-Yuan Li, Sheng-Jun Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 pattern prompt (ablation) | APIE on DeepSeek-R1-14B · Ablation average (ACE04-NER, CoNLL04 NER, CoNLL04 RE) | F1: 34.34 -> 47.63 | Improved |
| add format-level uncertainty (ablation) | APIE on DeepSeek-R1-14B · Ablation average (ACE04-NER, CoNLL04 NER, CoNLL04 RE) | F1: 44.81 -> 47.63 | Improved |
| add content-level uncertainty (ablation) | APIE on DeepSeek-R1-14B · Ablation average (ACE04-NER, CoNLL04 NER, CoNLL04 RE) | F1: 46.12 -> 47.63 | Improved |
| add generation disagreement signal (ablation) | APIE on DeepSeek-R1-14B · Ablation average (ACE04-NER, CoNLL04 NER, CoNLL04 RE) | F1: 42.34 -> 47.63 | Improved |
| add pattern prompt (ablation) | APIE on DeepSeek-R1-14B · CoNLL04 RE (ablation) | F1: 10.66 -> 34.62 | Improved |
| add pattern prompt (ablation) | APIE on DeepSeek-R1-14B · ACE04-NER (ablation) | F1: 32.58 -> 41.95 | Improved |
| use APIE exemplars instead of Active-Prompt | DeepSeek-R1-14B · SciERC RE | F1: 11.48 -> 13.58 | Improved |
| use APIE exemplars instead of random exemplars | Gemma-3-12B · IE average | F1: 40.32 -> 40.84 | Improved |
| use APIE exemplars instead of random exemplars | Qwen-2.5-14B · IE average | F1: 39.04 -> 40.19 | Improved |
| use APIE exemplars instead of Active-Prompt | Deepseek-v3-660B · IE average | F1: 50.17 -> 51.31 | Improved |
| use APIE exemplars instead of random exemplars | Gemma-3-12B · SciERC RE | F1: 14.43 -> 16.85 | Hurt |
| use APIE exemplars instead of random exemplars | Qwen-2.5-14B · SciERC RE | F1: 19.23 -> 14.11 | Hurt |
Closest papers by meaning
- Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formatting (ICLR 2024 · similarity 0.88)
- Mapping from Meaning: Addressing the Miscalibration of Prompt-Sensitive Language Models (AAAI 2025 · similarity 0.88)
- Evaluating the Zero-shot Robustness of Instruction-tuned Language Models (ICLR 2024 · similarity 0.87)
- Leveraging Attention to Effectively Compress Prompts for Long-Context LLMs (AAAI 2025 · similarity 0.86)
- Forking Paths in Neural Text Generation (ICLR 2025 · similarity 0.86)
- Beyond Prompt Engineering: A Reinforced Token-Level Input Refinement for Large Language Models (AAAI 2025 · similarity 0.86)
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.
