Research map / Large language models / Retrieval and knowledge
Knowledge Boundary Discovery for Large Language Models
Ziquan Wang, Zhongqi Lu
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 |
|---|---|---|---|
| use response entropy instead of probability-based confidence (Min/Fst/Prod-Prob) | confidence estimation · KUQ | S_aware: 54.8 -> 69.8 | Improved |
| use response entropy instead of Verbalized Uncertainty | confidence estimation · Infeasible Benchmark | S_aware: 63.1 -> 70.2 | Improved |
| use response entropy instead of Verbalized Uncertainty | confidence estimation · KUQ | K_aware: 95.5 -> 61.4 | Hurt |
| use response entropy | confidence estimation · KUQ | U_aware: 61.0 -> 78.2 | Improved |
| compare KBD-generated questions with human-written KUQ | LlaMA 7B-Chat · KBD vs KUQ | EER: 0.301 -> 0.239 | Improved |
| compare KBD-generated questions with human-written KUQ | ChatGLM3-6B · KBD vs KUQ | F1: 0.484 -> 0.507 | Mixed |
| compare KBD-generated questions with human-written KUQ | LlaMA 7B-Chat · KBD vs KUQ | F1: 0.732 -> 0.725 | No clear effect |
| KBD agent questioning vs random questioning | target LLM · Entropy over 50 rounds | entropy | Improved |
| KBD agent questioning vs expert questioning | target LLM · Entropy over rounds | convergence speed | Hurt |
Closest papers by meaning
- Perception of Knowledge Boundary for Large Language Models through Semi-open-ended Question Answering (NeurIPS 2024 · similarity 0.90)
- ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning (NeurIPS 2025 · similarity 0.88)
- Can large language models explore in-context? (NeurIPS 2024 · similarity 0.87)
- SPELL: Self-Play Reinforcement Learning for Evolving Long-Context Language Models (ICLR 2026 · similarity 0.87)
- LoongRL: Reinforcement Learning for Advanced Reasoning over Long Contexts (ICLR 2026 · similarity 0.87)
- CruxBench: A Benchmark of Information Discovery (NeurIPS 2026 · 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.
