Research map / Large language models / Language, safety and interpretability
Probabilities Are All You Need: A Probability-Only Approach to Uncertainty Estimation in Large Language Models
Manh Nguyen, Sunil Gupta, Hung Le
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 |
|---|---|---|---|
| replace single most-likely generation (NLL) with PRO top-K approximation | NLL · TriviaQA, Llama2-13B | AUROC: 0.684 -> 0.802 | Improved |
| replace NLL with PRO (ablation) | NLL · NQ | AUROC: +0.3% average | Improved |
| PRO versus semantic density (SD) | SD · TriviaQA | AUROC | No clear effect |
| PRO versus semantic density (SD) | SD · NQ | AUROC: +6.5% | Improved |
| use adaptive threshold alpha instead of fixed K=1 (ablation) | PRO · TriviaQA, Gemma-2B | AUROC: 0.806 -> 0.819 | Improved |
| use adaptive threshold alpha instead of best fixed K (ablation) | PRO · TriviaQA, Llama2-13B | AUROC: 0.809 (K=8) vs 0.802 | No clear effect |
| use fixed K=2 instead of K=1 (ablation) | NLL · TriviaQA, Gemma-7B | AUROC: 0.812 -> 0.834 | Improved |
| larger K up to 10 on fixed top-K (ablation) | fixed K · TriviaQA, Gemma-2B | AUROC: 0.806 -> 0.788 (K=10) | Hurt |
| PRO versus baselines under varying ROUGE-L correctness thresholds (ablation) | PRO · SciQ, Gemma-7B | AUROC: 0.5: 0.832 vs NLL 0.823 | Improved |
| larger model in same family | uncertainty estimators · TriviaQA/SciQ/NQ | AUROC | Mixed |
Closest papers by meaning
- Kernel Language Entropy: Fine-grained Uncertainty Quantification for LLMs from Semantic Similarities (NeurIPS 2024 · similarity 0.93)
- Inv-Entropy: A Fully Probabilistic Framework for Uncertainty Quantification in Language Models (NeurIPS 2025 · similarity 0.92)
- FLUE: Streamlined Uncertainty Estimation for Large Language Models (AAAI 2025 · similarity 0.91)
- Efficient semantic uncertainty quantification in language models via diversity-steered sampling (NeurIPS 2025 · similarity 0.91)
- Distinguishing the Knowable from the Unknowable with Language Models (ICML 2024 · similarity 0.91)
- Semantic Density: Uncertainty Quantification for Large Language Models through Confidence Measurement in Semantic Space (NeurIPS 2024 · similarity 0.91)
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.
