Research map / Large language models / Language, safety and interpretability
Addressing Pitfalls in the Evaluation of Uncertainty Estimation Methods for Natural Language Generation
Mykyta Ielanskyi, Kajetan Schweighofer, Lukas Aichberger, Sepp Hochreiter
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 |
|---|---|---|---|
| select correctness function adversarially for UE methods (ablation) | UE methods (e.g. G-NLL, Min Token Log-Prob) · QA datasets | Top-3 frequency: G-NLL 0.375 -> 0.688; Min Token Log-Probability 0.125 -> 0.500 | Improved |
| marginalize over multiple LLM judges (SP-MoJI, 4 judges) instead of a single judge (ablation) | UE evaluation · QA datasets | std of performance estimator: SD up to 0.04 with one judge; twofold lower with 4 judges | Improved |
| add more judge calls beyond about 10 (ablation) | SP-MoJI · QA datasets | std of performance estimator | No clear effect |
| apply length normalization to UE scores | Predictive Entropy / Semantic Entropy / Perplexity · QA, code, constrained text | AUROC / Elo | Hurt |
| apply length normalization to UE scores | Perplexity, Semantic Entropy · Perturbation detection | Elo | Improved |
| adjust token likelihoods with semantic importance (TokenSAR/SAR) | SAR family · Multiple tasks | Elo | Hurt |
| evaluate pretrained instead of instruction-tuned models | UE methods · Elo partitions | Elo | Hurt |
Closest papers by meaning
- Improving Uncertainty Estimation through Semantically Diverse Language Generation (ICLR 2025 · similarity 0.94)
- To Believe or Not to Believe Your LLM: Iterative Prompting for Estimating Epistemic Uncertainty (NeurIPS 2024 · similarity 0.91)
- Forking Paths in Neural Text Generation (ICLR 2025 · similarity 0.90)
- Efficient semantic uncertainty quantification in language models via diversity-steered sampling (NeurIPS 2025 · similarity 0.90)
- FLUE: Streamlined Uncertainty Estimation for Large Language Models (AAAI 2025 · similarity 0.90)
- Kernel Language Entropy: Fine-grained Uncertainty Quantification for LLMs from Semantic Similarities (NeurIPS 2024 · similarity 0.90)
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.
