analiza aiqbench
ProbabilityBench introduces a novel benchmark using simulated domains with analytically derived ground-truth probabilities to rigorously evaluate model calibration and uncertainty metrics. By overcoming the limitation of single-label observations in real-world data, it enables direct measurement of reported probability quality via KL divergence across 16 models. This work advances the field by providing a meta-evaluation framework specifically for refining how AI systems quantify their own confidence levels.
ProbabilityBench: Model Evaluation and Calibration with Exact Ground-Truth Uncertainty
arXiv:2610.17016v1 Announce Type: cross Abstract: Evaluating model uncertainty on real-world data is challenging because each instance provides only a single observed label, leaving true conditional class probabilities unknown. To address this, we introduce ProbabilityBench, a benchmark that provides exact, instance-level class probabilities across three simulated text-based domains: peer review, covert information transfer, and credit risk. By explicitly defining the full data-generating process for each environment, we analytically compute both exact class probabilities and Bayes error. Furthermore, we provide a theoretical guarantee that rendering these cases into natural-language memos strictly preserves their underlying generative probabilities. This exactness enables direct measurement of reported probability quality via KL divergence. Across 16 evaluated models, most models on the observed cost-quality Pareto frontiers are decision models, including Jev and Perplexity Decider. Finally, ProbabilityBench serves as a meta-evaluation framework, leveraging known ground-truth values to benchmark the accuracy of standard calibration metrics and Bayes-error estimators themselves.
Abstract:Evaluating model uncertainty on real-world data is challenging because each instance provides only a single observed label, leaving true conditional class probabilities unknown. To address this, we introduce ProbabilityBench, a benchmark that provides exact, instance-level class probabilities across three simulated text-based domains: peer review, covert information transfer, and credit risk. By explicitly defining the full data-generating process for each environment, we analytically compute both exact class probabilities and Bayes error. Furthermore, we provide a theoretical guarantee that rendering these cases into natural-language memos strictly preserves their underlying generative probabilities. This exactness enables direct measurement of reported probability quality via KL divergence. Across 16 evaluated models, most models on the observed cost-quality Pareto frontiers are decision models, including Jev and Perplexity Decider. Finally, ProbabilityBench serves as a meta-evaluation framework, leveraging known ground-truth values to benchmark the accuracy of standard calibration metrics and Bayes-error estimators themselves.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
arXiv:2610.17016 [cs.LG]
(or arXiv:2610.17016v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.17016 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 244 cuvinte
Subiecte conexe
Always ask your agent to be honest
LessWrong · 16 sept. 2026
Introducing CARE-X: Towards Clinically Useful Radiology VLMs with Auxiliary Supervision, Reward-Aligned Learning, and Tool-Augmented Measurement
Microsoft Research · 11 aug. 2026
LLM-as-a-Judge: Score AI Agent Outputs Automatically
OpenRouter · 14 sept. 2026
Jev thinks "I don't know'', but doesn't say it: Introducing Sys1Cal-v1 Dataset for Probability Calibration
arXiv · 29 sept. 2026