TECP: Token-Entropy Conformal Prediction for LLMs
It addresses a critical challenge for trustworthy AI by providing a principled solution for uncertainty quantification in black-box LLM settings, though it is incremental as it builds on conformal prediction.
The paper tackled uncertainty quantification for open-ended language generation in black-box LLMs by introducing Token-Entropy Conformal Prediction (TECP), which achieved reliable coverage and compact prediction sets, outperforming prior methods on benchmarks like CoQA and TriviaQA.
Uncertainty quantification (UQ) for open-ended language generation remains a critical yet underexplored challenge, especially under black-box constraints where internal model signals are inaccessible. In this paper, we introduce Token-Entropy Conformal Prediction (TECP), a novel framework that leverages token-level entropy as a logit-free, reference-free uncertainty measure and integrates it into a split conformal prediction (CP) pipeline to construct prediction sets with formal coverage guarantees. Unlike existing approaches that rely on semantic consistency heuristics or white-box features, TECP directly estimates epistemic uncertainty from the token entropy structure of sampled generations and calibrates uncertainty thresholds via CP quantiles to ensure provable error control. Empirical evaluations across six large language models and two benchmarks (CoQA and TriviaQA) demonstrate that TECP consistently achieves reliable coverage and compact prediction sets, outperforming prior self-consistency-based UQ methods. Our method provides a principled and efficient solution for trustworthy generation in black-box LLM settings.