| title | The First Token Knows: Single-Decode Confidence for Hallucination Detection |
|---|---|
| authors | Mina Gabriel |
| date | 2026-05-06 |
| venue | arXiv (preprint, not peer reviewed) |
| id | arXiv:2605.05166 |
| affiliation | Temple University |
| upvotes | 1 (Hugging Face Daily Papers, 2026-11-01) |
| links | abs pdf hf |
| abstract | Self-consistency detects hallucinations by generating multiple sampled answers to a question and measuring agreement, but this requires repeated decoding and can be sensitive to lexical variation. Semantic self-consistency improves this by clustering sampled answers by meaning using natural language inference, but it adds both sampling cost and external inference overhead. We show that first-token confidence, phi_first, computed from the normalized entropy of the top-K logits at the first content-bearing answer token of a single greedy decode, matches or modestly exceeds semantic self-consistency on closed-book short-answer factual question answering. Across three 7-8B instruction-tuned models and two benchmarks, phi_first achieves a mean AUROC of 0.820, compared with 0.793 for semantic agreement and 0.791 for standard surface-form self-consistency. A subsumption test shows that phi_first is moderately to strongly correlated with semantic agreement, and combining the two signals yields only a small AUROC improvement over phi_first alone. These results suggest that much of the uncertainty information captured by multi-sample agreement is already available in the model's initial token distribution. We argue that phi_first should be reported as a default low-cost baseline before invoking sampling-based uncertainty estimation. |
cite
@misc{gabriel2026first,
title={The First Token Knows: Single-Decode Confidence for Hallucination Detection},
author={Mina Gabriel},
year={2026},
eprint={2605.05166},
archivePrefix={arXiv},
url={https://arxiv.org/abs/2605.05166}
}