Tag
This paper introduces a temporal multi-signal fusion method for detecting token-level hallucinations in language models using sequence labeling over fused features, achieving improved cross-model performance without access to model internals.
ExtractConf is a confidence estimation method for LLM-based document field extraction that uses two structurally different calls (field-guided and document-guided) to derive disagreement signals, achieving 0.928 ROC AUC on DocILE invoices and enabling reliable selective prediction for high-stakes automation.