From RAG to Reality: Coarse-Grained Hallucination Detection via NLI Fine-Tuning

We present our submission to SciHal Subtask 1: coarse-grained hallucination detection for scientific question answering.We frame hallucination detection as an NLI-style three-way classification (entailment, contradiction, unverifiable) and show that simple fine-tuning of NLIadapted encoder models on task data outperforms more elaborate feature-based pipelines and large language model prompting.In particular, DeBERTa-V3-large, a model pretrained on five diverse NLI corpora, achieves one of the highest weighted F1 scores on the public leaderboard.We additionally explore a pipeline combining joint claim-reference embeddings and NLI softmax probabilities fed into a classifier, but find its performance consistently below direct encoder fine-tuning.Our findings demonstrate that, for reference-grounded hallucination detection, targeted encoder fine-tuning remains a competitive approach.