HausaNLP at SemEval-2025 Task 3: Towards a Fine-Grained Model-Aware Hallucination Detection
This work addresses the challenge of identifying subtle hallucinations in LLM outputs for NLP researchers, but it is incremental with low performance scores.
The paper tackled the problem of detecting hallucinations and overgeneration errors in large language model outputs, achieving an Intersection over Union score of 0.032 and a correlation score of 0.422 using a fine-tuned ModernBERT model on a synthetic dataset.
This paper presents our findings of the Multilingual Shared Task on Hallucinations and Related Observable Overgeneration Mistakes, MU-SHROOM, which focuses on identifying hallucinations and related overgeneration errors in large language models (LLMs). The shared task involves detecting specific text spans that constitute hallucinations in the outputs generated by LLMs in 14 languages. To address this task, we aim to provide a nuanced, model-aware understanding of hallucination occurrences and severity in English. We used natural language inference and fine-tuned a ModernBERT model using a synthetic dataset of 400 samples, achieving an Intersection over Union (IoU) score of 0.032 and a correlation score of 0.422. These results indicate a moderately positive correlation between the model's confidence scores and the actual presence of hallucinations. The IoU score indicates that our model has a relatively low overlap between the predicted hallucination span and the truth annotation. The performance is unsurprising, given the intricate nature of hallucination detection. Hallucinations often manifest subtly, relying on context, making pinpointing their exact boundaries formidable.