Hallucination Detection for Grounded Instruction Generation
Lingjun Zhao, Khanh Nguyen, Hal Daumé · 2023
We investigate the problem of generating instructions to guide humans to navigate in simulated residential environments.A major issue with current models is hallucination: they generate references to actions or objects that are inconsistent with what a human follower would perform or encounter along the described path.We develop a model that detects these hallucinated references by adopting a model pretrained on a large corpus of image-text pairs, and fine-tuning it with a contrastive loss that separates correct instructions from instructions containing synthesized hallucinations.Our final model outperforms several baselines, including using word probability estimated by the instruction-generation model, and supervised models based on LSTM and Transformer.