Hallucination Detection for Grounded Instruction Generation

Lingjun Zhao, Khanh Xuan Nguyen, Hal Daumé III

Submission Type: Regular Short Paper

Submission Track: Language Grounding to Vision, Robotics and Beyond

Keywords: Hallucination detection, multimodality, natural language generation

Abstract: 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 pre-trained 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.

Submission Number: 4161