On the Benefits of Fine-Grained Loss Truncation: A Case Study on Factuality in Summarization

Anonymous

Abstract: Text summarization and simplification are widely used applications of AI. However, such models are often prone to hallucination, which can result from training models on unaligned data. One of the prominent approaches to address this issue has been Loss Truncation (LT) (Kang and Hashimoto, 2020), an approach to modify the standard log loss to adaptively remove noisy examples during training. However, we find that LT alone yields a considerable number of hallucinated entities on various datasets. We study the behavior of the underlying losses between factual and non-factual examples, to understand and refine the performance of LT. We demonstrate that LT's performance is limited when the underlying assumption that noisy targets have higher NLL loss is not satisfied, and find that word-level NLL among entities provides better signal for distinguishing factuality. We then leverage this to propose a fine-grained NLL loss and fine-grained data cleaning strategies, and observe improvements in hallucination reduction across some datasets.

Paper Type: short

Research Area: Summarization

Contribution Types: Model analysis & interpretability

Languages Studied: English

Consent To Share Submission Details: On behalf of all authors, we agree to the terms above to share our submission details.