PairReranker: Pairwise Reranking for Natural Language Generation
This addresses performance bottlenecks in NLG tasks like summarization and translation, offering a flexible solution that can enhance even large models like GPT-3, though it is incremental as it builds on existing reranking ideas.
The paper tackles the problem of suboptimal outputs in natural language generation tasks by proposing PairReranker, a method that uses pairwise reranking to select the best candidate from multiple decoding results, achieving improvements such as 24.55% on CommonGen and 11.35% on WMT18 zh-en.
Pre-trained language models have been successful in natural language generation (NLG) tasks. While various decoding methods have been employed, they often produce suboptimal results. We first present an empirical analysis of three NLG tasks: summarization, machine translation, and constrained text generation. We found that selecting the best output from the results of multiple decoding methods can significantly improve performance. To further improve reranking for NLG tasks, we proposed a novel method, \textsc{PairReranker}, which uses a single encoder and a pairwise loss function to jointly encode a source input and a pair of candidates and compare them. Experiments on three NLG tasks demonstrated the effectiveness and flexibility of \textsc{PairReranker}, showing strong results, compared with previous baselines. In addition, our \textsc{PairReranker} can generalize to significantly improve GPT-3 (text-davinci-003) results (e.g., 24.55\% on CommonGen and 11.35\% on WMT18 zh-en), even though our rerankers are not trained with any GPT-3 candidates.