PairReranker: Pairwise Reranking for Natural Language Generation

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, textscPairReranker, 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 textscPairReranker, showing strong results, compared with previous baselines. In addition, our textscPairReranker 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.