Reinforcement learning from human feedback (RLHF) plays a crucial role in aligning language models with human preferences. While the significance of dataset quality is generally recognized, explicit investigations into its impact within the RLHF framework, to our knowledge, have been limited. This paper addresses the issue of text quality within the preference dataset by focusing on direct preference optimization (DPO), an increasingly adopted reward-model-free RLHF method. We confirm that text quality significantly influences the performance of models optimized with DPO more than those optimized with reward-model-based RLHF. Building on this new insight, we propose an extension of DPO, termed filtered direct preference optimization (fDPO). fDPO uses a trained reward model to monitor the quality of texts within the preference dataset during DPO training. Samples of lower quality are discarded based on comparisons with texts generated by the model being optimized, resulting in a more accurate dataset. Experimental results demonstrate that fDPO enhances the final model performance. Our code is available at https://github.com/CyberAgentAILab/filtered-dpo.

One of the most important challenges in text generation systems is to produce outputs that are not only correct but also diverse.Recently, Minimum Bayes-Risk (MBR) decoding has gained prominence for generating sentences of the highest quality among the decoding algorithms. However, existing algorithms proposed to generate diverse outputs are predominantly based on beam search or random sampling, thus their output quality is capped by these underlying decoding algorithms. In this paper, we investigate an alternative approach – we develop diversity-promoting decoding algorithms by enforcing diversity objectives to MBR decoding.We propose two variants of MBR; (i) Diverse MBR (DMBR) that adds a diversity penalty to the decoding objective and (ii) k-medoids MBR (KMBR) that reformulates the decoding task as a clustering problem.We evaluate DMBR and KMBR on a variety of directed text generation tasks using encoder-decoder models and a language model with prompting. The experimental results show that the proposed method achieves a better trade-off than the diverse beam search and sampling algorithms overall.

Minimum Bayes-risk (MBR) decoding has recently gained renewed attention in text generation.MBR decoding considers texts sampled from a model as pseudo-references and selects the text with the highest similarity to the others.Therefore, sampling is one of the key elements of MBR decoding, and previous studies reported that the performance varies by sampling methods.From a theoretical standpoint, this performance variation is likely tied to how closely the samples approximate the true distribution of references.However, this approximation has not been the subject of in-depth study.In this study, we propose using anomaly detection to measure the degree of approximation.We first closely examine the performance variation and then show that previous hypotheses about samples do not correlate well with the variation, but our introduced anomaly scores do.The results are the first to empirically support the link between the performance and the core assumption of MBR decoding.

Non-autoregressive (NAR) language models are known for their low latency in neural machine translation (NMT). However, a performance gap exists between NAR and autoregressive models due to the large decoding space and difficulty in capturing dependency between target words accurately. Compounding this, preparing appropriate training data for NAR models is a non-trivial task, often exacerbating exposure bias. To address these challenges, we apply reinforcement learning (RL) to Levenshtein Transformer, a representative edit-based NAR model, demonstrating that RL with self-generated data can enhance the performance of edit-based NAR models. We explore two RL approaches: stepwise reward maximization and episodic reward maximization. We discuss the respective pros and cons of these two approaches and empirically verify them. Moreover, we experimentally investigate the impact of temperature setting on performance, confirming the importance of proper temperature setting for NAR models’ training.