Reinforcement Learning for Edit-Based Non-Autoregressive Neural Machine Translation

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.

Filtered Direct Preference Optimization

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.

Regularized Best-of-N Sampling to Mitigate Reward Hacking for Language Model Alignment

Best-of-N (BoN) sampling with a reward model has been shown to be an effective strategy for aligning Large Language Models (LLMs) to human preferences at the time of decoding. BoN sampling is susceptible to a problem known as reward hacking. Because the reward model is an imperfect proxy for the true objective, over-optimizing its value can compromise its performance on the true objective. A common solution to prevent reward hacking in preference learning techniques is to optimize a reward using proximity regularization (e.g., KL regularization), which ensures that the language model remains close to the reference model. In this research, we propose Regularized Best-of-N (RBoN), a variant of BoN that aims to mitigate reward hacking by incorporating a proximity term in response selection, similar to preference learning techniques. We evaluate two variants of RBoN on the AlpacaFarm dataset and find that they outperform BoN, especially when the proxy reward model has a low correlation with the true objective.

On the True Distribution Approximation of Minimum Bayes-Risk Decoding

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.

Return-Aligned Decision Transformer

Traditional approaches in offline reinforcement learning aim to learn the optimal policy that maximizes the cumulative reward, also known as return. However, as applications broaden, it becomes increasingly crucial to train agents that not only maximize the returns, but align the actual return with a specified target return, giving control over the agent's performance. Decision Transformer (DT) optimizes a policy that generates actions conditioned on the target return through supervised learning and is equipped with a mechanism to control the agent using the target return. Despite being designed to align the actual return with the target return, we have empirically identified a discrepancy between the actual return and the target return in DT. In this paper, we propose Return-Aligned Decision Transformer (RADT), designed to effectively align the actual return with the target return. Our model decouples returns from the conventional input sequence, which typically consists of returns, states, and actions, to enhance the relationships between returns and states, as well as returns and actions. Extensive experiments show that RADT reduces the discrepancies between the actual return and the target return of DT-based methods.

Generating Diverse and High-Quality Texts by Minimum Bayes Risk Decoding

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 for generating diverse outputs are predominantly based on beam search or random sampling, thus their output quality is capped by these underlying methods. 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, Diverse MBR (DMBR) and $k$-medoids MBR (KMBR), methods to generate a set of sentences with high quality and diversity. We evaluate DMBR and KMBR on a variety of directed text generation tasks using encoder-decoder models and a large 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.

Model-Based Minimum Bayes Risk Decoding

Minimum Bayes Risk (MBR) decoding has been shown to be a powerful alternative to beam search decoding in a variety of text generation tasks. MBR decoding selects a hypothesis from a pool of hypotheses that has the least expected risk under a probability model according to a given utility function. Since it is impractical to compute the expected risk exactly over all possible hypotheses, two approximations are commonly used in MBR. First, it integrates over a sampled set of hypotheses rather than over all possible hypotheses. Second, it estimates the probability of each hypothesis using a Monte Carlo estimator. While the first approximation is necessary to make it computationally feasible, the second is not essential since we typically have access to the model probability at inference time. We propose Model-Based MBR (MBMBR), a variant of MBR that uses the model probability itself as the estimate of the probability distribution instead of the Monte Carlo estimate. We show analytically and empirically that the model-based estimate is more promising than the Monte Carlo estimate in text generation tasks. Our experiments show that MBMBR outperforms MBR in several text generation tasks, both with encoder-decoder models and with large language models.

Policy Gradient with Kernel Quadrature

Reward evaluation of episodes becomes a bottleneck in a broad range of reinforcement learning tasks. Our aim in this paper is to select a small but representative subset of a large batch of episodes, only on which we actually compute rewards for more efficient policy gradient iterations. We build a Gaussian process modeling of discounted returns or rewards to derive a positive definite kernel on the space of episodes, run an "episodic" kernel quadrature method to compress the information of sample episodes, and pass the reduced episodes to the policy network for gradient updates. We present the theoretical background of this procedure as well as its numerical illustrations in MuJoCo and causal discovery tasks.

On the Depth between Beam Search and Exhaustive Search for Text Generation

Beam search and exhaustive search are two extreme ends of text decoding algorithms with respect to the search depth. Beam search is limited in both search width and depth, whereas exhaustive search is a global search that has no such limitations. Surprisingly, beam search is not only computationally cheaper but also performs better than exhaustive search despite its higher search error. Plenty of research has investigated a range of beam widths, from small to large, and reported that a beam width that is neither too large nor too small is desirable. However, in terms of search depth, only the two extreme ends, beam search and exhaustive search are studied intensively. In this paper, we examine a range of search depths between the two extremes to discover the desirable search depth. To this end, we introduce Lookahead Beam Search (LBS), a multi-step lookahead search that optimizes the objective considering a fixed number of future steps. Beam search and exhaustive search are special cases of LBS where the lookahead depth is set to $0$ and $\infty$, respectively. We empirically evaluate the performance of LBS and find that it outperforms beam search overall on machine translation tasks. The result suggests there is room for improvement in beam search by searching deeper. Inspired by the analysis, we propose Lookbehind Heuristic Beam Search, a computationally feasible search algorithm that heuristically simulates LBS with 1-step lookahead. The empirical results show that the proposed method outperforms vanilla beam search on machine translation and text summarization tasks.

Why Guided Dialog Policy Learning performs well? Understanding the role of adversarial learning and its alternative

Dialog policies, which determine a system's action based on the current state at each dialog turn, are crucial to the success of the dialog. In recent years, reinforcement learning (RL) has emerged as a promising option for dialog policy learning (DPL). In RL-based DPL, dialog policies are updated according to rewards. The manual construction of fine-grained rewards, such as state-action-based ones, to effectively guide the dialog policy is challenging in multi-domain task-oriented dialog scenarios with numerous state-action pair combinations. One way to estimate rewards from collected data is to train the reward estimator and dialog policy simultaneously using adversarial learning (AL). Although this method has demonstrated superior performance experimentally, it is fraught with the inherent problems of AL, such as mode collapse. This paper first identifies the role of AL in DPL through detailed analyses of the objective functions of dialog policy and reward estimator. Next, based on these analyses, we propose a method that eliminates AL from reward estimation and DPL while retaining its advantages. We evaluate our method using MultiWOZ, a multi-domain task-oriented dialog corpus.