Dynamic Reward Adjustment in Multi-Reward Reinforcement Learning for Counselor Reflection Generation

In this paper, we study the problem of multi-reward reinforcement learning to jointly optimize for multiple text properties for natural language generation. We focus on the task of counselor reflection generation, where we optimize the generators to simultaneously improve the fluency, coherence, and reflection quality of generated counselor responses. We introduce two novel bandit methods, DYNAOPT and C-DYNAOPT, which rely on the broad strategy of combining rewards into a single value and optimizing them simultaneously. Specifically, we employ non-contextual and contextual multi-arm bandits to dynamically adjust multiple reward weights during training. Through automatic and manual evaluations, we show that our proposed techniques, DYNAOPT and C-DYNAOPT, outperform existing naive and bandit baselines, demonstrating their potential for enhancing language models.