Triviality Corrected Endogenous Reward

TL;DR

This paper introduces TCER, a novel reward mechanism for reinforcement learning in open-ended text generation that mitigates triviality bias and improves output diversity and quality across tasks.

Contribution

We propose TCER, a new endogenous reward method that rewards information gain relative to a reference policy, addressing triviality bias in open-ended text generation.

Findings

TCER improves diversity and content quality in open-ended text generation.

TCER outperforms baseline methods across multiple benchmarks.

TCER effectively transfers to mathematical reasoning tasks.

Abstract

Reinforcement learning for open-ended text generation is constrained by the lack of verifiable rewards, necessitating reliance on judge models that require either annotated data or powerful closed-source models. Inspired by recent work on unsupervised reinforcement learning for mathematical reasoning using confidence-based endogenous rewards, we investigate whether this principle can be adapted to open-ended writing tasks. We find that directly applying confidence rewards leads to Triviality Bias: the policy collapses toward high-probability outputs, reducing diversity and meaningful content. We propose TCER (Triviality Corrected Endogenous Reward), which addresses this bias by rewarding the relative information gain between a specialist policy and a generalist reference policy, modulated by a probability-dependent correction mechanism. Across multiple writing benchmarks and model architectures, TCER achieves consistent improvements without external supervision. Furthermore, TCER also transfers effectively to mathematical reasoning, validating the generality of our approach across different generation tasks.