From the Inside Out: Progressive Distribution Refinement for Confidence Calibration

Read full text Summarise with VIRAT
Authors: Xizhong Yang, Yinan Xia, Huiming Wang, Mofei Song
Cited by: 0
References: 27

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Leveraging the model's internal information as the self-reward signal in Reinforcement Learning (RL) has received extensive attention due to its label-free nature. While prior works have made significant progress in applying the Test-Time Scaling (TTS) strategies to RL, the discrepancy in internal information between test and training remains inadequately addressed. Moreover, Test-Time Training based on voting-based TTS strategies often suffers from reward hacking problems. To address these issues, we propose DistriTTRL, which leverages the distribution prior of the model's confidence during RL to progressively optimize the reward signal, rather than relying solely on single-query rollouts. Additionally, we mitigate the phenomenon of consistent reward hacking caused by the voting-based TTS strategies through diversity-targeted penalties. Benefiting from this training mechanism where model capability and self-reward signals complement each other, and the mitigation of reward hacking, DistriTTRL has achieved significant performance improvements across multiple models and benchmarks.

Fields of Study & Subjects

Reinforcement Learning in Robotics Domain Adaptation and Few-Shot Learning Adversarial Robustness in Machine Learning

Publication details

Journal / source
arXiv (Cornell University)
Publication date
17 March 2026
Language
EN
Document type
Preprint