Return to Issue Details Information-Bottleneck Policy Optimization for Stable Language-Model Reinforcement Learning Download Download PDF