Proximal Policy Optimization (PPO)

Adaptive KL Penalty

PPO variant that dynamically adjusts the KL penalty strength based on the observed divergence between policies, ensuring controlled updates.

← Back