Official implementation of ERPO: Entropy-Regularized Rank-Masked Policy Optimization for Test-Time RL in Code Generation (EMNLP 2026 Oral).
-
Updated
Oct 16, 2026 - Python
Official implementation of ERPO: Entropy-Regularized Rank-Masked Policy Optimization for Test-Time RL in Code Generation (EMNLP 2026 Oral).
Tensorflow framework implementing classical and advanced state-space filtering methods, including Kalman filters, particle filters, particle flow (EDH/LEDH, invertible PF-PF, kernel-embedded), and differentiable particle filtering with entropy-regularized optimal transport (Sinkhorn).
Soft Actor-Critic (SAC) maximum entropy policy value and soft Bellman target evaluator
Entropy-regularized Soft Actor-Critic (SAC) exploration engine balancing expected return and policy entropy to prevent early convergence.
To associate your repository with the entropy-regularization topic, visit your repo's landing page and select "manage topics."