Abstract:We introduce a new architecture for Distributional Reinforcement Learning (DistRL) that models return distributions using normalizing flows. This approach enables flexible, unbounded support for return distributions, in contrast to categorical approaches like C51 that rely on fixed or bounded representations. It also offers richer modeling capacity to capture multi-modality, skewness, and tail behavior than quantile based approaches. Our method is significantly more parameter-efficient than categorical approaches. Standard metrics used to train existing models like KL divergence or Wasserstein distance either are scale insensitive or have biased sample gradients, especially when return supports do not overlap. To address this, we propose a novel surrogate for the Cram\`er distance, that is geometry-aware and computable directly from the return distribution's PDF, avoiding the costly CDF computation. We test our model on the ATARI-5 sub-benchmark and show that our approach outperforms PDF based models while remaining competitive with quantile based methods.