pantheonrl.algos.adap.util.kl_divergence

kl_divergence(dist_true, dist_pred)[source]

Wrapper for the PyTorch implementation of the full form KL Divergence :param dist_true: the p distribution :param dist_pred: the q distribution :return: KL(dist_true||dist_pred)

Parameters:
  • dist_true (Distribution) –

  • dist_pred (Distribution) –

Return type:

Tensor