evidential_nignll_loss

probly.losses.torch.evidential_nignll_loss(inputs: dict[str, Tensor], targets: Tensor) → Tensor[source]

Evidential Normal-Inverse-Gamma regression loss from [ASSR20].

Implements the negative log-likelihood term used in Deep Evidential Regression with a Normal-Inverse-Gamma (NIG) distribution.

Parameters:

  • inputs – Dictionary containing NIG distribution parameters with keys "gamma", "nu", "alpha", and "beta", each of shape (B,).

  • targets – Ground-truth regression targets, shape (B,).

Returns:

Scalar NIG negative log-likelihood loss averaged over the batch.