probly.train.evidential.torch.evidential_nignll_loss¶

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

Evidence-based Normal-Inverse-Gamma (NIG) regression loss.

Implements the negative log-likelihood term used in Deep Evidential Regression as proposed by Amini et al. (2020).

Reference:: Amini et al., “Deep Evidential Regression”, NeurIPS 2020. https://arxiv.org/abs/1910.02600

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.