eazygrad.nll_loss¶ eazygrad.nll_loss(predicted: _Tensor, target: _Tensor) → _Tensor[source]¶ Compute the negative log-likelihood loss from log-probabilities. Parameters: predicted (_Tensor) – Log-probabilities of shape (N, C). target (_Tensor) – Integer class indices of shape (N,). Returns: _Tensor – Scalar loss tensor. See also torch.nn.functional.nll_loss