Mentions légales du service

Skip to content

Use torch.exp() instead of numpy.exp()

Since the param_vals here(returned by the network) are a torch tensor

Fixes a bug when running prediction using GPUs, since to call np.exp() on a Tensor we'd have to copy it to the CPU first (which could also be done), but there's no reason to numpify it.

Edited by E. Madison Bray

Merge request reports