Closed ayushkarnawat closed 4 years ago
Generally, I think this is fine, however the user has to be careful. Instead of
# Example: two different modules point to the same variable name; can lead to errors
import torch
from profit.models import torch
# This will not work
X = torch.Tensor([3.4, 8.6, 9.0])
# Traceback (most recent call last):
# File "<stdin>", line 1, in <module>
# AttributeError: module 'profit.models.torch' has no attribute 'Tensor'
# This will work
model = torch.SequenceVAE(seqlen=56, vocab_size=20, hdim=50, latent_size=20)
the user should import local torch models as the following
import torch
from profit.models.torch import SequenceVAE
# This will now work
X = torch.Tensor([3.4, 8.6, 9.0])
# So will this
model = SequenceVAE(seqlen=56, vocab_size=20, hdim=50, latent_size=20)
What does this PR do?
Renames all instances of "pytorch" to "torch".