-
Notifications
You must be signed in to change notification settings - Fork 559
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update min PyTorch version to 1.9, resolve deprecation warnings #1835
Conversation
c0e05e3
to
01f2987
Compare
9767b7d
to
eb023c4
Compare
It seems a bit aggressive to jump from 1.8.1 min req to 1.10 skipping 1.9, and just one month after 1.10 was released. This could mean a lot of folks that are using this in a prod setting and who may be able to upgrade gpytorch but not torch would not be able to use this for a while. Are there any warnings that were issued starting with 1.9? If so coudl we split those out and make a release that requires 1.9, and then merge the rest of this? |
Okay yeah that's fair. There was only one minor 1.10-specific warning. I'll split that out |
eb023c4
to
1c46ed6
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
couple of nits, o/w lgtm!
row_index_block = torch.div(row_index, self.base_lazy_tensor.size(-2), rounding_mode="floor") | ||
col_index_block = torch.div(col_index, self.base_lazy_tensor.size(-1), rounding_mode="floor") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
row_index_block = torch.div(row_index, self.base_lazy_tensor.size(-2), rounding_mode="floor") | |
col_index_block = torch.div(col_index, self.base_lazy_tensor.size(-1), rounding_mode="floor") | |
row_size, col_size = self.base_lazy_tensor.shape[-2:] | |
row_index_block = torch.div(row_index, row_size, rounding_mode="floor") | |
col_index_block = torch.div(col_index, col_size, rounding_mode="floor") |
row_index = torch.div(row_index, self.base_lazy_tensor.size(-3), rounding_mode="floor") | ||
col_index = torch.div(col_index, self.base_lazy_tensor.size(-3), rounding_mode="floor") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
row_index = torch.div(row_index, self.base_lazy_tensor.size(-3), rounding_mode="floor") | |
col_index = torch.div(col_index, self.base_lazy_tensor.size(-3), rounding_mode="floor") | |
block_dim = self.base_lazy_tensor.size(-3) | |
row_index = torch.div(row_index, block_dim, rounding_mode="floor") | |
col_index = torch.div(col_index, block_dim, rounding_mode="floor") |
oops so much for the suggestions, auto-merge |
There are a number of new deprecation warnings due to PyTorch 1.10 (plus some other warnings that we've previously missed). This PR addresses all warnings that pop up in the test suite.
This should hopefully be an easy merge.