Skip to content

Avoid table construction in KumoRelational relative-time normalization #1068

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.

Sign up for GitHub

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

Open
aw471 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Avoid table construction in KumoRelational relative-time normalization #1068

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 4 additions & 6 deletions sdm/models/kumo/relational/model.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -590,13 +590,11 @@ def _get_rel_time(
rel_time.nanmean(dim=-2, keepdim=True).nan_to_num(0.0),
rel_time,
)
rel_time = standardizer.fit_transform(
TableTensor.from_tensor(rel_time)
).numerical
standardizer._fit_tensor(rel_time)
standardizer._set_fitted(rel_time.device)
rel_time = standardizer._transform_tensor(rel_time)
else:
rel_time = standardizer.transform(
TableTensor.from_tensor(rel_time)
).numerical
rel_time = standardizer._transform_tensor(rel_time)
rel_time[na_mask] = 0.0

return rel_time
16 changes: 12 additions & 4 deletions sdm/processing/numerical/standardize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@
# SPDX-License-Identifier: Apache-2.0

import torch
from torch import Tensor

from sdm import Stype, TableTensor
from sdm.processing import InvertibleMixin, Processor
Expand Down Expand Up @@ -40,9 +41,12 @@ def _fit(
generator: torch.Generator | None = None,
) -> None:

finite = _isfinite(table.numerical)
self._fit_tensor(table.numerical)

def _fit_tensor(self, numerical: Tensor) -> None:
finite = _isfinite(numerical)
count = finite.sum(dim=-2, keepdim=True)
finite_or_nan = table.numerical.masked_fill(~finite, torch.nan)
finite_or_nan = numerical.masked_fill(~finite, torch.nan)

self.mean = finite_or_nan.nansum(-2, keepdim=True).div_(count)
self.mean.masked_fill_(self.mean.isnan(), 0.0)
Expand All @@ -59,8 +63,12 @@ def _fit(
self.scale += self.eps

def _transform(self, table: TableTensor) -> TableTensor:
numerical = (table.numerical - self.mean).div_(self.scale)
return table.replace_blocks(numerical=numerical)
return table.replace_blocks(
numerical=self._transform_tensor(table.numerical)
)

def _transform_tensor(self, numerical: Tensor) -> Tensor:
return (numerical - self.mean).div_(self.scale)

def _inverse_transform(self, table: TableTensor) -> TableTensor:
dtype = table.numerical.dtype
Expand Down