Closed craymichael closed 2 weeks ago
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request was exported from Phabricator. Differential Revision: D65618314
This pull request has been merged in pytorch/captum@7a12f69cdc6bf6c71b4f1e9bd7785e39189d873d.
Summary: PyTorch 2.6.0 is more strict about safe globals with weights-only pickle loading. To resolve we need to add certain safe globals from the NumPy library.
Disable test_contribution in insights
test_multi_features
for now as it fails with PyTorch 2.6.0.Differential Revision: D65618314