Skip to content

Conversation

@cicichen01
Copy link
Contributor

Summary:
As titled.

This changes helps keeping the tests to be an exact mapped of the implementations.
A future follow up could be a consolidation of helpers. (there could be some helpers universal for both attribution and influence)

Differential Revision: D55270852

@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D55270852

Summary:

As titled. The get_random_model_and_data() method is used to construct testing data for influence and it is reported as too complex by flake8 (https://www.flake8rules.com/rules/C901.html). This series of diff will split the method and abstract the common parts.

This diff isolate the model part for different gpu usage settings. It also eliminate the mix usage of bool and str.

Differential Revision: D55165054
Summary:

As titled. use_cpu is supposed to be a bool for datasets. However, it was combined with gpu settings with more informations inside. We separate the variable to clarify the logic.

This change also removes the duplicated data adjust for gpu.

Differential Revision: D55167566
…ry (meta-pytorch#1260)

Summary:

As titled. 

This changes helps keeping the tests to be an exact mapped of the implementations.
A future follow up could be a consolidation of helpers. (there could be some helpers universal for both attribution and influence)

Differential Revision: D55270852
@facebook-github-bot
Copy link
Contributor

This pull request was exported from Phabricator. Differential Revision: D55270852

@facebook-github-bot
Copy link
Contributor

This pull request has been merged in c6e9c22.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants