Skip to content

incorrect model for CelebA #48

@euphoria0-0

Description

@euphoria0-0

@scaldas Hi!
I have a question about the model for CelebA dataset.
In the paper, it said that

For the CelebA experiments, we use 10% of the total clients and the same model we described above for FEMNIST.
...
For the FEMNIST experiments in the same figure, we subsample 5% of the data, and use a model with two convolutional layers followed by pooling, and a final dense layer with 2048 units.

However, the code contains 4 convolutional layers and etc.
What is the correct version?
Thanks.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions