configuration of the pure convolution-based EfficientMod

[Mayner0220]

Thank you for your great ideas and work.

I want to know the configuration(ex. depths, mlp_ratio, context_size) of the pure convolution-based EfficientMod used in your experiments deploying and optimizing EfficientMod for mobile device in your paper.

Please share not only the configuration of EfficientMod-xxs-Conv used in the experiments but also any configurations not disclosed in the paper or repository.

Looking forward to your response. Thank you!