EMMeT

Embedding Mixup for Meta-Training (EMMeT), a novel regularization technique that creates new episodic batches through an embedding shuffling and averaging regime.

Weights for the CONV4 and DINO-S models are provided for EMMeT evaluation.

For usage and instructions, please refer to the Github repository: https://github.com/Brikwerk/EMMeT

Model Date

February 2023