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