Open RogerQi opened 1 year ago
Since the model was made for few-shot learning with a very small number of samples I don't know if there is any point in providing a pre-trained one right? Correct me if I am wrong I am not experienced in few-shot learning.
Is there a plan to release pre-trained weights so that sampling/evaluation can be done without re-training everything from scratch?