Regarding the hyperparameter settings for training mini-ImageNet episodes from scratch #18
Description
Dear author, thank you for your prompt response! However, the code you provided does not include the hyperparameter settings for training mini-ImageNet episodes from scratch, which has caused some obstacles in my attempt to reproduce your results. Could you kindly provide the detailed hyperparameter settings for achieving the 1-shot 63.03% and 5-shot 78.01% accuracy on mini-ImageNet as reported in Table 9, when training from scratch? Thank you very much!