yaoyao-liu/meta-transfer-learning

Questions about hard batch

hortune opened this issue · 3 comments

hortune commented

Hi,
I have seen the previous issue that hard batch is not included in the paper. However, there is some questions to ask. What would be the ratio for miniImagenet for the hard batch and normal batch? Do you combine both of them to backpropogate or calculate them separately?

yaoyao-liu commented

Thanks for your interest in our work.

The detailed settings for HT meta batch are available in the extended version on arXiv.

If you have any further questions, feel free to add more comments.

ecada commented

Hello, thank you for sharing the implementation. When do you plan to share the hard batch part of the code?

yaoyao-liu commented

Hello, thank you for sharing the implementation. When do you plan to share the hard batch part of the code?

Thanks for your interest in our work. Currently, I am working on another project so I am not able to promise you an exact date to release the updated code. I'll keep you informed if it is available.