abdulfatir/prototypical-networks-tensorflow

Reproducing Paper's Results

alirezazareian opened this issue · 1 comments

Dear Abdul,

Your implementation of the mini-imagenet experiment yields an accuracy of 0.61 while it is reported to be 0.68 in the paper. What do you think the issue is? Have you tried to close the gap?

Thanks;
Ali

Hi Ali,

I didn't spend much time trying to reproduce results. The results in this repo are from a random run. Also, note that the mini-Imagenet used by me is different from the one used in the paper.