Inferencing on feedback model

[ash2703]

Hey,
Really interesting results with the ROIPool model, how do we reproduce the results for the feedback model
Is it possible to share the weight file for the same?

Thankyou

[ash2703]

[UPDATE]
I found the FM model weights, although i am not sure how the weights are smaller than ROIpool model?
Am i missing some implementation details