We have been reading your paper named TB-Network recently and tested with your code using the model you provided. But we found that the l1 loss we tested is far less than the data provided in your paper. The question is how to measure l1 loss in your paper, are all images normalized to between 0 and 1, or between -1 and 1? Thanks.
We have been reading your paper named TB-Network recently and tested with your code using the model you provided. But we found that the l1 loss we tested is far less than the data provided in your paper. The question is how to measure l1 loss in your paper, are all images normalized to between 0 and 1, or between -1 and 1? Thanks.