Open m-ali-awan opened 3 years ago
Right now compute_label_confidence is using self.dl.items. This works fine for from folder structure, but once we are using csvs them dl.items is actually a df, so for filenames we need to do dl.items['filename'], so I think that's why with test_dl with df plot_label_confidence do not work fine.
Kindly correct me if I am wrong,
Thanks
As can be seen in your tutorial nbs, there are percentages on the bars, showing confidence level.. But in my implementation I couldn't find them...
Please help me with this....
Thanks