The code for the paper "Understanding Dataset Difficulty with V-Usable Information" #11
Description
I attempted to use a text classification dataset for my experiments. However, in my final prediction file, the predicted_label is always 1, and the PVI for most instances is the same value. Could you help me understand this issue?