Open YuxiaoWang-AI opened 11 months ago
Another question is, on which dataset (HOT-Annotated, HOT-Generated, Full Set) were the model checkpoints you provided?
Hi,
1) Yes, the reported results are evaluated on the test split. 2) The model checkpoints we released are trained on HOT-Annotated. We will clarify this on the webpage. Thanks for pointing out.
hello, are the results in the paper based on the test dataset?