Questions about notebook example code

[HelloWorldLTY]

Hi, thanks for your interesting work. According to the example notebook, we can have a print_metrics function to help us visualize prediction results. However, I tried this method but nothing happened:

You can see that if I directly visualize linprobe_eval_metrics, there is not issue.

Furthermore, I think in the example code, the dataset is only splitted into training and testing datasets. Does this mean you do not use validation dataset in the application stage? If so, how to tune the hyper-parameter of downstream classifier?

Thanks a lot.