Partial dependence plots for classification

[Mosen111]

Hello. I am using h2o r package, deep learning. Would appreciate it if you could answer these questions:

1- For a classification task, I expected to get partial dependence plots where the probability of belonging to a category is shown on the y-axis, but the label of the y-axis is "mean-response", which does not make sense to me.

2- For binary classification, where the categories are represented by numbers (e.g., 0 and 1), in addition to the problem mentioned above, it is not clear to me membership in which category membership is being plotted.

3- For multinomial tasks, to get a partial plot, you need to specify a target category. My question is, what reference category is used? The specified category is plotted against all other categories, or a single reference category?

4- Can you explain the difference between the h2o.pd_plot and h2o.partialPlot functions? The manual says that both produce partial dep plots, why do we need 2 separate functions?

thanks