Section 16.7 - MLP results differ from the book and are similar to TempCNN results, contradicting the text

[biancazadrozny]

This is the result I get from running the MLP code in Python:

This is very similar to the TempCNN results, with the SoyCotton prediction for 2004. In this case, the random seed is set, so this is not the problem.