That option should have been passed in the SVM case, but it doesn’t seem to be used right now:
-classifier.svm.k <string> SVM Kernel Type [linear/rbf/poly/sigmoid] (mandatory, default value is linear)
Random seed used for training
It’s just a seed used to generate random numbers in a reproducible way. Randomness is used for splitting the in-situ data into training and validation sets, and for choosing which pixels to use for training. If you change it you will obtain different results, but it doesn’t fundamentally change the quality of the results.
Ratio between validation and training points
Did you mean “polygons”? That one affects the training / validation mentioned above. With the default of 0.75, 75% of the in-situ data polygons will be used for training, and the rest of them will be used for validation.
Or if there is technical document that puts detailed explanation on each parameter settings of Sen2agro Configurator?
They should be described in the manual, but it’s somewhat lacking in explanations.