I set up a neural network model with a training and test data set and saved the model with the "Save Model" - Widget. I wonder what kind of information the saved pickle-file contains? Does it only contain an neural-network-algorithm with all set parameters or does it depends on the used training data set?
As far as i know, saved model contains the parameters of the trained model. This means, that:
- No, it does not contain the algorithm itself, however it contains all the parameters to run it.
- The values of these parameters definitely depend on the used training set.
- If later loaded, can be used with another data set, which contains "compatible attributes".