Working with multiple experiments

Easy experimenting is a key feature of the Peltarion Platform. You can experiment and optimize your AI-models in many ways. A great way to do this by working with multiple experiments. Iterate on an experiment to resume training with different settings, or Duplicate the experiment and make deeper changes to its architecture.

How to iterate on an experiment

Go back to the Modeling view while a model is running or has already finished its training, and click on Iterate.

Continue training

Continue training lets you train the same model for more epochs starting from the select Checkpoint, and let you change some of the Settings.

Reuse part of model

Reuse part of model creates a new experiment with a single block that contains the model you just trained. This is useful to build another model around the current one.
For instance, you might train an autoencoder, then reuse just the encoding part of this model as a single block inside a new model.

If a model has one or more Output blocks, you will be able to decide whether you want the new model block to output values from one of these blocks or from the Target block.

How to duplicate experiment

You can also make bigger changes to your existing model. Go back to the Modeling view, and click on Duplicate.

This will create a copy of your current model that you can edit, but training progress will be lost.

Continue experimenting in the new experiment and then keep duplicating, tweaking and running until you are happy with the result!

Experiments in the Modeling view
Figure 1. Experiments in the Modeling view

A duplicated model will have a link to its parent experiment in the status bar.

Link to parent experiment
Was this page helpful?
YesNo