Question about re-fiting with Auto* models for multiple experimental time series

[Teculos]

Preamble
I'm trying to fit models for biological systems where we have a large amount of experimental data, ergo I have multiple multivariate trajectories from the same biological system that I'm trying to fit with a single model; To give some intuition, this could be a model for gene expression of a T-cell over time where we measure this across multiple genetically identical T-cells. The solution that I've arrived at is to iteratively train models on individual experimental trajectories but I can't figure out how re-training works for the optimized variant of models.

Question
If I were to fit and then re-fit a collection of the Auto* models would this effectively fine-tune the models on incoming data or would it re-start the optimization for each new experiment I try to fit on?

PS
If there was a way to include multiple experiments from the same system as separate trajectories when fitting a model this would avoid the issue but I couldn't figure out if that was possible.

Any help would be greatly appreciated <3

[elephaint]

I'm struggling to understand the use-case and question; could you provide a minimal code example explaining what it is you're doing / trying to achieve?

In general if you execute model.fit() where model is an Auto* model the optimization is restarted.