If we already have 50 data points, is there a difference between training the model on all data at once versus using an active learning loop to select batches from the same data？？

[lin-199612]