How to define probability in a multi-dataset training

[mark1ng123]

Hey there,
Currently working on agnostic - segmentation model, i have multiple datasets but for the example lets say i have two different datasets.
Dataset1 and Dataset2, both datasets are annotated with the same classes, but the number of images in Dataset1 is significantly larger.
My goal is to achieve a bigger sampling proportion from Dataset2, can i set the probabilty of dataset2 higher the dataset1 when it comes to picking an image sample for training?
Thanks in advance,
Mark