Defined values on train and test images

[tacouto]

On script eval_satnerf.py, function predefined_val_ts(img_id), what is the meaning of the atributed values in the test and train images names?
For example,

    elif aoi_id == "JAX_214":
        d = {"JAX_214_020_RGB": 0,
            "JAX_214_006_RGB": 8,
            "JAX_214_001_RGB": 18,
            "JAX_214_008_RGB": 2}
    elif aoi_id == "JAX_260":
        d = {"JAX_260_015_RGB": 0,
             "JAX_260_006_RGB": 3,
             "JAX_260_004_RGB": 10}

For now, it was noticeable that the first image name, with value 0, it is the first image name from train.txt, and the remaining the images are the ones present in test.txt, although i can understand the corresponding values.
What is the procedure to define these values and how can i do it in a personal dataset?