What is the point of the hrs_each_step variable? about climax

[tung-nd]

Hi Stephen,

Thank you for your interest in ClimaX. Please checkout the Forecast dataset https://github.com/microsoft/ClimaX/blob/main/src/climax/pretrain/dataset.py#L67 for a better understanding of the arguments. Basically, hrs_each_step indicates the time difference between two consecutive recorded data points in the dataset, and is used to compute the lead_times vector correctly. Therefore, this value is a characteristic of a dataset and is a fixed value (1 for ERA5 and 6 for CMIP6). To specify how far the model should forecast into the future, you should use max_predict_range instead. For example, if you want to make a 24-hr forecast, then you should set max_predict_range=24 and hrs_each_step=1.