jjjj32481

I suggest you to add this sentence to change the dtype of the variable which needs to be sent to the graph model: x = x.to(torch.float16)

Replace the code "day_of_year = data["timestamps"][0].dayofyear / 366.0" in the __iter__() function of the XrDataset with the following lines can tackle this issue. date = data["timestamps"][0] day_of_year = ((date-datetime.datetime(date.year,1,1)).days+1) /...

When the tensorflow and keras version are higher than the requirements provided by author. Please make sure the code in resnet.py: Change **if K.image_dim_ordering() == 'tf'** towards **if K.image_data_format() ==...