openclimatefix/nimrod-uk-1km-validation · Really donnot know how to use these files? hope someone guild me read these file and dive into the training!

Hi, sorry for the delay. For these files, as they are in Parquet format, you can load them using

from datasets import load_dataset

dataset = load_dataset("openclimatefix/nimrod-uk-1km-validation")

for batch in iter(dataset):
    out = model(batch)

with the datasets library. The examples are 256x256 tiles over the UK of the precipitation rainfall rate at 5 minute intervals, and in chunks of 24 frames per example. This is the validation set from the DeepMind Deep Generative Model of Radar paper and publicly released dataset. There is also the test dataset available under openclimatefix/nimrod-uk-1km-test. The training set is still being processed, and we haven't converted it to Parquet yet. But all the splits are available under openclimatefix/nimrod-uk-1km.