Understanding uncertainties of Bayesan ConvLSTM output

[jekmasto]

Thanks very much for sharing the code, great job.
I'm trying to use your example regarding the prediction of the Lorenz 84 model with Bayesan ConvLSTM.
The example works without any problems, but I didn't understand how to have access to the uncertainties about the model fitness.
Are uncertainties the third output of this line? last_pred, _, _ = model(y_pred,timestep, training=False)

Thank you so much in advance