When using the iid_test_split e.g. then logging the metric wont work because the data loader is only loading context in the first place.
[ ] Suggest quick fix first in spatiotemporal loop (only call metric if t(pred) <= t(targ)
[ ] Also: spatio_temporal.py", line 182, in
json.dump([{k: v if isinstance(v, str) else v.item() for k,v in test_step_outputs[i][j][l].items()} for i in range(len(test_step_outputs)) for j in range(len(test_step_outputs[i])) for l in range(len(test_step_outputs[i][j]))], fp)
AttributeError: 'float' object has no attribute 'item' --> i.e.: this is only save for string and tensors, not for floats (as from ENS)
When using the iid_test_split e.g. then logging the metric wont work because the data loader is only loading context in the first place.