AnsweredAssumed Answered

Deep Learning error Pro 2.5: Unable to initialize python raster function with scalar arguments

Question asked by anneley_hadland on Apr 3, 2020
Latest reply on Apr 6, 2020 by anneley_hadland

Hi, I'm trying to find the most up to date version of the libraries to install at pro 2.5 to enable the machine learning capabilities. I will be running on GPU in future but just at present I'm trying a quick test on CPU. I'm using the library combination from here: Install deep learning frameworks for ArcGIS—ArcGIS Pro | Documentation 

but this may be a bit out of date now. I'm getting errors with applying the models. Exporting and training seem OK. I saw some other posts saying to try to revert to tensorflow 1.12.0 and this hasn't helped. I guess it may be a CPU compatibility thing?

 

I would like to check that everything is compatible before seeing if it's an error in the processing. I'm getting the following errors:

 

RuntimeError: Error(s) in loading state_dict for DynamicUnet:
size mismatch for layers.11.0.weight: copying a param with shape torch.Size([9, 99, 1, 1]) from checkpoint, the shape in current model is torch.Size([8, 99, 1, 1]).
size mismatch for layers.11.0.bias: copying a param with shape torch.Size([9]) from checkpoint, the shape in current model is torch.Size([8]).

Unable to initialize python raster function with scalar arguments.

Outcomes