Running on Win10 laptop w/ NVidia card

[csherwood-usgs]

1. Update the driver for the NVidia card
   I have a Quadro P2000 and installed 471.11
2. If you don't have MS VisualStudio, install that.
   I installed the 2019 Community Version with no workloads or other options.
3. Install Cuda.
   https://developer.nvidia.com/cuda-downloads?target_os=Windows&target_arch=x86_64
4. Follow instructions here to enable long path names in Win10
   https://knowledge.autodesk.com/support/autocad/troubleshooting/caas/sfdcarticles/sfdcarticles/The-Windows-10-default-path-length-limitation-MAX-PATH-is-256-characters.html
5. Install the environment.yml that Rich posted in previous issue. But use conda install instead of mamba. When I used mamba, I got tf 2.3
6. Install conda install jupyter. Not sure if this is missing from the env or the install got messed up.
7. Modify the notebook OWG_train slightly: remove the magic command in cell 3. It does not work in this context:
   #%tensorflow_version 2.x. I get tf 2.5.0 when this cell executes.
   Also add a %%time command to the top of cell 19 that does the training. Mine took an hour to train.

[dbuscombe-usgs]

Thanks. I didn't realize long path names were an issue so I can change that part of the code next time I look at it. I'll also test and incorporate Rich's yaml file but haven't yet looked into the relative merits of mamba and conda

[csherwood-usgs]

The long path name problem arose during the Python environment installation. I don't think you need to revise the code here.

[dbuscombe-usgs]

Ah, gotcha