Everything runs smooth until I start the train_network function. It loads the data and the parameters, and the line sticks on “Starting training…” for about 10-15 minutes before aborting without any error message given. Furthermore, it also seems like it kicks me out of python entirely.
Running windowsCPU with python 3.69
This coulde be several tings. Either TensorFlow issue. Can you perhaps change the settings do display every step during training (to see whether there is any progress in 10minutes). For testing you could also reduce the global_scale variable to make training much faster (to see if your machine is perhaps not strong enough). Perhaps also check this link: https://github.com/AlexEMG/DeepLabCut/wiki/Troubleshooting-Tips
Thank you for helping out.
I reduced global_scale to 0.6 for the time being and will do some testing. But how do I make it display the training steps?
You can set these: displayiters=100,saveiters=15000
Here: https://github.com/AlexEMG/DeepLabCut/blob/master/docs/functionDetails.md#g-train-the-network