Several times when I use a smaller dataset (few hundred particles) to test the training, I get the following error:
ValueError: Cannot feed value of shape (74, 64, 64, 1) for Tensor u'Placeholder_2:0', which has shape '(100, 64, 64, 1)'
In this case, I can either manually change the model_input_size variable to 74, 64, 64, 1 from the default 100, 64, 64, 1 set in the script, or use a larger dataset. I never run into this issue when training on large (1000s of particles) datasets. Do you know what could be causing this issue?
I'm using CUDA 9 with the latest tensorflow. I did have to change a few of the api call names/variables, but over all it works well.
Several times when I use a smaller dataset (few hundred particles) to test the training, I get the following error:
ValueError: Cannot feed value of shape (74, 64, 64, 1) for Tensor u'Placeholder_2:0', which has shape '(100, 64, 64, 1)'
In this case, I can either manually change the model_input_size variable to 74, 64, 64, 1 from the default 100, 64, 64, 1 set in the script, or use a larger dataset. I never run into this issue when training on large (1000s of particles) datasets. Do you know what could be causing this issue?
I'm using CUDA 9 with the latest tensorflow. I did have to change a few of the api call names/variables, but over all it works well.