for horizontal in range(-10, 11, 1):
for vertical in range(-10, 11, 1):
predicted_values = 0
predict_model = theano.function(inputs=[index],
outputs=layer3.errors(y),
givens={layer0.input:
t_test_set_x[index * 500: (index + 1) * 500],
y: test_set_y[index * 500:
(index + 1) * 500]})
for batch_value in range(0, 20, 1):
temp_predicted_values = predict_model(batch_value)
predicted_values = temp_predicted_values + predicted_values
predicted_values = predicted_values/20
t_test_set_x = theano_translation(test_set_x, horizontal, vertical,
borrow=True)
* I have 3 for loops, and i want to update the test set image through
translation. However, the processing speed decreases as epoch number increses.*
*I think the reason is that numerous theano functions were created and took the
GPU memory.*
*I have checked theano.clone , however, it will still create new theano
function.*
*Is there any good way to update the test set without creating new function?*
*Thanks*
--
---
You received this message because you are subscribed to the Google Groups
"theano-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
