Closed merceyz closed 7 years ago
After adding some more training data DeepCL started crashing during training, turns out batch * batchSize caused an integer overflow.
batch * batchSize
This PR fixes that by using a 64bit int instead of a 32bit one
@hughperkins Anything preventing this from getting merged?
I didnt notice it basically :-) Thanks! :-)
After adding some more training data DeepCL started crashing during training, turns out
batch * batchSize
caused an integer overflow.This PR fixes that by using a 64bit int instead of a 32bit one