<Tanja3>
Hi guys, I have a problem that my training takes too much RAM. I need to implemen CNN for traffic sign recognition and I have transformed my data images to .csv format like is done with MNIST dataset. My question is: Is training implemented in a way that it allocate memory for whole dataset at once or is it alocating respectfully with batches with
<Tanja3>
batch_size defined in optimizer?
rcurtin_matrixor has joined #mlpack
Tanja3 has quit [Quit: Client closed]
rcurtin_matrixor has quit [Quit: You have been kicked for being idle]