My computer is CPU i5 4590,gtx750,RAM 8G,only has 640 Stream processors.
I want to train a model of ocr which the datasets has 3.6 million images.
There are 3.6 million images in total, with the image resolution of 280*32 and the file size of about 8.6GB.
If I trained with my pc.How long would it take? One year ?I just dont konw.
You should write the training routine using your desired model, profile a few iterations, and try to estimate how long the entire training might take. Everything else would be just a wild guess as we don’t even know which kind of model you are using, what preprocessing is applied etc.