Parameter to choose the amount of data to train?

Part 1 (2018)
1

data = ImageClassifierData.from_paths(PATH, tfms=tfms)
learn = ConvLearner.pretrained(arch, data, precompute=True)

Here all the images in the PATH will be used to train the model.

@jeremy

Is there any parameter to choose the percentage of images from PATH for training.

This might be helpful for some quick testing or to those working on CPU.

2

You could always use a python script to copy a random subset of the images in PATH to some other PATH.