In this quick tutorial we'll set up python on a mac os x in order to run ansble from the current folder.

I recently wrote this utility... a mere function, really, to resize and scale a python image at the center. This way, I don't have to worry about how my images are being fed into an AI pipeline, if they all get scaled and cropped the same.

Additionally, while previously I used bash to rename files in a folder to sequential names such as 1.jpg , the current implementation ignores filenames, which saves me time when massaging the data.

From Selectstar.

Popular Public Image Datasets

1. Image Net

Used For: Classification and Detection

By Alexander Van de Kleut. But you may also want to read the VAE by Reza Kalantar.

Motivation

The problem was naming a local module with the same name as an imported module. I had a file called ./tokenize.py and moving it to ./run_tokenize.py solved the problem.

If you see an error similar to this one:

RuntimeError: CUDA out of memory. Tried to allocate 11.88 MiB (GPU 4; 15.75 GiB total capacity; 10.50 GiB already allocated; 1.88 MiB free; 3.03 GiB cached)

then, try reducing the batch size.

    train_loader = DataLoader(
        train_dataset,
        shuffle=False,
        pin_memory=False,
        batch_size=2, # was: 16
        num_workers=12,
    )

ML is taking the world over by storm in 2023. Here is an easy-to follow tutorial in pure python that explains the basics.