felixfeiyu / instruct-pix2pix-dataset Goto Github PK

This repository provides utilities to a minimal dataset for InstructPix2Pix like training for Diffusion models.

1. Download the original dataset as discussed here. I used this version: clip-filtered-dataset. Note that the download can take as long as 24 hours depending on the internet bandwidth. The dataset also requires at least 600 GB of storage.

2. Then run:

   python make_dataset.py --data_root clip-filtered-dataset --num_samples_to_use 1000

3. The make_dataset.py was specifically designed to obtain a 🤗 dataset. So, it's the most useful when you push the minimal dataset to the 🤗 Hub. You can do so by setting push_to_hub while running make_dataset.py.

https://huggingface.co/datasets/sayakpaul/instructpix2pix-1000-samples

The full version of the CLIP filtered dataset used for InstructPix2Pix training can be found here: https://huggingface.co/datasets/timbrooks/instructpix2pix-clip-filtered

With the dataset being on the 🤗 Hub, one can do load the dataset with two lines of code:

from datasets import load_dataset

dataset = load_dataset("timbrooks/instructpix2pix-clip-filtered", split="train")

And voila 🤗

The structure of make_dataset.py is inspired by Nate Raw's notebook.