README.md

metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: text
      dtype: string
    - name: colors
      dtype:
        array2_d:
          shape:
            - 2
            - 3
          dtype: float32
    - name: conditioning_image
      dtype: image
  splits:
    - name: train
      num_bytes: 66204989
      num_examples: 100000
  download_size: 46802201
  dataset_size: 66204989

Dataset Card for "fill100k"

Kinda similar to https://huggingface.co/datasets/offchan/fill50k but with more images, lower resolution, and uniform color randomization. This is generated using my own script entirely.