This dataset contains approximately 137,800 Midjourney 5.1 images from Vodka's publicly released training dataset.
The primary benefit of this dataset is that the content within it will never change.
It is a permanent repository of a large volume of randomly sampled MJ v5.1 images, which may become harder to obtain over time.
Each image in PNG format is available in the split_train directory.
162G split_train - training data / images
How… See the full description on the dataset page: https://huggingface.co/datasets/bghira/mj51-data.
