datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
GPT-Image-Edit-1.5M
GPT-Image-Edit-1.5M A Million-Scale, GPT-Generated Image Dataset
📃Arxiv | 🌐 Project Page | 💻Github
GPT-Image-Edit-1.5M is a comprehensive image editing dataset that is built upon HQ-Edit, UltraEdit, OmniEdit and Complex-Edit, with all output images regenerated with GPT-Image-1.
📣 News
[2025.08.20] 🚀 We provide a script for multi-process downloading. See Multi-process Download.
[2025.07.27] 🤗 We release GPT-Image-Edit, a state-of-the-art image editing model with… See the full description on the dataset page: https://huggingface.co/datasets/UCSC-VLAA/GPT-Image-Edit-1.5M.gpt4o_captions_1k5samples
PACO
WebDataset export for PACO-style localized caption data.
Summary
Samples: 1500
Shards: 2
Payload format inside each shard: pickle
Split: train
Config: PACO
Layout
Media files are stored in WebDataset tar shards.
Each sample key is stable and becomes __key__ in the dataset viewer.
Hugging Face will infer columns such as jpg, pickle, json, __key__, and __url__ from the shard contents.
Manifest file: PACO/annotations.json