
by Jasper
MONET is a dataset of 103.8 million English image-text pairs for training text-to-image models. It was distilled from 2.9 billion raw pairs through safety, domain and duplicate filtering.
DatasetFree
Step 1: Open the dataset page on Hugging Face and read the card to choose between the parquet and webdataset configs.
Step 2: Install the webdataset and huggingface_hub packages with pip.
Step 3: Load the parquet config for quick browsing with thumbnails, or the webdataset config for full-resolution images.
Step 4: Pick the caption, embedding and score fields you need, and filter by source or by quality scores.
Step 5: Train or fine-tune your text-to-image model on the selected subset.
It contains 103.8 million image-text pairs, selected from 2.9 billion raw pairs across nine open sources.
The card lists Apache-2.0 for the dataset. It also lists the upstream license for each source dataset.
Each image has its original alt-text where available and up to four synthetic captions, ranging from short concept-level text to long fine-grained descriptions.
The card lists English as the language.
Jasper AgentsJasper is an AI marketing platform that combines purpose-built agents, brand context and governance. Its agents run end-to-end marketing workflows such as GEO, research and translation.
Clipdrop APIClipdrop API uses AI for image-processing and generative APIs for background removal, cleanup, relighting, upscaling, and image generation.
JasperJasper is an AI platform for marketing teams that combines brand context, agents and a content editor to produce on-brand campaigns.