Paired images and prompts for RAG
A dataset of 4,000 pairs of PNG images and English prompts for multimodal retrieval.
- English
Hugging Face datasets with plain-language descriptions: what's inside and what models to train on them. 4,788 in the catalog so far, with new ones added every hour.
Answered by Gemini 4 · the chat is kept only in this tab
A dataset of 4,000 pairs of PNG images and English prompts for multimodal retrieval.
1.4M English text-image pairs with high artistic style consistency.
A collection of school math problems with images in Chinese for training and evaluating multimodal models.
A small dataset of Pokemon images (under 1,000 items) re-captioned with GPT-4 Turbo for image generation.
A subset of 500 English question-answer pairs for document scans to evaluate multimodal retrieval and VQA.
A collection of 4,172 scanned engravings and drawings from old books with text captions for image generation and analysis.
Hundreds of thousands of images with detailed texts averaging 200 words and original speech transcripts.
Between 100K and 1M synthetic portraits from Microsoft FaceSynthetics with SPIGA facial landmark annotations and BLIP captions.
English image and metadata dataset from an ICML 2024 paper for safety fine-tuning of VLMs.
A small set of multimodal feedback for aligning vision-language models.
An array of 1 to 10 million pairs of SFW anime-style images and text descriptions.
Boson thermal camera image dataset for image-to-image and geolocation tasks.
10 to 100 thousand frames from long multi-camera videos with frame-by-frame annotation of falls and actions, including wheelchair users.
A corpus of 1,457 images with Hayashi scale acne severity assessment and nearly 19k lesion boxes.
Recompressed Danbooru 2023 version: cleaned images up to 4 MP for training generative models.
Over 170k pairs of geometric images and texts for training multimodal models.
1,639 English website screenshots with instructions and pixel targets for web navigation verification.
Formatted corpus with images and text for visual SFT fine-tuning via TRL.
Large image-text pair corpus with detailed instructions for training CV models for zero-shot generalization.
Set of 3,600 global images with reference descriptions in 36 languages.
A curated set of images and text for testing VLMs without multiple-choice inflation or noise.
Over a million records with sky images and tabular data on solar energy generation for short-term forecasting.
Generated metadata and vector representations for 12 million images from the DataComp subset.
A large dataset of artworks from the Metropolitan Museum for multi-label classification by style and materials.