Multimodal reasoning on 1.8M examples
1.8M multimodal examples with detailed reasoning distilled from Qwen3-VL-235B-Thinking.
- Visual question answering
- Question answering
- Text generation
- 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
1.8M multimodal examples with detailed reasoning distilled from Qwen3-VL-235B-Thinking.
173,451 training examples across 11 tasks, including 56,971 new Dronemillion images.
6k+ top-voted NSFW images from CivitAI with prompt and NSFW-level metadata.
Data from Alibaba and Chinese universities for training and evaluating models processing simultaneous video and voice streams.
2M synthetic examples for map QA and path tracing in English.
Printed sheet music images with region annotations for testing optical music recognition systems.
Annotated dataset with photos of early school exam pages for evaluating multimodal OCR and VQA.
Collection of multimodal training annotations for the Nemotron Super 3.5 VL model.
400,476 videos with previews, view counts, explicit flags, and captions for video-text models.
Image and caption pairs with identical words in different orders to test vision-language models.
Over a million images of complex mathematical expressions for training OCR and image-to-text systems.
Large multimodal dataset of math problems with text and images in English and Chinese.
Image-based questions to evaluate atomic world knowledge in multimodal models.
A large-scale STAIR Captions dataset with 820,310 Japanese image captions for text-to-image and image-to-text generation tasks.
114 days of one person's daily life with video frames, gaze tracking, audio, and activity annotations.
A large updated English image-text pair dataset used to improve InternVL3.5 models.
Benchmark for evaluating the accuracy of structured data extraction from resumes by JSON Schema.
Multimodal English captions for image-text alignment and training captioning models.
Image-text question-answer pairs in nine languages, including Russian, for evaluating and training multimodal models.
Synthetic business PDFs and nearly 10,000 questions on tables, charts, forms, and text for evaluating Document QA systems.
PubMed Central page OCR markup with bounding boxes for layout-aware tasks.
A collection of 38,870 cleaned and verified image-question-answer pairs for training reasoning multimodal models.
A compact set of challenging vision-language tasks with reference answers in four languages.
Dataset with videos and text questions from NVIDIA for road anomaly analysis, accepted at NeurIPS 2026.