Dataset · osunlp
Multimodal dataset for Mind2Web web agents
osunlp/Multimodal-Mind2Web
HTML pages with screenshots and instructions for training web agents to perform tasks on websites.
What's inside
Each HTML document from Mind2Web is paired with a screenshot of the corresponding webpage. Data is in English, size ranges from 10K to 100K examples; this is a multimodal version of Mind2Web. It eliminates the need to download images from the full raw dump of about 300 GB.
What you can do with it
- Train web agents that follow language instructions
- Evaluate multimodal models on website navigation tasks
- Train LoRA for agents with HTML and screenshots
- Languages
- English
- Size
- 10K–100K rows
- Downloads this month
- 12.4K
- Likes
- 102
Models trained on it
- mradermacher/groundhog-v1-i1-GGUF 690
- mradermacher/groundhog-v1-GGUF 624
- KeytechSRL/TernaryPilot-v2-Computer-Browser-Use-LoRAImage text → text 159
- yp9211/julia-1-mind2web-finetunedText classification 133
- adriandj3/TernaryPilot-v1-Computer-Browser-Use-LoRAImage text → text 88
- lukbit/devisionVisual question answering 68
- senpaicq/m2w-gemma4-e2b-pointer 30
- decido-ai/decido-1-web-26b-a4bImage text → text 24
How to load it
from datasets import load_dataset
ds = load_dataset("osunlp/Multimodal-Mind2Web")