Skip to content

Dataset · McGill-NLP

Real-world website navigation in dialogue (WebLINX)

McGill-NLP/weblinx-browsergym

English dataset for training and evaluating web agents in BrowserGym and Agentlab.

What's inside

Adaptation of the WebLINX benchmark for use within the BrowserGym and Agentlab ecosystems. Contains multi-step navigation data on real websites with English dialog instructions, oriented toward text generation and image processing. Corpus size and storage format are not specified in the description.

What you can do with it

  • Training multimodal web agents to perform browser actions via instructions
  • Testing and comparing autonomous agents in a standardized BrowserGym environment
Tasks
Image → text, Text generation
Languages
English
Downloads this month
74.5K
Likes
5

How to load it

from datasets import load_dataset

ds = load_dataset("McGill-NLP/weblinx-browsergym")