SKU RLE-1101 · Sold by External
WebArena
Product specifications
| SKU | RLE-1101 |
|---|---|
| Data type | RL environments |
| Volume | 812 tasks (from 241 intent templates) |
| Size on disk | Not published by source |
| Format | JSON task configs + Python (gym) + self-hosted Docker sites |
| Access model | PUBLIC LICENSE |
| Pricing | Free · open-source license |
| Quality score | — |
| License | Apache-2.0 |
WebArena is a standalone, self-hostable web environment for building and evaluating autonomous agents across five functional websites (e-commerce/Magento, a Postmill social forum, GitLab, a Magento CMS admin, and an OpenStreetMap-based maps site) plus utilities like a scratchpad, calculator and Wikipedia mirror. Agents act through a gym-style browser action space (click, type, hover, scroll, key press, tab management, goto URL) with observations from the DOM/accessibility tree and screenshots. It ships 812 tasks instantiated from 241 intent templates, each scored by execution-based functional-correctness checks (string_match, url_match, program_html).