# WebRL

Training pipeline that evolves web tasks from prior attempts and uses an outcome reward model to refine browser policies.

- Catalog URL: https://awesomecua.github.io/resources/github-thudm-webrl/
- Original source: https://github.com/THUDM/WebRL
- Kind: framework
- Review: curated
- Scope: core
- Environments: web
- Topics: reinforcement-learning, web-navigation, data-generation
- Evidence checked: 2026-10-17T02:02:06Z
- ID: github:thudm/webrl

## Limitations

Training code and checkpoint links are released. No root code-license declaration was found in the inspected repository metadata; bundled dependency licenses do not establish a license for the whole project. Weights have separate Hub terms.

- Record JSON: https://awesomecua.github.io/resources/github-thudm-webrl/record.json

## License declarations

- code: not recorded
- data: not recorded
- document: not recorded
- weights: not recorded

## Primary evidence

- https://raw.githubusercontent.com/THUDM/WebRL/HEAD/README.md (retrieved 2026-10-17T02:02:06Z; SHA-256 eb217b96c54c3c2ccfd3a27255c6fd740df40c2d6743ae811858339bf7f740b0)
- https://api.github.com/repos/THUDM/WebRL (retrieved 2026-10-17T02:02:05Z; SHA-256 42496cf0cff06a7f9e794f4313a10a670bc48df9909ec6d57763cb902bb229ea)
- https://api.github.com/repos/THUDM/WebRL/git/trees/main?recursive=1 (retrieved 2026-10-17T02:02:05Z; SHA-256 bb1683c6fb5ff658c89f0ebf0f8baf8026bb703840e174c9e46b223c85a16856)