# Language Models can Self-Improve at State-Value Estimation for Better Search

Introduces Self-Taught Lookahead (STL), which trains a value model to predict actions and resulting states for more efficient search, with evaluation on web-agent tasks.

- Catalog URL: https://awesomecua.com/resources/arxiv-2503-02878/
- Original source: https://arxiv.org/abs/2503.02878
- Kind: paper
- Review: curated
- Scope: core
- Environments: web
- Topics: planning, efficiency, world-models
- Evidence checked: 2026-11-05T23:01:33Z
- ID: arxiv:2503.02878

## Limitations

The annotation and scope were reviewed against the complete retained arXiv abstract (v3). The abstract also reports evaluations on multi-hop question answering and math puzzles. Experiments, reported performance and artifact availability have not been independently reproduced.

- Record JSON: https://awesomecua.com/resources/arxiv-2503-02878/record.json

## Authors

Ethan Mendes, Alan Ritter

## Citation exports

- BibTeX: https://awesomecua.com/resources/arxiv-2503-02878/citation.bib
- CSL JSON: https://awesomecua.com/resources/arxiv-2503-02878/citation.csl.json

## Recorded arXiv snapshot

- Version: 3
- Version source: https://arxiv.org/abs/2503.02878v3
- First submission: 2025-03-04T18:58:11Z
- Version updated: 2025-10-29T22:28:23Z
- Abstract SHA-256: 50cf772cf321e8daac4124f5e55e2d295fb751b53e10a73b4fb36f87f1854b35
- Abstract hash normalization: sha256-utf8-whitespace-collapsed-v1

## License declarations

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

## Primary evidence

- https://export.arxiv.org/api/query?id_list=2501.12485%2C2502.01456%2C2502.03214%2C2502.03971%2C2502.06776%2C2502.12561%2C2502.13417%2C2502.17419%2C2502.20694%2C2503.02878%2C2503.16365%2C2503.20036%2C2504.07491%2C2504.12682%2C2505.03733%2C2505.05262%2C2505.05895%2C2505.07062%2C2505.10978%2C2505.17629%2C2505.20075%2C2505.23678%2C2505.24298%2C2505.24876%2C2506.00520%2C2506.01952%2C2506.06287%2C2506.14079%2C2506.19998%2C2506.24045%2C2507.01006%2C2507.04370%2C2507.06261%2C2507.10202%2C2507.10644%2C2507.11662%2C2507.23773%2C2508.09736%2C2508.16271%2C2508.16763%2C2508.18265%2C2508.19828%2C2508.20018%2C2508.21475%2C2509.02547%2C2509.08826%2C2509.08827%2C2509.11067%2C2509.13313%2C2509.21733&max_results=50 (retrieved 2026-11-05T21:13:58Z; SHA-256 62b85cd4bee4c668046d5f6eb67dedef915ba699d6765891016403c863691a73)
- https://export.arxiv.org/api/query?id_list=2503.02878%2C2503.06378%2C2503.06580%2C2503.07365%2C2503.07701%2C2503.08026%2C2503.08175%2C2503.10613%2C2503.10809%2C2503.11579%2C2503.12228%2C2503.12461%2C2503.12687%2C2503.13843%2C2503.14495%2C2503.15426%2C2503.16036%2C2503.16248%2C2503.16365%2C2503.16975%2C2503.17671%2C2503.18102%2C2503.20036%2C2503.21889%2C2503.21949%2C2503.23803%2C2504.01328%2C2504.01377%2C2504.03206%2C2504.03601%2C2504.06821%2C2504.07491%2C2504.09723%2C2504.10147%2C2504.11301%2C2504.11543%2C2504.11703%2C2504.12609%2C2504.12682%2C2504.14233%2C2504.14348%2C2504.17892%2C2504.18580%2C2504.20965%2C2505.00019%2C2505.00693%2C2505.01578%2C2505.02094%2C2505.03574%2C2505.03733&max_results=50 (retrieved 2026-11-05T23:01:33Z; SHA-256 b7001e189c17d6fa072573558661e37a7cb4c7709d33080603d72a969cc4ba39)