# ARPO UI-TARS-1.5-7B checkpoint

UI-TARS-1.5-7B-derived checkpoint trained with ARPO on OSWorld desktop tasks.

- Catalog URL: https://awesomecua.github.io/resources/url-db6fc5f29f186d13991c/
- Original source: https://huggingface.co/Fanbin/ARPO_UITARS1.5_7B
- Kind: model
- Review: curated
- Scope: core
- Environments: desktop, linux
- Topics: reinforcement-learning, action-prediction
- Evidence checked: 2026-10-18T04:27:12Z
- ID: url:db6fc5f29f186d13991c

## Limitations

The public Hub manifest lists four Safetensors shards, an index and configuration/tokenizer files. The model card reports a 15-step evaluation setting; no performance result is independently reproduced. No weight license is declared, despite Apache-2.0 being established for the separate code repository.

- Record JSON: https://awesomecua.github.io/resources/url-db6fc5f29f186d13991c/record.json

## License declarations

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

- weights review: The inspected model card contains no checkpoint-license declaration. The code repository Apache license is not transferred to weights.
  Sources: https://huggingface.co/Fanbin/ARPO_UITARS1.5_7B/raw/2c5604d86c529843ef346698edcf33b170c1e817/README.md

## Primary evidence

- https://raw.githubusercontent.com/JIA-Lab-research/ARPO/HEAD/README.md (retrieved 2026-10-18T04:27:12Z; SHA-256 839b2de0e0c614fe8d2df4782d0f6a6e7d33f5c60d8020c9d4a01c61d533f39e)
- https://huggingface.co/api/models/Fanbin/ARPO_UITARS1.5_7B (retrieved 2026-10-18T04:33:21Z; SHA-256 6e19a5339c4e52ca3753cce26ac2034883b96f5cb26383a4b41bc87a5fbf5bf6)
- https://huggingface.co/Fanbin/ARPO_UITARS1.5_7B/raw/2c5604d86c529843ef346698edcf33b170c1e817/README.md (retrieved 2026-10-18T04:35:16Z; SHA-256 b6957fad37a7995e9dde4c7d4a4e0f6e2153b85354b2d234587c2a94ff95d0bd)