CUA-Gym

Summary

CUA-Gym is a collection of verifiable computer-use agent tasks for reinforcement learning with verifiable rewards (RLVR). Each task pairs a natural-language instruction with executable setup artifacts and a Python reward function that checks task completion programmatically. For details, see the paper [CUA-Gym: Scaling Verifiable Training Environments and Tasks for Computer-Use Agents](https://arxiv.org/abs/2605.25624).

Keywords

hf-dataset reinforcement-learning · -language-modeling parquet tabular text datasets pandas polars mlcroissant computer-use-agents gui-agents desktop-agents web-agents rlvr verifiable-rewards programmatic-reward synthetic-data osworld webarena has-paper

Topics

Reinforcement Learning, Language Modeling

Research notes

• downloads=802; likes=22