Skip to content

Dataset · xlangai

Computer agent tasks for RLVR

xlangai/CUA-Gym

A collection of verifiable tasks for training computer agents using reinforcement learning.

What's inside

The dataset contains a collection of verifiable English tasks for computer agents. Each task combines a textual instruction, executable setup artifacts, and a Python function for programmatic reward verification (RLVR). The size of the dataset ranges from 10K to 100K records.

What you can do with it

  • Training RL agents for computer use
  • Developing systems with verifiable rewards
Tasks
Reinforcement learning, Text generation
Languages
English
Size
10K–100K rows
Downloads this month
6.8K
Likes
32

How to load it

from datasets import load_dataset

ds = load_dataset("xlangai/CUA-Gym")