Dataset · meta-agents-research-environments
Benchmark for evaluating AI agents in dynamic environments
meta-agents-research-environments/gaia2
800 English scenarios for testing agents in continuous time conditions.
What's inside
The English text dataset includes 800 scenarios for testing AI agents in simulated environments where time flows continuously and events occur dynamically. The set evaluates seven key skills, including multi-step planning, information retrieval, and adaptation to changes.
What you can do with it
- Benchmarking autonomous AI agents in complex changing conditions
- Testing planning systems and reinforcement learning
- Tasks
- Reinforcement learning
- Languages
- English
- Size
- up to 1K rows
- Downloads this month
- 17.6K
- Likes
- 48
How to load it
from datasets import load_dataset
ds = load_dataset("meta-agents-research-environments/gaia2")