IteratedStagHunt

Description

IteratedStagHunt is an environment for evaluating agents on coordination and cooperation in a game with multiple equilibria. This environment wraps the IteratedStagHunt implementation from TextArena, a framework for text-based game environments.

Capabilities

• Coordination problem solving
• Risk assessment between safe and cooperative strategies
• Trust building over multiple rounds
• Competitive gameplay against an LLM opponent

Compute Requirements

IteratedStagHunt does not require a sandbox. It has minimal compute requirements.

License

MIT.

Tasks

There are two splits: train (300 tasks) and test (300 tasks). Each split contains 50 tasks across each of 6 variants:

• IteratedStagHunt-v0
• IteratedStagHunt-v0-train
• IteratedStagHunt-v0-raw
• IteratedStagHunt-v0-randomized
• IteratedStagHunt-v0-randomized-train
• IteratedStagHunt-v0-randomized-raw

Each task is seeded for reproducibility.

Reward Structure

This is a sparse reward environment. Rewards are mapped from TextArena's native range of {-1, 0, 1} to {0.0, 0.5, 1.0} via (raw + 1) / 2.

We do not use LLM graders for this environment; reward is determined programmatically.

Data

Game state is generated procedurally by the TextArena engine using seeded randomness. No external data files are required.

Tools

Agents are given a single tool:

• hunt(target): Choose to hunt either 'stag' or 'hare'. Stag requires cooperation but gives highest reward. Hare is safe but gives lower reward.

Time Horizon

IteratedStagHunt is a multi-turn environment.

Environment Difficulty

Medium - requires balancing the risk of cooperation (hunting stag) with the safety of defection (hunting hare), while modeling opponent trustworthiness and maintaining credible cooperative signals.

Other Environment Requirements

This environment requires an OpenAI API key (passed via secrets) to power the LLM opponent.

Safety

Agents in IteratedStagHunt interact only with a game theory simulation and have no access to external systems, the internet, or sensitive data. The environment does not present safety risks.

Citations

@software{textarena2024,
  author    = {Guertler, Leon and Banting, Wilfried and Pignatelli, Eduardo},
  title     = {TextArena},
  year      = {2024},
  publisher = {GitHub},
  url       = {https://github.com/LeonGuertler/TextArena}
}