FrozenLake Reinforcement Learning Environment Data

Featuring data from the FrozenLake-v1 environment with the slippery parameter set to False. It was created by AntoineLB and last updated in April 2022. The specific row count, column structure, and data format are unknown.

Use Cases

• Train reinforcement learning agents on the deterministic FrozenLake-v1 environment to benchmark policy learning algorithms.
• Analyze state transitions and reward structures within the non-slippery grid world for environment verification.
• Compare agent performance metrics between the deterministic (slippery=False) and standard stochastic versions of the FrozenLake environment.

Strengths

• Data originates from a well-known and standardized OpenAI Gym environment, FrozenLake-v1.
• The environment configuration (slippery=False) provides a deterministic variant for controlled experimentation.
• Dataset is tagged for use with common data libraries like Polars, Pandas, and the Hugging Face `datasets` library.

Limitations

• The dataset's size, specific contents (rows/columns), and file format are not documented.
• The data is from a simulated environment and does not represent real-world observations or events.
• The last update was in 2022, which may indicate the dataset is not actively maintained.

Provenance

Source

OpenAI Gym FrozenLake-v1 environment.

Collection Method

Simulation data generated from the reinforcement learning environment.

Freshness

Last updated on 2022-04-26.