comodoro's picture
Push Q-Learning agent to Hub
42ba144
metadata
tags:
  - FrozenLake-v1-4x4-no_slippery
  - q-learning
  - reinforcement-learning
  - custom-implementation
model-index:
  - name: q-FrozenLake-v1-4x4-noSlippery
    results:
      - metrics:
          - type: mean_reward
            value: 1.00 +/- 0.00
            name: mean_reward
        task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: FrozenLake-v1-4x4-no_slippery
          type: FrozenLake-v1-4x4-no_slippery

Q-Learning Agent playing FrozenLake-v1

This is a trained model of a Q-Learning agent playing FrozenLake-v1 .

Usage

model = load_from_hub(repo_id="comodoro/q-FrozenLake-v1-4x4-noSlippery", filename="q-learning.pkl")

# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])

evaluate_agent(env, model["max_steps"], model["n_eval_episodes"], model["qtable"], model["eval_seed"])