README.md

metadata

tags:
  - FrozenLake-v1-4x4-no_slippery
  - q-learning
  - reinforcement-learning
  - custom-implementation
model-index:
  - name: q-FrozenLake-v1-4x4-noSlippery
    results:
      - task:
          type: reinforcement-learning
          name: reinforcement-learning
        dataset:
          name: FrozenLake-v1-4x4-no_slippery
          type: FrozenLake-v1-4x4-no_slippery
        metrics:
          - type: mean_reward
            value: 1.00 +/- 0.00
            name: mean_reward
            verified: false

Q-Learning Agent playing1 FrozenLake-v1

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

Usage

import gymnasium as gym
from huggingface_sb3 import load_from_hub
import numpy as np
import pickle

# Load the Q table
env_name = "FrozenLake-v1"
model_name = "q-FrozenLake-v1-4x4-noSlippery"
model_path = load_from_hub(repo_id="ch-bz/" + model_name, filename="q-learning.pkl")
Qtable = pickle.load(open(model_path, "rb"))["qtable"]

# Run the demonstration of the result
env = gym.make("FrozenLake-v1", map_name="4x4", is_slippery=False, render_mode="human")
state, info = env.reset()

while True:
    action = np.argmax(Qtable[state][:])
    state, reward, terminated, truncated, info = env.step(action)
    env.render()
    
    if terminated or truncated:
        state, info = env.reset()