Q-Learning on FrozenLake-v1 (4x4, No Slippery)

This project implements a Q-Learning agent trained on the FrozenLake-v1 environment as part of the Hugging Face Deep Reinforcement Learning course.

---

🎥 Agent in Action

---

📌 Environment

• FrozenLake-v1
• 4x4 grid
• No slippery (deterministic)

---

⚙️ Algorithm

Tabular Q-Learning with ε-greedy exploration.

Q-update rule:

Q(s,a) = Q(s,a) + α (r + γ max Q(s',a') - Q(s,a))

---

📊 Results

The trained agent consistently reaches the goal in the deterministic FrozenLake environment.

---

🤗 Hugging Face Model

https://huggingface.co/BhushanGatty/q-FrozenLake-v1-4x4-noSlippery

---

👨‍💻 Author

Bhushan Gatty
Robotics Engineering Student
Deep Reinforcement Learning Enthusiast