🎮 Reinforcement Learning Playground

This repository contains two separate RL experiments in one place:

1. Tic-Tac-Toe RL Agent — a gentle introduction where the agent learns not to play like a potato.
2. Atari Pitfall RL Agent — pure chaos mode, where the agent struggles against holes, ladders, and the meaning of existence.

Each experiment lives in its own notebook. Run them independently.

---

1) 🤝 Tic-Tac-Toe RL Agent

Notebook: tic-tac-toe.ipynb

This notebook trains an agent to play Tic-Tac-Toe using reinforcement learning.

What Happens Here

• Agent initially places moves at random.
• Learns that losing feels bad.
• Eventually begins to block and win (sometimes intentionally).

How to Run

1. Open the notebook in Jupyter / Colab.
2. Run cells in order.
3. Observe the transition from “huh???” to “oh, I get it!”

---

2) 🏃‍♂️💥 Atari Pitfall RL Agent

Notebook: atari_pitfall.ipynb

This notebook attempts to teach an agent how to play Pitfall, a game where falling into holes is less of a possibility and more of a destiny.

What Happens Here

• Agent tries to move and jump.
• Agent discovers gravity the hard way.
• Training is slow but sometimes rewarding.

How to Run

1. Make sure you have gym + Atari environments installed.
2. Preferably train using GPU (CPU training may take your entire lifespan).
3. Run notebook step-by-step.

---

🛠 Requirements

• Python
• Jupyter Notebook
• gym (with Atari dependencies for Pitfall)
• Torch / TensorFlow (depending on your implementation)
• Patience (mandatory)

---

⚠️ Disclaimer

If the agent performs well, you may celebrate.
If it fails spectacularly — that’s part of the show.
We learn. We fall. We fall again. And then maybe we jump correctly once.

---