README.md

---
tags:
- reinforcement-learning
- "2048"
- wandb
library_name: generic
---

# 🎮 2048 RL Agent (W&B Serverless)

This agent was trained using **W&B Serverless RL** on CoreWeave infrastructure.
It learned to play 2048 via reinforcement learning loops.

## 📂 Files
- **adapter_model.safetensors**: The trained LoRA weights.
- **evaluation_script.py**: Script to evaluate the agent.

## 🚀 Usage
Check `evaluation_script.py` for inference details.