How eligrealish/tensorflow-deep-reinforcement-learning works
This repository describes a TensorFlow-based deep reinforcement learning agent for 2048. Its principal workflow is trial-and-error learning: a DQN agent develops moves for the tile-matching game. The final agent adds double, dueling, and prioritized-memory improvements. The repository contains only a notebook and README, with no source excerpts, so the graph reflects documented capabilities rather than verified internal implementation details.
Main entry points: main.ipynb DQN Agent
Components 4
2048 Game
- 2048 Game
main.ipynb
Deep RL Agent
- DQN Agent
main.ipynb - DQN Improvements
main.ipynb
Other components
- TensorFlow
Connections 3
- DQN Agent → 2048 Game: learns moves
README.md - TensorFlow → DQN Agent: builds agent
README.md - DQN Improvements → DQN Agent: improves
README.md