How ranasinghe843/minesweeperrl works
MinesweeperRL contains parallel Minesweeper reinforcement-learning implementations: tabular Q-learning at the repository root and under Minesweeper_Qlearning, plus a neural-network/replay-based variant under MinesweeperDQN. The principal workflow is to train from simulated games, use the learned policy to play more games, then report outcomes; the DQN runner also plots evaluation metrics. Source excerpts show the game mechanics, root Q-learning runner, and parts of the DQN path. Some DQN model wiring is only indicated by imports, so those links are marked unverified.
Main entry points: minesweeper.py DQN engine · MinesweeperDQN/minesweeper_qlearning.py DQN runner · minesweeper_qlearning.py Q-learning runner · Minesweeper_Qlearning/minesweeper_qlearning.py Legacy runner
Components 14
Tabular Q-learning
- Q-learning runner
minesweeper_qlearning.py - Q-learning engine
minesweeper.py - Legacy runner
Minesweeper_Qlearning/minesweeper_qlearning.py - Legacy engine
Minesweeper_Qlearning/minesweeper.py
DQN learning
- DQN runner
MinesweeperDQN/minesweeper_qlearning.py - DQN engine
MinesweeperDQN/minesweeper.py - Experience replay
MinesweeperDQN/ExpReplay.py - Neural network
MinesweeperDQN/NN1.py - MLP classifier
MinesweeperDQN/neural_network.py - Training data
MinesweeperDQN/train_data.npy - Training labels
MinesweeperDQN/train_labels.npy
Evaluation
- Metric plots
MinesweeperDQN/minesweeper_qlearning.py - Game outcomes
MinesweeperDQN/minesweeper_qlearning.py
Other components
- User
Connections 12
- User → Q-learning runner: runs
minesweeper_qlearning.py - Q-learning runner → Q-learning engine: trains and plays
minesweeper_qlearning.py - User → Legacy runner: runs
- Legacy runner → Legacy engine: trains and plays
Minesweeper_Qlearning/minesweeper_qlearning.py - User → DQN runner: runs
MinesweeperDQN/minesweeper_qlearning.py - DQN runner → DQN engine: trains and plays
MinesweeperDQN/minesweeper_qlearning.py - DQN engine → Experience replay: stores samples
MinesweeperDQN/minesweeper.py - DQN engine → Experience replay: gets batches
MinesweeperDQN/minesweeper.py - DQN engine → Neural network: uses model
MinesweeperDQN/minesweeper.py - DQN engine → MLP classifier: imports model
MinesweeperDQN/minesweeper.py - DQN runner → Metric plots: plots metrics
MinesweeperDQN/minesweeper_qlearning.py - DQN runner → Game outcomes: computes outcomes
MinesweeperDQN/minesweeper_qlearning.py