This repository is built referred to deep learning from scratch 4
| folder name | explanation |
|---|---|
| DQN | DQN with PyTorch |
| Dynamic Planning | value iteration with dynamic planning |
| Monte Carlo | value iteration with Monte Carlo |
| NN_Qlearning | Qlearining and Neural network with PyTorch |
| Temporal Defference | value iteration with dynamic palanning |
| common | some functions and gridworld |
What's difference from deep learining from scratch 4?
- Maze shape I expanded the shape of maze from 34 to 45 and added to 2 blocks and 1 penalty area.
- Framework for deep learning. I used PyTorch for solve maze instead of the original framework DeZero made in deep learining from scratch 3, which is used in deep learning from scratch 4.