david-parkk / Tetris_DQN_PyTorch

DQN to play tetris game [PyTorch]

[PyTorch] Tetris with DQN(Deep Q-Network)

• Presentation video: YouTube

Results

Epoch	84	1185	2042	3651
total number of rows removed	1	13	42	312
final score	26	201	781	7515

Details

✔ Definition of State, Reward, Action in Tetris

✔ DQN algorithm (Q-function)

How to run codes

Requirements

imageio==2.13.5
importlib-metadata==4.8.2
matplotlib==3.5.0
numpy==1.18.1
opencv-python==4.5.4.60
Pillow==8.4.0
tensorboard==1.15.0
tensorboardX==2.4.1
tensorflow==1.15.5
torch==1.10.0
torchvision==0.11.1
tqdm==4.62.3

Installing Packages

$ pip install -r requirements.txt

Train & Save model

$ python3 train.py

Test & Save the results in gif format

$ python3 test.py

---

소스코드 참고: https://github.com/uvipen/Tetris-deep-Q-learning-pytorch