rl

Stuffs related to reinforement learning experiments.

connect4-rl1-conv3.ipynb connect four AI using CNN
my2048-NN.ipynb 2048 AI using neural network for strategy, currently can reach 2048 in 40% of games.
more 2048 NN AI can be found at https://github.com/tjwei/2048-NN a network can 2048 95% of the time and a reinforcemnt learning ipynb.

Name		Name	Last commit message	Last commit date
Latest commit History 29 Commits
.ipynb_checkpoints		.ipynb_checkpoints
img		img
README.md		README.md
c4-conv3.pkl		c4-conv3.pkl
c4-conv7(200000 midterm avgloss=0.240524 mc100 vs nn: [4, 14, 82], nn vs mc100 [3, 83, 14]).pkl		c4-conv7(200000 midterm avgloss=0.240524 mc100 vs nn: [4, 14, 82], nn vs mc100 [3, 83, 14]).pkl
c4.pkl		c4.pkl
c4client.py		c4client.py
connect4-rl.ipynb		connect4-rl.ipynb
connect4-rl1-Copy1.ipynb		connect4-rl1-Copy1.ipynb
connect4-rl1-Copy2.ipynb		connect4-rl1-Copy2.ipynb
connect4-rl1-conv3.ipynb		connect4-rl1-conv3.ipynb
connect4-rl1.ipynb		connect4-rl1.ipynb
connect4-rl2.ipynb		connect4-rl2.ipynb
connect4.ipynb		connect4.ipynb
control_linear.js		control_linear.js
control_linear2.js		control_linear2.js
control_table.js		control_table.js
game2048.py		game2048.py
model-0.npz		model-0.npz
my2048-NN.ipynb		my2048-NN.ipynb
ox-nn.ipynb		ox-nn.ipynb
ox.ipynb		ox.ipynb

Provide feedback