reinforcement-learning
reinforcement-learning copied to clipboard
rlcode
Fail to converge for Breakout dqn
I have train the model for around 4 days, episode now is 14278, while the score is 40 ~ 50, what's the problem?
Did you change any parameters? Changing parameters affect performance very much.
if you give us any other information (like tensorboard graph) than we can figure out more things. But now we don't have any idea. Sorry.
Sorry for late replay. These are images. I don't change any parameters
2017-10-10 22:49 GMT+08:00 youngmoo [email protected]:
Did you change any parameters? Changing parameters affect performance very much.
if you give us any other information (like tensorboard graph) than we can figure out more things. But now we don't have any idea. Sorry.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/rlcode/reinforcement-learning/issues/66#issuecomment-335497918, or mute the thread https://github.com/notifications/unsubscribe-auth/AVnNGcALls8xU-ZgvL1DOLinyHR-chffks5sq4PlgaJpZM4PzP9X .
Hello, I have been troubled by this problem for a long time. How can I solve it
Use a better implementation.
e.g. https://github.com/tensorpack/tensorpack/tree/master/examples/DeepQNetwork
