Implementation of deep Q network in conventional language

[Jiří Vahala]

Hello, 

I am trying to implement my own deep Q net and I am struggeling with convergence. It seems that system converges but then suddently something happens and the whole learning is broken)

I would like to have my own deep Q network, unfortunately, I don't understnad lua and it's library torch so I can't read the original paper about playing atari games using deep Q network.

Do you know about some implementation in conventional language like c/cpp/java?  

thank you for help

[Tambet Matiisen]

There is quite compact implementation in C++ using Caffe:
https://github.com/muupan/dqn-in-the-caffe

It hasn't been updated for long time though...

  Tambet

--
You received this message because you are subscribed to the Google Groups "Deep Q-Learning" group.
To unsubscribe from this group and stop receiving emails from it, send an email to deep-q-learni...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/deep-q-learning/5b8f49dc-c2fc-4792-bf06-6e3017adbd8f%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.