search

kuz / DeepMind-Atari-Deep-Q-Learner

The original code from the DeepMind article + my tweaks
http://www.nature.com/nature/journal/v518/n7540/full/nature14236.html
1.83k stars 533 forks source link

Is there "trained" model? #16

Open gaoteng-git opened 8 years ago

gaoteng-git commented 8 years ago

Since it costs a long time training (30hours on GPU) to get the smart model, would you please provide an already trained model file that I can run without training from beginning? I only have a CPU laptop so it can't afford such a long time training. Thanks a lot in advance!

baicaigithub commented 8 years ago

I've been running for ~8 hours on my GTX 970 and it can achieve high score of 10. The network file has grew to 400MiB.

baicaigithub commented 8 years ago

btw, it seems I cannot resume to train the model, even though I changed related fields in run_gpu.

baicaigithub commented 8 years ago

My bad, I set the save_freq too small, so that the network is overwritten before being analyzed.

jtgebert commented 8 years ago

Could you specify how you resumed training on a network after stopping? Thanks

scm-ns commented 7 years ago

So , do we have a trained model under some non commercial or open source license ?

kertansul commented 7 years ago

@jtgebert have you found anyway to resume training ?

jtgebert commented 7 years ago

Nope, I set this aside for a bit.