wojzaremba / trpo

99 stars 52 forks source link

Can't reproduce result on RepeatCopy #4

Closed ruotianluo closed 8 years ago

ruotianluo commented 8 years ago

Hi, I tried your code and ran it for multiple times. My agents turn to stuck at 4 after even more than 10k iterations. Do you have any insights what the problem could be?

wojzaremba commented 8 years ago

Sorry for that. I haven't played with this code for a while. Hyperparameters like kl, and batchsize make a difference. You can increase batchsize, and it should help (it will make computation slower though).