search

ShangtongZhang / DeepRL

Modularized Implementation of Deep RL Algorithms in PyTorch
MIT License
3.21k stars 684 forks source link

Simple Question #44

Closed Sungtae-Lee closed 5 years ago

Sungtae-Lee commented 6 years ago

It seems like the code gets slower epoch by epoch. Is there any way I can fix this? (After exploration steps which is 50000)

ShangtongZhang commented 6 years ago

Could you share more details, e.g., logs, GPU/CPU, memory size? Before 50000 steps there is no training, so it should be faster. And for some game maybe it may take more time to render after reaching a certain level, I'm not sure about this. Lack of memory may also be a reason.