search

muupan / async-rl

Replicating "Asynchronous Methods for Deep Reinforcement Learning" (http://arxiv.org/abs/1602.01783)
MIT License
400 stars 81 forks source link

Color transform is incorrect. #24

Open siemanko opened 8 years ago

siemanko commented 8 years ago

Hey,

so the color transform that you use is incorrect (for example in Seaquest it sometimes causes the fishes to disappear).

https://github.com/muupan/async-rl/blob/12dac595b2ad9d99c19e82e92adb825359a3a3eb/ale.py#L67-L68

You can get the correct one from the repo Deepmind uses: https://github.com/torch/image/blob/9f65c30167b2048ecbe8b7befdc6b2d6d12baee9/generic/image.c#L2105)

Let me know how much this improves your results if you decide to adopt this. For me the difference was massive.

muupan commented 8 years ago

Thank you for pointing it out! I'll check the performance with that transformation.