The atari's state.shape is (210,160,3),and u define the net's in_channel as env.shape[0]

higgsfield / RL-Adventure

Pytorch Implementation of DQN / DDQN / Prioritized replay/ noisy networks/ distributional values/ Rainbow/ hierarchical RL

2.99k stars 587 forks source link

The atari's state.shape is (210,160,3),and u define the net's in_channel as env.shape[0] #23

Open WonderSeven opened 5 years ago

WonderSeven commented 5 years ago

This mainly causes the size error like ''Calculated padded input size per channel: (160 x 3). Kernel size: (8 x 8). Kernel size can't be greater than actual input size...'', so why not adjust the state's dim,maybe the code is :state = np.transpose(state, (2, 0, 1))or adjust the Net, but we may need to pay attention to this. 0.0

WonderSeven commented 5 years ago

Is there anyone find the same issue or could answer for me, thanks a lot!

krishnaw14 commented 5 years ago

In the script common/wrappers.py , the state.shape is changed to (3, 160, 210) when the function wrap_pytorch() is called. Hope this answers your question.

WonderSeven commented 5 years ago

Thanks for your answer. This really help me solve the problem.