doihye/ear

Official pytorch implementation of the paper [Environment Agnostic Representation for Visual Reinforcement learning]

ear's Issues

Unable to replicate the results using hyperparameters provided

Hi,
Could you provide a list of instructions to replicate your results? Unfortunately, the agent does not learn with the list of hyper-parameters provided, such as Mutual loss coefficient = 0.01 and recon_loss coefficient=1. It seems the problem arises in the EAF encoder defined as self.di. Also, could you comment if the use of Inv module that predicts the future action in the loss function is necessary as it is not mentioned in the paper.

Thanks in advance

Follow up the code

It's an interesting work. I am wondering when the code will be published.