Open daihuiao opened 1 year ago
Hi, why the Q loss cannot covergence is that we just normalize the state and did not normalize the next state in the buffer.py file. We've updated the source code to fix this fault. Thank you for your careful check and valuable comments.
@Lei-Kun ,hi, since the next state is normalized ,the loss and score don't show the effect as we expect
Is this normal, or the environment is not installed properly?