The following commands all result in a NaN error when the robot begins to train in the environment. Generally, this occurs in the training when the robot reaches either the gap or the start of the stair.
I tried each command several times, but the error cannot be reproduced on my machine (using the most recent code on github). Is there anything else I should be aware of?
The following commands all result in a
NaN
error when the robot begins to train in the environment. Generally, this occurs in the training when the robot reaches either the gap or the start of the stair.