Agent is not learning.

[maxchu-aa]

I tried to train the model with the command "python .\dreamerv3.py --env "ALE/Breakout-v5", but the agent is not learning. The average reward will increase initially to around 2, and then it will just stand there without any improvement.

[naivoder]

Hi @maxchu-aa, apologies for having wasted your time! I have never quite gotten this implementation right... if you're looking for an implementation that is easier to understand than the official one, I'd really recommend you check out https://github.com/InexperiencedMe/NaturalDreamer

[maxchu-aa]

@naivoder no worries, your code still helps me a lot for learning how to implement Dreamer V3. I actually have corrected all the errors (to the best of my knowledge) and i have put the corrected version in my personal GitHub repo (https://github.com/MaxChu719/Dreamer_V3).