[Hussain04]

How does the learner forget old data? If a model's price drops by 50%, or a new model is swapped into an existing rung, the historical weights learned from the old pricing/capability will actively harm routing decisions. Is there a decay factor on the weights or something else?

[rezat]

You are right, I will take that think about how to improve it. Thank you very much

[rezat]

It is now fixed, it rehydrates on each turn

[Hussain04]

Awesome!