Dueling Netowrk is getting worse over training episodes

Dear community, could you tell me, what is going wrong, if my reward is getting worse over time and the loss is increasing?

I am using a simple dueling network architecture with linear layers.