Performance & Stability
        
        How Is the Reward Function Structured to Prevent Unwanted Agent Behaviors?
        
         
        
        
          
        
        
      
        
     
        
        A reward function prevents unwanted behavior by encoding penalties for negative side effects and unintended actions directly into the agent's core optimization objective.

 
  
  
  
  
 