You are viewing a single comment's thread from:RE: Teaching AI to Play Chrome Dino Game: Reinforcement LearningView the full contextView the direct parentmateodm03 (70)Geek Verificadoin Geek Zone • 8 days ago I have now checked the 7B model I had used.
That explains it. It's really weak with just 7 billion parameters compared to the best 671 billion parameters.