• 7
  • 109
  • 8
  • 1
  • 165
  • 12
  • 2
  • This topic is deleted!

    1
  • 5
  • 16
  • 5
  • 44
  • 44
  • 4
  • 9
  • 10
  • 11
  • 7
  • 5
  • 183

Recent Posts

  • @kindwind interested to see your progress 🙂

    read more
  • B

    @kindwind said in Game Engine Rules (AI Training):

    This is pretty hard for a guy who doesn't really code. I am sure my code is a mess but it works so there is that.

    RightArm.png salute.png

    Rock On !!!.png

    read more
  • @ubernaut that's the plan. I am going to do an LLM native ai. My final deliverable is a person takes their revised .tsvg game save of where they are at gets advice and then tells the lmm to play the other side. I think it's neat. Others not so much. I was thinking that people might like a Markov chain battle calculator too in the final deliverable too.

    This is pretty hard for a guy who doesn't really code. I am sure my code is a mess but it works so there is that.

    Trying to just have Claude play Claude calling my engine and a database seemed like a good idea to which I think he was doing too but the cost was like 1.20 a game and for 100,000 games the math didn't math. I tried using Qwen but it's not smart enough. Then there was another problem about distillation from Claude

    I am stuck with ML for now and training it that way until I get an ok player and then a good player and then a super human player. It might not even work either no one has done this before made one work. I think I can get there just a wish right now.

    Pretty sure I can get a really good heuristic model if this fails.

    read more
  • read more