Wire post by Administrator

Even subscription querys would run on only 50 billion parameter models,to save generation tokens, so 300 billion parameter model will come up with some awesomely more detailed query outputs, I THEORIZE…I could be wrong