3 comments

  • ilaksh10 minutes ago
    This is not an LLM obviously , it's just for generating random names. But interesting to think of the possibilities of truly tiny language models if there were connected together.
  • throwa35626236 minutes ago
    And the 5 years old AMD Ryzen 5 5600H is doing 7M?<p>Am I reading this right? Then I need to try this on Strix Halo
  • fwip16 minutes ago
    Model is 4K parameters - I don&#x27;t know enough about that size of model to know if this impressive or not.