• NGC2346
    link
    fedilink
    English
    arrow-up
    13
    ·
    1 year ago

    Its fine, i got my own LlaMa at home, it does almost the same as GPT

      • NGC2346
        link
        fedilink
        English
        arrow-up
        1
        ·
        edit-2
        1 year ago

        uncensored vicuna 7b if i recall

        nvidia rtx 3060, 20 core intel cpu, 32gb ram

      • BetaDoggo_@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        1 year ago

        In terms of flexibility oobabooga’s text generation webui is the best as it supports nearly every model as ggml, gptq and transformers.

        For the model itself Nous-hermes-llama2 is one the best for easy use.

    • average650@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      1 year ago

      The biggest issue I find is it’s just not as easy to use as chatgpt. I’m surprised no one has sold an easy to use consumer version.