Ask HN: Selfhosted ChatGPT and Stable-diffusion like alternatives?

1 point by null4bl3 1 year ago | 2 comments
I know this has been asked before, but things are moving so quickly in this realm and people here seem to have a good insight, so I am asking again.

The speed of answers and computation is not really an issue, and I know that most selfhosted solutions obviously is in no way fully on-par with services like Chatgpt or Stable-diffusion.

I do have somewhat modest resources.

16 GB RAM NVIDIA GPU with 4 GB vRAM.

Is there any options that means I can run it selfhosted?

  • firebaze 1 year ago
    https://github.com/invoke-ai/InvokeAI should work on your machine. For LLM models, the smaller ones should run using llama.cpp, but I don't think you'll be happy comparing them to ChatGPT.
    • null4bl3 1 year ago
      Thanks. I will check that out.

      I tried out gpt4all with the mistral orca and atleast that was able to answer my basic use cases which are not that advanced and very factual based. So I am hoping another alternative could do the same.