3 ms·

We already have ollama,LocalAI.

• trzy 3y ago
  Neither of which run LLaVA as far as I can tell. Use this until that happens. Also, this is way easier to set up. No dependencies. Easy.
  • v3ss0n 3y ago
    why not? both uses llamacpp as backend
    • trzy 3y ago
      llama.cpp does not have interactive mode support for LLaVA nor the web server yet. So if they're just invoking the app, there is a lot of overhead for each call until they add that in. This server doesn't have that issue.