Closed MohamedBassem closed 5 months ago
Just landed the ollama support, it'll be included in the next release (happing later today)!
Yes!!!!
Ollama support is now available in the 0.10.0
release! 🎉 You can check the new configuration arguments here!
Basically add the following to the env file (make sure you remove the OPENAI_API_KEY
otherwise it takes precedence):
OLLAMA_BASE_URL=<ollama_addr>
INFERENCE_TEXT_MODEL=mistral
INFERENCE_IMAGE_MODEL=llava
Make sure that those two models are pulled in ollama with ollama pull mistral
and ollama pull llava
.
Id recommend the openchat 0106 model for text, that thing is impressive for a 7b model!
This issue to track supporting ollama as an alternative to openai.