search

drndos / hass-openai-custom-conversation

Conversation support for home assistant using vicuna local llm
Apache License 2.0
85 stars 10 forks source link

mixing pipelines? #4

Closed Anto79-ops closed 5 months ago

Anto79-ops commented 12 months ago

hi there.

Thanks for the integration. I've been looking for integrations that incorporate LocalAI into HA and there are only 2, or maybe just yours depending on how you count the other.

There was a time where I was using your integration and this other integration here

to try and get my working LocalAI into HA. I experienced an unusual problem where I had to delete both integrations, but its not clear which integration caused it so I will report here for now.

I had the other integration setup to point to my LocalAI instance, and then setup yours to point to it also. I also have the Core HA Integration for OpenAI using the API key.

But, what ended up happening was I was using the OpenAI HA Core integration as the voice assistant, but it was submitting the queries to my LocalAI instance.

I don't think I had changed the OpenAI integration to point to my LocalAI for the conversation agent...but maybe it was? Anyway could it be possible that there was some sort of pipeline crossing with your integration?

thanks :)

Anto79-ops commented 11 months ago

hi, ok I canfirm that its when I add this integrations that pipelines gets mixed up.

Here what do/see:

I have the the core OpenAI integration setup and also have the other integration setup. They are both working well, independtly. I Install this integration and point it to my LocalAI server. Your integration works well, except, when I try to use core OpenAI integration, that now dumps all the queries to my LocalAI instance and not to OpenAI.

any ideas what could be goin on?

Anto79-ops commented 5 months ago

Moved on from this. Thanks