Prompt format for llama3

[tannisroot]

I'm currently playing with llama3:instruct, and in the logs I'm seeing this:
Model response did not end on a stop token (unfinished sentence)
I'm using Alpaca Prompt format. Is this expected behavior with this prompt format, or am I using the wrong prompt format with llama3?

[acon96]

If you turn on debug logging for the component, it will output whatever is being sent to the backend. Add this to your configuration.yml

logger:
  logs:
    custom_components.llama_conversation: debug

Also I'm hoping to get the official Llama3 system prompt added to the next release soon.

[tannisroot]

I do have that already, it's how I got the warning.
Didn't know llama3 uses a different prompt format. Will await a new release!