Problem talking to the backend

[andreas-bulling]

Installation went fine but I get the following error when trying to invoke the assistant:

Sorry, there was a problem talking to the backend: RuntimeError('llama_decode returned 1')

[acon96]

Can you provide more information? What model were you using? How many entities do you have exposed? Were there any errors or warnings in the HA logs?

[Teagan42]

llama_decode is a method in llama.cpp: https://llama-cpp-python.readthedocs.io/en/latest/api-reference/#llama_cpp.llama_cpp.llama_decode

Does your LLM model work if you call it directly (i.e. no Home-Assistant, use the CLI or API of the runner)?