Is there a way to adjust the response timeout when using llama.cpp?

[stash763]

Is there a way to adjust the response timeout when using llama.cpp? When submitting a question that references a lot of code it will timeout waiting for a response. Its not a problem with smaller models that can fit in a GPU, but when using something like deepseek r1 split between cpu and gpu the amount of time it can take before it starts responding can be a couple of minutes.

Along the same lines if a request takes too long while it is in the thinking stage it will automatically cancel the request instead of allowing it to finish.