The behaviour of rollama can be controlled through
options(). Specifically, the options below can be set.
Arguments
Details
rollama_server
This controls the default server where Ollama is expected to run. It assumes
that you are running Ollama locally in a Docker container.
default:
"http://localhost:11434"
rollama_model
The default model is llama3.1, which is a good overall option with reasonable
performance and size for most tasks. You can change the model in each
function call or globally with this option.
default:
"llama3.1"
rollama_verbose
Whether the package tells users what is going on, e.g., showing a spinner
while the models are thinking or showing the download speed while pulling
models. Since this adds some complexity to the code, you might want to
disable it when you get errors (it won't fix the error, but you get a
better error trace).
default:
TRUE
rollama_config
The default configuration or system message. If NULL, the system message
defined in the used model is employed.