The encoding to use by tiktoken when tokenize is called.
Ollama does not provide any API to count tokens, so we use tiktoken to get an estimation of the number of tokens in a prompt.
String encoding;