- run-with-backend args backendprocedure
Generates text as described by args with backend. Starting from the beginning-of-sequence token, the prompt tokens are fed in first; after them each next token is sampled from the logits. Tokens are printed as they are produced, followed by the generation rate.