https://promptfoo.dev/ logo
different LLM providers for running prompt v/s eva...
# ❓|questions
r
Is there a way to use to use different LLM providers for running prompts and getting output to evaluate v/s for evaluating the ouput?
in other words is there a way I can execute the prompt and then provide the prompt, LLM output and any other optional data to promptfoo to run evaluation using a LLM provider which is different than the LLM the executed the initial prompt
i noticed that in the command line promptfoo eval there is an option --model-outputs . I am using the node package. How can I pass the model output with the node package?
i
yes, see "Overriding the LLM grader" here