I would like to know which parameter to pass to the model to return the generated tokens and associated logprobs? As I am doing a comparison of these with OpenAI's models.
Apologies if I have missed something obvious here, but I am using a vLLM deployment of `Mistral7B-V.01' in GCPs Model Garden.
I would like to know which parameter to pass to the model to return the generated tokens and associated
logprobs
? As I am doing a comparison of these with OpenAI's models.Apologies if I have missed something obvious here, but I am using a vLLM deployment of `Mistral7B-V.01' in GCPs Model Garden.