Testing my models text generation in the Playground UI always performs significantly better than when inferencing using the URL. I add all of the same parameters from the playground. Is there a way to view the hidden code (for the stop sequences, streaming, etc.) used for the playground? Thank you in advance!