Token usage

[alfresanz]

Hi, I would like to use guidance to obtain easy to parse completions. However, I was wondering how guidance actually works when calling the OpenAI's API (I tried looking at the source code, but I'm just an amateur, so I couldn't arrive at any conclusion). Particularly, I would like to know if guidance calls OpenAI's API iteratively for each {{gen}} item. Because if that is the case, then a prompt of 100 tokens and 3 {{gen}} items would result in three calls of the OpenAI's API, so the token consumption would be 300+. Since the pricing is based on tokens, you can understand my interest on this, and I couldn't find any discussion or reference on the readme.

Thank you very much.