Is there a chance that something like this might be implemented?
While it's not a major headache, it is indeed wasteful to issue a request to the LLM when one merely intends to go to the next line while typing. I face this almost every day now and sometimes multiple times a day. Often, I have to hurry and hit the 'stop' button, after which I edit the prompt which I had accidentally submitted instead of going to the next line.
This is wastes money if some tokens have been added to the billed quota and is also wasteful in terms of the resources spent on processing the incomplete prompt.
I suppose it ought to be possible to measure how often the users are editing the prompt shortly after hitting the stop button? Or perhaps using the edit button to only add more lines instead of editing existing text? That way, one could measure the degree to which this affects users and subsequently make an informed decision on whether it might be worth fixing.
