How do you format your feedback to the GPT?

[JeremyDudet]

I'm sorry for the basic question.
When I input plain text it often returns an ERROR: Invalid JSON

but how do you format your JSON?
I notice that depending on the command that the GPT is executing the key/value will change.
what is the key you use to submit feedback? instead of returning a simple 'y' or 'n'

[JeremyDudet]

I have been playing with this more, and I'm finding that for some reason, when my feedback is longer, it doesn't return the ERROR: Invalid JSON.

[JeremyDudet]

Here is an example of an ERROR: Invalid JSON when I simply input a 'y'.
This error also appeared when my inputs were one sentence long. Maybe I was simply confusing the AI with a response that required more context for it to understand.

Ultimately this is not a big deal because this error is self-correcting by the AI. I started this thread in case there was something obvious I was missing. But now it seems like this error is somewhat random and not very concerning.

[Anders-Anderson]

I'm encountering the same thing. Short inputs return "Error: Invalid JSON" and long inputs (about 250chars) seem to work fine.

[meesterdude]

i get this less when i run with --gpt3only

[ALB13]

Same issue. Once it happens, it doesn't matter what I do it's always there for every response I provide. Also, it gets stuck here as well:
NEXT ACTION: COMMAND = Error: ARGUMENTS = Invalid JSON

[mikecavallo]

I am getting the same error as well when searching for longer responses. Is it possible to chunk it and split the docs?