Hello! How does everyone handle processing content that exceeds the 4000 character limits? For example reviewing of a longer form document that you want summarized? Is the only way to chunk in smaller sizes and summarize those chunks THEN summarize the summaries?
Looking for other creative approaches… Thank you for any/all feedback and help
Yeah, the token limits (which actually don’t quite correspond to characters) What are tokens and how to count them? | OpenAI Help Center are based on the model itself and the way it takes a “fragment” of text and “completes” it (which is why the limit applies to input+output combined). So there really isn’t a way to send in any more.
Your approach of summarizing the summaries is a pretty good one. I’ve had some luck in other applications breaking the problem down into smaller steps where the completion from one prompt can be used as helpful data in the next. I’d definitely suggest you keep exploring that direction.