Large input summarized text

Prompting
1

I am having large set of questions and answers and I need to summarize them…
the positive news is that it’s working really well and generates at least 80% summarized text well. But, I have been facing a major challenge while doing the summarization for question/answers in OpenAI and this gets even more problematic when the input increases due to the large number of question/answers. In this case, it throws an error about Invalid token length(please see attached). We can do exception handling but is there a way where OpenAI can handle large input and provide the summarized text at once??..also please suggest if I can use any other application(bit cheap though) on the top of my software to get large input summarized text at ONCE?

I have researched about fine tuning option but I think it wouldn’t be appropriate for my use case and there are no examples for now that my model needs to learn from.

I have been curious and researching about this for a few days now but haven’t found a perfect solution for this. I wonder if anyone could please help me with this issue???

2

It would be helpful if you provided examples of what you’re trying to achieve. It’s not entirely clear from your description what problem you’re trying to overcome.

3

My solution to this problem when I was experimenting was to build out a list of text sections capped at x tokens, then have a function that iterates over that list and builds a new list of responses. It’s not a perfect solution, especially because of the potential discontinuity between outputs, but it’s one example of a token limit workaround.

You could also potential have GPT-3 break up the text into various sections, each assigned a value for it’s perceived relevance. Then from there you could use the same list iteration strategy