How to generate different texts through chatgpt and combine them
Status: Closed · Asked by Ramon Tissler on · 0 views
I wonder what the best way is to generate an outline of an article title through chatgpt and then generate paragraphs for each outline element and finally connect them together. I just need tips to how to loop the outline results?! Really appreciate any help
Hi @rtissler,
You can follow the steps given in the workflow shared below to achieve this use case.
https://connect.pabbly.com/workflow/share/D0dYagJVBmVRG1M5AlFQdw8bBAAIUgE8UksFFwEOAXkBT1YTBUQNZ10RUzkESwFgAElVP1EOVDQBFVFVCF5WJVdNAxEEBwZ8BEkAM1QdWGsPUVhwAmM#
No, this is not solving the problem. The outline generated consists of 2-15 maybe more sentences, so I cannot map the different elements of an outline. I therefore need to split this outline into the different elements and for each element I need to generate the text. But I do not know how much elements there are. Any solutions to this?
I am afraid the use case you are describing is currently not possible.
This is probably one of the weekest points in pabbly then, compared to zapier. Zapier has these wonderful functions to better then handle json results or results through line itemizer. Would be great if you could consider that!
We do have an Iterator that essentially serves the same purpose and loops through the data, but I am not sure how that will be useful for your particular use case as the outline will be generated in different ChatGPT action steps.
Maybe this is sth that needs more explanation. I am generating one jason file with all data. When I split this data, I get them in x separate fields. How can I then loop through these x textx and do sth with it, e.g. generate a text through chatgpt for each one of them? And again, after that, how can I loop through these results and get it all in one file? Maybe a more in-depth video would be great to see all options in the iterator
As mentioned previously, this particular use case is currently not possible.
I guess this will apply for many use cases. Hope you can have a solution for this in the near future; even with a workaround