Implement the a way to "continue generating" with the API

ChatGPT Feature requests
,
1

I think it would be awesome to have an easy way (with the API) to continue generating when the generation stopped for length reasons.

For instance, I’m translating a very long json file, and it seems very complicated (even impossible?) to continue the generation where it stopped. With the user interface, the “continue generation” button made it possible.

I can see that ChatGPT UI uses a custom API that takes an “action” parameter (which value is “next” or “continue”) along with a conversation identifier. Could it be possible to have the same with the Chat completion API?

Thanks for your great work

6 Likes
2

I’m also waiting for this problem, many of my requests are timed out because no API feedback on this feature and in fact, I can’t get any results, and the tokens are counted down, I put everything on hold in waiting for a solution…

2 Likes