Same thing happened 4 days ago with “GPT-4” but it was fixed 2 days later. Now its both models. Just raising the flag so it can get fixed.
1 Like
Example of an output (GPT-4-Preview):
"Heading Application Head PR Samino:aging Expert in Communication Strategic
: a background TV, creation and planning art culture, am to Head PR atl.dk My in diverse and teams achieve outcomesips uniquely narr your compelling and significant fores"
Hi!
Can you describe what parameters you’re using?
streaming? temperature? what environment?
we had a similar issue with axios streaming where a programming bug caused the system to miss some SSE frames. If your issue is that tokens are being skipped, it’s possible that you have an efficiency problem in your application.
Thank you for the answer.
Our robots work when i change the API key to GPT-3.5, which is why i dont think the problem is from our end.
Also, we had similar problem 4 days ago, which got solved after 2 days automatically and OpenAI later reported an issue in their API.
We are experiencing the same issue with GPT-4 only
We had to change the code, so the text is not being written fluently anymore. I think that is the safest approach for now.