#playground vs API call

2 messages · Page 1 of 1 (latest)

ripe dust
#

Hi all, I would like to know if any of you are experiencing discrepancies between the output generated from playground and the output generated by using api ChatCompletion. Specifically From playground I get the desired output, while from API call the output (coin same prompt and settings) is much shorter. Where am I going wrong? Has anyone faced this problem before?

keen heart
#

Hi there!
I'm also having issues trying to get ChatCompletion API calls to output like it does in Playground, but not as far as the output being shorter.
Sounds like you confirmed this, but the max_tokens and temperature from Playground should be reflected in your API call. Make sure the model is the same of course.
Not sure it will, but hope that helps. 😊