I have developed a langchain summarization using chatgpt api 3.5 turbo , when i use different type of keys for the summrization the structre of the summary is changing. how can i short out this problem ?
What temperature are you using? A temperature of 0 gives the most consistent responses between requests.
i use tempreture as 0 but still my responses are getting different when using different type of apis
Did you originally mean to say models? Or API keys?
Different API keys ( I used Different type of Azure Open AI APIs)