As title, when I used the GPT API in my tests, I've observed variations in computation time.
For instance, processing 420 data took 25 minutes, while 954 data took 9 minutes.
Have you ever encountered a similar situations?
What is a reasonable amount of time for inference?
As title, when I used the GPT API in my tests, I've observed variations in computation time. For instance, processing 420 data took 25 minutes, while 954 data took 9 minutes. Have you ever encountered a similar situations? What is a reasonable amount of time for inference?