Using GPT-4-Turbo I am not always getting the same output when using the same image, same prompt, and a temperature of 0. How come?
#Response not the same with 0 temperature
placid mist
stable flumeBOT
Hi @placid mist. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:
The behavior you're experiencing might be due to several factors. Even though you're using the same prompt and a temperature of 0, there are other parameters that could affect the output of the model.
Firstly, ensure that your prompt is exactly the same each time. Even slight differences, such as an extra space or newline character, can lead to different outputs.
Secondly, make sure you're using the same parameters in all cases. For instance, the model parameter set to gpt-3.5-turbo and gpt-4 will produce different completions even with the same prompt, because gpt-4 is a newer and more capable instruction-following model.
Lastly, it's important to note that not all models support reproducible outputs. For instance, only the new GPT-4 Turbo models support reproducible outputs. If you're using a model that doesn't support this feature, you might experience different outputs even with the same parameters.
If you've double-checked all of these things and are still seeing discrepancies, you can ask for help on the Community Forum (https://community.openai.com/), where users may have experienced similar issues or may be able to assist in troubleshooting your specific case.
Sources:
- Help Article > En > Articles (https://help.openai.com/en/articles/6643200-why-am-i-getting-different-completions-on-playground-vs-the-api)
- Forum Post: No system fingerprints for GPT-4-0613 > Official Answer (https://community.openai.com/t/648338)