structured output and finetunned models

[fossil hare]

If I use a finetunned gpt-4o-mini model can I ude structured output?

[dusty atlas]

Hi @fossil hare. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

---

Yes, you can use structured outputs with a fine-tuned gpt-4o-mini model. The gpt-4o-mini model supports structured outputs, which can be utilized in two forms: function calling and json_schema response format. This capability is available for gpt-4o-mini and its snapshots, such as gpt-4o-mini-2024-07-18.

For more information on structured outputs, you can refer to the Structured Outputs guide.

If you are interested in fine-tuning, you can follow the process outlined in the fine-tuning guide.

If you need to implement structured outputs in your application, you can use the following example code for structured data extraction:

from pydantic import BaseModel
from openai import OpenAI

client = OpenAI()

class Step(BaseModel):
    explanation: str
    output: str

class MathReasoning(BaseModel):
    steps: list[Step]
    final_answer: str

completion = client.beta.chat.completions.parse(
    model="gpt-4o-2024-08-06",
    messages=[
        {"role": "system", "content": "You are a helpful math tutor. Guide the user through the solution step by step."},
        {"role": "user", "content": "how can I solve 8x + 7 = -23"}
    ],
    response_format=MathReasoning,
)

math_reasoning = completion.choices[0].message.parsed

This example demonstrates how to use structured outputs with a model, which can be adapted for a fine-tuned gpt-4o-mini model.

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon: