Fine-Tuning
Easily fine-tune a model using your request logs inside of Helicone.
Fine-tuning allows you to reduce your costs and improve your applications performance - all using the data already inside of Helicone. Read more about it here: OpenAI’s fine-tuning guide
1
Create a dataset
Navigate to the request page and use filters to get your desired dataset
![Filtering requests on the request page to create a dataset in Helicone.](https://mintlify.s3-us-west-1.amazonaws.com/helicone/images/features/fine-tuning/filters.png)
Press Create Dataset
and name it
![Creating and name your dataset in Helicone.](https://mintlify.s3-us-west-1.amazonaws.com/helicone/images/features/fine-tuning/create-dataset.png)
You need at least 10 requests within a dataset to fine-tune.
2
Start a fine-tuning job
Navigate to the fine-tune page and creete a new fine-tuning job.
![Starting a fine-tuning job in Helicone.](https://mintlify.s3-us-west-1.amazonaws.com/helicone/images/features/fine-tuning/fine-tune-job.png)
3
Replace your model name
Once your fine-tuning job is completed, all you have to do is replace the model name and you can start using the new fine-tuned model!
![Python integration example showing how to replace the model name after fine-tuning in Helicone.](https://mintlify.s3-us-west-1.amazonaws.com/helicone/images/features/fine-tuning/model.png)
Example Python integration