OpenAI’s GPT-4o is getting a fascinating new upgrade. The company has unveiled a new AI model fine-tuning tool that enables developers and organizations to train the model using customized datasets.
OpenAI announced the release of the updated features in a post, highlighting that for certain use cases, the tool will enable developers and organizations to enjoy faster performance at lower costs. Now that developers can use their own datasets to fine-tune GPT-4o, the model can be optimized in terms of both tone and structure of responses, making it possible to adjust it to specific needs. The company claimed that even with only a few dozen cases, significant speed improvements can be achieved.
“Developers can now fine-tune GPT-4o with custom datasets to get higher performance at a lower cost for their specific use cases,” Reads OpenAI’s post. “Fine-tuning enables the model to customize structure and tone of responses or to follow complex domain-specific instructions. Developers can already produce strong results for their applications with as little as a few dozen examples in their training dataset.”
In this context, fine-tuning refers to essentially curating certain datasets to make the large language model (LLM) adaptable to specialized processes while still utilizing the LLM’s full processing capabilities. It functions similarly to AI agents or GPTs, but because it has infinite processing power, it can respond more quickly and with more accuracy.
Users can click on Create in the fine-tuning dashboard that appears in a new window, choose “gpt-4o-2024-08-06” from the base model drop-down menu, and then fine-tune GPT-4o. They will need to use the “gpt-4o-mini-2024-07-18” base model in order to do the same for the mini model. Only developers with paid OpenAI subscriptions will have access to these AI models.
OpenAI makes sure that developers have total control over the fine-tuned models and full ownership of all business data. No data is shared or utilized for training other models. The company has also put in place several safety measures, such as automatic safety assessments and monitoring to ensure that usage guidelines are being followed.
- DJI’s Controversial Move: Ending Drone Geofencing Restrictions - January 15, 2025
- ChatGPT’s Latest Feature: A Smarter Way to Manage Reminders and To-Dos - January 15, 2025
- Microsoft Establishes New Internal AI Organization Aimed at Developers - January 14, 2025
