News
OpenAI Unveils New Fine-Tuning API Features and Expands Custom Model Program for Enhanced AI Development
- By John K. Waters
- 04/05/2024
OpenAI has announced new features for its fine-tuning API that promise to give developers greater command over the customization of AI models. This innovation follows the company's initial introduction of a self-service option for fine-tuning its GPT-3.5 model.
With these enhancements, developers can now create epoch-based checkpoints, reducing the need for repeat trainings, and utilize a comparative playground for direct model assessment. The API also offers integration with third-party platforms and improved validation metrics that provide a fuller picture of a model’s quality.
Highlighting the practical applications of these features, Indeed, the online employment platform, has leveraged the API to refine its job recommendation algorithms, leading to a notable increase in operational efficiency.
OpenAI has also expanded its Custom Model program to include assisted fine-tuning support, aiding organizations in creating models that cater specifically to their unique datasets and operational needs. As a case in point, South Korea's SK Telecom worked with OpenAI to enhance its AI's understanding of customer service inquiries in Korean, resulting in a marked improvement in both accuracy and customer satisfaction.
The company further underscores the significance of developing custom AI models for organizations with substantial proprietary data. This bespoke approach is exemplified by Harvey, an AI-driven legal tool, which collaborated with OpenAI to craft a highly specialized model for the legal domain, achieving a significant boost in performance.
As artificial intelligence becomes more tailored to specific business needs, OpenAI’s suite of development tools stands ready to support a new era of customized AI applications.
Developers interested in these new capabilities are being invited to consult the fine-tuning API documentation or contact OpenAI for more information on custom model partnerships.
About the Author
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at jwaters@converge360.com.