OpenAI and Scale Partnership for Enterprise Fine-Tuning
OpenAI and Scale have established a partnership to help enterprises customize advanced AI models through fine-tuning. This collaboration enables companies to leverage Scale's data expertise to improve model performance, steerability, and customization for production-grade AI deployments.
Enterprise Fine-Tuning Capabilities
OpenAI provides the capability to securely customize models on proprietary data. As of August 2023, fine-tuning is available for GPT-3.5 Turbo, with plans to extend this capability to GPT-4 in the fall of 2023.
Key data privacy guarantees include:
- Data sent to and from the fine-tuning API is owned by the customer.
- Customer data used for fine-tuning is not used by OpenAI or any other organization to train other models.
Role of Scale as a Preferred Partner
Scale serves as a preferred partner to assist enterprises in the rigorous process of data enrichment and model evaluation required for enterprise-grade functionality. Scale customers can fine-tune OpenAI models using the same methods as they would directly through OpenAI, while gaining access to Scale's enterprise AI expertise and its Data Engine.
Real-World Application
Scale has already applied these capabilities to deliver value for specific customers, including the fine-tuning of GPT-3.5 for Brex.