news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

News on August 21, local time on August 20,OpenAIIt announced that it will launch a "personalized customization service" for enterprises, and enterprises can use their own data to fine-tune and "customize" OpenAI.

According to John Allard, a software engineer at OpenAI who is responsible for model customization, fine-tuning training takes an average of one to two hours, and initially users can only fine-tune the model using text-based data, not images or other content.

Fine-tuning is an important means of AI model optimization. Its core is to optimize the model through targeted training while retaining a wide range ofknowledge baseBased on this, it has the ability to quickly adapt and optimize the processing of specific tasks to improve the completion and accuracy of the tasks.

OpenAI said,GPT-4The fine-tuning function is one of the most anticipated features for developers. Developers can now fine-tune GPT-4o using custom datasets, which can achieve higher performance improvements at a lower cost and accurately meet the needs of various specific business scenarios.

It is reported that GPT-4o and its predecessor GPT-4 have not provided fine-tuning functions before. As the competition among startups in the field of commercial AI products becomes increasingly fierce, corporate customers are increasingly looking forward to the return on investment of AI. OpenAI's new move also shows that the competition among major AI model development companies for corporate customers is intensifying. (Xianxian)

This article is from NetEase Technology Report. For more information and in-depth content, follow us.