谷歌近日推出了一款名为 ASPIRE 的模型训练框架，专门为大语言模型设计。该框架旨在提高 AI 的选择性预测能力，使得 AI 能够自主判断其输出内容的正确性。谷歌的研究人员表示，通过 ASPIRE 框架的训练，大语言模型的输出准确率得到了显著提升。即使是对较小规模的模型进行微调，也能够帮助它们做出“准确且有自信”的预测。这一进展对于推动 AI 在内容生成领域的应用具有重要意义，尤其是在提升信息准确性和可靠性方面。
Title: Google’s ASPIRE Framework Enhances AI Content Accuracy
Keywords: Google AI, ASPIRE Framework, Content Accuracy
News content:
Google recently launched a model training framework called ASPIRE, specifically designed for large language models. This framework aims to enhance AI’s selective prediction capabilities, allowing AI to autonomously assess the correctness of its output content. Researchers at Google claim that through ASPIRE framework training, the output accuracy of large language models has been significantly improved. Even for smaller models, fine-tuning with ASPIRE can help them make “accurate and confident” predictions. This advancement is significant for advancing the application of AI in content generation, particularly in enhancing the accuracy and reliability of information.
【来源】https://www.ithome.com/0/746/717.htm
Views: 1