据卡内基梅隆大学的研究人员与 BerriAI 的企业研究人员最近的一项研究，谷歌Gemini Pro模型在所有任务中的准确度与OpenAI的GPT-3.5 Turbo当前版本相比，虽然达到可比但稍差的准确度。
谷歌发言人回应称，内部研究显示Gemini Pro的性能优于GPT-3.5，而即将于2024年初推出的更强大版本Gemini Ultra将进一步提高得分，超过GPT-3.5。
该研究结果引发了业界关注，人们纷纷期待谷歌在未来能够推出更加先进的模型，以满足不断变化的市场需求。
英文标题：Google’s Gemini Model Performs Worse Than GPT-3.5 Turbo
关键词：Google，Gemini，GPT-3.5 Turbo
新闻内容：
Recently, a group of researchers from Carnegie Mellon University and a company called BerriAI conducted a study and found that the Google Gemini model is not as good as the OpenAI GPT-3.5 Turbo at all. The study was published on December 19th, 2023. According to the research, the Gemini Pro model performs similarly to the GPT-3.5 Turbo but with a slightly lower accuracy.
Google released a response, stating that the internal research indicates that the Gemini Pro model performs better than the GPT-3.5, and the more powerful version of Gemini, called Gemini Ultra, will be released in early 2024, with an even higher score.
The findings of this study have caused concern among the public, who are looking forward to Google’s future releases to meet the ever-changing market demands.
【来源】https://venturebeat.com/ai/google-gemini-is-not-even-as-good-as-gpt-3-5-turbo-researchers-find/
Views: 1