The Stanford AI team admitted to plagiarizing the Tsinghua model, publicly apologized and pulled down the controversial project

Global Tech

2024-06-04 14:08Published in Beijing

On June 4, it was reported that recently, a controversy about the plagiarism of open source models was set off on the Internet. The Llama3-V model launched by the Stanford University AI team is accused of plagiarizing the MiniCPM-Llama3-V 2.5, an open-source model jointly developed by Tsinghua University and Facewall Intelligence in China. As soon as this matter was exposed, it immediately attracted widespread attention in the technology community.

As the incident unfolded, two core members of the Stanford Llama3-V team, Siddharth Sharma and Aksh Garg, issued an official apology on social platforms. They acknowledged academic misconduct during the development of the Llama3-V model and sincerely apologized to the MiniCPM team.

Aksh Garg said in a statement: "First of all, we would like to apologize to the original author of MiniCPM. I, Siddharth Sharma, and Mustafa were co-responsible for the release of Llama3-V. Mustafa was in charge of writing the code for the project, but we have not been able to contact him since yesterday. Siddharth and I were mainly responsible for assisting Mustafa in promoting the model. Both of us had consulted the latest research papers to verify the novelty of the project, but we were not informed or aware of any previous work on OpenBMB. We are deeply sorry for this and are disappointed that we have not been able to fully verify the originality of the project. We are willing to take full responsibility for what has happened and have decided to remove the Llama 3-V model. Apologies again to everyone for our actions. ”

In addition, Christopher David Manning, director of the Stanford Artificial Intelligence Lab, also publicly condemned the plagiarism incident and spoke highly of China's MiniCPM open-source model.

View original image 20K

•