"Today we learned from the Internet that the review report 'A Roadmap for Big Model' (Big Model Roadmap) published by Zhiyuan Research Institute on the pre-print website arXiv is suspected of plagiarism. In this case, the institute immediately organized an internal investigation, confirmed that there were problems in some articles, and then launched the invitation of third-party experts to carry out independent review and relevant accountability. On the evening of April 13, the Zhiyuan Research Institute publicly responded to the recent hotly discussed large-scale academic review report on Sino-foreign cooperation suspected of plagiarism.
"A Roadmap for Big Model" is a 200-page academic review report jointly completed by many universities and enterprises at home and abroad, with as many as 100 authors, from domestic universities such as Tsinghua University and Peking University, foreign universities such as Columbia University and University of Montreal, byteDance, Huawei, JD.com, Tencent and other enterprises, as well as institutions such as the Chinese Academy of Sciences and Beijing Zhiyuan.
"On April 13, we learned that Google researcher Nicholas Carlini pointed out on his personal blog that the report had plagiarized several paragraphs of their paper, as well as other paragraphs and statements that had plagiarized other papers. We checked this item by item, and after checking and confirming that the 179 words in section 3.1 of article 2, the 74 words in section 3.1 of article 8, the 55 words in section 2.3 of article 12, the 159 words in section 2 of article 14, and the 146 words in section 1 of article 16 were duplicated with other papers and should be plagiarized. We have decided to remove the content from the report immediately, and the report revision will be submitted to arXiv for update today. Authors of all articles have been notified to conduct a comprehensive review of all content, and new versions will be released after rigorous review. In a public apology letter, the Zhiyuan Research Institute introduced the current findings.
According to the Zhiyuan Research Institute, the report is a review of the field of large models, hoping to cover all important literature in this field at home and abroad as much as possible, led by the Zhiyuan Research Institute, responsible for the framework design and manuscript summary, and invited 100 researchers at home and abroad to write 16 independent special articles, each article invited a group of authors to write and sign individually, a total of 257 pages. After the report was released, it continued to be revised and improved according to the feedback, and by April 2 on the arXiv website it had been updated to the third edition.
In addition to re-examining the article, Zhiyuan Research Institute said that it will next launch an independent review of the report by inviting third-party experts to conduct an independent review of the report and hold relevant responsible persons accountable based on the formal findings. At the same time, we will further improve the system management, and through a more stringent review mechanism and more clear disciplinary measures, strengthen the education of study style for the scientific research personnel within and supported by the institute to prevent the recurrence of similar incidents.
Attached: Letter of apology from Beijing Zhiyuan Artificial Intelligence Research Institute on the issue of the "A Roadmap for Big Model" roundup report
Today we learned from the Internet that the review report "A Roadmap for Big Model" (Big Model Roadmap) published by Zhiyuan Research Institute on the pre-print website arXiv is suspected of plagiarism. In this case, the institute immediately organized an internal investigation, confirmed that there were problems in some articles, and then launched the invitation of third-party experts to carry out independent review and relevant accountability.
We feel deeply guilty about this problem. As a scientific research institution, Zhiyuan Research Institute attaches great importance to academic norms, encourages academic innovation and academic exchanges, and has zero tolerance for academic misconduct. Hereby, we would like to express our sincere apologies to the authors of the original texts and colleagues and friends in academia and industry.
The preliminary results of the internal investigation of Zhiyuan Research Institute are as follows:
1. The report is a review of the field of large models, hoping to cover all important literature in this field at home and abroad as much as possible, led by The Wisdom Source Research Institute, responsible for framework design and manuscript summary, and invited 100 researchers at home and abroad to write 16 independent special articles, each article invited a group of authors to write and sign individually, a total of 257 pages. After the report was released, it continued to be revised and improved according to the feedback, and by April 2 on the arXiv website it had been updated to the third edition.
2. On April 13, we learned that Google researcher Nicholas Carlini pointed out on his personal blog that the report had plagiarized several paragraphs of their paper, along with other paragraphs and statements that had plagiarized other papers. We checked this item by item, and after checking and confirming that the 179 words in section 3.1 of article 2, the 74 words in section 3.1 of article 8, the 55 words in section 2.3 of article 12, the 159 words in section 2 of article 14, and the 146 words in section 1 of article 16 were duplicated with other papers and should be plagiarized. We have decided to remove the content from the report immediately, and the report revision will be submitted to arXiv for update today. Authors of all articles have been notified to conduct a comprehensive review of all content, and new versions will be released after rigorous review.
3. As the organizer of the report, Zhiyuan should strictly review all the contents of each article, and such a problem is hard to blame. We deeply blame ourselves for this, and especially thank our friends in academia and the media for helping us identify problems. We will deeply learn lessons, rectify the scientific research management and paper publication process, and hope that friends from all walks of life will supervise our work.
In the next step, Zhiyuan Research Institute will take this as a warning and take practical measures to strengthen the construction of scientific research integrity and study style:
(1) Initiate an independent review of the report by inviting third-party experts to conduct an independent review of the report on the same day, and hold the relevant responsible persons accountable based on the formal investigation results.
(2) Further improve the system management, through a more stringent review mechanism and more clear disciplinary measures, strengthen the education of study style for the scientific research personnel within and supported by the institute, and prevent the recurrence of similar incidents.
We welcome friends from all walks of life to continue to strictly supervise our work in the future, and to criticize and correct possible omissions and deficiencies in our work.
Beijing Zhiyuan Artificial Intelligence Research Institute
April 13, 2022
![20 years of old friends reunited! Zhang Hongjiang joined Tsinghua AIR led by Zhang Yaqin as an outstanding visiting professor[fig]](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACwAAAAAAQABAAACAkQBADs=)