Large Language Models
暫譯: 大型語言模型

您是否渴望探索人工智慧的最新突破，特別是大型語言模型（LLMs）領域？這本書是您理解LLMs核心基礎和進階技術的最佳指南。

這本全面的資源提供了對LLM發展的完整理解，從預訓練到微調。它詳細說明了經典的Transformer架構、其在LLMs中的適應性，以及完整的訓練過程，包括數據收集、清理和準備。讀者還可以從書中學習如何微調LLMs以遵循人類指令，並與人類的價值觀和意圖對齊，確保更安全和更具倫理的AI行為。此外，它幫助讀者發現有效的提示策略，例如上下文學習和思維鏈，以增強LLM的能力並解決複雜任務。

這本書適合初學者和經驗豐富的專業人士，是導航LLMs動態領域的寶貴資源，提供了對該主題的簡明而全面的探索。

翻譯最初是使用人工智慧完成的。隨後進行了全面的人類修訂，以確保整本書內容的準確性和一致性。

Wayne Xin Zhao is a professor at Gaoling School of Artificial Intelligence, Renmin University of China. His research areas include natural language processing, information retrieval, and data mining, with a particular focus on large language models. Xin graduated from Harbin Institute of Technology in 2008 and earned his PhD from Peking University in 2014. He has published more than 200 technical papers in top international conferences and journals, accumulating more than 29,000 citations according to Google Scholar. His contributions have been honored with awards, such as the ECIR 2021 Test-of-time award and EACL 2024 Evaluation and Model Insight Award. Xin has also regularly served as the area chair or senior program committee member for prominent conferences. He is the lead author of the survey paper "A survey of large language models," which provides a comprehensive overview of the field.

Kun Zhou obtained his Ph.D degree at School of Information, Renmin University of China in 2024. His research interests encompass natural language processing and multimodal systems, with focuses on large language models and their applications in complex scenarios. Kun has published more than 40 papers at top conferences and journals, gathering more than 9,000 citations according to Google Scholar. Kun has been awarded by MSRA Fellowship, Baidu Scholarship, Bytedance Scholarship, Baosteel Scholarship, and EACL 2024 Evaluation and Model Insight Award.

Junyi Li is a postdoctoral researcher at School of Computing, National University of Singapore, Singapore. His research interests center around natural language processing and multi-modal systems, with an emphasis on large language models and their applications. Junyi received his PhD degree from Renmin University of China, supervised by Prof. Xin Zhao and a second PhD degree from Université de Montréal, advised by Prof. Jian-Yun Nie. He has published several technical papers at top international conferences and journals including ACL, SIGIR, EMNLP, and NAACL, accumulating more than 6,500 citations according to Google Scholar. Junyi has been awarded National Scholarship at 2019 and 2021 and 2024 Outstanding Graduates. Junyi has also served as the program committee member for several prominent conferences and journals, including ACL, EMNLP, AAAI, and ACM Computing Survey.

Tianyi Tang is a senior algorithm engineer at the Qwen Team, Alibaba Group. His research interests include natural language processing and large language models. He received both his M.E. and B.E. degrees from Renmin University of China, under the supervision of Prof. Wayne Xin Zhao. Tianyi has authored over 20 research papers in top journals and conferences such as ACM Computing Surveys, ACL, EMNLP, and NAACL, amassing more than 6,900 citations according to Google Scholar. He leads the LLMBox project, a comprehensive code library that provides researchers with a convenient and effective toolkit for training and utilizing large language models. Additionally, he has achieved four silver medals in ACM-ICPC contests.

Ji-Rong Wen is a full professor, and Executive Dean of the Gaoling School of Artificial Intelligence at Renmin University of China. With extensive experience in big data and AI, he has an impressive publication record in renowned international conferences and journals, amassing than 41,000 citations. Prof. Wen served as the PC Chair of SIGIR 2020 and was the Associate Editor of ACM TOIS and IEEE TKDE. He spent 14 years at Microsoft Research Asia (MSRA), where he was a Senior Researcher and Group Manager of the Web Search and Mining Group. In 2013, he joined Renmin University of China to lead the big data and AI research, especially interdisciplinary research between AI and social sciences & humanities. He was elected as a National Distinguished Expert in 2013 and Beijing's Distinguished Young Scientist in 2018. Prof. Wen also holds the position of Chief Scientist at the Beijing Academy of Artificial Intelligence.

韋恩·辛·趙（Wayne Xin Zhao）是中國人民大學高靈人工智慧學院的教授。他的研究領域包括自然語言處理、資訊檢索和資料挖掘，特別專注於大型語言模型。辛於2008年畢業於哈爾濱工業大學，並於2014年獲得北京大學的博士學位。他在頂級國際會議和期刊上發表了超過200篇技術論文，根據Google Scholar的資料，累計引用次數超過29,000次。他的貢獻獲得了多項獎項的表彰，如ECIR 2021的終身成就獎和EACL 2024的評估與模型洞察獎。辛教授還定期擔任重要會議的領域主席或高級程序委員會成員。他是調查論文《大型語言模型的調查》的主要作者，該論文提供了該領域的全面概述。

周坤（Kun Zhou）於2024年在中國人民大學資訊學院獲得博士學位。他的研究興趣涵蓋自然語言處理和多模態系統，專注於大型語言模型及其在複雜場景中的應用。周坤在頂級會議和期刊上發表了超過40篇論文，根據Google Scholar的資料，累計引用次數超過9,000次。他獲得了MSRA獎學金、百度獎學金、字節跳動獎學金、寶鋼獎學金以及EACL 2024的評估與模型洞察獎。

李俊毅（Junyi Li）是新加坡國立大學計算學院的博士後研究員。他的研究興趣集中在自然語言處理和多模態系統，強調大型語言模型及其應用。李俊毅在中國人民大學獲得博士學位，指導教授為辛·趙教授，並在蒙特利爾大學獲得第二個博士學位，指導教授為聶建雲教授。他在ACL、SIGIR、EMNLP和NAACL等頂級國際會議和期刊上發表了多篇技術論文，根據Google Scholar的資料，累計引用次數超過6,500次。李俊毅於2019年和2021年獲得國家獎學金，並於2024年被評為優秀畢業生。他還擔任多個重要會議和期刊的程序委員會成員，包括ACL、EMNLP、AAAI和ACM Computing Survey。

唐天怡（Tianyi Tang）是阿里巴巴集團Qwen團隊的高級算法工程師。他的研究興趣包括自然語言處理和大型語言模型。他在中國人民大學獲得碩士和學士學位，指導教授為韋恩·辛·趙教授。唐天怡在ACM Computing Surveys、ACL、EMNLP和NAACL等頂級期刊和會議上發表了超過20篇研究論文，根據Google Scholar的資料，累計引用次數超過6,900次。他領導LLMBox項目，這是一個綜合代碼庫，為研究人員提供方便有效的工具包，用於訓練和利用大型語言模型。此外，他在ACM-ICPC比賽中獲得了四枚銀牌。

溫基榮（Ji-Rong Wen）是中國人民大學高靈人工智慧學院的全職教授及執行院長。他在大數據和人工智慧方面擁有豐富的經驗，在知名國際會議和期刊上發表了大量論文，累計引用次數超過41,000次。溫教授曾擔任SIGIR 2020的程序委員會主席，並擔任ACM TOIS和IEEE TKDE的副編輯。他在微軟亞洲研究院（MSRA）工作了14年，擔任高級研究員和網路搜尋與挖掘小組的組長。2013年，他加入中國人民大學，領導大數據和人工智慧的研究，特別是人工智慧與社會科學及人文學科之間的跨學科研究。他於2013年被評選為國家傑出專家，並於2018年被評為北京市傑出青年科學家。溫教授還擔任北京人工智慧學院的首席科學家。