As a crucial step to enhance LLMs alignment with human intentions, Instruction Fine-Tuning (IFT) has a high demand on dataset quality. However, existing IFT datasets often contain knowledge that is inconsistent with LLMs' internal knowledge learned from the pre-training phase, which can greatly affect the efficacy of IFT. To address this issue, we introduce NILE (iNternal consIstency aLignmEnt) framework, aimed at optimizing IFT datasets to unlock LLMs' capability further. NILE operates by eliciting target pre-trained LLM's internal knowledge corresponding to instruction data. The internal knowledge is leveraged to revise the answer in IFT datasets. Additionally, we propose a novel Internal Consistency Filtering (ICF) method to filter training samples, ensuring its high consistency with LLM's internal knowledge. Our experiments demonstrate that NILE-aligned IFT datasets sharply boost LLM performance across multiple LLM ability evaluation datasets, achieving up to 66.6% gain on Arena-Hard and 68.5% on Alpaca-Eval V2. Further analysis confirms that each component of the NILE}framework contributes to these substantial performance improvements, and provides compelling evidence that dataset consistency with pre-trained internal knowledge is pivotal for maximizing LLM potential.
Internal Consistency and Self-Feedback in Large Language Models: A Survey
School of Information, Renmin University of China(中国人民大学信息学院);Large Language Model Center, Institute for Advanced Algorithms Research(高级算法研究院大语言模型中心);School of Computer Science and Technology, Beijing Institute of Technology(北京理工大学计算机科学与技术学院);State Key Laboratory of Media Convergence Production Technology and Systems, Xinhua News Agency(新华社媒体融合生产技术与系统国家重点实验室)
Towards Fully Exploiting LLM Internal States to Enhance Knowledge Boundary Perception
CAS Key Lab of Network Data Science and Technology, ICT, CAS(中国科学院信息科技研究院网络数据科学与技术重点实验室);State Key Laboratory of AI Safety(人工智能安全国家重点实验室);University of Chinese Academy of Sciences(中国科学院大学)
Bridging Internal Probability and Self-Consistency for Effective and Efficient LLM Reasoning
National Key Laboratory for Novel Software Technology, Nanjing University(南京大学计算机软件新技术国家重点实验室);School of Artificial Intelligence, Nanjing University(南京大学人工智能学院);School of Intelligence Science and Technology, Nanjing University(南京大学智能科学与技术学院)
Do LLMs Need to Think in One Language? Correlation between Latent Language and Task Performance
NAIST(日本国立信息与科技研究所);NII LLMC(日本信息处理学会大语言模型中心);MBZUAI(微软亚洲人工智能研究院);RIKEN(日本研究机构);Tohoku University(东北大学);The University of Tokyo(东京大学);Nagoya Institute of Technology(名古屋技术大学)
On the Entity-Level Alignment in Crosslingual Consistency
Center for Information and Language Processing, LMU Munich(信息与语言处理中心,慕尼黑大学);Munich Center for Machine Learning (MCML)(慕尼黑机器学习中心(MCML));Sorbonne Université, CNRS, ISIR, France(索邦大学,CNRS,ISIR,法国)
How Do Language Models Process Ethical Instructions? Deliberation, Consistency, and Other-Recognition Across Four Models
Research Institute of Criminal Psychiatry / Sex Offender Medical Center(刑事精神病学研究所/性犯罪者医疗中心);Department of Neuropsychiatry, Kyoto University(京都大学神经精神医学系)