4 FALL 2024 NO.87 / THOUGHT LEADERSHIP BRIEF Read all HKUST IEMS Thought Leadership Briefs at http://iems.ust.hk/tlb CONCLUSION InvestLM shows strong capabilities in understanding financial text and typically arrives at a more concise logical investment conclusion compared to state-of-the-art commercial LLMs. Furthermore, we discover that using a small yet high-quality instruction dataset to fine-tune a large foundational model, yields a promising approach for crafting domain-specific LLMs. Finally, we find that generic instructions, like those used in Alpaca can detrimentally impact the performance of instruction-tuned models on domain tasks. This emphasizes the importance of curating domain specific instructions. Together, our findings provide insights into how to fine-tune a foundation model for a specific domain. We release the parameters of InvestLM and adopt the same licensing terms as LLaMA. Yi Yang is an Associate professor in the Department of Information Systems, Business Statistics and Operations Management at Hong Kong University of Science and Technology. He is the Director of the Center for Business and Social Analytics (CBSA). He received his PhD in computer science from Northwestern University. He designs machine learning methods in his research to solve challenging business and Fintech problems. His work has been published in business discipline journals such as Information Systems Research, Management Information Systems Quarterly, Journal of Marketing, Contemporary Accounting Research and INFORMS Journal on Computing. His work has also been published in top-tier machine learning and natural language processing conferences such as Annual Meeting of the Association for Computational Linguistics (ACL), Conference on Empirical Methods in Natural Language Processing (EMNLP) and International Conference on Artificial Intelligence and Statistics (AISTATS). Kar Yan Tam is Vice-President for Administration and Business and Chair Professor of Information Systems, Business Statistics and Operations Management at the Hong Kong University of Science and Technology (HKUST). Prof Tam is known for his contributions in information systems and the diffusion of innovations in organizations. According to Google Scholar, his publications have received over 23,000 citations. Prof Tam is currently serving on the editorial board of a number of academic journals. Prof Tam also plays an active role in public services. He is a member of the Hong Kong Exchange Fund Advisory Committee of the Hong Kong Monetary Authority and the Chairperson of the Hong Kong Committee for the Pacific Economic Collaboration. Yixuan Tang is an MPhil student in Information Systems at the Hong Kong University of Science and Technology. Yixuan has cultivated her interest in Natural language processing (NLP) in Finance. She is particularly passionate about adapting Large Language Models in the Finance domain and mining finance signals from text embedding. She has published in machine learning conferences such as the Conference on Empirical Methods in Natural Language Processing (EMNLP) and the Conference on Language Modeling (COLM). T: (852) 3469 2215 E: iems@ust.hk W: http://iems.ust.hk A: Lo Ka Chung Building, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon With Support from
RkJQdWJsaXNoZXIy NDk5Njg=