Kaikai Wang
π Dalian University of Technology Β· School of Software Engineering
π Dalian, China
π§ Email: sazeracwang@gmail.com
π GitHub | Google Scholar | LinkedIn
π§βπ» About Me
Hello! My name is Kaikai Wang, and I’m currently studying Software Engineering at Dalian University of Technology. My research interests are primarily focused on Large Language Model (LLM) Fine-tuning and Optimization, Social Network Simulation, and Knowledge Retrieval and Augmented Generation (RAG).
I am dedicated to combining deep learning techniques with humanities knowledge to explore the social value of AI in new media.
π Research Areas
- Large Language Models (LLMs)
- Fine-tuning and adaptation of the LLaMA series models
- Efficient parameter fine-tuning methods such as LoRA and VeRA
- Retrieval-Augmented Generation (RAG)
- Vector database retrieval combined with LLM question-answering
- LangChain & multi-turn dialogue systems
- Social Network Simulation
- Simulating the impact of social hot topics
- Building personalized agent models
π Research & Project Experience
π¬ Large Language Model Fine-tuning & Optimization
- Used HuggingFace Transformers + PEFT for LoRA fine-tuning.
- Implemented distributed training (torchrun), quantization (QLoRA), and Flash Attention.
- Explored weight fusion (VeRA) technology to improve model generalization and adaptation capabilities.
𧩠Conversational Q&A System (LangChain)
- Developed a system based on the LangChain framework, combining Tongyi Qianwen with a vector database.
- Implemented prompt templating, historical record management, and multi-turn dialogues.
- Applied the system to assist with scientific literature and educational Q&A scenarios.
π Honors & Awards
- π National Scholarship
- π Academic Excellence Scholarship
- π “Merit Student” Honor
βοΈ Publications
N/A
π Last Updated: August 25, 2025