AV名湿

Zekun is a Ph.D. student involved in content analysis within the framework of iSAT's Strand 1 projects. His work primarily focuses on implementing domain adaptation methods tailored to the needs of iSAT's domain data. Some of his notable efforts include fine-tuning Large Language Models (LLMs) with adapters for efficiency, leveraging language models for enhanced performance, and employing techniques such as parsing and back-generation for data augmentation. Additionally, Zekun has contributed to creating knowledge graphs, drawing from sentence-level Abstract Meaning Representations (AMRs), and utilizing sensor immersion curriculum materials to produce document-level insights.