In April 2025, Jiaqi Li developed LoRA support for WanModel within the ModelTC/LightX2V repository, focusing on parameter-efficient fine-tuning for deep learning workflows. He designed and implemented a LoRA wrapper, a dedicated handling module, and integrated LoRA weight loading into the model’s runtime path. By updating the main script’s argument parsing and model loading logic, he enabled seamless application of LoRA weights, allowing for faster experimentation and reduced computational requirements. Working primarily in Python and leveraging skills in deep learning, machine learning, and model adaptation, Jiaqi’s work laid the foundation for more efficient model variant development and testing.