Abstract: This study attempts to explore and implement a lightweight optimization strategy for large language models in resource-constrained environments. Specifically, it uses model distillation ...