1 paper · 1 filter
Weijie Lv, Xuan Xia, Sheng-Jun Huang
Large language models (LLMs) have shown great potential in code-related tasks, yet open-source models lag behind their closed-source counterparts. To bridge this performance gap, e…