Merge and Branch: Efficient Tree-Structured Supervised Fine-Tuning of LLMs for Recommendation
LLM-based generative recommendation reads long, text-rich interaction histories but generates only short semantic identifiers (SIDs). This asymmetry makes supervised fine-tuning inefficient: encoding the history is expensive, yet each history usually provides supervision for only one target. We present Merge and Branch...
Yun-Qin Zhu, Li-Yang Shu, Yu-Si Zhang et al.
· 0 citations