Efficient Cold-Start Recommendation via BPE Token-Level Embedding Initialization with LLM

Open in new window