Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings

Open in new window