llama 3 · 1 topics

transforming llms into efficient cross encoders via knowledge distillation for rag reranking