opensearch-project / opensearch-neural-sparse-encoding-doc-v2-mini

Overall, the v2 series of models have better search relevance, efficiency and inference speed than the v1 series. The specific advantages and disadvantages may vary across different datasets.