Impressive LMEB performance with only 137M parameters
#3
by Yuki131 - opened
Congratulations on the excellent LMEB results! With only 137M parameters, opensearch-neural-sparse-encoding-doc-v3-gte achieves a mean score of 56.99, outperforming or matching many substantially larger models. This is a very impressive balance of effectiveness and efficiency—great work!
https://huggingface.co/proxy/mteb-leaderboard.hf.space/benchmark/LMEB
