MTEB evaluation
Hi! We have opened PR for integrating your odel to the mteb, but can't reproduce results. Can you share your script for evalution or look to the PR https://github.com/embeddings-benchmark/mteb/pull/2478?
For reproducing our results (including the performance in the paper and the MTEB leaderboard), please refer to the code at HITsz-TMG/KaLM-Embedding.
It is important to note that the results we provide only include MTEB(en), MTEB(zh), MTEB(fr), MTEB(pl) as mentioned in the model's README.
The results on MMTEB (Multilingual, v1) are not officially evaluated by us, and the specific source and reproduction process may require contacting Hugging Face staff.
We are also in the process of independently evaluating the results on MMTEB, and once the evaluation is completed, we will compare them with the results on Hugging Face.
For reproducing our results (including the performance in the paper and the MTEB leaderboard), please refer to the code at HITsz-TMG/KaLM-Embedding.
For all the instructions used in our evaluation, please refer to the code.