Spaces:
Running
Running
Add new multilangue reranker to the benchmark
#1
by
arnauld-a
- opened
I really love your work Antoine Louis. However I found a multi-reranker which seems to be very efficient. Could you add it to the benchmark so that we could check how it performs compare to the other model ?
Model : BAAI/bge-reranker-v2-m3
I found a multi-reranker which seems to be very efficient. Could you add it to the benchmark so that we could check how it performs compare to the other model ?
Model : BAAI/bge-reranker-v2-m3
Done! Indeed, the best cross-encoder so far on mMARCO-fr. I also added the new "gte-multilingual-reranker-base" model from Alibaba-NLP.
antoinelouis
changed discussion status to
closed