I've used "raphaelsty/neural-cherche-colbert" model for reranking of results and the results are good. But it is slow , an Index made using this model would be a perfect solution.
Also if anyone could please help me in using all_mpnet_base_v2 transformer model as the base model for colbert indexing instead of bert_base_uncased, all_mpnet_base_v2 is just too good , it should be the default choice when making index and retrieving results.
I've used "raphaelsty/neural-cherche-colbert" model for reranking of results and the results are good. But it is slow , an Index made using this model would be a perfect solution.
Also if anyone could please help me in using all_mpnet_base_v2 transformer model as the base model for colbert indexing instead of bert_base_uncased, all_mpnet_base_v2 is just too good , it should be the default choice when making index and retrieving results.
Thanks