Open alex2awesome opened 1 year ago
Hi,
I understand that there are reasons why we only want to do indexing once, since there are corpus-level statistics that need to be calculated.
But is there any way to index a huge batch of documents, then index a few more, assuming they are from the same distribution?
Alex
It's not possibile atm. I will consider adding this feature in the future.
Elias
It's not possibile atm. I will consider adding this feature in the future. Elias
Hello, when will this feature be added?
Hi,
I understand that there are reasons why we only want to do indexing once, since there are corpus-level statistics that need to be calculated.
But is there any way to index a huge batch of documents, then index a few more, assuming they are from the same distribution?
Alex