ywmario

5 ms·

1. 1.▲
   by ywmario 3y ago

   Judging from the qps and cost columns, milvus/zilliz is much more efficient! It's probably a good choice for large scale data. It would be nice to have another column to compare the largest scales these DBs can support.
2. 2.▲
   by ywmario 3y ago

   probably required by some laws
3. 3.▲
   by ywmario 3y ago

   I have been under the impression that the embedded vector is the one actually matters. Token is just another format.
4. 4.▲
   by ywmario 4y ago

   Yes, there will be a lot of 10x engineer
5. 5.▲
   by ywmario 4y ago

   Milvus standalone is pretty neat. I made it successfully run on the first try.
6. 6.▲
   by ywmario 4y ago

   I'm very sure milvus supports it. Milvus calls it hybrid search though: https://milvus.io/docs/hybridsearch.md#Conduct-a-hybrid-vect...