Dmitry Kan

Hey guys, I wanted to ask for your advice: not sure if something is misconfigured on my side, but some of the algorithms run into timeout issue. Here is one...

I tweaked the sleeping schedule of my machine -- still same issue. Will upgrade docker next from 3.3.1 to more recent one. Also noticed stable ram insufficiency for Puffin algorithm....

Hello! Is there some progress on releasing the new framework? Specifically I'm interested to see the I/O libraries to handle loading of 100M or 1B vectors. When using for instance...

sounds great! Looking forward to test it.

in 1.10.0 there is no num_client. On your dashboard screenshot I can see 8 processes?

btw, that does not seem to work on Python 3.8.8

Hi @nishithbenhur If you use BERT to compute document vector in preprocessing phase, you can store such vectors in a search engine, like Solr or Elasticsearch. You have two major...

@nishithbenhur take a look at this plugin, it returns the score in the range [0, 1]: https://github.com/saaay71/solr-vector-scoring

@nishithbenhur here is the code behind cosine calculation: https://github.com/saaay71/solr-vector-scoring/blob/master/src/com/github/saaay71/solr/VectorScoreQuery.java#L71 It is normalized by norms of query and document's vectors => it is bounded [0, 1] value.

@nishithbenhur I think you can achieve this by using cosine similarity and simply controlling what to show on the client side. Does this make sense? Also, I have just published...