Can we include the computationally less expensive l1 space?

arvindmsd · February 3, 2021, 6:40pm

Will it make sense to have l1 distance score as well in this other than l2, cosine, hamming bit?

Vijay · February 4, 2021, 8:22pm

Hi @arvindmsd , Absolutely, knn will support l1 distance from 1.13.0 onwards!

arvindmsd · February 8, 2021, 8:37am

That sounds great @Vijay . Is there a tentative date for the release?
Can you point me to the link where I can find out about the set of new features?

Vijay · February 8, 2021, 6:01pm

Sure. You can track release information here and our 1.13.0 release note here .

Vijay · February 18, 2021, 12:44am

@arvindmsd we released odfe -1.13.0 today!

arvindmsd · February 22, 2021, 4:43am

@Vijay Great news. I’ll check it out today

Topic		Replies	Views
The right space_type value for inner product similarity in ANN k-NN	1	761	June 16, 2021
Cosine Similarity Formula k-NN	17	3602	December 29, 2020
Support Inner Product Distance Metrict k-NN	1	634	January 22, 2021
Multiple distance types in same index? k-NN	2	573	January 19, 2021
Is the K-NN comparable with the latest Elasticsearch vector search feature? Open Source Elasticsearch and Kibana	2	1000	October 2, 2019

Can we include the computationally less expensive l1 space?

Related Topics