Femi

Assuming everything is satisfactory and gets merged I can also take up #6094 @dbczumar, since it looks stale.

#6731 #6732 Have been merged

I believe you can do something like this to create a deployment with 3 replicas kubectl create deployment test --replicas 3

Is using the `approx-0_5` feature in ndarray a potential solution to the approx duplication?

Otherwise, the approx bump is coming in v0.16.0 of ndarray

with guidance I am willing to do a blas vs non blas comparison for linfa-clustering.

@YuhanLiin Linfa Clustering results attached as zip. Conditions: Each test was run with computer fully charged, plugged in, and without it being in use. Commands: 1st run - ` cargo...

Also for linfa_linear is the expectation that the written benchmark just use the diabetes dataset (n=1599)? Or do we want to do something akin to a small,mediu, large benchmark say...

> We should go with the latter. > > BTW, for the `linfa-clustering` benchmarks, the "previous benchmark" is the one with BLAS enabled right? Okay I'll start a benchmarking PR...

The zip file is also attached. Will dump flamegraph in #262. [criterion.zip](https://github.com/rust-ml/linfa/files/9967661/criterion.zip) ### Context - Laptop fully charged - Laptop not in use during test - Laptop not over heating...