Compatibility with Ann-Benchmarks project about elastiknn HOT 4 CLOSED

Some notes on how to work with the ann-benchmarks project:

• Useful to wrap this line with timing: https://github.com/erikbern/ann-benchmarks/blob/master/ann_benchmarks/runner.py#L77
• Run the benchmark locally (i.e. not in docker container): python run.py --dataset kosarak-jaccard --algorithm elastiknn-exact --local
• Generate plots once there results directory is populated: python plot.py --dataset kosarak-jaccard -o out.png. If you generated the results in a docker image, you might need to run sudo chown -R alex:alex results because the permissions get set to root.
• Original project use c5.4xlarge instance

from elastiknn.

Once #125 is completed, I can use py4j to call models without worry about setting up the elasticsearch server in the same container. That was a major pain and bottleneck the first time I tried this integration.

from elastiknn.

Working on this in #150

from elastiknn.

Closing this since there was a pr merged over in ann-benchmarks recently.

from elastiknn.