Run a NERDD module as a
service that consumes input molecules and produces prediction tuples.

pip install -U nerdd-link

When a class inherits from nerdd_module.AbstractModel (see NERDD Module Github page), it can be used to create a Kafka service.

# run a Kafka service for NerddModel on localhost:9092
run_nerdd_server package.path.to.NerddModel

# modify broker url, input topic and batch size
run_nerdd_server package.path.to.NerddModel \
  --broker-url my-cluster-kafka-bootstrap.kafka:9092 \
  --input-topic examples \
  --batch-size 10

# more information via --help
run_nerdd_server --help

If the model class is called ExamplePredictionModel, the server will read input tuples from the input topic example-prediction-inputs in batches of size 100 and write results to the results topic. The batch size specifies the number of input tuples that are given to the model at once.