sail-experiments-hpo.yaml

settings:
  name: sail-hyperparameter-optimization-experiment
  description: Sail model implemented in TF using Hyperparameter Optimization
training_references: 
- name: model1
  training_definition_url: https://ibm-watson-ml.mybluemix.net/v3/ml_assets/training_definitions/<REPLACE WITH UUID>
  command: python3 sail_network.py --dataFile ${DATA_DIR}/sail.csv.gz
  compute_configuration:
      name: k80
  hyper_parameters_optimization:
    method:
      name: random
      parameters:
      - name: objective
        string_value: accuracy
      - name: maximize_or_minimize
        string_value: maximize
      - name: num_optimizer_steps
        int_value: 4
    hyper_parameters:
    - name: learning_rate
      double_range:
        min_value: 0.005
        max_value: 0.01
        step: 0.001
    - name: NUM_LAYER1_NODES
      int_range: 
        min_value: 5
        max_value: 6
        power: 2
    - name: training_iters
      int_range:
        min_value: 100000
        max_value: 300000
        step: 1000
    - name: dropout
      double_range:
        min_value: 0.65
        max_value: 0.85
        step: 0.01
training_results_reference:
  name: sail_training_results
  connection: 
    endpoint_url: <INSERT endpoint_url found in your cloud storage endpoint>
    access_key_id: <INSERT access_key_id found in your cloud storage credentials>
    secret_access_key: <INSERT secret_access_key found in your cloud storage credentials>
  target:
    bucket: <INSERT name of training_results bucket found in your cloud storage>
  type: s3