method: bayes
metric:
  name: objective/score
  goal: maximize
command:
  - ${env}
  - python
  - -m
  - engine.train
parameters:
  algo:
    value: sac
  total_timesteps:
    values: [50000, 80000, 120000]
  seed:
    values: [13, 42, 77]
  alpha:
    distribution: uniform
    min: 0.15
    max: 0.55
  n_products:
    values: [8, 10, 12]
  lambda_coi:
    distribution: uniform
    min: 0.05
    max: 0.5
  robust_radius:
    distribution: uniform
    min: 0.05
    max: 0.3
  robust_points:
    values: [3, 5, 7]
  info_value:
    distribution: uniform
    min: 0.5
    max: 2.0
  revenue_weight:
    values: [0.005, 0.01, 0.02]
  learning_rate:
    distribution: log_uniform_values
    min: 3.0e-5
    max: 1.0e-3
  gamma:
    values: [0.98, 0.99, 0.995]
  buffer_size:
    values: [50000, 100000, 200000]
  batch_size:
    values: [128, 256, 512]
  tau:
    values: [0.002, 0.005, 0.01]
  train_freq:
    values: [1, 4, 8]
  learning_starts:
    values: [1000, 3000, 5000]