145928b26625096f10645a52545179b08e96e6c5,model/arguments.py,,parse_arguments,#,20

Before Change

      default=1e-5,
      type=float,
      help="network regularization multiplicative")
  parser.add_argument(
      "--alpha",
      default=1.0,
      type=float,
      help="inverse gamma shape")
  parser.add_argument(
      "--beta",
      default=1.0,
      type=float,

After Change

      default=1e-5,
      type=float,
      help="network regularization multiplicative")
  parser.add_argument(
      "--transition_bias",
      default=None,
      type=float,
      help="The value of p0, corresponding to Eq. (6) in the "
           "paper. If the value is given, we will fix to this value. If the "
           "value is None, we will estimate it from training data "
           "using Eq. (13) in the paper.")
  parser.add_argument(
      "--crp_alpha",
      default=1.0,
      type=float,
      help="The value of alpha for the Chinese restaurant process (CRP), "
           "corresponding to Eq. (7) in the paper. In this open source "
           "implementation, currently we only support using a given value "
           "of crp_alpha.")
  parser.add_argument(
      "--sigma2",
      default=None,
      type=float,
      help="The value of sigma squared, corresponding to Eq. (11) in the "
           "paper. If the value is given, we will fix to this value. If the "
           "value is None, we will estimate it from training data.")
  parser.add_argument(
      "--sigma_alpha",
      default=1.0,
      type=float,
      help="The inverse gamma shape for estimating sigma2. This value is only "
           "meaningful when sigma2 is not given, and estimated from data.")
  parser.add_argument(
      "--sigma_beta",
      default=1.0,
      type=float,
      help="The inverse gamma scale for estimating sigma2. This value is only "
           "meaningful when sigma2 is not given, and estimated from data.")

  // training/testing configurations
  parser.add_argument(
      "--optimizer",

Instances