Mf tpe
neps.optimizers.bayesian_optimization.mf_tpe
#
MultiFidelityPriorWeightedTreeParzenEstimator
#
MultiFidelityPriorWeightedTreeParzenEstimator(
pipeline_space: SearchSpace,
use_priors: bool = True,
prior_num_evals: float = 2.5,
good_fraction: float = 0.3334,
random_interleave_prob: float = 0.0,
initial_design_size: int = 0,
prior_as_samples: bool = True,
pending_as_bad: bool = True,
fidelity_weighting: Literal[
"linear", "spearman"
] = "spearman",
surrogate_model: str = "kde",
good_model_bw_factor: int = 1.5,
joint_kde_modelling: bool = False,
threshold_improvement: bool = True,
promote_from_acq: bool = True,
acquisition_sampler: (
str | AcquisitionSampler
) = "mutation",
prior_draws: int = 1000,
prior_confidence: Literal[
"low", "medium", "high"
] = "medium",
surrogate_model_args: dict = None,
soft_promotion: bool = True,
patience: int = 50,
logger=None,
budget: None | int | float = None,
loss_value_on_error: None | float = None,
cost_value_on_error: None | float = None,
)
Bases: BaseOptimizer
PARAMETER | DESCRIPTION |
---|---|
pipeline_space |
Space in which to search
TYPE:
|
prior_num_evals |
[description]. Defaults to 2.5.
TYPE:
|
good_fraction |
[description]. Defaults to 0.333.
TYPE:
|
random_interleave_prob |
Frequency at which random configurations are sampled instead of configurations from the acquisition strategy.
TYPE:
|
initial_design_size |
Number of 'x' samples that are to be evaluated before selecting a sample using a strategy instead of randomly. If there is a user prior, we can rely on the model from the very first iteration.
TYPE:
|
prior_as_samples |
Whether to sample from the KDE and incorporate that way, or
TYPE:
|
pending_as_bad |
Whether to treat pending observations as bad, assigning them to
TYPE:
|
prior_draws |
The number of samples drawn from the prior if there is one. This
TYPE:
|
patience |
How many times we try something that fails before giving up.
TYPE:
|
budget |
Maximum budget |
loss_value_on_error |
Setting this and cost_value_on_error to any float will supress any error during bayesian optimization and will use given loss value instead. default: None
TYPE:
|
cost_value_on_error |
Setting this and loss_value_on_error to any float will supress any error during bayesian optimization and will use given cost value instead. default: None
TYPE:
|
logger |
logger object, or None to use the neps logger
DEFAULT:
|
Source code in neps/optimizers/bayesian_optimization/mf_tpe.py
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 |
|
__call__
#
__call__(
x: Iterable,
asscalar: bool = False,
only_lowest_fidelity=True,
only_good=False,
) -> ndarray | Tensor | float
Return the negative expected improvement at the query point
Source code in neps/optimizers/bayesian_optimization/mf_tpe.py
get_cost
#
Calls result.utils.get_cost() and passes the error handling through. Please use self.get_cost() instead of get_cost() in all optimizer classes.
Source code in neps/optimizers/base_optimizer.py
get_learning_curve
#
Calls result.utils.get_loss() and passes the error handling through. Please use self.get_loss() instead of get_loss() in all optimizer classes.
Source code in neps/optimizers/base_optimizer.py
get_loss
#
Calls result.utils.get_loss() and passes the error handling through. Please use self.get_loss() instead of get_loss() in all optimizer classes.