How to use the deephyper.search.models.base.param.discrete function in deephyper
To help you get started, we’ve selected a few deephyper examples, based on popular ways it is used in public projects.
Secure your code as it's written. Use Snyk Code to scan source code in minutes - no build needed - and fix issues immediately.
deephyper / deephyper / deephyper / benchmarks_hps / reutersmlp / space.py View on Github 
def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 128, step.GEOMETRIC, 2),
dropout,
param.discrete("epochs", 2, 20, step.ARITHMETIC, 1),
param.discrete("max_words", 800, 1300, step.ARITHMETIC, 1),
param.discrete("nunits", 128, 1024, step.ARITHMETIC, 128),
optimizer,
param.discrete("skip_top", 0,30, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 32, 0.5, 5, 1000, 512, 'adam', 0]def __init__(self):
self.space = [
param.discrete('batch_size', 8, 512, step.GEOMETRIC, 2),
param.discrete('epochs', 5, 60, step.ARITHMETIC, 1),
param.discrete('nhidden', 1, 20, step.ARITHMETIC, 1),
param.discrete('nunits', 1, 1000, step.ARITHMETIC, 1),
activation,
activation1,
activation2,
dropout,
optimizer
]
self.params = [param.name for param in self.space]
self.starting_point = [128, 20, 1, 512, 'relu', 0.2, 'adam']def __init__(self):
self.space = [
param.discrete('margin', 1, 3, step.ARITHMETIC, 2),
param.discrete('units', 32, 512, step.ARITHMETIC, 1),
param.discrete('batch_size', 32, 512, step.ARITHMETIC, 1),
param.discrete('epochs', 10, 45, step.ARITHMETIC, 1),
activation,
dropout,
optimizer
]
self.params = [param.name for param in self.space]
self.starting_point = [1, 128, 128, 20, 'relu', 0.3, 'adam']def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 1024, step.GEOMETRIC, 2),
param.non_ordinal("data_augmentation", [False, True]),
dropout,
param.discrete("epochs", 50, 200, step.ARITHMETIC, 1),
param.discrete("f1_size", 1, 5, step.ARITHMETIC, 1),
param.discrete("f2_size", 1, 5, step.ARITHMETIC, 1),
param.discrete("f1_units", 8, 64, step.GEOMETRIC, 2),
param.discrete("f2_units", 8, 64, step.GEOMETRIC, 2),
param.continuous("nunits", 1, 1000),
optimizer,
param.discrete("p_size", 1, 4, step.ARITHMETIC, 1),
param.non_ordinal("padding_c1", ["same", "valid"]),
param.non_ordinal("padding_c2", ["same", "valid"]),
param.non_ordinal("padding_p1", ["same", "valid"]),
param.non_ordinal("padding_p2", ["same", "valid"]),
param.discrete("stride1", 1, 4, step.ARITHMETIC, 1),
param.discrete("stride2", 1, 4, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 32, True, 0.25, 100, 3, 3, 16, 16, 512, 'adam', 2, 'valid', 'valid', 'same', 'same', 1, 1]def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 1024, step.GEOMETRIC, 2),
dropout,
param.discrete("epochs", 5, 35, step.ARITHMETIC, 1),
param.discrete("f1_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f2_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f1_units", 8, 64, step.GEOMETRIC, 2),
param.discrete("f2_units", 8, 64, step.GEOMETRIC, 2),
param.non_ordinal("max_pool", [True, False]),
param.discrete("nunits", 1, 1000, step.ARITHMETIC, 1),
optimizer,
param.non_ordinal("padding_c1", ["valid", "same"]),
param.non_ordinal("padding_c2", ["valid", "same"]),
param.non_ordinal("padding_p1", ["valid", "same"]),
param.non_ordinal("padding_p2", ["valid", "same"]),
param.discrete('p_size', 1, 5, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 128, 0.3, 12, 3, 3, 16, 16, True, 512, 'adam', 'valid', 'valid', 'same', 'same', 3]def __init__(self):
self.space = [
activation,
param.discrete('batch_size', 8, 128, step.GEOMETRIC, 2),
param.continuous('base_lr', 1e-3, 1e-1),
param.discrete('epochs', 100, 300),
param.continuous('lr80', 1e-3, 1e-1),
param.continuous('lr120', 1e-4, 1e-1),
param.continuous('lr160', 1e-4, 1e-2),
param.continuous('lr180', 1e-5, 1e-2),
param.discrete('num_filters', 16, 64, step.GEOMETRIC, 2),
param.discrete('num_filters_in', 16, 64, step.GEOMETRIC, 2),
optimizer
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 32, 1e-3, 200, 1e-1, 1e-2, 1e-3, 0.5e-3, 16, 16, 'adam']def __init__(self):
self.space = [
activation1,
activation2,
param.discrete("batch_size", 8, 1024, step.GEOMETRIC, 2),
dropout,
param.discrete("embedding_dims", 40, 60, step.ARITHMETIC, 1),
param.discrete("epochs", 1, 5, step.ARITHMETIC, 1),
param.discrete("filters", 200, 300, step.ARITHMETIC, 1),
param.discrete("hidden_dims", 20, 30, step.ARITHMETIC, 1),
param.discrete("kernel_size", 1, 5, step.ARITHMETIC, 1),
param.discrete("max_features", 3000, 7000, step.ARITHMETIC, 1),
param.discrete("maxlen", 300, 500, step.ARITHMETIC, 1),
optimizer,
padding,
param.discrete("strides", 1, 4, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 32, 0.2, 50, 3, 250, 25, 3, 5000, 400, 'adam', 'same', 1]def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 1024, step.GEOMETRIC, 2),
dropout,
param.discrete("epochs", 5, 35, step.ARITHMETIC, 1),
param.discrete("f1_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f2_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f1_units", 8, 64, step.GEOMETRIC, 2),
param.discrete("f2_units", 8, 64, step.GEOMETRIC, 2),
param.non_ordinal("max_pool", [True, False]),
param.discrete("nunits", 1, 1000, step.ARITHMETIC, 1),
optimizer,
param.non_ordinal("padding_c1", ["valid", "same"]),
param.non_ordinal("padding_c2", ["valid", "same"]),
param.non_ordinal("padding_p1", ["valid", "same"]),
param.non_ordinal("padding_p2", ["valid", "same"]),
param.discrete('p_size', 1, 5, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 128, 0.3, 12, 3, 3, 16, 16, True, 512, 'adam', 'valid', 'valid', 'same', 'same', 3]def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 1024, step.GEOMETRIC, 2),
dropout,
param.discrete("epochs", 5, 35, step.ARITHMETIC, 1),
param.discrete("f1_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f2_size", 1, 7, step.ARITHMETIC, 1),
param.discrete("f1_units", 8, 64, step.GEOMETRIC, 2),
param.discrete("f2_units", 8, 64, step.GEOMETRIC, 2),
param.non_ordinal("max_pool", [True, False]),
param.discrete("nunits", 1, 1000, step.ARITHMETIC, 1),
optimizer,
param.non_ordinal("padding_c1", ["valid", "same"]),
param.non_ordinal("padding_c2", ["valid", "same"]),
param.non_ordinal("padding_p1", ["valid", "same"]),
param.non_ordinal("padding_p2", ["valid", "same"]),
param.discrete('p_size', 1, 5, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 128, 0.3, 12, 3, 3, 16, 16, True, 512, 'adam', 'valid', 'valid', 'same', 'same', 3]def __init__(self):
self.space = [
activation1,
activation2,
activation3,
activation4,
activation5,
param.discrete("batch_size", 8, 128, step.GEOMETRIC, 2),
dropout,
param.discrete("epochs", 2, 20, step.ARITHMETIC, 1),
param.discrete("max_words", 800, 1300, step.ARITHMETIC, 1),
param.discrete("nunits", 128, 1024, step.ARITHMETIC, 128),
optimizer,
param.discrete("skip_top", 0,30, step.ARITHMETIC, 1)
]
self.params = [param.name for param in self.space]
self.starting_point = ['relu', 32, 0.5, 5, 1000, 512, 'adam', 0]deephyper
Scalable asynchronous neural architecture and hyperparameter search for deep neural networks
Package Health Score
75 / 100Popular deephyper functions
- deephyper.benchmarks.keras_cmdline
- deephyper.benchmarks.keras_cmdline.create_parser
- deephyper.benchmarks.keras_cmdline.fill_missing_defaults
- deephyper.benchmarks.keras_cmdline.return_optimizer
- deephyper.benchmarks.util
- deephyper.benchmarks.util.resume_from_disk
- deephyper.benchmarks.util.save_meta_data
- deephyper.benchmarks_hps.cliparser.build_base_parser
- deephyper.benchmarks_hps.util
- deephyper.benchmarks_hps.util.get_activation_instance
- deephyper.benchmarks_hps.util.get_optimizer_instance
- deephyper.benchmarks_hps.util.handle_cli
- deephyper.model.arch
- deephyper.search.models.base.param.discrete
- deephyper.search.nas.model.space.node.ConstantNode
- deephyper.search.nas.model.space.node.VariableNode
- deephyper.search.nas.model.space.op.op1d.Dense
- deephyper.search.nas.operation.mlp.MLP
- deephyper.search.util
- deephyper.search.util.conf_logger