def get_knob_config():
return {
'max_epochs': FixedKnob(10),
'learning_rate': FloatKnob(1e-5, 1e-2, is_exp=True),
'batch_size': CategoricalKnob([16, 32, 64, 128]),
'max_image_size': CategoricalKnob([32, 64, 128, 224]),
}
def get_knob_config():
return {
'n_estimators': IntegerKnob(50, 200),
'oob_score': CategoricalKnob([True, False]),
'max_depth': IntegerKnob(10, 100),
'max_features': CategoricalKnob(['auto', 'sqrt', 'log2'])
}
def get_knob_config():
return {
'max_depth': IntegerKnob(1, 32),
'splitter': CategoricalKnob(['best', 'random']),
'criterion': CategoricalKnob(['gini', 'entropy']),
'max_image_size': CategoricalKnob([16, 32])
}
def get_knob_config():
return {
'epochs': FixedKnob(15),
'batch_size': CategoricalKnob([32, 64, 128]),
'l_rate': FloatKnob(0.0001, 0.001, 0.01),
'max_image_size': CategoricalKnob([28, 32])
}
def get_knob_config():
return {
'model_class': CategoricalKnob(['resnent101_mnist']),
# Learning parameters
'lr': FixedKnob(0.0001), ### learning_rate
'weight_decay': FixedKnob(0.0),
'drop_rate': FixedKnob(0.0),
'max_epochs': FixedKnob(30),
'batch_size': CategoricalKnob([200]),
'max_iter': FixedKnob(20),
'optimizer': CategoricalKnob(['adam']),
'scratch': FixedKnob(True),
# Data augmentation
'max_image_size': FixedKnob(32),
'share_params': CategoricalKnob(['SHARE_PARAMS']),
'tag': CategoricalKnob(['relabeled']),
'workers': FixedKnob(8),
'seed': FixedKnob(123456),
'scale': FixedKnob(512),
'horizontal_flip': FixedKnob(True),
# Hyperparameters for PANDA modules
# Self-paced Learning and Loss Revision
'enable_spl': FixedKnob(False),
'spl_threshold_init': FixedKnob(16.0),
'spl_mu': FixedKnob(1.3),
'enable_lossrevise': FixedKnob(False),
'lossrevise_slop': FixedKnob(2.0),
# Label Adaptation
'enable_label_adaptation': FixedKnob(False), # error occurs
# GM Prior Regularization
'enable_gm_prior_regularization': FixedKnob(True),
'gm_prior_regularization_a': FixedKnob(0.001),
'gm_prior_regularization_b': FixedKnob(0.0001),
'gm_prior_regularization_alpha': FixedKnob(0.5),
'gm_prior_regularization_num': FixedKnob(4),
'gm_prior_regularization_lambda': FixedKnob(0.0001),
'gm_prior_regularization_upt_freq': FixedKnob(100),
'gm_prior_regularization_param_upt_freq': FixedKnob(50),
# Explanation
'enable_explanation': FixedKnob(False),
'explanation_gradcam': FixedKnob(True),
'explanation_lime': FixedKnob(False),
# Model Slicing
'enable_model_slicing': FixedKnob(True),
'model_slicing_groups': FixedKnob(0),
'model_slicing_rate': FixedKnob(1.0),
'model_slicing_scheduler_type': FixedKnob('randomminmax'),
'model_slicing_randnum': FixedKnob(1),
# MC Dropout
'enable_mc_dropout': FixedKnob(False),
'mc_trials_n': FixedKnob(10)
}
def get_knob_config():
return {
'max_iter': FixedKnob(20),
'kernel': CategoricalKnob(['rbf', 'linear', 'poly']),
'gamma': CategoricalKnob(['scale', 'auto']),
'C': FloatKnob(1e-4, 1e4, is_exp=True),
'max_image_size': CategoricalKnob([16, 32])
}
def get_knob_config():
return {
'penalty': CategoricalKnob(['l1', 'l2']),
'tol': FloatKnob(0.0001, 0.001),
'C': IntegerKnob(4, 15),
'fit_intercept': CategoricalKnob([True, False]),
'solver': CategoricalKnob(['lbfgs', 'liblinear']),
}
def get_knob_config():
return {
# Learning parameters
'lr': FixedKnob(0.0001),
'weight_decay': FixedKnob(0.0),
'drop_rate': FixedKnob(0.0),
'max_epochs': FixedKnob(10), # original 5
'batch_size': CategoricalKnob([96]), # original 32
'max_iter': FixedKnob(20),
'optimizer': CategoricalKnob(['adam']),
'scratch': FixedKnob(True),
# Data augmentation
'max_image_size': FixedKnob(32),
'share_params': CategoricalKnob(['SHARE_PARAMS']),
'tag': CategoricalKnob(['relabeled']),
'workers': FixedKnob(8),
'seed': FixedKnob(123456),
'scale': FixedKnob(512),
'horizontal_flip': FixedKnob(True),
# Self-paced Learning and Loss Revision
'enable_spl': FixedKnob(True),
'spl_threshold_init': FixedKnob(16.0),
'spl_mu': FixedKnob(1.3),
'enable_lossrevise': FixedKnob(False),
'lossrevise_slop': FixedKnob(2.0),
# Label Adaptation
'enable_label_adaptation': FixedKnob(False),
# GM Prior Regularization
'enable_gm_prior_regularization': FixedKnob(False),
'gm_prior_regularization_a': FixedKnob(0.001),
'gm_prior_regularization_b': FixedKnob(0.0001),
'gm_prior_regularization_alpha': FixedKnob(0.5),
'gm_prior_regularization_num': FixedKnob(4),
'gm_prior_regularization_lambda': FixedKnob(0.0001),
'gm_prior_regularization_upt_freq': FixedKnob(100),
'gm_prior_regularization_param_upt_freq': FixedKnob(50),
# Explanation
'enable_explanation': FixedKnob(False),
'explanation_method': FixedKnob('lime'),
# Model Slicing
'enable_model_slicing': FixedKnob(False),
'model_slicing_groups': FixedKnob(0),
'model_slicing_rate': FixedKnob(1.0),
'model_slicing_scheduler_type': FixedKnob('randomminmax'),
'model_slicing_randnum': FixedKnob(1),
# SelectiveNet
'selectionheadloss_weight': FixedKnob(0.5),
'target_coverage': FixedKnob(0.8),
'lamda': FixedKnob(32)
}
def get_knob_config():
return {
'alpha': FloatKnob(0.001, 0.01),
'normalize': CategoricalKnob([True, False]),
'copy_X': CategoricalKnob([True, False]),
'tol': FloatKnob(1e-05, 1e-04),
'solver': CategoricalKnob(['svd', 'sag']),
'random_state': IntegerKnob(1, 123)
}
def get_knob_config():
return {
'criterion': CategoricalKnob(['mse', 'mae']),
'splitter': CategoricalKnob(['best', 'random']),
'min_samples_split': IntegerKnob(2, 5),
'max_features': CategoricalKnob(['auto', 'sqrt']),
'random_state': IntegerKnob(1, 123),
'min_impurity_decrease': FloatKnob(0.0, 0.2),
'min_impurity_split': FloatKnob(1e-07, 1e-03)
}
def get_knob_config():
return {
'max_epochs': FixedKnob(10),
'hidden_layer_count': IntegerKnob(1, 2),
'hidden_layer_units': IntegerKnob(2, 128),
'learning_rate': FloatKnob(1e-5, 1e-1, is_exp=True),
'batch_size': CategoricalKnob([16, 32, 64, 128]),
'max_image_size': CategoricalKnob([16, 32, 48]),
'quick_train':
PolicyKnob('EARLY_STOP') # Whether early stopping would be used
}
def get_knob_config():
return {
'C': IntegerKnob(2, 3),
'kernel': CategoricalKnob(['poly', 'rbf', 'linear']),
'degree': IntegerKnob(2, 3),
'gamma': CategoricalKnob(['scale', 'auto']),
'coef0': FloatKnob(0.0, 0.1),
'shrinking': CategoricalKnob([True, False]),
'tol': FloatKnob(1e-03, 1e-01, is_exp=True),
'decision_function_shape': CategoricalKnob(['ovo', 'ovr']),
'probability': CategoricalKnob([True, False]),
}
def get_knob_config():
return {
'C': FloatKnob(1.0, 1.5),
'tol': FloatKnob(1e-03, 1e-01, is_exp=True),
'validation_fraction': FloatKnob(0.01, 0.1),
'n_iter_no_change': IntegerKnob(3, 5),
'shuffle': CategoricalKnob([True, False]),
'loss': CategoricalKnob(['hinge', 'squared_hinge']),
'random_state': IntegerKnob(1, 2),
'warm_start': CategoricalKnob([True, False]),
'average': IntegerKnob(1, 5),
}
def get_knob_config():
return {
'epochs': FixedKnob(15),
'learning_rate': FloatKnob(0.001, 0.07),
'decay_rate': FloatKnob(5e-5, 1e-4, is_exp=True),
'momentum': FloatKnob(0.1, 0.3, 0.6),
'batch_size': CategoricalKnob([32, 64, 128]),
'max_image_size': FixedKnob(28)
}
def get_knob_config():
return {
'epochs': FixedKnob(1),
'word_embed_dims': IntegerKnob(16, 128),
'word_rnn_hidden_size': IntegerKnob(16, 128),
'word_dropout': FloatKnob(1e-3, 2e-1, is_exp=True),
'learning_rate': FloatKnob(1e-2, 1e-1, is_exp=True),
'batch_size': CategoricalKnob([16, 32, 64, 128]),
}
def get_knob_config():
return {
'trial_epochs': FixedKnob(300),
'lr': FloatKnob(1e-4, 1, is_exp=True),
'lr_decay': FloatKnob(1e-3, 1e-1, is_exp=True),
'opt_momentum': FloatKnob(0.7, 1, is_exp=True),
'opt_weight_decay': FloatKnob(1e-5, 1e-3, is_exp=True),
'batch_size': CategoricalKnob([32, 64, 128]),
'drop_rate': FloatKnob(0, 0.4),
'max_image_size': FixedKnob(32),
'share_params': PolicyKnob('SHARE_PARAMS'),
# Affects whether training is shortened by using early stopping
'quick_train': PolicyKnob('EARLY_STOP'),
'early_stop_train_val_samples': FixedKnob(1024),
'early_stop_patience_epochs': FixedKnob(5)
}
def get_knob_config():
return {
'epoch': IntegerKnob(5, 10),
'learning_rate': FloatKnob(1e-3, 1e-1, is_exp=True),
'layer_dim': CategoricalKnob([50, 100, 250])
}
def get_knob_config():
return {
'n_neighbors': IntegerKnob(3, 4, 6),
'metric': CategoricalKnob(['minkowski', 'euclidean']),
'p': IntegerKnob(1, 2),
}