-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathmodel_configs.py
285 lines (233 loc) · 12.4 KB
/
model_configs.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
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
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
from copy import deepcopy
from tabpfniml.tabpfn_interpret.priors.utils import uniform_int_sampler_f
from tabpfniml.tabpfn_interpret.priors.differentiable_prior import DifferentiableHyperparameter
from ConfigSpace import hyperparameters as CSH
import torch
from tabpfniml.tabpfn_interpret.priors.differentiable_prior import replace_differentiable_distributions
import ConfigSpace as CS
def get_general_config(max_features, bptt, eval_positions=None):
""""
Returns the general PFN training hyperparameters.
"""
config_general = {
"lr": CSH.UniformFloatHyperparameter('lr', lower=0.0001, upper=0.00015, log=True),
"dropout": CSH.CategoricalHyperparameter('dropout', [0.0]),
"emsize": CSH.CategoricalHyperparameter('emsize', [2 ** i for i in range(8, 9)]), ## upper bound is -1
"batch_size": CSH.CategoricalHyperparameter('batch_size', [2 ** i for i in range(6, 8)]),
"nlayers": CSH.CategoricalHyperparameter('nlayers', [12]),
"num_features": max_features,
"nhead": CSH.CategoricalHyperparameter('nhead', [4]),
"nhid_factor": 2,
"bptt": bptt,
"eval_positions": None,
"seq_len_used": bptt,
"sampling": 'normal',#hp.choice('sampling', ['mixed', 'normal']), # uniform
"epochs": 80,
"num_steps": 100,
"verbose": False,
"mix_activations": False,
"pre_sample_causes": True,
"multiclass_type": 'rank'
}
return config_general
def get_flexible_categorical_config(max_features):
""""
Returns the configuration parameters for the tabular multiclass wrapper.
"""
config_flexible_categorical = {
"nan_prob_unknown_reason_reason_prior": CSH.CategoricalHyperparameter('nan_prob_unknown_reason_reason_prior', [0.5]),
"categorical_feature_p": CSH.CategoricalHyperparameter('categorical_feature_p', [0.0, 0.1, 0.2]),
"nan_prob_no_reason": CSH.CategoricalHyperparameter('nan_prob_no_reason', [0.0, 0.1]),
"nan_prob_unknown_reason": CSH.CategoricalHyperparameter('nan_prob_unknown_reason', [0.0]),
"nan_prob_a_reason": CSH.CategoricalHyperparameter('nan_prob_a_reason', [0.0]),
# "num_classes": lambda : random.randint(2, 10), "balanced": False,
"max_num_classes": 2,
"num_classes": 2,
"noise_type": CSH.CategoricalHyperparameter('noise_type', ["Gaussian"]), # NN
"balanced": True,
"normalize_to_ranking": CSH.CategoricalHyperparameter('normalize_to_ranking', [False]),
"set_value_to_nan": CSH.CategoricalHyperparameter('set_value_to_nan', [0.5, 0.2, 0.0]),
"normalize_by_used_features": True,
"num_features_used":
{'uniform_int_sampler_f(3,max_features)': uniform_int_sampler_f(1, max_features)}
# hp.choice('conv_activation', [{'distribution': 'uniform', 'min': 2.0, 'max': 8.0}, None]),
}
return config_flexible_categorical
def get_diff_flex():
""""
Returns the configuration parameters for a differentiable wrapper around the tabular multiclass wrapper.
"""
diff_flex = {
# "ordinal_pct": {'distribution': 'uniform', 'min': 0.0, 'max': 0.5},
# "num_categorical_features_sampler_a": hp.choice('num_categorical_features_sampler_a',
# [{'distribution': 'uniform', 'min': 0.3, 'max': 0.9}, None]),
# "num_categorical_features_sampler_b": {'distribution': 'uniform', 'min': 0.3, 'max': 0.9},
"output_multiclass_ordered_p": {'distribution': 'uniform', 'min': 0.0, 'max': 0.5}, #CSH.CategoricalHyperparameter('output_multiclass_ordered_p', [0.0, 0.1, 0.2]),
"multiclass_type": {'distribution': 'meta_choice', 'choice_values': ['value', 'rank']},
}
return diff_flex
def get_diff_gp():
""""
Returns the configuration parameters for a differentiable wrapper around GP.
"""
diff_gp = {
'outputscale': {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 10., 'min_mean': 0.00001, 'round': False,
'lower_bound': 0},
'lengthscale': {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 10., 'min_mean': 0.00001, 'round': False,
'lower_bound': 0},
'noise': {'distribution': 'meta_choice', 'choice_values': [0.00001, 0.0001, 0.01]}
}
return diff_gp
def get_diff_causal():
""""
Returns the configuration parameters for a differentiable wrapper around MLP / Causal mixture.
"""
diff_causal = {
#"mix_activations": {'distribution': 'meta_choice', 'choice_values': [True, False]},
#"num_layers": {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 6, 'min_mean': 1, 'round': True,
# 'lower_bound': 2},
"num_layers": {'distribution': 'meta_gamma', 'max_alpha': 2, 'max_scale': 3, 'round': True,
'lower_bound': 2},
# Better beta?
#"prior_mlp_hidden_dim": {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 130, 'min_mean': 5,
# 'round': True, 'lower_bound': 4},
"prior_mlp_hidden_dim": {'distribution': 'meta_gamma', 'max_alpha': 3, 'max_scale': 100, 'round': True, 'lower_bound': 4},
"prior_mlp_dropout_prob": {'distribution': 'meta_beta', 'scale': 0.6, 'min': 0.1, 'max': 5.0},
# This mustn't be too high since activations get too large otherwise
"noise_std": {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': .3, 'min_mean': 0.0001, 'round': False,
'lower_bound': 0.0},
"init_std": {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 10.0, 'min_mean': 0.01, 'round': False,
'lower_bound': 0.0},
#"num_causes": {'distribution': 'meta_trunc_norm_log_scaled', 'max_mean': 12, 'min_mean': 1, 'round': True,
# 'lower_bound': 1},
"num_causes": {'distribution': 'meta_gamma', 'max_alpha': 3, 'max_scale': 7, 'round': True,
'lower_bound': 2},
"is_causal": {'distribution': 'meta_choice', 'choice_values': [True, False]},
"pre_sample_weights": {'distribution': 'meta_choice', 'choice_values': [True, False]},
"y_is_effect": {'distribution': 'meta_choice', 'choice_values': [True, False]},
"sampling": {'distribution': 'meta_choice', 'choice_values': ['normal', 'mixed']},
"prior_mlp_activations": {'distribution': 'meta_choice_mixed', 'choice_values': [
torch.nn.Tanh
, torch.nn.Identity
, torch.nn.ReLU
]},
"block_wise_dropout": {'distribution': 'meta_choice', 'choice_values': [True, False]},
"sort_features": {'distribution': 'meta_choice', 'choice_values': [True, False]},
"in_clique": {'distribution': 'meta_choice', 'choice_values': [True, False]},
#'pre_sample_causes': {'distribution': 'meta_choice', 'choice_values': [True, False]},
}
return diff_causal
def get_diff_prior_bag():
""""
Returns the configuration parameters for a GP and MLP / Causal mixture.
"""
diff_prior_bag = {
'prior_bag_exp_weights_1': {'distribution': 'uniform', 'min': 2.0, 'max': 10.0},
# MLP Weight (Biased, since MLP works better, 1.0 is weight for prior number 0)
}
return diff_prior_bag
def get_diff_config():
""""
Returns the configuration parameters for a differentiable wrapper around GP and MLP / Causal mixture priors.
"""
diff_prior_bag = get_diff_prior_bag()
diff_causal = get_diff_causal()
diff_gp = get_diff_gp()
diff_flex = get_diff_flex()
config_diff = {'differentiable_hyperparameters': {**diff_prior_bag, **diff_causal, **diff_gp, **diff_flex}}
return config_diff
def get_prior_config(config_type):
if config_type == 'causal':
return get_prior_config_causal()
elif config_type == 'gp':
return get_prior_config_gp()
elif config_type == 'bnn':
return get_prior_config_bnn()
def get_prior_config_gp(max_features=100):
config_general = get_general_config(max_features, 50, eval_positions=[30])
config_general_real_world = {**config_general}
config_flexible_categorical = get_flexible_categorical_config(max_features)
config_flexible_categorical_real_world = {**config_flexible_categorical}
config_gp = {}
config_diff = get_diff_config()
config = {**config_general_real_world, **config_flexible_categorical_real_world, **config_diff, **config_gp}
config['differentiable_hyperparameters']['prior_bag_exp_weights_1'] = {'distribution': 'uniform', 'min': 0.0,
'max': .01} # Never select MLP
def get_prior_config_bnn(max_features=100):
config_general = get_general_config(max_features, 50, eval_positions=[30])
config_general_real_world = {**config_general}
config_flexible_categorical = get_flexible_categorical_config(max_features)
config_flexible_categorical_real_world = {**config_flexible_categorical}
config_gp = {}
config_mlp = {}
config_diff = get_diff_config()
config = {**config_general_real_world, **config_flexible_categorical_real_world, **config_diff, **config_gp,
**config_mlp}
config['differentiable_hyperparameters']['prior_bag_exp_weights_1'] = {'distribution': 'uniform',
'min': 1000.0,
'max': 1001.0} # Always select MLP
def get_prior_config_causal(max_features=100):
config_general = get_general_config(max_features, 50, eval_positions=[30])
config_general_real_world = {**config_general}
config_flexible_categorical = get_flexible_categorical_config(max_features)
config_flexible_categorical_real_world = {**config_flexible_categorical}
config_flexible_categorical_real_world[
'num_categorical_features_sampler_a'] = -1.0 # Categorical features disabled by default
config_gp = {}
config_mlp = {}
config_diff = get_diff_config()
config = {**config_general_real_world, **config_flexible_categorical_real_world, **config_diff, **config_gp,
**config_mlp}
return config
def sample_differentiable(config):
""""
Returns sampled hyperparameters from a differentiable wrapper, that is it makes a non-differentiable out of
differentiable.
"""
# config is a dict of dicts, dicts that have a 'distribution' key are treated as distributions to be sampled
result = deepcopy(config)
del result['differentiable_hyperparameters']
for k, v in config['differentiable_hyperparameters'].items():
s_indicator, s_hp = DifferentiableHyperparameter(**v, embedding_dim=None,
device=None)() # both of these are actually not used to the best of my knowledge
result[k] = s_hp
return result
def list_all_hps_in_nested(config):
""""
Returns a list of hyperparameters from a neszed dict of hyperparameters.
"""
if isinstance(config, CSH.Hyperparameter):
return [config]
elif isinstance(config, dict):
result = []
for k, v in config.items():
result += list_all_hps_in_nested(v)
return result
else:
return []
def create_configspace_from_hierarchical(config):
cs = CS.ConfigurationSpace()
for hp in list_all_hps_in_nested(config):
cs.add_hyperparameter(hp)
return cs
def fill_in_configsample(config, configsample):
# config is our dict that defines config distribution
# configsample is a CS.Configuration
hierarchical_configsample = deepcopy(config)
for k, v in config.items():
if isinstance(v, CSH.Hyperparameter):
hierarchical_configsample[k] = configsample[v.name]
elif isinstance(v, dict):
hierarchical_configsample[k] = fill_in_configsample(v, configsample)
return hierarchical_configsample
def evaluate_hypers(config, sample_diff_hps=False):
""""
Samples a hyperparameter configuration from a sampleable configuration (can be used in HP search).
"""
if sample_diff_hps:
# I do a deepcopy here, such that the config stays the same and can still be used with diff. hps
config = deepcopy(config)
replace_differentiable_distributions(config)
cs = create_configspace_from_hierarchical(config)
cs_sample = cs.sample_configuration()
return fill_in_configsample(config, cs_sample)