-
Notifications
You must be signed in to change notification settings - Fork 1
/
utils.py
279 lines (243 loc) · 10.8 KB
/
utils.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
from lightning.pytorch.callbacks import TQDMProgressBar
from tqdm import tqdm
import sys
import numpy as np
import pandas as pd
import json
# for tqdm on screen so it can use ascii
class MyProgressBar(TQDMProgressBar):
def init_sanity_tqdm(self) -> tqdm:
bar = tqdm(
desc=self.sanity_check_description,
position=(2 * self.process_position),
disable=self.is_disabled,
leave=False,
dynamic_ncols=True,
file=sys.stdout, ascii=True,
)
return bar
def init_train_tqdm(self) -> tqdm:
bar = tqdm(
desc=self.train_description,
#initial=self.train_batch_idx,
position=(2 * self.process_position),
disable=self.is_disabled,
leave=True,
dynamic_ncols=True,
file=sys.stdout,
smoothing=0, ascii=True,
)
return bar
def init_predict_tqdm(self) -> tqdm:
bar = tqdm(
desc=self.predict_description,
#initial=self.train_batch_idx,
position=(2 * self.process_position),
disable=self.is_disabled,
leave=True,
dynamic_ncols=True,
file=sys.stdout,
smoothing=0, ascii=True,
)
return bar
def init_validation_tqdm(self) -> tqdm:
# The main progress bar doesn't exist in `trainer.validate()`
has_main_bar = self.trainer.state.fn != "validate"
bar = tqdm(
desc=self.validation_description,
position=(2 * self.process_position + has_main_bar),
disable=self.is_disabled,
leave=not has_main_bar,
dynamic_ncols=True,
file=sys.stdout, ascii=True,
)
return bar
def init_test_tqdm(self) -> tqdm:
bar = tqdm(
desc="Testing",
position=(2 * self.process_position),
disable=self.is_disabled,
leave=True,
dynamic_ncols=True,
file=sys.stdout, ascii=True,
)
return bar
class config_ffnn:
def __init__(self, id, num_layers, num_neurons, activation_funct, learning_rate, batch_size, neg_slope_leakyrelu=None, alpha_loss=None):
self.id = id
self.costs = []
self.epochs = []
self.num_layers = num_layers
self.num_neurons = num_neurons
self.activation_funct = activation_funct
self.learning_rate = learning_rate
self.batch_size = batch_size
self.neg_slope_leakyrelu = neg_slope_leakyrelu
self.alpha_loss = alpha_loss
def add_cost(self, cost):
self.costs.append(cost)
def add_epoch(self, epoch):
self.epochs.append(epoch)
def print(self):
print("id: ",self.id,"\nnum_layers: ", self.num_layers)
print("num_neurons: ", self.num_neurons[:self.num_layers])
print("activation_functions: ", self.activation_funct[:self.num_layers])
print("lr: ", self.learning_rate)
print("bs: ", self.batch_size)
print("Costs: ", self.get_best_cost())
print("Epochs: ",self.get_best_epoch())
def config_dict(self):
config_dict = {
"ID": self.id,
"batch_size": self.batch_size,
"learning_rate_init": self.learning_rate,
"n_layer": self.num_layers,
"neg_slope_leakyrelu": self.neg_slope_leakyrelu}
for i in range(1, self.num_layers+1):
config_dict["n_neurons_"+str(i)] = self.num_neurons[i-1]
config_dict["activation_"+str(i)] = self.activation_funct[i-1]
return config_dict
def get_best_cost(self):
return min(self.costs)
def get_best_epoch(self):
return self.epochs[self.costs.index(min(self.costs))]
class config_transformer:
def __init__(self, id, num_layers, embed_dim, hidden_dim, learning_rate, batch_size, num_heads, dropout):
self.id = id
self.costs = []
self.epochs = []
self.num_layers = num_layers
self.embed_dim = embed_dim
self.hidden_dim = hidden_dim
self.learning_rate = learning_rate
self.batch_size = batch_size
self.num_heads = num_heads
self.dropout = dropout
def add_cost(self, cost):
self.costs.append(cost)
def add_epoch(self, epoch):
self.epochs.append(epoch)
def print(self):
print("id: ",self.id,"\nnum_layers: ", self.num_layers," - embed_dim: ", self.embed_dim," - hidden_dim: ", self.hidden_dim," - num_heads: ",self.num_heads, " - lr: ", self.learning_rate," - bs: ", self.batch_size," - dropout: ", self.dropout)
print("Costs: ", self.costs)
print("Epochs: ",self.epochs)
def config_dict(self):
config_dict = {
"ID": self.id,
"batch_size": self.batch_size,
"learning_rate": self.learning_rate,
"embed_dim": self.embed_dim,
"hidden_dim": self.hidden_dim,
"num_layers": self.num_layers,
"num_heads": self.num_heads,
"dropout": self.dropout,}
return config_dict
def get_best_cost(self):
return min(self.costs)
def get_best_epoch(self):
return self.epochs[self.costs.index(min(self.costs))]
class config_xgboost:
def __init__(self, id, eta, n_estimators, gamma, max_depth, min_child_weight, max_delta_step, subsample, colsample_bytree, colsample_bylevel, colsample_bynode,):
self.id = id
self.costs = []
self.epochs = []
self.eta = eta
self.n_estimators = n_estimators
self.gamma = gamma
self.max_depth = max_depth
self.min_child_weight = min_child_weight
self.max_delta_step = max_delta_step
self.subsample = subsample
self.colsample_bytree = colsample_bytree
self.colsample_bylevel = colsample_bylevel
self.colsample_bynode = colsample_bynode
def add_cost(self, cost):
self.costs.append(cost)
def print(self):
print("id: ",self.id,"\neta: ", self.eta,"\nn_estimators: ", self.n_estimators," - gamma: ", self.gamma," - max_depth: ", self.max_depth," - min_child_weight: ",self.min_child_weight, " - max_delta_step: ", self.max_delta_step," - subsample: ", self.subsample," - colsample_bytree: ", self.colsample_bytree," - colsample_bylevel: ", self.colsample_bylevel," - colsample_bynode: ", self.colsample_bynode)
print("Costs: ", self.costs)
def config_dict(self):
config_dict = {
"eta": self.eta,
"n_estimators": self.n_estimators,
"gamma": self.gamma,
"max_depth": self.max_depth,
"min_child_weight": self.min_child_weight,
"max_delta_step": self.max_delta_step,
"subsample": self.subsample,
"colsample_bytree": self.colsample_bytree,
"colsample_bylevel": self.colsample_bylevel,
"colsample_bynode": self.colsample_bynode,}
return config_dict
def get_best_cost(self):
return min(self.costs)
# this function is used to extract best smac_values
def BestSMAC_FFNN(file_name):
# read jsonfile
filepath= file_name+"/0/runhistory.json"
jsonfile = open(filepath, 'r')
jsondata = jsonfile.read()
jsonobj = json.loads(jsondata)
configs_names=list(jsonobj["configs"]["1"].keys())
# search for all strings starting with "activation_fun"
max_num_layers=len([s for s in configs_names if "activation_" in s])
runs_history = []
for id in jsonobj["configs"]:
activations=[]
num_neurons=[]
for j in range(1,max_num_layers+1):
activations.append(str(jsonobj["configs"][id]["activation_"+str(j)]))
num_neurons.append(int(jsonobj["configs"][id]["n_neurons_"+str(j)]))
single_config = config_ffnn(id,jsonobj["configs"][id]["n_layer"],num_neurons,activations,jsonobj["configs"][id]["learning_rate_init"],jsonobj["configs"][id]["batch_size"],neg_slope_leakyrelu=jsonobj["configs"][id]["neg_slope_leakyrelu"] if "neg_slope_leakyrelu" in jsonobj["configs"][id] else None,alpha_loss=jsonobj["configs"][id]["alpha_loss"] if "alpha_loss" in jsonobj["configs"][id] else None)
runs_history.append(single_config)
for i in jsonobj["data"]:
runs_history[i[0]-1].add_cost(i[4])
runs_history[i[0]-1].add_epoch(np.ceil(i[3]).astype(int))
all_best_costs=[]
for i in range(len(runs_history)):
all_best_costs.append(runs_history[i].get_best_cost())
id_sorted=np.argsort(all_best_costs)
best_config_id = id_sorted[0]
best_config_dict = runs_history[best_config_id].config_dict()
return best_config_dict
def BestSMAC_Transformer(file_name):
# read jsonfile
filepath= file_name+"/0/runhistory.json"
jsonfile = open(filepath, 'r')
jsondata = jsonfile.read()
jsonobj = json.loads(jsondata)
# search for all strings starting with "activation_fun"
runs_history = []
for id in jsonobj["configs"]:
single_config = config_transformer(id,jsonobj["configs"][id]["num_layers"],jsonobj["configs"][id]["embed_dim"],jsonobj["configs"][id]["hidden_dim"],jsonobj["configs"][id]["learning_rate"],jsonobj["configs"][id]["batch_size"],jsonobj["configs"][id]["num_heads"],jsonobj["configs"][id]["dropout"])
runs_history.append(single_config)
for i in jsonobj["data"]:
runs_history[i[0]-1].add_cost(i[4])
runs_history[i[0]-1].add_epoch(np.ceil(i[3]).astype(int))
all_best_costs=[]
for i in range(len(runs_history)):
all_best_costs.append(runs_history[i].get_best_cost())
id_sorted=np.argsort(all_best_costs)
best_config_id = id_sorted[0]
best_config_dict = runs_history[best_config_id].config_dict()
return best_config_dict
def BestSMAC_XGBoost(file_name):
# read jsonfile
filepath= file_name+"/0/runhistory.json"
jsonfile = open(filepath, 'r')
jsondata = jsonfile.read()
jsonobj = json.loads(jsondata)
# search for all strings starting with "activation_fun"
runs_history = []
for id in jsonobj["configs"]:
single_config = config_xgboost(id,jsonobj["configs"][id]["eta"],jsonobj["configs"][id]["n_estimators"],jsonobj["configs"][id]["gamma"],jsonobj["configs"][id]["max_depth"],jsonobj["configs"][id]["min_child_weight"],jsonobj["configs"][id]["max_delta_step"],jsonobj["configs"][id]["subsample"],jsonobj["configs"][id]["colsample_bytree"],jsonobj["configs"][id]["colsample_bylevel"],jsonobj["configs"][id]["colsample_bynode"])
runs_history.append(single_config)
for i in jsonobj["data"]:
runs_history[i[0]-1].add_cost(i[4])
all_best_costs=[]
for i in range(len(runs_history)):
all_best_costs.append(runs_history[i].get_best_cost())
id_sorted=np.argsort(all_best_costs)
best_config_id = id_sorted[0]
best_config_dict = runs_history[best_config_id].config_dict()
return best_config_dict