-
Notifications
You must be signed in to change notification settings - Fork 0
/
utils.py
442 lines (375 loc) · 14.9 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
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
import json
import os
import pickle
import random
from argparse import Namespace
from typing import Optional, List
import numpy as np
import pandas as pd
import torch
import yaml
from sklearn.metrics import accuracy_score, precision_recall_fscore_support, r2_score, f1_score, roc_auc_score
from sklearn.preprocessing import LabelEncoder
from torch.nn import functional as F
from yaml.loader import SafeLoader
from models import create_mil_model
SKLEARN_LABEL_ENCODER = LabelEncoder()
REGRESSION_LABELS = [
"care",
"purity",
"equality",
"proportionality",
"loyalty",
"authority",
"fairness",
"honor",
"dignity",
"face"
]
def set_seed(seed: int) -> None:
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
if torch.cuda.device_count() > 0:
torch.cuda.manual_seed_all(seed)
def get_mse(model, test_generator, device):
model.eval()
loss = torch.nn.MSELoss()
all_y = []
all_y_hat = []
for x, y, _ in test_generator:
x = x.to(device)
output = model(x)
all_y_hat.append(output)
all_y.append(y)
yt = torch.cat(all_y_hat).squeeze().to("cpu")
yt_hat = torch.cat(all_y).squeeze().to("cpu")
# print(yt_hat.shape, yt.shape)
mse = loss(yt_hat, yt).item()
return mse
def get_crossentropy(model, dataloader, device):
loss = torch.nn.CrossEntropyLoss()
all_y = []
all_y_hat = []
for batch in dataloader:
x, y = batch[0], batch[1]
x = x.to(device)
output = model(x)
all_y_hat.append(output)
all_y.append(y)
# all_y.append(y[:, 0])
yt_hat = torch.cat(all_y_hat).squeeze().to("cpu")
yt = torch.cat(all_y).squeeze()
crossentropy_loss = loss(yt_hat, yt).item()
return crossentropy_loss
def get_r2_score(model, dataloader, device, min_clip, max_clip):
model.eval()
all_y = []
all_y_hat = []
for batch in dataloader:
x, y = batch[0], batch[1]
x = x.to(device)
output = model(x)
all_y_hat.append(output.cpu().detach().numpy())
all_y.append(y.cpu().detach().numpy())
yt = np.concatenate(all_y).squeeze()
yt_hat = np.concatenate(all_y_hat).squeeze()
score = r2_score_with_clip(yt, yt_hat, min_clip, max_clip)
return score
def r2_score_with_clip(yt, yt_hat, min_clip, max_clip):
yt_hat[yt_hat < min_clip] = min_clip
yt_hat[yt_hat > max_clip] = max_clip
score = r2_score(yt, yt_hat)
return score
def get_classification_metrics(model, dataloader, device, average='macro', detailed=False):
model.eval()
all_y = []
all_y_hat = []
all_y_hat_prob = []
for batch in dataloader:
x, y = batch[0], batch[1]
x = x.to(device)
output = model(x)
output = F.softmax(output, dim=1)
y_prob = torch.softmax(output, dim=1)
y_hat = torch.argmax(output, dim=1).cpu()
all_y_hat.extend(y_hat.tolist())
y = y.to(torch.int64)
all_y.extend(y.tolist())
all_y_hat_prob.extend(y_prob.cpu().detach().tolist())
precision, recall, f1, _ = precision_recall_fscore_support(
all_y, all_y_hat, average=average, zero_division=0
)
f1_micro = f1_score(all_y, all_y_hat, average="micro")
all_y, all_y_hat_prob= np.array(all_y), np.array(all_y_hat_prob)
# print(all_y)
# print(all_y_hat_prob)
# print(all_y.shape, all_y_hat_prob.shape)
if all_y_hat_prob.shape[1] == 2:
auc = roc_auc_score(all_y, all_y_hat_prob[:, 1], average='macro')
else:
auc = roc_auc_score(all_y, all_y_hat_prob, average='macro', multi_class='ovr')
metrics = {
"acc": accuracy_score(all_y, all_y_hat),
"precision": precision,
"recall": recall,
"f1": f1,
"f1_micro": f1_micro,
"auc": auc,
}
if not detailed:
return metrics
return metrics, all_y, all_y_hat, all_y_hat_prob
class AverageMeter(object):
"""Computes and stores the average and current value"""
def __init__(self):
self.count = None
self.sum = None
self.avg = None
self.val = None
self.reset()
def reset(self):
self.val = 0
self.avg = 0
self.sum = 0
self.count = 0
def update(self, val, n=1):
self.val = val
self.sum += val * n
self.count += n
self.avg = self.sum / self.count
def save_pickle(file, data):
with open(file, "wb") as f:
pickle.dump(data, f)
def load_pickle(file):
with open(file, "rb") as f:
dataset = pickle.load(f)
return dataset
def load_yaml_file(file):
with open(file, "r") as f:
data = yaml.load(f, Loader=SafeLoader)
return data
def save_json(path, data):
with open(path, "w") as f:
json.dump(data, f)
def load_json(path):
with open(path, "r") as f:
data = json.load(f)
return data
def get_data_directory(dataset: str, data_embedded_column_name: str, random_seed: int) -> str:
return os.path.join(
os.path.dirname(__file__), "data", f"seed_{random_seed}", dataset, data_embedded_column_name
)
def get_model_save_directory(dataset: str, data_embedded_column_name: str, embedding_model_name: str,
target_column_name: str, bag_size: int, baseline: str,
autoencoder_layers: Optional[List[int]], random_seed, dev, task_type, prefix, multiple_runs=False) -> str:
model_name = get_model_name(baseline, autoencoder_layers)
folder_name = "runs"
if multiple_runs:
folder_name = "multiple_" + folder_name
if dev:
folder_name = "dev_runs"
if prefix:
model_save_directory = os.path.join(
os.path.dirname(__file__),
folder_name,
task_type,
f"seed_{random_seed}",
dataset,
data_embedded_column_name,
embedding_model_name,
target_column_name,
f"bag_size_{bag_size}",
model_name,
prefix,
)
else:
model_save_directory = os.path.join(
os.path.dirname(__file__),
folder_name,
task_type,
f"seed_{random_seed}",
dataset,
data_embedded_column_name,
embedding_model_name,
target_column_name,
f"bag_size_{bag_size}",
model_name,
)
if not os.path.exists(model_save_directory):
os.makedirs(model_save_directory)
return model_save_directory
def get_model_name(baseline: str, autoencoder_layers: Optional[List[int]]) -> str:
if autoencoder_layers:
return baseline + "_" + "_".join([str(layer) for layer in autoencoder_layers])
else:
return baseline
def read_data_split(data_dir, embedding_model, split):
df = pd.read_pickle(os.path.join(data_dir, embedding_model, f"{split}.pickle"))
df = df.reset_index(drop=True)
return df
def create_bag_masks(df, bag_size, bag_embedded_column_name):
bag_masks = torch.zeros(
(df.shape[0], df[bag_embedded_column_name][0].shape[0]), dtype=torch.bool
)
bag_masks[:, :bag_size] = 1
return bag_masks
def preprocess_dataframe(
df: pd.DataFrame,
dataframe_set: str,
label: str,
train_dataframe_mean: Optional[float],
train_dataframe_median: Optional[float],
train_dataframe_std: Optional[float],
task_type: str,
extra_columns: Optional[List[str]] = [],
):
# from IPython import embed; embed()
df = df[["bag_embeddings", "bag", "bag_mask", label] + extra_columns]
df = df.dropna()
df = df.reset_index(drop=True)
label2id = None
id2label = None
if task_type == "regression":
new_label = label
else:
if label in REGRESSION_LABELS:
new_label = f"{label}_encoded"
# 1 std before mean classifies as 0, between 1 std before and 1 std after mean classifies as 1, 1 std after
# mean classifies as 2
# 0 if below mean and 1 if above or equal to mean
# df[new_label] = df[label].apply(lambda x: 0 if x < train_dataframe_mean else 1)
# 0 if below median and 1 if above or equal to median
df[new_label] = df[label].apply(lambda x: 0 if x < train_dataframe_median else 1)
else:
new_label = f"{label}_encoded"
if dataframe_set == "train":
df[new_label] = SKLEARN_LABEL_ENCODER.fit_transform(df[label])
else:
df[new_label] = SKLEARN_LABEL_ENCODER.transform(df[label])
label2id = {label: idx for idx, label in enumerate(SKLEARN_LABEL_ENCODER.classes_.tolist())}
id2label = {idx: label for idx, label in enumerate(SKLEARN_LABEL_ENCODER.classes_.tolist())}
df = df[["bag_embeddings", "bag", "bag_mask", new_label] + extra_columns]
df = df.rename(columns={new_label: "labels"})
return df, label2id, id2label
def get_df_mean_median_std(df, label):
"""
Receives a dataframe and a label and returns the mean, median and std of the label in the dataframe.
:param df: dataframe
:param label: label
:return: mean, median and std
"""
if label in REGRESSION_LABELS:
return df[label].dropna().mean(), df[label].dropna().median(), df[label].dropna().std()
return None, None, None
def create_preprocessed_dataframes(train_dataframe: pd.DataFrame, val_dataframe: pd.DataFrame,
test_dataframe: pd.DataFrame, label: str, task_type: str, extra_columns: Optional[List[str]] = []):
train_dataframe_mean, train_dataframe_median, train_dataframe_std = get_df_mean_median_std(
train_dataframe, label
)
train_dataframe, label2id, id2label = preprocess_dataframe(df=train_dataframe, dataframe_set="train", label=label,
train_dataframe_mean=train_dataframe_mean,
train_dataframe_median=train_dataframe_median,
train_dataframe_std=train_dataframe_std,
task_type=task_type,
extra_columns=extra_columns)
val_dataframe, _, _ = preprocess_dataframe(df=val_dataframe, dataframe_set="val", label=label,
train_dataframe_mean=train_dataframe_mean,
train_dataframe_median=train_dataframe_median,
train_dataframe_std=train_dataframe_std, task_type=task_type,
extra_columns=extra_columns)
test_dataframe, _, _ = preprocess_dataframe(df=test_dataframe, dataframe_set="test", label=label,
train_dataframe_mean=train_dataframe_mean,
train_dataframe_median=train_dataframe_median,
train_dataframe_std=train_dataframe_std, task_type=task_type,
extra_columns=extra_columns)
return train_dataframe, val_dataframe, test_dataframe, label2id, id2label
class EarlyStopping:
def __init__(self, models_dir: str, save_model_name, trace_func, patience: int = 7, verbose: bool = False,
delta: float = 0.0, descending: bool = True):
self.patience = patience
self.verbose = verbose
self.counter = 0
self.best_score = None
self.early_stop = False
self.val_loss_min = np.Inf
self.delta = delta
self.models_dir = models_dir
self.trace_func = trace_func
self.descending = descending
if save_model_name:
self.model_address = os.path.join(self.models_dir, save_model_name)
def __call__(self, val_loss, model):
if self.descending:
score = -val_loss
else:
score = val_loss
if self.best_score is None:
self.best_score = score
self.save_checkpoint(val_loss, model)
elif score <= self.best_score + self.delta:
self.counter += 1
self.trace_func(
f"EarlyStopping counter: {self.counter} out of {self.patience}"
)
if self.counter >= self.patience:
self.early_stop = True
else:
self.best_score = score
self.save_checkpoint(val_loss, model)
self.counter = 0
def save_checkpoint(self, val_loss, model):
"""Saves model when validation loss decrease."""
if self.verbose:
self.trace_func(
f"Score changed ({self.val_loss_min:.6f} --> {val_loss:.6f}). Saving model ..."
)
self.val_loss_min = val_loss
if self.model_address:
torch.save(model.state_dict(), self.model_address)
def get_model(model_path: str, ensemble: bool = False):
if ensemble:
model_path = os.path.join(model_path, "only_ensemble_loss_sweep_best_rl_model.pt")
p_model_state_dict = torch.load(model_path, map_location=torch.device("cpu"))
model_state_dict = {}
for k in p_model_state_dict.keys():
if k.startswith("task_model."):
model_state_dict[k.split("task_model.")[1]] = p_model_state_dict[k]
else:
model_path = os.path.join(model_path, "best_model.pt")
model_state_dict = torch.load(model_path, map_location=torch.device("cpu"))
model_name = model_path.split("/")[-2]
baseline = model_name.split("_")[0]
autoencoder_layers = list(map(int, model_name.split("_")[1:]))
if "MLP" in model_name:
args = Namespace(
**{
"dropout_p": 0.5,
"input_dim": model_state_dict["mlp.0.weight"].size()[1],
"hidden_dim": model_state_dict["mlp.0.weight"].size()[0],
"number_of_classes": model_state_dict["mlp.3.bias"].size()[0],
"autoencoder_layer_sizes": autoencoder_layers,
"baseline": baseline,
}
)
else:
args = Namespace(
**{
"input_dim": autoencoder_layers[-1],
"dropout_p": 0.5,
"n_hidden_sets": model_state_dict["fc1.weight"].size()[1],
"n_elements": model_state_dict["Wc"].size()[1] // model_state_dict["fc1.weight"].size()[1],
"number_of_classes": model_state_dict["fc2.bias"].size()[0],
"autoencoder_layer_sizes": autoencoder_layers,
"baseline": baseline,
}
)
model = create_mil_model(args)
model.load_state_dict(model_state_dict)
return model, args
def get_balanced_weights(labels):
label_set = list(set(labels))
label_set.sort()
perfect_balance_weights = [len(labels)/labels.count(element) for element in label_set]
sample_weights = [perfect_balance_weights[t] for t in labels]
return sample_weights