-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathframework.py
167 lines (139 loc) · 6.76 KB
/
framework.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
from tqdm import tqdm
import torch
import torch.nn as nn
from transformers import get_linear_schedule_with_warmup, AdamW
from torch.utils.data import DataLoader
import os
from seqeval.metrics import f1_score, classification_report
import random
import numpy as np
def set_seed(seed):
random.seed(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
set_seed(2020)
class Framework:
def __init__(self, args):
self.args = args
def train_step(self, batch_data, model):
model.train()
batch = tuple(t.to(self.args.device) for t in batch_data)
input_ids, attention_mask, pred_mask, labels = batch
predicted, loss = model(
input_ids=input_ids,
attention_mask=attention_mask,
pred_mask=pred_mask,
input_labels=labels
)
loss.backward()
torch.nn.utils.clip_grad_norm_(model.parameters(), self.args.max_grad_norm)
self.optimizer.step()
self.scheduler.step() # Update learning rate schedule
model.zero_grad()
return loss.item()
def train(self, train_dataset, dev_dataset, model, labels):
train_dataloader = DataLoader(train_dataset, batch_size=self.args.train_batch_size, shuffle=True)
# get optimizer schedule and loss
no_decay = ["bias", "LayerNorm.weight"]
optimizer_grouped_parameters = [
{
"params": [p for n, p in model.named_parameters() if not any(nd in n for nd in no_decay)],
"weight_decay": 0.0,
},
{"params": [p for n, p in model.named_parameters() if any(nd in n for nd in no_decay)], "weight_decay": 0.0},
]
self.optimizer = AdamW(optimizer_grouped_parameters, lr=self.args.learning_rate, eps=self.args.adam_epsilon)
self.scheduler = get_linear_schedule_with_warmup(
self.optimizer, num_warmup_steps=self.args.warmup_steps, num_training_steps=len(train_dataloader) * self.args.num_train_epochs
)
# Train!
print("***** Running training *****")
print(" Num examples = ", len(train_dataset))
print(" Num Epochs = ", self.args.num_train_epochs)
global_step = 0
# Check if continuing training from a checkpoint
best_result = 0
early_stop = 0
model.to(self.args.device)
for epoch in range(0, int(self.args.num_train_epochs)):
for step, batch in enumerate(train_dataloader):
loss = self.train_step(batch, model)
if step % 20 == 0:
print('Train Epoch[{}] Step[{} / {}] - loss: {:.6f} '.format(epoch+1, step+1, len(train_dataloader), loss)) # , accuracy, corrects
global_step += 1
if (self.args.evaluate_step > 0 and global_step % self.args.evaluate_step == 0) or (epoch==int(self.args.num_train_epochs)-1 and step == len(train_dataloader)-1):
early_stop += 1
result = self.evaluate(dev_dataset, model, labels)
print("best f1: %.2f, current f1: %.2f" % (best_result, result))
if best_result <= result:
best_result = result
print("Saving model checkpoint to %s"%self.args.save_model)
torch.save(model.state_dict(), self.args.save_model)
early_stop = 0
print()
if early_stop >= 5:
return
def evaluate(self, dev_dataset, model, all_labels):
print("\n Evaluating ...")
dev_dataloader = DataLoader(dev_dataset, batch_size=self.args.dev_batch_size, shuffle=False)
model.eval()
total_loss = 0
predicted_list = []
labels_list = []
for step, batch in enumerate(tqdm(dev_dataloader)):
batch = tuple(t.to(self.args.device) for t in batch)
with torch.no_grad():
input_ids, attention_mask, pred_mask, labels = batch
predicted, loss = model(
input_ids=input_ids,
attention_mask=attention_mask,
pred_mask=pred_mask,
input_labels=labels
)
total_loss += loss.item()
if self.args.crf:
predicted = [seq[seq>=0].tolist() for seq in predicted]
else:
predicted = [seq[mask==1].tolist() for seq, mask in zip(predicted, pred_mask)]
groud_labels = [seq[mask==1].tolist() for seq, mask in zip(labels, pred_mask)]
for tl, pl in zip(groud_labels, predicted):
labels_list.append([all_labels[l] for l in tl])
predicted_list.append([all_labels[l] for l in pl])
print("Dev Loss: {:.6f}".format(total_loss / len(dev_dataloader)))
class_report = classification_report(labels_list, predicted_list, digits=4)
print(class_report)
return f1_score(labels_list, predicted_list)
def test(self, test_dataset, model, all_labels):
test_dataloader = DataLoader(test_dataset, batch_size=self.args.dev_batch_size, shuffle=False)
model.to(self.args.device)
model.eval()
predicted_list = []
labels_list = []
for step, batch in enumerate(tqdm(test_dataloader)):
batch = tuple(t.to(self.args.device) for t in batch)
with torch.no_grad():
input_ids, attention_mask, pred_mask, labels = batch
predicted = model(
input_ids=input_ids,
attention_mask=attention_mask,
pred_mask=pred_mask
)
predicted = predicted[0]
if self.args.crf:
predicted = [seq[seq>=0].tolist() for seq in predicted]
else:
predicted = [seq[mask==1].tolist() for seq, mask in zip(predicted, pred_mask)]
groud_labels = [seq[mask==1].tolist() for seq, mask in zip(labels, pred_mask)]
for tl, pl in zip(groud_labels, predicted):
labels_list.append([all_labels[l] for l in tl])
predicted_list.append([all_labels[l] for l in pl])
with open(self.args.output_dir, "w", encoding="utf-8") as f:
for labels in predicted_list:
for l in labels:
f.write(l+"\n")
f.write("\n")
class_report = classification_report(labels_list, predicted_list, digits=4)
print(class_report)
with open(self.args.output_dir+"_report", "w", encoding="utf-8") as f:
f.write(class_report)