/
model.py
188 lines (169 loc) · 7.02 KB
/
model.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
import pytorch_lightning as pl
from pytorch_lightning.core.step_result import TrainResult,EvalResult
from pytorch_lightning import Trainer
from torch.utils.data import SequentialSampler,RandomSampler
from torch import nn
import numpy as np
import math
import torch
from torch.utils.data.dataloader import DataLoader
from torch.utils.data import RandomSampler
from torch.utils.data import DataLoader,RandomSampler
from torch.optim import AdamW
from transformers import get_linear_schedule_with_warmup
from transformers import AutoTokenizer,AutoModel
import functools
class MCQAModel(pl.LightningModule):
def __init__(self,
model_name_or_path,
args):
super().__init__()
self.init_encoder_model(model_name_or_path)
self.args = args
self.batch_size = self.args['batch_size']
self.dropout = nn.Dropout(self.args['hidden_dropout_prob'])
self.linear = nn.Linear(in_features=self.args['hidden_size'],out_features=1)
self.ce_loss = nn.CrossEntropyLoss()
self.save_hyperparameters()
def init_encoder_model(self,model_name_or_path):
self.tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
self.model = AutoModel.from_pretrained(model_name_or_path)
def prepare_dataset(self,train_dataset,val_dataset,test_dataset=None):
"""
helper to set the train and val dataset. Doing it during class initialization
causes issues while loading checkpoint as the dataset class needs to be
present for the weights to be loaded.
"""
self.train_dataset = train_dataset
self.val_dataset = val_dataset
if test_dataset != None:
self.test_dataset = test_dataset
else:
self.test_dataset = val_dataset
def forward(self,input_ids,attention_mask,token_type_ids):
outputs = self.model(input_ids=input_ids,
attention_mask=attention_mask,
token_type_ids=token_type_ids)
pooled_output = outputs[1]
pooled_output = self.dropout(pooled_output)
logits = self.linear(pooled_output)
reshaped_logits = logits.view(-1,self.args['num_choices'])
return reshaped_logits
def training_step(self,batch,batch_idx):
inputs,labels = batch
for key in inputs:
inputs[key] = inputs[key].to(self.args["device"])
logits = self(**inputs)
loss = self.ce_loss(logits,labels)
result = TrainResult(loss)
result.log('train_loss', loss, on_epoch=True)
return result
def test_step(self, batch, batch_idx):
inputs,labels = batch
for key in inputs:
inputs[key] = inputs[key].to(self.args["device"])
logits = self(**inputs)
loss = self.ce_loss(logits,labels)
result = EvalResult(loss)
result.log('test_loss', loss, on_epoch=True)
result.log('logits',logits,on_epoch=True)
result.log('labels',labels,on_epoch=True)
self.log('test_loss', loss)
return result
def test_epoch_end(self, outputs):
avg_loss = outputs['test_loss'].mean()
predictions = torch.argmax(outputs['logits'],axis=-1)
labels = outputs['labels']
self.test_predictions = predictions
correct_predictions = torch.sum(predictions==labels)
accuracy = correct_predictions.cpu().detach().numpy()/predictions.size()[0]
result = EvalResult(checkpoint_on=avg_loss,early_stop_on=avg_loss)
result.log_dict({"test_loss":avg_loss,"test_acc":accuracy},prog_bar=True,on_epoch=True)
self.log('avg_test_loss', avg_loss)
self.log('avg_test_acc', accuracy)
return result
def validation_step(self, batch, batch_idx):
inputs,labels = batch
for key in inputs:
inputs[key] = inputs[key].to(self.args["device"])
logits = self(**inputs)
loss = self.ce_loss(logits,labels)
result = EvalResult(loss)
result.log('val_loss', loss, on_epoch=True)
result.log('logits',logits,on_epoch=True)
result.log('labels',labels,on_epoch=True)
self.log('val_loss', loss)
return result
def validation_epoch_end(self, outputs):
avg_loss = outputs['val_loss'].mean()
predictions = torch.argmax(outputs['logits'],axis=-1)
labels = outputs['labels']
correct_predictions = torch.sum(predictions==labels)
accuracy = correct_predictions.cpu().detach().numpy()/predictions.size()[0]
result = EvalResult(checkpoint_on=avg_loss,early_stop_on=avg_loss)
result.log_dict({"val_loss":avg_loss,"val_acc":accuracy},prog_bar=True,on_epoch=True)
self.log('avg_val_loss', avg_loss)
self.log('avg_val_acc', accuracy)
return result
def configure_optimizers(self):
optimizer = AdamW(self.parameters(),lr=self.args['learning_rate'],eps=1e-8)
scheduler = get_linear_schedule_with_warmup(
optimizer,
num_warmup_steps=100,
num_training_steps=(self.args['num_epochs'] + 1) * math.ceil(len(self.train_dataset) / self.args['batch_size']),
)
return [optimizer],[scheduler]
def process_batch(self,batch,tokenizer,max_len=32):
expanded_batch = []
labels = []
context = None
for data_tuple in batch:
if len(data_tuple) == 4:
context,question,options,label = data_tuple
else:
question,options,label = data_tuple
question_option_pairs = [question+' '+option for option in options]
labels.append(label)
if context:
contexts = [context]*len(options)
expanded_batch.extend(zip(contexts,question_option_pairs))
else:
expanded_batch.extend(question_option_pairs)
tokenized_batch = tokenizer.batch_encode_plus(expanded_batch,truncation=True,padding="max_length",max_length=max_len,return_tensors="pt")
return tokenized_batch,torch.tensor(labels)
def train_dataloader(self):
train_sampler = RandomSampler(self.train_dataset)
model_collate_fn = functools.partial(
self.process_batch,
tokenizer=self.tokenizer,
max_len=self.args['max_len']
)
train_dataloader = DataLoader(self.train_dataset,
batch_size=self.batch_size,
sampler=train_sampler,
collate_fn=model_collate_fn)
return train_dataloader
def val_dataloader(self):
eval_sampler = SequentialSampler(self.val_dataset)
model_collate_fn = functools.partial(
self.process_batch,
tokenizer=self.tokenizer,
max_len=self.args['max_len']
)
val_dataloader = DataLoader(self.val_dataset,
batch_size=self.batch_size,
sampler=eval_sampler,
collate_fn=model_collate_fn)
return val_dataloader
def test_dataloader(self):
eval_sampler = SequentialSampler(self.test_dataset)
model_collate_fn = functools.partial(
self.process_batch,
tokenizer=self.tokenizer,
max_len=self.args['max_len']
)
test_dataloader = DataLoader(self.test_dataset,
batch_size=self.batch_size,
sampler=eval_sampler,
collate_fn=model_collate_fn)
return test_dataloader