最近用transformers用的比较多,用transformers加载预训练模型,然后做预测的教程挺多的,但是加载自己的数据进行训练的挺少的,我这里分享一下我的实现:
数据准备
数据来自于kaggle上面情感分析的数据,地址为:
https://www.kaggle.com/lava18/google-play-store-apps?select=googleplaystore_user_reviews.csv
导入函数
import transformers
from transformers import BertModel, BertTokenizer, AdamW, get_linear_schedule_with_warmup
import torch
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.metrics import confusion_matrix, classification_report
from collections import defaultdict
from textwrap import wrap
from torch import nn, optim
from torch.utils.data import Dataset, DataLoader
import torch.nn.functional as F
RANDOM_SEED = 42
np.random.seed(RANDOM_SEED)
torch.manual_seed(RANDOM_SEED)
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
加载数据,预处理
df = pd.read_csv("archive/googleplaystore_user_reviews.csv")
df=df.dropna()
def to_sentiment(rating):
if rating == 'Positive':
return 2
elif rating == 'Neutral':
return 1
return 0
df['sentiment'] = df.Sentiment.apply(to_sentiment)
class_names=["Negative","Neutral","Positive"]
df_train, df_test = train_test_split(df, test_size=0.1, random_state=RANDOM_SEED)
df_val, df_test = train_test_split(df_test, test_size=0.5, random_state=RANDOM_SEED)
创建dataset和dataloader
class GPReviewDataset(Dataset):
def __init__(self, reviews, targets, tokenizer, max_len):
self.reviews = reviews
self.targets = targets
self.tokenizer = tokenizer
self.max_len = max_len
def __len__(self):
return len(self.reviews)
def __getitem__(self, item):
review = str(self.reviews[item])
target = self.targets[item]
encoding = self.tokenizer.encode_plus(
review,
add_special_tokens=True,
max_length=self.max_len,
return_token_type_ids=False,
pad_to_max_length=True,
return_attention_mask=True,
return_tensors='pt',
)
# print(target)
return {
'review_text': review,
'input_ids': encoding['input_ids'].flatten(),
'attention_mask': encoding['attention_mask'].flatten(),
'targets': torch.tensor(target, dtype=torch.long)
}
def create_data_loader(df, tokenizer, max_len, batch_size):
ds = GPReviewDataset(
reviews=df.Translated_Review.to_numpy(),
targets=df.sentiment.to_numpy(),
tokenizer=tokenizer,
max_len=max_len
)
return DataLoader(
ds,
batch_size=batch_size,
num_workers=4
)
BATCH_SIZE = 16
MAX_LEN = 160
PRE_TRAINED_MODEL_NAME = 'bert-base-uncased'
tokenizer = BertTokenizer.from_pretrained(PRE_TRAINED_MODEL_NAME)
train_data_loader = create_data_loader(df_train, tokenizer, MAX_LEN, BATCH_SIZE)
val_data_loader = create_data_loader(df_val, tokenizer, MAX_LEN, BATCH_SIZE)
test_data_loader = create_data_loader(df_test, tokenizer, MAX_LEN, BATCH_SIZE)
创建基于BERT的情感分析模型
class SentimentClassifier(nn.Module):
def __init__(self, n_classes):
super(SentimentClassifier, self).__init__()
self.bert = BertModel.from_pretrained(PRE_TRAINED_MODEL_NAME)
self.drop = nn.Dropout(p=0.3)
self.out = nn.Linear(self.bert.config.hidden_size, n_classes)
def forward(self, input_ids, attention_mask):
_, pooled_output = self.bert(
input_ids=input_ids,
attention_mask=attention_mask
)
output = self.drop(pooled_output)
return self.out(output)
训练和验证
model = SentimentClassifier(len(class_names))
model = model.to(device)
EPOCHS = 10
optimizer = AdamW(model.parameters(), lr=2e-5, correct_bias=False)
total_steps = len(train_data_loader) * EPOCHS
scheduler = get_linear_schedule_with_warmup(
optimizer,
num_warmup_steps=0,
num_training_steps=total_steps
)
loss_fn = nn.CrossEntropyLoss().to(device)
训练函数
def train_epoch(
model,
data_loader,
loss_fn,
optimizer,
device,
scheduler,
n_examples
):
model = model.train()
losses = []
correct_predictions = 0
for d in data_loader:
input_ids = d["input_ids"].to(device)
attention_mask = d["attention_mask"].to(device)
targets = d["targets"].to(device)
outputs = model(
input_ids=input_ids,
attention_mask=attention_mask
)
_, preds = torch.max(outputs, dim=1)
loss = loss_fn(outputs, targets)
correct_predictions += torch.sum(preds == targets)
losses.append(loss.item())
loss.backward()
nn.utils.clip_grad_norm_(model.parameters(), max_norm=1.0)
optimizer.step()
scheduler.step()
optimizer.zero_grad()
return correct_predictions.double() / n_examples, np.mean(losses)
验证函数:
def eval_model(model, data_loader, loss_fn, device, n_examples):
model = model.eval()
losses = []
correct_predictions = 0
with torch.no_grad():
for d in data_loader:
input_ids = d["input_ids"].to(device)
attention_mask = d["attention_mask"].to(device)
targets = d["targets"].to(device)
outputs = model(
input_ids=input_ids,
attention_mask=attention_mask
)
_, preds = torch.max(outputs, dim=1)
loss = loss_fn(outputs, targets)
correct_predictions += torch.sum(preds == targets)
losses.append(loss.item())
return correct_predictions.double() / n_examples, np.mean(losses)
调用他们来进行训练:
history = defaultdict(list)
best_accuracy = 0
for epoch in range(EPOCHS):
print(f'Epoch {epoch + 1}/{EPOCHS}')
print('-' * 10)
train_acc, train_loss = train_epoch(
model,
train_data_loader,
loss_fn,
optimizer,
device,
scheduler,
len(df_train)
)
print(f'Train loss {train_loss} accuracy {train_acc}')
val_acc, val_loss = eval_model(
model,
val_data_loader,
loss_fn,
device,
len(df_val)
)
print(f'Val loss {val_loss} accuracy {val_acc}')
print()
history['train_acc'].append(train_acc)
history['train_loss'].append(train_loss)
history['val_acc'].append(val_acc)
history['val_loss'].append(val_loss)
if val_acc > best_accuracy:
torch.save(model.state_dict(), 'best_model_state.bin')
best_accuracy = val_acc
结束,虽然预测只需要加载模型,然后把输入变成id就行了。但是训练过程还是比预测过程麻烦很多,我这里只提供了一种基础方法,还有很多需要完善的地方,欢迎改进。
参考文献
[1].Google Play Store Apps. https://www.kaggle.com/lava18/google-play-store-apps?select=googleplaystore_user_reviews.csv
[2].Sentiment Analysis with BERT and Transformers by Hugging Face using PyTorch and Python. https://curiousily.com/posts/sentiment-analysis-with-bert-and-hugging-face-using-pytorch-and-python/
[3].Sentiment Analysis with BERT. https://github.com/curiousily/Getting-Things-Done-with-Pytorch/blob/master/08.sentiment-analysis-with-bert.ipynb