train.py

import torch
import torch.nn as nn
import torch.optim as optim
import pandas as pd
import numpy as np
import urllib.request
import zipfile
import os

from torch.utils.data import Dataset, DataLoader
from sklearn.model_selection import train_test_split
from transformers import BertTokenizer, BertModel
from model import SentimentClassifier
# Download dataset
url = "https://archive.ics.uci.edu/ml/machine-learning-databases/00331/sentiment%20labelled%20sentences.zip"
filename = "sentiment.zip"

if not os.path.exists(filename):
    urllib.request.urlretrieve(url, filename)

# Extract dataset
with zipfile.ZipFile(filename, 'r') as zip_ref:
    zip_ref.extractall()

# Load dataset
filepath_dict = {'yelp':   'sentiment labelled sentences/yelp_labelled.txt',
                 'amazon': 'sentiment labelled sentences/amazon_cells_labelled.txt',
                 'imdb':   'sentiment labelled sentences/imdb_labelled.txt'}

df_list = []
for source, filepath in filepath_dict.items():
    df = pd.read_csv(filepath, names=['sentence', 'label'], sep='\t')
    df['source'] = source
    df_list.append(df)

df = pd.concat(df_list)

# Split dataset into train and test sets
sentences = df['sentence'].values
labels = df['label'].values
train_sentences, test_sentences, train_labels, test_labels = train_test_split(
    sentences, labels, test_size=0.25)

# Define tokenizer
tokenizer = BertTokenizer.from_pretrained(
    'bert-base-uncased', do_lower_case=True)

# Define dataset


class SentimentDataset(Dataset):
    def __init__(self, sentences, labels, tokenizer, max_len):
        self.sentences = sentences
        self.labels = labels
        self.tokenizer = tokenizer
        self.max_len = max_len

    def __len__(self):
        return len(self.sentences)

    def __getitem__(self, item):
        sentence = str(self.sentences[item])
        label = self.labels[item]

        encoding = self.tokenizer.encode_plus(
            sentence,
            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'
        )

        return {'sentence': sentence,
                'input_ids': encoding['input_ids'].flatten(),
                'attention_mask': encoding['attention_mask'].flatten(),
                'label': torch.tensor(label, dtype=torch.long)}

# Define model


# Set device
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')

# Define hyperparameters
MAX_LEN = 100
BATCH_SIZE = 16
EPOCHS = 5

# Define dataloaders
train_dataset = SentimentDataset(
    train_sentences, train_labels, tokenizer, MAX_LEN)
test_dataset = SentimentDataset(
    test_sentences, test_labels, tokenizer, MAX_LEN)
train_dataloader = DataLoader(
    train_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=4)
test_dataloader = DataLoader(
    test_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=4)

# Define model and optimizer
model = SentimentClassifier(2)
model = model.to(device)
optimizer = optim.Adam(model.parameters(), lr=2e-5)

# Define loss function
criterion = nn.CrossEntropyLoss()

# Train model
for epoch in range(EPOCHS):
    print('Epoch:', epoch+1)
    train_loss = 0
    train_acc = 0

    model.train()
    for batch in train_dataloader:
        input_ids = batch['input_ids'].to(device)
        attention_mask = batch['attention_mask'].to(device)
        labels = batch['label'].to(device)

        optimizer.zero_grad()

        outputs = model(input_ids=input_ids, attention_mask=attention_mask)

        loss = criterion(outputs, labels)
        loss.backward()
        optimizer.step()

        train_loss += loss.item()
        train_acc += (outputs.argmax(1) == labels).sum().item()

    train_loss /= len(train_dataloader)
    train_acc /= len(train_dataset)

    print('Train loss:', train_loss, 'Train accuracy:', train_acc)

    model.eval()
    test_loss = 0
    test_acc = 0

    with torch.no_grad():
        for batch in test_dataloader:
            input_ids = batch['input_ids'].to(device)
            attention_mask = batch['attention_mask'].to(device)
            labels = batch['label'].to(device)

            outputs = model(input_ids=input_ids, attention_mask=attention_mask)

            loss = criterion(outputs, labels)

            test_loss += loss.item()
            test_acc += (outputs.argmax(1) == labels).sum().item()

    test_loss /= len(test_dataloader)
    test_acc /= len(test_dataset)

    print('Test loss:', test_loss, 'Test accuracy:', test_acc)

torch.save(model.cpu().state_dict(), 'sentiment_model.pth')