File size: 5,170 Bytes
2086153
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
import logging

from datetime import datetime

import re
from collections import Counter

import pandas as pd
import numpy as np

import torch
from torch.nn import CrossEntropyLoss
from torch.utils.data import Dataset, DataLoader

from transformers import (
    BertConfig,
    BertForSequenceClassification,
    BertTokenizer,
    Trainer,
    TrainingArguments,
    EarlyStoppingCallback,
)

from sklearn.model_selection import train_test_split
from sklearn.metrics import (
    accuracy_score,
    f1_score,
    precision_score,
    recall_score,
    confusion_matrix,
)
from sklearn.utils.class_weight import compute_class_weight

tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
config = BertConfig.from_pretrained("bert-base-uncased", num_labels=2)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

class WeightedBertForSequenceClassification(BertForSequenceClassification):
    def __init__(self, config, class_weights):
        super().__init__(config)
        self.class_weights = class_weights

    def forward(self, input_ids=None, attention_mask=None, labels=None, **kwargs):
        outputs = super().forward(input_ids=input_ids, attention_mask=attention_mask, labels=None, **kwargs)
        logits = outputs.logits
        loss = None
        if labels is not None:
            loss_fct = CrossEntropyLoss(weight=self.class_weights)
            loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1))
        return {"loss": loss, "logits": logits}

class SMSClassificationDataset(Dataset):
    def __init__(self, encodings, labels):
        self.encodings = encodings
        self.labels = torch.tensor(labels, dtype=torch.long)

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

    def __getitem__(self, idx):
        item = {key: val[idx] for key, val in self.encodings.items()}
        item["labels"] = self.labels[idx]
        return item

def compute_metrics(eval_pred):
    logits, labels = eval_pred
    predictions = torch.argmax(torch.tensor(logits), dim=1)

    acc = accuracy_score(labels, predictions)
    precision = precision_score(labels, predictions, average="weighted", zero_division=0)
    recall = recall_score(labels, predictions, average="weighted")
    f1 = f1_score(labels, predictions, average='weighted')
    cm = confusion_matrix(labels, predictions)

    print("Confusion Matrix:\n", cm)

    return {
        'accuracy': acc,
        'precision': precision,
        'recall': recall,
        'f1': f1
    }

def train():
    df = pd.read_csv('data/spam.csv', encoding='iso-8859-1')[['label', 'text']]

    label_mapping = {'spam': 1, 'ham': 0}
    df['label'] = df['label'].map(label_mapping)

    train_texts, val_texts, train_labels, val_labels = train_test_split(
        df['text'].tolist(), df['label'].tolist(), test_size=0.25, random_state=42)

    class_weights = compute_class_weight(
        class_weight='balanced',
        classes=np.unique(train_labels),
        y=train_labels
    )
    class_weights = torch.tensor(class_weights, dtype=torch.float).to(device)

    model = WeightedBertForSequenceClassification(config, class_weights=class_weights)

    loggers = [logging.getLogger(name) for name in logging.root.manager.loggerDict]
    for logger in loggers:
        if "transformers" in logger.name.lower():
            logger.setLevel(logging.ERROR)

    model.load_state_dict(BertForSequenceClassification.from_pretrained("bert-base-uncased", num_labels=2, use_safetensors=True, return_dict=False, attn_implementation="sdpa").state_dict(), strict=False)
    model.to(device)

    train_encodings = tokenizer(train_texts, truncation=True, padding=True, return_tensors="pt")
    val_encodings = tokenizer(val_texts, truncation=True, padding=True, return_tensors="pt")

    train_dataset = SMSClassificationDataset(train_encodings, train_labels)
    val_dataset = SMSClassificationDataset(val_encodings, val_labels)

    training_args = TrainingArguments(
        output_dir='./models/pretrained',
        num_train_epochs=5,
        per_device_train_batch_size=8,
        per_device_eval_batch_size=16,
        warmup_steps=500,
        weight_decay=0.01,
        logging_dir='./logs',
        logging_steps=10,
        eval_strategy="epoch",
        report_to="none",
        save_total_limit=1,
        load_best_model_at_end=True,
        save_strategy="epoch",
    )

    trainer = Trainer(
        model=model,
        args=training_args,
        train_dataset=train_dataset,
        eval_dataset=val_dataset,
        compute_metrics=compute_metrics,
        callbacks=[EarlyStoppingCallback(early_stopping_patience=3)]
    )

    trainer.train()

    logs = trainer.state.log_history
    df_logs = pd.DataFrame(logs)

    timestamp = datetime.now().strftime('%Y%m%d_%H%M%S')
    df_logs.to_csv(f"logs/training_logs_{timestamp}.csv", index=False)

    tokenizer.save_pretrained('./models/pretrained')
    model.save_pretrained('./models/pretrained')

if __name__ == "__main__":
    train()