Model

# Step 1: Install Required Libraries

# Uncomment the following lines if you are running this in a new environment

# !pip install transformers torch datasets pandas



# Step 2: Import Required Libraries

import pandas as pd

import torch

from transformers import BertTokenizer, BertForSequenceClassification, Trainer, TrainingArguments

from datasets import load_dataset



# Step 3: Prepare Your Dataset

# Create a sample dataset (replace this with your actual dataset)

data = {

    'text': [

        "అయ్యో, నిన్ను చూసి చాలా కాలమైంది!",  # 0

        "నువ్వు ఈ రోజు రాంబాబు ఇంటికి వెళ్ళావా?",  # 1

        "హే, నువ్వు ఈ కొత్త సినిమా చూసావా?",  # 0

        "ఈ ప్రాజెక్ట్ మీద పని ఎలా జరుగుతోంది?",  # 1

        "సూపర్ మాస్! బాగా ఎంజాయ్ చేశా.",  # 0

        "చాలా కష్టంగా ఉంది, కానీ మామూలుగా ఉంది."  # 1

    ],

    'label': [0, 1, 0, 1, 0, 1]  # Example labels

}



# Create a DataFrame

df = pd.DataFrame(data)



# Split into train and test sets

train_texts = df['text'][:4].tolist()

train_labels = df['label'][:4].tolist()

test_texts = df['text'][4:].tolist()

test_labels = df['label'][4:].tolist()



# Step 4: Tokenization

tokenizer = BertTokenizer.from_pretrained('bert-base-multilingual-cased')  # Use a multilingual model

train_encodings = tokenizer(train_texts, truncation=True, padding=True)

test_encodings = tokenizer(test_texts, truncation=True, padding=True)



# Step 5: Create a Dataset Class

class ColloquialDataset(torch.utils.data.Dataset):

    def _init_(self, encodings, labels):

        self.encodings = encodings

        self.labels = labels



    def _getitem_(self, idx):

        item = {key: torch.tensor(val[idx]) for key, val in self.encodings.items()}

        item['labels'] = torch.tensor(self.labels[idx])

        return item



    def _len_(self):

        return len(self.labels)



# Create dataset objects

train_dataset = ColloquialDataset(train_encodings, train_labels)

test_dataset = ColloquialDataset(test_encodings, test_labels)



# Step 6: Load the Model

model = BertForSequenceClassification.from_pretrained('bert-base-multilingual-cased', num_labels=2)



# Step 7: Set Up Training Arguments

training_args = TrainingArguments(

    output_dir='./results',          # output directory

    num_train_epochs=3,              # total number of training epochs

    per_device_train_batch_size=2,   # batch size per device during training

    per_device_eval_batch_size=2,    # batch size for evaluation

    warmup_steps=10,                  # number of warmup steps for learning rate scheduler

    weight_decay=0.01,                # strength of weight decay

    logging_dir='./logs',            # directory for storing logs

    logging_steps=10,

)



# Step 8: Initialize the Trainer

trainer = Trainer(

    model=model,

    args=training_args,

    train_dataset=train_dataset,

    eval_dataset=test_dataset

)



# Step 9: Train the Model

trainer.train()



# Step 10: Evaluate the Model

eval_results = trainer.evaluate()

print("Evaluation Results:", eval_results)



# Step 11: Save the Fine-Tuned Model

model.save_pretrained('./fine_tuned_model')

tokenizer.save_pretrained('./fine_tuned_model')



print("Model and tokenizer saved successfully!")