train.py

# Import necessary libraries
from datasets import load_dataset
from transformers import AutoTokenizer, AutoModelForSequenceClassification, Trainer, TrainingArguments

# Load your dataset (assuming you uploaded it to Hugging Face)
dataset = load_dataset("romanurdu_dataset")

# Load pre-trained mBERT tokenizer and model
tokenizer = AutoTokenizer.from_pretrained('bert-base-multilingual-cased')
model = AutoModelForSequenceClassification.from_pretrained('bert-base-multilingual-cased', num_labels=2)

# Tokenize the dataset (adjust based on your dataset's structure)
def tokenize_function(examples):
    return tokenizer(examples['Text'], padding="max_length", truncation=True)

# Tokenize the datasets
tokenized_datasets = dataset.map(tokenize_function, batched=True)

# Split into train and test datasets (if not already split)
train_dataset = tokenized_datasets['train']
test_dataset = tokenized_datasets['test']

# Define training arguments
training_args = TrainingArguments(
    output_dir='./results',          # output directory for model checkpoints
    evaluation_strategy="epoch",     # evaluate after each epoch
    learning_rate=2e-5,              # learning rate
    per_device_train_batch_size=16,  # batch size for training
    per_device_eval_batch_size=64,   # batch size for evaluation
    num_train_epochs=3,              # number of epochs
    weight_decay=0.01,               # strength of weight decay
    logging_dir='./logs',            # directory to store logs
)

# Initialize Trainer
trainer = Trainer(
    model=model,                         # the model to be trained
    args=training_args,                  # training arguments
    train_dataset=train_dataset,         # training dataset
    eval_dataset=test_dataset            # evaluation dataset
)

# Train the model
trainer.train()

# Save the model to Hugging Face Model Hub
model.push_to_hub("SentimentAnalysisRomanUrdu")
tokenizer.push_to_hub("SentimentAnalysisRomanUrdu")