fine_tune_model.py

import pandas as pd
from transformers import AutoModelForCausalLM, AutoTokenizer, Trainer, TrainingArguments
from datasets import load_dataset

# Load the dataset
dataset = load_dataset('csv', data_files='path/to/your/dataset.csv')  # Update this path

# Load pre-trained model and tokenizer
model_name = "gpt2"  # or another model
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

# Tokenize the dataset
def tokenize_function(examples):
    return tokenizer(examples['text'], truncation=True)

tokenized_datasets = dataset.map(tokenize_function, batched=True)

# Set training arguments
training_args = TrainingArguments(
    output_dir='./results',
    evaluation_strategy='epoch',
    learning_rate=2e-5,
    per_device_train_batch_size=2,
    num_train_epochs=3,
    weight_decay=0.01,
)

# Create Trainer
trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_datasets['train'],
    eval_dataset=tokenized_datasets['test']
)

# Fine-tune the model
trainer.train()

# Save the model
model.save_pretrained('./fine_tuned_model')
tokenizer.save_pretrained('./fine_tuned_model')

print("Model fine-tuned and saved successfully.")