fine_tune.py

from transformers import AutoTokenizer, AutoModelForCausalLM, Trainer, TrainingArguments
from datasets import load_dataset

# Load the dataset
dataset = load_dataset("csv", data_files="qa_dataset.csv")

# Load the tokenizer and model
model_name = "gpt2"  # Base model
tokenizer = AutoTokenizer.from_pretrained(model_name)
# Add padding token to the tokenizer
tokenizer.pad_token = tokenizer.eos_token # Set the padding token to the end-of-sequence token
model = AutoModelForCausalLM.from_pretrained(model_name)


# Prepare the dataset
def preprocess_function(examples):
    inputs = [f"Q: {q} A:" for q in examples["question"]]
    outputs = examples["answer"]
    model_inputs = tokenizer(inputs, text_target=outputs, max_length=512, padding ='longest', truncation=True)
    return model_inputs

tokenized_dataset = dataset["train"].map(preprocess_function, batched=True)

training_args = TrainingArguments(
    output_dir="./results",
    evaluation_strategy="no",
    learning_rate=5e-5,
    per_device_train_batch_size=8,
    num_train_epochs=3,
    weight_decay=0.01,
    logging_dir='./logs',
    logging_steps=10,
    report_to="tensorboard",
    run_name="gpt2-finetuning"
)

trainer = Trainer(
    model=model,
    args=training_args,
    train_dataset=tokenized_dataset,
    tokenizer=tokenizer
)

# Fine-tune the model
trainer.train()

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