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

# Load model and tokenizer
model_name = "distilgpt2"
model = AutoModelForCausalLM.from_pretrained(model_name)
tokenizer = AutoTokenizer.from_pretrained(model_name)
tokenizer.pad_token = tokenizer.eos_token

# Load dialogue dataset
dataset = load_dataset("HuggingFaceH4/ultrachat", split="train[:1%]")  # Use 1% for demo

# Preprocess dataset
def preprocess(examples):
    prompts = [f"User: {ex['prompt']} Assistant: {ex['response']}" for ex in examples]
    return tokenizer(prompts, truncation=True, padding="max_length", max_length=512)

tokenized_dataset = dataset.map(preprocess, batched=True)

# Training arguments
training_args = TrainingArguments(
    output_dir="./evo_finetuned",
    per_device_train_batch_size=4,
    num_train_epochs=3,
    save_steps=1000,
    save_total_limit=2,
)

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

# Fine-tune
trainer.train()

# Save model
model.save_pretrained("evo_finetuned")
tokenizer.save_pretrained("evo_finetuned")