app.py

import torch
from transformers import GPT2LMHeadModel, GPT2Tokenizer


def select_model():
    while True:
        print("\nAvailable GPT-2 Models:")
        print("1. gpt2 (Small)")
        print("2. gpt2-medium (Medium)")
        print("3. gpt2-large (Large)")
        print("4. gpt2-xl (Extra Large)")
        choice = input("Select a model (1/2/3/4): ")

        if choice == "1":
            return "gpt2"
        elif choice == "2":
            return "gpt2-medium"
        elif choice == "3":
            return "gpt2-large"
        elif choice == "4":
            return "gpt2-xl"
        else:
            print("Invalid choice. Please select a valid model.")


def enhance_prompt(prompt, model_name, max_length=50, num_return_sequences=1):
    # Load the selected pre-trained GPT-2 model and tokenizer
    model = GPT2LMHeadModel.from_pretrained(model_name)
    tokenizer = GPT2Tokenizer.from_pretrained(model_name)

    # Tokenize the prompt
    input_ids = tokenizer.encode(prompt, return_tensors="pt")

    # Generate text based on the prompt
    output = model.generate(
        input_ids,
        max_length=max_length,
        num_return_sequences=num_return_sequences,
        no_repeat_ngram_size=2,  # Avoid repetitive phrases
        top_k=50,  # Limit choices to top-k tokens
        top_p=0.95,  # Control diversity with nucleus sampling
        temperature=0.7  # Adjusts the randomness of the output
    )

    # Decode and return the generated text
    enhanced_prompts = [tokenizer.decode(output_item, skip_special_tokens=True) for output_item in output]
    return enhanced_prompts


if __name__ == "__main__":
    while True:
        model_name = select_model()
        prompt = input("Enter a prompt (or 'exit' to quit): ")
        if prompt.lower() == "exit":
            break

        enhanced_prompts = enhance_prompt(prompt, model_name)
        print("\nEnhanced Prompts:")
        for idx, enhanced_prompt in enumerate(enhanced_prompts):
            print(f"Enhanced Prompt {idx + 1}: {enhanced_prompt}\n")