demo.py

#!/usr/bin/env python3
"""
T5 Prompt Enhancer V0.3 Demo Script
Quick test of all four instruction types
"""

import torch
from transformers import T5Tokenizer, T5ForConditionalGeneration

def load_model():
    """Load the T5 V0.3 model"""
    print("🤖 Loading T5 Prompt Enhancer V0.3...")
    
    tokenizer = T5Tokenizer.from_pretrained(".")
    model = T5ForConditionalGeneration.from_pretrained(".")
    
    if torch.cuda.is_available():
        model = model.cuda()
        print("✅ Model loaded on GPU")
    else:
        print("✅ Model loaded on CPU")
    
    return model, tokenizer

def enhance_prompt(model, tokenizer, text, style="clean"):
    """Generate enhanced prompt with style control"""
    
    style_prompts = {
        "clean": f"Enhance this prompt (no lora): {text}",
        "technical": f"Enhance this prompt (with lora): {text}",
        "simplify": f"Simplify this prompt: {text}",
        "standard": f"Enhance this prompt: {text}"
    }
    
    prompt = style_prompts[style]
    inputs = tokenizer(prompt, return_tensors="pt", max_length=256, truncation=True)
    
    if torch.cuda.is_available():
        inputs = {k: v.cuda() for k, v in inputs.items()}
    
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_length=80,
            num_beams=2,
            repetition_penalty=2.0,
            no_repeat_ngram_size=3,
            pad_token_id=tokenizer.pad_token_id
        )
    
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

def main():
    """Demo all four instruction types"""
    
    # Load model
    model, tokenizer = load_model()
    
    # Test prompts
    test_prompts = [
        "woman in red dress",
        "cat on chair", 
        "cyberpunk cityscape",
        "masterpiece, best quality, ultra-detailed render of a fantasy dragon with golden scales"
    ]
    
    styles = ["standard", "clean", "technical", "simplify"]
    
    print("\n🎨 T5 Prompt Enhancer V0.3 Demo")
    print("="*60)
    
    for prompt in test_prompts:
        print(f"\n📝 Input: '{prompt}'")
        print("-" * 40)
        
        for style in styles:
            try:
                result = enhance_prompt(model, tokenizer, prompt, style)
                print(f"{style:>10}: {result}")
            except Exception as e:
                print(f"{style:>10}: ERROR - {e}")
        
        print()

if __name__ == "__main__":
    main()