llm_processing.py

def generate_email(about_text, recipient_name, project_name, key_features, email_style):
    # Define email style prompts
    style_prompts = {
        "Casual": "Compose a friendly and casual email.",
        "Professional": "Draft a formal and professional email.",
        "Cold": "Write a concise and attention-grabbing cold email."
    }
    style_intro = style_prompts.get(email_style, "Compose an email")

    # Construct the prompt for the LLM
    prompt = (
        f"{style_intro} Addressing {recipient_name}, based on this profile: {about_text}. "
        f"Discuss {project_name} with its key features: {', '.join(key_features)}."
    )
    
    try:
        # Tokenize the prompt
        inputs = tokenizer(prompt, return_tensors='pt')
        
        # Generate the response
        with torch.no_grad():
            outputs = model.generate(
                inputs['input_ids'],
                max_length=500,  # Adjust the length as needed
                num_beams=5,     # Beam search parameters for better quality
                early_stopping=True
            )
        
        # Decode the generated text
        response_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
        return response_text
    
    except Exception as e:
        return f"Error: {str(e)}"