paraphraser.py

# paraphraser.py
import torch
from model_loader import paraphrase_model, paraphrase_tokenizer

def paraphrase_comment(comment):
    """
    Paraphrase a toxic comment using the Granite 3.2-2B-Instruct model.
    Returns the paraphrased comment.
    """
    # Define the paraphrasing prompt with system instruction, guidelines, examples, and the task
    prompt = (
        "You are a content moderator tasked with rewriting toxic comments into neutral and constructive ones while maintaining the original meaning.\n"
        "Guidelines:\n"
        "- Remove explicit hate speech, personal attacks, or offensive language.\n"
        "- Keep the response neutral and professional.\n"
        "- Ensure the rewritten comment retains the original intent but in a constructive tone.\n"
        "Examples:\n"
        "Toxic: \"You're so dumb! You never understand anything!\"\n"
        "Neutral: \"I think there's some misunderstanding. Let's clarify things.\"\n"
        "Toxic: \"This is the worst idea ever. Only an idiot would suggest this.\"\n"
        "Neutral: \"I don't think this idea works well. Maybe we can explore other options.\"\n"
        "Now, rewrite this comment: \"{comment}\""
    )

    # Format the prompt with the input comment
    prompt = prompt.format(comment=comment)

    # Tokenize the prompt
    inputs = paraphrase_tokenizer(prompt, return_tensors="pt", truncation=True, padding=True, max_length=512)

    # Generate the paraphrased output
    with torch.no_grad():
        outputs = paraphrase_model.generate(
            **inputs,
            max_length=512,
            num_return_sequences=1,
            do_sample=True,
            top_p=0.95,
            temperature=0.7
        )

    # Decode the generated output
    paraphrased_comment = paraphrase_tokenizer.decode(outputs[0], skip_special_tokens=True)

    # Remove the prompt part from the output (if the model includes it)
    paraphrased_comment = paraphrased_comment.replace(prompt, "").strip()

    return paraphrased_comment