Polite Guard is an open-source NLP language model developed by Intel, fine-tuned from BERT for text classification tasks. It is designed to classify text into four categories: polite, somewhat polite, neutral, and impolite. This model, along with its accompanying dataset and source code, is available on Hugging Face* and GitHub* to enable both communities to contribute to developing more sophisticated and context-aware AI systems. In this project, you will:
- Generate your own synthetic data, and
- Fine-tune a language model on your dataset using Intel technology.
- Key Contributions
- Repository Structure
- Articles
- Model and Dataset
- Getting Started
- Usage
- Join the Community
- License
Polite Guard provides a scalable model development pipeline and methodology, making it easier for developers to create and fine-tune their own models. Other contributions of the project include:
- Improved Robustness: Polite Guard enhances the resilience of systems by providing a defense mechanism against adversarial attacks. This ensures that the model can maintain its performance and reliability even when faced with potentially harmful inputs.
- Benchmarking and Evaluation: The project introduces the first politeness benchmark, allowing developers to evaluate and compare the performance of their models in terms of politeness classification. This helps in setting a standard for future developments in this area.
- Enhanced Customer Experience: By ensuring respectful and polite interactions on various platforms, Polite Guard can significantly boost customer satisfaction and loyalty. This is particularly beneficial for customer service applications where maintaining a positive tone is crucial.
This repository consists of two main packages:
- data-generator: A package for generating synthetic datasets.
- fine-tuner: A package for fine-tuning language models for text classification utilizing Al accelerators.
To learn more about the implementation of the data generator and fine-tuner packages, refer to
- Synthetic Data Generation with Language Models: A Practical Guide, and
- How to Fine-Tune Language Models: First Principles to Scalable Performance.
For more AI development how-to content, visit Intel® AI Development Resources.
The Polite Guard model and dataset are available on Hugging Face. You can find them at the following links:
Clone the repository:
git clone https://github.com/intel/polite-guard.gitCreate and activate a virtual environment. Then install the dependencies for each package:
# For data-generator
cd polite-guard/data-generator
pip install -r requirements.txt
# For fine-tuner
cd ../fine-tuner
pip install -r requirements.txtThe data-generator package is responsible for generating and synthetic datasets with specified labels and categories. Refer to the data-generator README for detailed instructions on how to run this synthetic data generator on the Intel® Tiber™ AI Cloud.
The fine-tuner package is used to fine-tune language models utilizing Al accelerators such as Intel® Gaudi®. Refer to the fine-tuner README for detailed instructions on how to use this package.
You can use Polite Guard directly with a pipeline for categorizing text into classes polite, somewhat polite, neutral, and impolite.
from transformers import pipeline
classifier = pipeline("text-classification", "Intel/polite-guard")
texts = [
"I sincerely apologize for the inconvenience you've experienced. Please allow me a moment to resolve this for you as quickly as possible.",
"I understand this isn't ideal, but could we move forward with this solution?",
"The product specifications are as follows.",
"You must be new here; you clearly don't know what you're doing.",
]
outputs = classifier(texts)
for text, output in zip(texts, outputs):
print(f'"{text}": {output["label"]}')Output
"I sincerely apologize for the inconvenience you've experienced. Please allow me a moment to resolve this for you as quickly as possible.": polite
"I understand this isn’t ideal, but could we move forward with this solution?": somewhat polite
"The product specifications are as follows.": neutral
"You must be new here; you clearly don’t know what you’re doing.": impolite
The (train, validation, and test) dataset(s) can be downloaded as follows.
from datasets import load_dataset
dataset = load_dataset("Intel/polite-guard", split="train")If you are interested in exploring other models, join us in the Intel and Hugging Face communities. These models simplify the development and adoption of Generative AI solutions, while fostering innovation among developers worldwide. Here are some ways you can contribute:
If you find this project valuable, please give it a star ★ on GitHub and share it with your network. Your support helps us grow the community and reach more contributors.
Help us improve and expand the project by contributing:
- Code: Fix bugs, optimize performance, or add new features.
- Documentation: Enhance the documentation to make it more accessible and user-friendly.
Check out the Contributing Guide to get started.
Run the software on your Intel hardware and share your experience. Report issues, suggest improvements, or request new features through the issues tab on GitHub.
Use this project as a foundation for your own work. Build new applications or integrate it with other tools and libraries. Let us know what you create--we'd love to feature your work!
Help us amplify our message by blogging, tweeting, or presenting about the project at conferences or meetups. Tag us and use our official hashtag so we can share your content with the community.
This project is licensed under the MIT License. See the LICENSE file for details.
*Other names and brands may be claimed as the property of others.