Blenderbot 400M Distill

🚀 Chatbot Model for Open - Domain Conversation

This project focuses on building an open - domain chatbot. It demonstrates that with appropriate training data and generation strategies, large - scale models can learn essential conversation skills. The provided models and code are publicly available, and human evaluations show their superiority in multi - turn dialogues.

🚀 Quick Start

No specific quick - start steps are provided in the original document.

✨ Features

• Skill - based Conversation: The chatbot can blend multiple conversation skills such as providing engaging talking points, listening, asking and answering questions, and appropriately displaying knowledge, empathy, and personality.
• Scalable Models: Variants of the model are built with 90M, 2.7B, and 9.4B parameter neural models.
• Publicly Available: Both the models and the code are publicly accessible.
• Human - evaluated: Human evaluations show that the best models outperform existing approaches in multi - turn dialogues in terms of engagingness and humanness.

📚 Documentation

Model description

• Paper: Recipes for building an open - domain chatbot
• Original PARLAI Code

Abstract

Building open - domain chatbots is a challenging area for machine learning research. While prior work has shown that scaling neural models in the number of parameters and the size of the data they are trained on gives improved results, we show that other ingredients are important for a high - performing chatbot. Good conversation requires a number of skills that an expert conversationalist blends in a seamless way: providing engaging talking points and listening to their partners, both asking and answering questions, and displaying knowledge, empathy and personality appropriately, depending on the situation. We show that large scale models can learn these skills when given appropriate training data and choice of generation strategy. We build variants of these recipes with 90M, 2.7B and 9.4B parameter neural models, and make our models and code publicly available. Human evaluations show our best models are superior to existing approaches in multi - turn dialogue in terms of engagingness and humanness measurements. We then discuss the limitations of this work by analyzing failure cases of our models.

📄 License

The model is licensed under the apache - 2.0 license.