DeepPavlov

DeepPavlov is an open source conversational AI framework for building chatbots, virtual assistants and complex dialogue systems, and for research in natural language processing. The core DeepPavlov Library provides pretrained deep learning models, including BERT based models, for tasks such as text classification, named entity recognition and question answering, and lets developers assemble them into production ready conversational skills. The project serves both beginners and experienced NLP engineers. Beginners can start with tutorials that run without installation, while experts follow guides to install, configure and extend the framework. Components run from Python code, a command line interface, an API or Docker, and prepackaged containers are hosted on Nvidia NGC and Docker Hub. The wider stack includes DeepPavlov Agent for multi skill dialogue management and integration of skills as API services, the DREAM socialbot, the DFF dialogue flow framework, and the DeepPavlov Topics dataset. The team also runs a community forum, a blog, internships, published research and educational courses in deep learning for NLP, reinforcement learning and theoretical deep learning. DeepPavlov is free, open source software. The site publishes no pricing, paid plans or hosted service. The DeepPavlov project has been developed since June 2017 and grew out of an academic research team that has organised conversational AI competitions, including the ConvAI challenges, and taken part in the Alexa Prize. Much of the site reflects that research heritage, so buyers expecting a commercial product with support contracts should look elsewhere.