In recent years, deep learning techniques have contributed to the evolution of AI and in particular to that of natural language processing (NLP). So much so that it is quite possible to use it in the creation of software or applications. It is in this context that NLP Cloud, a company launched in January and based in Grenoble, launches an NLP programming interface allowing companies and their developers and / or data scientists to use NLP in their projects.
This natural language processing programming interface, based on the Hugging Face and spaCy open-source models, has been designed in such a way that it is ready and high-performance for any data scientist, programmer. or developer using it. 500 users were registered at the end of March. This NPA API allows you to do everything necessary in software development and application designs: from Entity Extraction (NER), Sentiment Analysis, Classification and Text Summarization, answering questions and morphosyntactic labeling (or POS tagging).
Users can then decide to use the ready-to-use pre-trained models or to integrate their own models. Currently, the software has a free version: it gives access to all pre-trained models, up to 3 requests per minute.
A paid version is also available for all those who need to generate more requests and have access to more advanced models or precisely, to integrate their own model. This second version makes it possible in particular to test the quality of its own models, with high levels of performance thanks to a dedicated GPU infrastructure. Ultimately, this NPA API should offer more languages, new features resulting from new integrated open-source models. The company anticipates revenue growth of 500% and plans to hire 5 people by the end of the year.
Julien Salinas, founder and CTO of NLP Cloud, explains: “As a developer, I have repeatedly encountered the difficulty of reliably and quickly putting NLP models into production. After seeing this pitfall on many projects, I decided to work on a powerful and easy-to-integrate NLP infrastructure, in order to save developers and data scientists weeks or even months of work.“