Active Learning with AutoNLP and Prodigy

Active learning in the context of Machine Learning is a process in which you iteratively add labeled data, retrain a model and serve it to the end user. It is an endless process and requires human interaction for labeling/creating the data. In this article, we will discuss how to use AutoNLP and Prodigy to build an active learning pipeline.

AutoNLP

AutoNLP is a framework created by Hugging Face that helps you to build your own state-of-the-art deep learning models on your own dataset with almost no coding at all. AutoNLP is built on the giant shoulders of Hugging Face's transformers, datasets, inference-api and many other tools.

With AutoNLP, you can train SOTA transformer models on your own custom dataset, fine-tune them (automatically) and serve them to the end-user. All models trained with AutoNLP are state-of-the-art and production-ready.

At the time of writing this article, AutoNLP supports tasks like binary classification, regression, multi class classification, token classification (such as named entity recognition or part of speech), question answering, summarization and more. You can find a list of all the supported tasks here. AutoNLP supports languages like English, French, German, Spanish, Hindi, Dutch, Swedish and many more. There is also support for custom models with custom tokenizers (in case your language is not supported by AutoNLP).

Prodigy

Prodigy is an annotation tool developed by Explosion (the makers of spaCy). It is a web-based tool that allows you to annotate your data in real time. Prodigy supports NLP tasks such as named entity recognition (NER) and text classification, but it's not limited to NLP! It supports Computer Vision tasks and even creating your own tasks! You can try the Prodigy demo: here.

Note that Prodigy is a commercial tool. You can find out more about it here.

We chose Prodigy as it is one of the most popular tools for labeling data and is infinitely customizable. It is also very easy to setup and use.