Deep learning, graph algorithms, and NLP are all hot topics among data and research scientists. To facilitate research on the application of deep learning on graphs to NLP, contributors from various institutions have released Graph4NLP, a user-friendly library that implements state-of-the-art graph learning models (e.g. Graph2Seq, Graph2Tree, and Graph2Graph) and includes a collection of NLP applications (e.g. text classification, named entity recognition, summarization). Graph4NLP, which leverages runtime libraries including DGL and from Huggingface, also includes interfaces to enable users to build customized models.