You may have been wondering what the team at CODAIT has been up to. Perhaps you have been wondering, “Where are my new models on the IBM Model Asset eXchange?” Well, the team has been pretty busy with our newly announced Data Asset eXchange, a trusted source for curated open datasets that will integrate with IBM Cloud and AI services. In fact, we’re still hard at work behind the scenes on Data Asset eXchange, so stay tuned!
, 2, 3 … NLP!
But don’t worry, we have put together a few model morsels for the MAX community. Recently, deep learning for natural language processing (NLP) has emerged as a rapidly advancing area of machine learning research. MAX already has a few models in key areas of NLP (e.g., Text Sentiment Classifier, Named Entity Tagger, Review Text Generator, and Word Embedding Generator).
Today, we’re pleased to announce a new batch of models for natural language processing tasks:
- Toxic Comment Classifier – This model detects whether a piece of text (typically a user comment on the Internet) contains various types of toxic content. Like our sentiment classifier, it is based on the state-of-the-art BERT architecture. Possible use case: Automated moderation of user comments on website articles or posts.
- Text Summarizer – Using this model, a summary can be generated for a given piece of text. Possible use case: Generating automated summaries or headlines for news articles.
- Chinese Phonetic Similarity Estimator – This model is able to estimate the phonetic distance between Chinese words and get similar sounding candidate words. Possible use case: Spellcheck functionality on social media.
The (code)pen is mightier than the sword
To make it even simpler to experiment with what MAX models are capable of, we’ve created a collection of pens on CodePen that illustrate how to send image, audio, or video data to the model-serving microservices and how to visualize the prediction results using our new MAX visualization NPM package.
Visit the Model Asset eXchange and check out the latest models and enhancements. As always, we welcome your comments and suggestions that help us improve and better serve the ML/DL community.
And watch this space for exciting developments to come soon with the Model Asset eXchange, the Data Asset eXchange, and the powerful combination of the two!