Language Models in Plain English

 View Only
When:  Dec 14, 2021 from 02:00 PM to 03:00 PM (ET)

About


Please join the IBM Data Science Community for our December 2021 virtual meetup, co-hosted by Data, Cloud and AI in Miami, FL and Data, Cloud and AI in Tampa, FL. Feel free to also RSVP on Meetup.com and help spread the word. Thanks!

Want to get involved in your local IBM Community meetup? Join our Slack: https://join.slack.com/t/icmeetupsorg/shared_invite/zt-zkrhyfd2-jvru9Dkmc~KYNpaPtfIUxA

Agenda

Recent advances in machine learning have lowered the barriers to creating and using ML models. But understanding what these models are doing has only become more difficult. In their recent report (https://ibm.biz/BdfLSv), authors Austin Eovito and Marina Danilevsky from IBM focus on how to think about neural network-based language model architectures. They guide you through various models (neural networks, RNN/LSTM, encoder-decoder, attention/transformers) to convey a sense of their abilities without getting entangled in the complex details. The report uses simple examples of how humans approach language in specific applications to explore and compare how different neural network-based language models work.

Join Austin and Marina as they discuss their collaboration in authoring their report. This session will summarize key insights from the report and present important learnings for the data scientist audience. And find out what bats have to do with understanding language models!

About the presenters

Austin Eovito is a senior data scientist on the Client Engineering Team at IBM. Austin’s current role focuses on understanding and solving complex enterprise AI problems for Segment 1 clients. His role covers areas such as natural language, computer vision, and capacity management. He received his MS in computational science from the Florida State University in 2019, and his BS in applied eco‐ nomics from the Florida State University in 2018. He lives in Tampa, Florida with his wife, two dogs and two cats.

Marina Danilevsky is a research staff member at IBM Research. Her work centers on understanding structured and unstructured text data, with a particular focus on model explainability and human-in- the-loop techniques. She received her PhD in Computer Science from the University of Illinois at Urbana-Champaign in 2014 and her BS in Mathematics from the University of Chicago in 2007. She lives in San Jose, California, with her husband and two children.

Photo credit

Photo by cafuego / CC BY-SA 2.0

#MachineLearning