Talks # 3: Lorenzo Ampil - Introduction to T5 for Sentiment Span Extraction

This is Episode 3 of Talks Series and please note that it is one hour before the normal time for Talks :) Title: Introduction to T5 for Sentiment Span Extraction Abstract: T5 is a recently released encoder-decoder model that reaches SOTA results by solving NLP problems with a text-to-text approach. This where text is used as both an input and an output for solving all types of tasks. I believe that the combination of text-to-text as a universal format for NLP tasks paired with multi-task learning (single model learning multiple tasks) will have a huge impact on how NLP deep learning is applied in practice. In this presentation I aim to give a brief overview of #T5, explain some of its implications for NLP in industry, and demonstrate how it can be used for sentiment span extraction on tweets. Speaker Bio: Lorenzo Ampil is a Machine Learning Product Manager and Data Scientist at Thinking Machines, a global AI consulting firm w/ operations in Singapore and Manila. He specializes in developing products that utilize deep learning and machine learning on NLP for various industries. Prior to this, he set up his own consulting practice where he provided end-to-end data science solutions for finance and tech companies in Southeast Asia and Australia. He also previously worked at Uber as an analyst, where he handled projects related to NLP, analytics, and automation for the APAC region’s community operations. ------ If you want to be a speaker and talk about your #MachineLearning #DeepLearning Projects, then please fill out this form: https://bit.ly/AbhishekTalks Follow me on: Twitter: https://twitter.com/abhi1thakur LinkedIn: https://www.linkedin.com/in/abhi1thakur/ Kaggle: https://kaggle.com/abhishek