Published on

Text Similarity for NLP


Sam Witteveen and I were invited to give an NLP workshop by the AI Professionals Association (AIP) which is "the first grassroot-driven Association for engineers and professionals working in AI-related roles founded in Singapore".

The workshop was part of an in-person event (which surprised us, but it was certainly nice to be able to meet some of our course participants face-to-face, whereas we'd only previously interacted with them via Zoom/Meet/Email/Colab), and part of their first AIP Developer Conference.

The content of the workshop that we proposed included a tour through some of the latest techniques being used in research and industry for comparing and measuring text similarity for tasks such as Search, Conversational AI, Paraphrasing and Summarization. We also explored how a variety of Transformers are being used for these tasks and compare the advantages and disadvantages of using them.

In terms of audience, we presumed that participants had a decent understanding of NLP and Transformers before taking the workshop (i.e. this was not aimed at beginners).

The slides for my talk are here :

Presentation Screenshot

If there are any questions about the presentation please ask below, or contact me using the details given on the slides themselves.

Presentation Content Example

In order to prepare these materials (and associated notebooks), we made use of Google Cloud - and their team generously provided us with some cloud credits to do so. These were particularly important because the Transformer-based models we were demonstrating need beefy machines to train, and good hardware at inference time too.