Neowin News Feed for: Bert

AI won't be taking up software engineering jobs any time soon, but it's getting there

Ather Fawaz — Sun, 23 May 2021 13:00:01 +0000

In a paper, researchers test the best language models (GPT-2/3/Neo) at solving programming questions from coding interviews. Results aren't particularly groundbreaking but show potential. Read more...

BLEURT is a metric for natural language generators boasting unprecedented accuracy

Ather Fawaz — Wed, 27 May 2020 07:50:01 +0000

Researchers at Google have developed BLUERT, an automatic metric that gauges the performance of natural language generation models and delivers SOTA performance on two academic benchmarks. Read more...

TextFooler tricks venerable NLP models like BERT into making wrong predictions

Ather Fawaz — Sat, 08 Feb 2020 07:08:01 +0000

Researchers at MIT have created a framework—TextFooler—that brought down the prediction accuracy of certain NLP models from 90% down to under 20% by simply using synonyms in place of certain words. Read more...

Nvidia Jarvis—a multi-modal AI SDK—fuses speech, vision, and other sensors into one system

Ather Fawaz — Tue, 22 Oct 2019 07:58:01 +0000

By harmonizing modules pertaining to different media, Nvidia Jarvis will help build context for accurately predicting and generating responses in conversation-based AI applications. Read more...