Vokenization! You Read That Right. Improving Language Understanding with Contextualized, Visual-Grounded Supervision. #Spock

Very interesting project. Researchers have been able to use tokenization on images (""vokenization"") and apply unsupervised transformer techniques on a large corpus of unlabeled images. This approach greatly improved natural language understanding for the NLP models.
Click the image to read the article

Find more #DSotD posts