Highlights from Machine Translation and Multilinguality in August 2022
Highlights from Machine Translation and Multilinguality in August 2022 | Jindřich’s blog
Read moreDeep Learning, NLP, NMT, AI, ML
Highlights from Machine Translation and Multilinguality in August 2022 | Jindřich’s blog
Read moreHere are my monthly highlights from paper machine translation and multilinguality. A preprint from the Nara Institute of Science and Technology shows that target-language-specific fully connected layers in the Transformer decoder improve multilingual and zero-shot MT compared to the current practice of using a special token to indicate what the target language is. A very similar idea is also in a preprint from Tianjin University, but in this case, they add language-specific parameters for the other part of the Transformer […]
Read moreFastText is a lightweight, open-source framework that enables users to learn text representations and classifiers. It runs on common, generic hardware. This model may be used to create an unsupervised or supervised learning technique for obtaining word vector representations. This module has acquired a lot of interest in the NLP community and might be a viable alternative to the genism package, which includes Word Vectors and other features. FastText differs in that word vectors, also known as word2vec, take […]
Read moreWe all know that the machine is able to understand our language. But have we ever tried to understand how is that possible and which codes or software or programs are made to run through it to make it happen? This blog, therefore, is devoted to Natural Language Processing (NLP) which is behind this hyper-intelligent technology.
Read morePhoto by Roberto
Read moreMuch of the Python code you’ll write is unconditional. That is, the code does not make any choices. Every line of code is executed in the order that it’s written or in the order that functions are called, with possible repetitions inside loops. In this course, you’ll learn how to use conditional logic to write programs that perform different actions based on different conditions. Paired with functions and loops, conditional logic allows you to write complex programs that can handle […]
Read moreMany programmers, regardless of their specialty, deal with text on a daily basis. For example, web developers work with text input from web forms. Data scientists process text to extract data and perform tasks like sentiment analysis, which can help identify and classify opinions in a body of text. Collections of text in Python are called strings. Special functions called string methods are used to manipulate strings. There are string methods for changing a string from lowercase to uppercase, removing […]
Read moreLast week I collaborated with a team of two other fellows from the University of North Florida in an AI for Good Hackathon organized by JEA and sponsored by Microsoft and FIS to solve one of the two presented problems for the hackathon — on Vehicle Route Optimization.
Read more