The ability of machine to interpret human language and perform decision making
What an exciting time to be alive
These are showcases of cutting edge deep learning based NLP applications
In each post you will find interpretations of their original papers
Code implementations and tutorials to walk you through how you can leverage these techniques in you own projects
Transformers in NLP
Here are some very interpreble articles and code implementations of best performing NLP models such as
BERT, RoBERTa , ALBERT, etc.
From basic theory on attention mechanism to transfer learning and fine-tuning, everything you need to nail your NLP project with transformers.
ULMFiT sentiment analysis
Universal Language Model FIne-Tuning
This is the first successful attempt of applying transfer learning in NLP tasks and it sets the ImageNet moment for NLP
NLP for non-English languages
As an NLP practitioner, it is important to know how to deal with
languages that are structurally and grammatically different from English
Named Entity Recognition
A technical guide in learning from both word-level and character-level
leveraging multi-task learning and the concept of conditional random field
We are not done yet
What's so cool about NLP is that it posseses the capability to process large corpus of text with unparalleled speed
We would be able to generate realistic and stylistic articles, draw insights from huge amount of data and perform dicision and even chat bot
For the state-of-the-art advances, deep learning based NLP still can't claim to have super-human performance(unlike in Computer Vision)
Yet, with the success of transfer learning in this field and rapidly iteration of architecture updates (such as BERT and XLNet) makes me very optimistic about future of NLP