Masked Language Modelling with Hugging Face - Microsoft Sentence Completion - Coding Tutorial

A practical Python Coding Guide - In this guide I use a hugging face language model on the Microsoft research sentence completion challenge! This is a two part video, in this part I explain what masked language modelling is and then show you how to apply a pre-trained language model to the Microsoft research sentence completion challenge and evaluate the models performance.

---- TUTORIAL NOTEBOOK

Intro and outline: 00:00:00
What is masked language modelling? 00:04:00
Colab notebook / Inspect data: 00:06:01
Applying the language model to MRSCC: 00:11:02
Evaluating the model: 00:43:08

This series attempts to offer a casual guide to Hugging Face and Transformer models focused on implementation rather than theory. Let me know if you enjoy them! Will be doing future videos on computer vision if that is something people are interested in, let me know in the comments :)

----- Research material for theory

[towhidultonmoy]

Good work! Very helpful! Hope to see more!

[AG_CFC]

Very helpful and well explained! Thank you!

[blacksirius]

I have a question, I trying to make model that will understand the mathematical reasoning given in the text. I am taking MLM is my starting point because in maths text both sequence and sentiment matters. So can you make some videos on NPL for maths

[pramodsingh]

Any idea?
why am i getting this error :"Must have equal len keys and value when setting with an ndarray" . While evaluating the model

[juancruzalric]

Great video! Is there a way to use transformers in computer vision problems? For example maybe the attention mechanism can be really good to get the most important features in an image. Like for example as how humans scan quickly an image and detect the most important features in it. 🤔

[venkateshr]

Please create a video on table to text generation

[MohsinAli-ydjs]

I was trying to implement this tutorial but I have an error in my code, I couldnt sort it out how to make it correct. so if you can help. the error is (KeyError: 'candidate_questions')