Training CycleGAN for season translation using tensorflow 2

After covering basic , taking a step further, we will explore an advanced GAN version i.e CycleGAN having some fascinating application in the field of image translation

What the heck is image translation?

It's basically tweaking the image such that the domain/style of the input image changes to another keeping core content intact. Like: Season changes for a landscape image, keeping other features in the image intact. …

GAN implementation on Fashion-MNIST using Tensorflow 2

Photo by on

As of now, we are done with

It's high time we explore GANs I guess !!

GANs or Generative Adversarial Network is a combination of two networks (assume them to be A & B for now) that helps us get to a system that produces, when fed with some random input (& I mean it), similar images as in the training data. Before deep-diving how this system of 2 networks works, we must understand what does an Adversarial system mean:


an important pre-requisite for GANs

Photo by on

After seeing the above image (apart from Inception & Leo Dicaprio), the first thing that comes to mind is Balance or Equilibrium !!

You must have heard of the Hollywood movie ‘A Beautiful Mind’ (2001) about the mathematician John Nash who actually came up with the Nash Equilibrium. We will be talking about this Nash Equilibrium. When you have a movie around a mathematical concept, you know it’s important !!

Setting tone for GANs part-3

So, we are already done with the & how & the reasons behind its failure on complex data.

So what to do to for generating complex data, especially images?

Recollecting some important points from the past 2 posts:

Generative models should have an element of randomness. Hence, data generated shouldn’t be copies of training data nor from some 3rd world.

Naive Bayes failed for complex data as 1) Neighbourhood pixels have no mechanism to have an association with other neighbors. So every pixel is independently leading to a complete…

setting tone for GANs

After covering generative modeling basics in the , this time I will be exploring how, our very own, Naive Bayes acts as a generative model. We all must have used it for some classification task. But data generation? maybe a big no.

Do read out about Naive Bayes if you are new to it.

But to a surprise, Naive Bayes has been amongst the earliest models used for data generation, though, not for data as complex as images. …

setting tone for GANS

GANs have taken the world by storm. The incoming of models like (generating images given any fantasy text), the concept of , or apps like , etc literally forced me to jump into generative modeling & at least have a taste of it. So, this time I may be penning down my longest series trying to unveil each & every possibility with GANs starting today. Though, taking a step back, I will start off with the basics of generative modeling

To begin with, On the basis of the learning approach on how a model learns, we have 2 types…

basic concepts & terminologies

Drifting from the world of models & algorithms for sometime, I will be deep diving on exploring Docker this time in a series of blogs.

But before moving onto Docker, let’s assume a real world scenario:

You are a developer who just coded out a new application/service on your local environment. The app looks great & works fine on your local environment & the next phase is to take this code to Staging environment & the finally to Production.

Once you move your code to Staging,

The code breaks down.

You must have heard this line a million times from devs:

‘But this…

DistilBERT & TinyBERT using knowledge distillation

The past 3 blogs have been a rollercoaster where we first dived in then to & then a few prominent . Continuing the trend, we will be discussing the concept of Knowledge Distillation & how this lead to the inception of some more BERT variants.

Note: You may wish to refer to the past blogs in this series for a better understanding

As discussed last time, there exists a few evident issues with BERT majorly it being very bulky & difficult to train & deploy due to numerous parameters(110 Million to be precise). …

Helping people to analyze COVID-19 situation in their city

2020 was the year when India first witnessed the community outbreak of the SARS-COV-2 virus AKA COVID-19. Though, around late 2020, the dropping number of fresh COVID-19 cases made us all believe we have finally won the battle & things will go back to normal by 2021 & our guards went down.

Little did we know the battle has just begun!

2021 has been anticlimactic given where we stand at the moment. The virus is not only deadly but also spreading at ~4x speed when compared with 2020. Hospitals are full, medicines are in scarce supply, people panicking & what…


It was one hell of a ride discussing the past 2 blogs on T & . Since the inception of BERT, a number of variants came up trying to cover a few issues found with the original BERT.

  • Consumes a lot of time & resources to train
  • Humongous in size due to 110 million parameters
  • High inference time
  • BERT is considered to be ‘less efficiently’ trained by some researchers. They think some better training techniques could have been used.

Note: If any terminology looks ghostly, to refer to &

People soon came out with variants as good as…

Mehul Gupta

Data Scientist@1mg| IIIT’ian| LinkedIn:

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store