diff --git a/tutorialposts/2021-10-08-dcgan-mnist.md b/tutorialposts/2021-10-08-dcgan-mnist.md
index 4d33911..7da930b 100644
--- a/tutorialposts/2021-10-08-dcgan-mnist.md
+++ b/tutorialposts/2021-10-08-dcgan-mnist.md
@@ -11,7 +11,7 @@ This is a beginner level tutorial for generating images of handwritten digits us
A GAN is composed of two sub-models - the **generator** and the **discriminator** acting against one another. The generator can be considered as an artist who draws (generates) new images that look real, whereas the discriminator is a critic who learns to tell real images apart from fakes.
-
+
The GAN starts with a generator and discriminator which have very little or no idea about the underlying data. During training, the generator progressively becomes better at creating images that look real, while the discriminator becomes better at telling them apart. The process reaches equilibrium when the discriminator can no longer distinguish real images from fakes.
@@ -24,9 +24,9 @@ This tutorial demonstrates the process of training a DC-GAN on the [MNIST datase
~~~
-
-
-
+
+
+
+\
After installing the libraries, load the required packages and functions:
```julia
using Base.Iterators: partition
@@ -59,7 +59,7 @@ using Flux.Losses: logitbinarycrossentropy
using MLDatasets: MNIST
using CUDA
```
-
+\
Now we set default values for the learning rates, batch size, epochs, the usage of a GPU (if available) and other hyperparameters for our model.
```julia
@@ -116,7 +116,6 @@ We will also apply the weight initialization method mentioned in the original DC
# sampled from a Gaussian distribution with μ=0 and σ=0.02
dcgan_init(shape...) = randn(Float32, shape) * 0.02f0
```
-
```julia
function Generator(latent_dim)
@@ -137,7 +136,7 @@ function Generator(latent_dim)
)
end
```
-
+\
Time for a small test!! We create a dummy generator and feed a random vector as a seed to the generator. If our generator is initialized correctly it will return an array of size (28, 28, 1, `batch_size`). The `@assert` macro in Julia will raise an exception for the wrong output size.
```julia
@@ -150,10 +149,7 @@ gen_image = generator(noise)
@assert size(gen_image) == (28, 28, 1, 3)
```
-
Our generator model is yet to learn the correct weights, so it does not produce a recognizable image for now. To train our poor generator we need its equal rival, the *discriminator*.
-
-
### Discriminator
@@ -187,7 +183,7 @@ discriminator = Discriminator()
logits = discriminator(gen_image)
@assert size(logits) == (1, 3)
```
-
+
Just like our dummy generator, the untrained discriminator has no idea about what is a real or fake image. It needs to be trained alongside the generator to output positive values for real images, and negative values for fake images.
## Loss functions for GAN
@@ -195,7 +191,6 @@ Just like our dummy generator, the untrained discriminator has no idea about wha
In a GAN problem, there are only two labels involved: fake and real. So Binary CrossEntropy is an easy choice for a preliminary loss function.
But even if Flux's `binarycrossentropy` does the job for us, due to numerical stability it is always preferred to compute cross-entropy using logits. Flux provides [logitbinarycrossentropy](https://fluxml.ai/Flux.jl/stable/models/losses/#Flux.Losses.logitbinarycrossentropy) specifically for this purpose. Mathematically it is equivalent to `binarycrossentropy(σ(ŷ), y, kwargs...).`
-
### Discriminator Loss
@@ -213,7 +208,6 @@ function discriminator_loss(real_output, fake_output)
return real_loss + fake_loss
end
```
-
### Generator Loss
The generator's loss quantifies how well it was able to trick the discriminator. Intuitively, if the generator is performing well, the discriminator will classify the fake images as real (or 1).
@@ -221,7 +215,7 @@ The generator's loss quantifies how well it was able to trick the discriminator.
```julia
generator_loss(fake_output) = logitbinarycrossentropy(fake_output, 1)
```
-
+\
We also need optimizers for our network. Why you may ask? Read more [here](https://towardsdatascience.com/overview-of-various-optimizers-in-neural-networks-17c1be2df6d5). For both the generator and discriminator, we will use the [ADAM optimizer](https://fluxml.ai/Flux.jl/stable/training/optimisers/#Flux.Optimise.ADAM).
## Utility functions
@@ -254,7 +248,7 @@ function train_discriminator!(gen, disc, real_img, fake_img, opt, ps, hparams)
return disc_loss
end
```
-
+\
We define a similar function for the generator.
```julia
@@ -268,8 +262,7 @@ function train_generator!(gen, disc, fake_img, opt, ps, hparams)
return gen_loss
end
```
-
-
+\
Now that we have defined every function we need, we integrate everything into a single `train` function where we first set up all the models and optimizers and then train the GAN for a specified number of epochs.
```julia
@@ -337,7 +330,7 @@ function train(hparams)
return nothing
end
```
-
+
Now we finally get to train the GAN:
```julia
@@ -359,10 +352,11 @@ images = load.(img_paths)
gif_mat = cat(images..., dims=3)
save("./output.gif", gif_mat)
```
-
-
-
-
+~~~
+
+
+