An implementation-focused walkthrough of CycleGAN, including generators, discriminators, adversarial loss, and cycle consistency.
A from-scratch implementation of DCGAN, covering generator and discriminator design, adversarial training, and sample visualization.
A dense-prediction walkthrough of FCN-style semantic segmentation with VGG backbones, skip connections, and upsampling.
A code-heavy walkthrough of FSRCNN, from low-resolution feature extraction to learned deconvolution for image upscaling.
A paper-to-code walkthrough of Grad-CAM, guided backpropagation, and guided Grad-CAM using pretrained AlexNet and VGG-16.
A character-level LSTM implementation that connects recurrent modeling theory with gated sequence generation in code.
A code-first walkthrough of neural style transfer using VGG feature losses, Gram matrices, and iterative image optimization.
A machine-translation walkthrough of encoder-decoder sequence modeling, attention, training, and evaluation.
A step-by-step implementation of Spatial Transformer Networks for distortion-invariant image classification.
A comparative reimplementation of VGG and ResNet, focused on depth, optimization, and residual learning.
