A bigram language model. This implementation shows how to create a simple character-level language model that can generate new words similar to examples it has seen.

An implementation of a small autograd engine, inspired by Andrej Karpathy's micrograd. This notebook demonstrates building neural networks from scratch with automatic differentiation.

Implementation of the Neural Probabilistic Language Model as introduced by Bengio et al. This notebook explores the foundations of neural language modeling.

Experiments with Batch Normalization in neural networks. This notebook explores how batch normalization affects training dynamics, convergence speed, and overall performance by comparing models with and without this technique.

Visualization and analysis of optimization algorithms including SGD, Momentum, and Adam.

A series of notebooks exploring convolutional neural networks:

• convolution_fundamentals: Understanding the mathematics and intuition behind convolution operations.
• mnist_linear: Implementation of linear and MLP models on the MNIST dataset.
• mnist_cnn: Implementation of CNNs on MNIST, demonstrating how convolutions improve performance.
• fashion_mnist: Implementation of CNN architectures on the Fashion-MNIST dataset, demonstrating performance on a more complex image classification task with clothing items.
• cat_and_dog: Comparative implementation of cat vs. dog image classification using three approaches: PyTorch, Keras, and an AlexNet architecture in PyTorch, showcasing framework differences and architecture choices.
• cifar_10: Comparison of advanced CNN architectures (AlexNet, ResNet, DenseNet) on the CIFAR-10 dataset, showcasing the evolution and improvements in convolutional network design
• german_traffic_sign

Exploration of object detection techniques. This directory contains implementations and experiments, demonstrating real-time object detection capabilities and applications.

• car-detection: Implementation of YOLO for detecting cars in images and video.

Natural language processing

• text_classification: Implementation of Text classification tasks with multiple ML algorithms
• token_classification: Implementation of Token classification (Named entity recognition)