PyTorch Fundamentals: A 7-Day Learning Journey

A hands-on PyTorch learning repository covering tensors, autograd, neural networks, data loading, and binary classification.

Overview

This repository documents my structured journey of learning PyTorch from the fundamentals to building a complete binary classification model.

Over seven days, I explored:

• Tensor Operations
• Automatic Differentiation (Autograd)
• Manual Gradient Descent
• Neural Networks using nn.Module
• Data Loading with DataLoader
• Binary Classification and Evaluation

The goal was to understand the complete deep learning workflow rather than simply using pre-built models.

Learning Roadmap

Day 1 → Tensors
Day 2 → Autograd
Day 3 → Manual Regression
Day 4 → nn.Module
Day 5 → DataLoader
Day 6 → Binary Classification
Day 7 → Documentation

Daily Progress

Day	Topic	Notebook
1	Tensor Operations	day1_tensors.ipynb
2	Autograd	day2_autograd.ipynb
3	Manual Linear Regression	day3_manual_regression.ipynb
4	Neural Networks with nn.Module	day4_nn_module.ipynb
5	Dataset & DataLoader	day5_dataloader.ipynb
6	Binary Classification	day6_binary_classifier.ipynb
7	Documentation	README.md

Technologies Used

• Python
• PyTorch
• NumPy
• Matplotlib
• Jupyter Notebook

Repository Structure

Pytorch/
│
├── day1_tensors.ipynb
├── day2_autograd.ipynb
├── day3_manual_regression.ipynb
├── day4_nn_module.ipynb
├── day5_dataloader.ipynb
├── day6_binary_classifier.ipynb
│
├── requirements.txt
├── results-day3_fit.png
├── day6_boundary.png
└── README.md

Standard PyTorch Training Workflow

for epoch in range(epochs):
    model.train()

    for X_batch, y_batch in train_loader:
        optimizer.zero_grad()

        predictions = model(X_batch)

        loss = criterion(predictions, y_batch)

        loss.backward()

        optimizer.step()

This workflow forms the foundation of most modern deep learning projects in PyTorch.

Results

Linear Regression (Day 3)

Binary Classification Decision Boundary (Day 6)

Key Outcomes

• Built a regression model from scratch
• Implemented gradient descent manually
• Developed a neural network classifier
• Achieved approximately 98% classification accuracy
• Visualized model predictions and decision boundaries

Skills Demonstrated

• Tensor Manipulation
• Automatic Differentiation
• Gradient Descent Optimization
• Linear Regression
• Neural Network Development
• Data Loading Pipelines
• Binary Classification
• Model Training
• Model Evaluation
• Deep Learning Fundamentals

Getting Started

Clone Repository

git clone https://github.com/AlqamahAnsari/Pytorch.git
cd Pytorch

Install Dependencies

pip install -r requirements.txt

Launch Jupyter Notebook

jupyter notebook

Open any notebook and run the cells sequentially.

Future Learning Goals

• Convolutional Neural Networks (CNNs)
• Recurrent Neural Networks (RNNs)
• Transfer Learning
• Transformers
• Natural Language Processing (NLP)
• Large Language Models (LLMs)
• Model Deployment

Author

Mohammad Alquamah Ansari

• GitHub: https://github.com/AlqamahAnsari
• Portfolio: https://alqamahansari.github.io/

Support

If you found this repository useful, consider giving it a star.
It helps motivate future open-source AI and Machine Learning projects.