Building Your First PyTorch Solution

Description

In this course, you will :

• Learn how to install PyTorch with pip and conda, as well as how to take advantage of GPU support.
• Learn how to hand-craft a linear regression model with a single neuron by defining your own loss function.
• how PyTorch optimizers can be used to make this process much smoother
• learn how to add different activation functions and dropout to PyTorch neural networks
• Investigate how to create classification models in PyTorch.

Syllabus :

1. Installing PyTorch on a Local Machine

• Prerequisites and Course Outline
• CUDA Support in PyTorch
• Exploring PyTorch Install Options on a Local Machine
• Setting up a Virtual Machine
• Installing PyTorch with CPU Support Using Conda
• Installing PyTorch with CPU Support Using Pip
• Adding GPU Support to the VM and Installing the CUDA Toolkit
• Installing PyTorch with GPU Support Using Conda
• Installing PyTorch with CUDA Support Using Pip

2. Understanding Linear Regression with a Single Neuron

• Linear Regression
• Finding the Best Fit Line
• Gradient Descent
• Training a Simple Neural Network with One Neuron
• Visualizing Regression Results and Compare with Regression Using scikit-learn
• Preventing Overfitting Using Regularization
• Performing Ridge Regression Using a Neural Network with One Neuron

3. Building a Regression Model Using PyTorch

• Training a Neural Network Forward and Backward Passes
• Optimizers
• Building a Neural Network Using PyTorch Layers
• Training a Neural Network Using Optimizers
• Dropout
• Epochs and Batches
• Exploring the Bike Sharing Dataset
• Using Datasets and Data Loaders in PyTorch
• Building and Train a Neural Network for Bike Sharing Demand Prediction
• Working with Different Neural Network Architectures

4. Building a Classification Model Using PyTorch

• Softmax and Cross Entropy
• Softmax and LogSoftmax
• Evaluating Classifiers
• Exploring the Graduate Admissions Dataset
• Preprocessing the Data
• Building a Custom Neural Network
• Training and Evaluating the Neural Network
• Customizing and Evaluating Different Models

Show more