This tutorial introduces the fundamental concepts of PyTorch through self-contained examples.
Type of Material:
Online Course
Recommended Uses:
This will be useful for in-class, homework, individual, team, and self-paced learning. Can also be adapted for lecture purposes.
Technical Requirements:
Basic Web browser.
Identify Major Learning Goals:
As stated in the starting page, the goals of the tutorials are:
Understand PyTorch’s Tensor library and neural networks at a high level.
Train a small neural network to classify images
Target Student Population:
College General Ed, College Lower Division
Prerequisite Knowledge or Skills:
A basic understanding of Python and the programming environment is needed.
Content Quality
Rating:
Strengths:
- It breaks the contents into four modules starting with the operation tensor with NumPy, autograd, convolutional neural network and training a classifier.
- In addition to the basic of forward passing and backpropagating, it also illustrates the use of torch.autograd using some pre-trained models and the work and commonly used dataset like CIFAR10.
- The content is appropriate for the beginners in CNN and PyTorch attaining its aim.
Potential Effectiveness as a Teaching Tool
Rating:
Strengths:
- The learning goal is clearly presented in the opening video.
- Apart from the introductory video, it does not provide some funny videos, instead it uses examples to illustrate the features in PyTorch.
- While it does pay much effort on the theories of neural networks, so the readers can see the expected output, but the detail of the explanation is sometimes missing, while it provides some further references to help learners to understand the theories, such as the calculus of backpropagation.
In summary, the material is well organized into simpler modules. Each module can be read and understood in about an hour. The use of a concrete example throughout the entire series of modules is very effective in helping the learning process. The final module puts everything together into solving a specific problem.
Concerns:
- It is more focused on coding, the potential problem of installation and the deployment is not elaborated across different development platforms.
- A few modules are more complicated than they need to be (then they will be accessible to a wider audience).
Ease of Use for Both Students and Faculty
Rating:
Strengths:
- The source codes are available from the github, so users can download the code for instant trial. At the same, the code snippets can also be copied and pasted in the viewer's Jupyter Notebook to follow step by step.
- The site has clear and accurate instructions for use, it has a clear and consistent layout, and the code provided is easy to understand and to run.
Creative Commons:
Search by ISBN?
It looks like you have entered an ISBN number. Would you like to search using what you have
entered as an ISBN number?
Searching for Members?
You entered an email address. Would you like to search for members? Click Yes to continue. If no, materials will be displayed first. You can refine your search with the options on the left of the results page.
Searching for Members?
You entered an email address. Would you like to search for members? Click Yes to continue. If no, materials will be displayed first. You can refine your search with the options on the left of the results page.