In this section, you take an extended look at a process to create a project in Deep Learning Studio (DLS). Here we will create one example project for MNIST dataset.
In this tutorial, we assume that you have basic understanding of Deep Learning. But it doesn't assume you're an expert in either.
To build a Deep Learning Project, you need to have Deep Learning Studio installed. If not, please follow the installation guide.
1) Start DLS and login and click on the "Projects" tab. c:
1) My Projects
2) Sample Projects
Note: Deep Learning Studio provides some example "Sample Projects". Click on Sample projects and copy any project by clicking on "Copy" icon.
2) Creating MNIST Handwritten Digit Classifier using recurrent neural network (RNN) project with Deep Learning Studio. And most importantly without writing code.
Go to the “Projects” tab from the left sidebar. Create a new project by clicking on the + icon.
Name of the project.
Choose project type, here you will get the two options “Custom Neural Network” and “AI App Module” as we are building a neural network. We will choose “Custom neural network” as the project type.
You can write a small description for project.
Press on right-click to save the project and open it.
3) Once opened the project, you will find these tabs:
1) Data Tab
You will see the "Data" tab first. Select MNIST dataset. DLS provides MNIST dataset by default in public datasets.
After selecting the dataset, choose the train/test/validation split. Click on the drop-down and choose the ratio split or you can also do a custom train/test/validation split. You will see the separation of data according to the total record.
Moving to the training tab, change the default run name if you want.
Save your weight in three ways: “End of epoch”,”Best Accuracy” and ”Lowest Loss”.
You can select your device “CPU” or “GPUs” and start your training.
It takes a few seconds to start the training process. While training is going on you will see the graph of Accuracy and Loss. After completion of every batch, you will see the change in Accuracy and Loss graph.