Skip to content
A collection of AWS Sagemaker notebooks for training different machine learning models with TuriCreate for delivery with Skafos.ai.
Jupyter Notebook
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
ImageClassification
TextClassification
assets
.gitignore
LICENSE
README.md

README.md

Skafos Example Models on AWS Sagemaker

A collection of AWS Sagemaker notebooks for training machine learning models with TuriCreate. At the end of each example notebook, you will be able to upload a trained CoreML model to Skafos.ai where you can deliver over-the-air to your iOS applications. In this repo, we provide two machine learning tasks to get you started.

Syncing with Sagemaker

Sagemaker Logo

  1. Open the Amazon Sagemaker console: https://console.aws.amazon.com/sagemaker
  2. Choose Notebook Instances on the left panel, and then click Create Notebook Instance
  3. Fill in all required settings on the Create Notebook Instance page. Choose an instance type.
    • If you plan on using a GPU for compute, select one of the ml.p2.xxxxxx instances.
  4. Open the Git repositories panel. In the Default repository section, click the dropdown and select "Clone a public Git repository to this notebook instance only".
  5. Copy and paste this repo URL into the field that pops up.
  6. Hit Create Notebook Instance at the bottom and wait for the instance to show InService.

Amazon will launch an ML compute instance running the Jupyter Notebook App with a set of pre-installed Anaconda libraries. This repository will be your working directory when you open your new instance.

  1. Open up the Jupyter Notebook (or JupyterLab if you prefer that UI over the traditional one).
  2. Navigate to one of the example models that interests you and select a kernel from the drop down.
    • If you plan on using a GPU for compute, select the conda_amazonei_mxnet_p36 kernel

Lastly, we recommend creating different instances for different models you plan on working with. You don't want to mix up training environments/dependencies between models.

Example Models

Each one of these example models links with an example app on the Skafos platform. You can easily deliver these models to iOS apps over-the-air by creating an account and heading over to https://dashboard.skafos.ai.

Text Classification

Classify chunks of user text into different categories.

Image Classification

Classify images into different categories.

Object Detection

Coming Soon

Need Help?

Didn't find something you need? Confused by something? Need more guidance?

Please contact us with questions or feedback! Here are two ways:

We will continue to update this repository with updated/new example models to get you started. Stay tuned.

You can’t perform that action at this time.