Comprehensive development platform for artificial intelligence
The importance of the cloud for development and the operation of services using artificial intelligence (AI) is growing continuously. Therefore, many cloud operators are making the switch to providing entire customized development environments as AI PaaS. These AI suites comprise everything an AI developer needs to generate their own models. With ModelArts, the Open Telekom Cloud offers a modern, easy-to-use and end-to-end development platform for AI that allows models to be trained and provided.
During the public beta phase, the (limited) use of ModelArts is free, with costs only being incurred for the used storage.
Good reasons to use ModelArts in the Open Telekom Cloud
Security made in Germany:
aproven European platform for AI
The Open Telekom Cloud offers companies that want to use AI an advantage: all data remains in Europe. AI models can thus be developed and trained in a protected environment – even if sensitive data from internal systems is required for the training. The Open Telekom Cloud either does not use the training data to improve internal algorithms. Thus it is a first-class development platform for companies that rely on AI and value the fact that the data used remains their property. And it offers full flexibility: After development, the final model can easily be operated on any other cloud.
Features and functions of ModelArts
ModelArts accompanies the entire AI development process, from data preparation up to the deployment of the finished service, and provides comprehensive life cycle management.
ModelArts – the one-stop development platform for AI developers
ModelArts provides support in data collection, data annotation and version management.
ModelArts contains preconfigured, ready-to-use algorithms and myriad popular open-source frameworks, such as TensorFlow, PyTorch, MXNet, Caffe, Spark ML, Scikit, XGBoost, as well as the MoXing Software Development Kit (SDK).
ModelArts supports various training methods, such as machine learning, deep learning and auto learning (via ExeML).
The finished AI services can be used online (via APIs), as batch jobs, or on end devices.
A marketplace where developers can offer their models, apps, APIs and data sets is planned.
Life cycle management
The individual process steps for AI development are integrated in an end-to-end life cycle management system which ensures that the development process is transparent and clear. This includes ExeML UI Wizard, Jupyter Notebook, management of training jobs and models, as well as the visualization of work flows.
ModelArts is the right development platform for every type of AI developer.
Do you have no experience in developing AI services? – No problem!
ModelArts is the right development platform for every type of AI developer. Whatever the path to your own model looks like – one-click training automatically provides you with the suitable computing resources including containerization. It has never been simpler to develop an AI service.
Do you have no experience in developing AI services? No problem: In ready-to-use mode, ModelArts provides pre-trained models which you can easily apply to your data – all without any programming yourself.
Do you just want to build a model quickly? – You can access preconfigured algorithms.
WIf you just want to build a model quickly, you can access preconfigured algorithms from ModelArts, such as ResNet50, Faster-RCNN, VGG-16 or SegNet-VGG. Trained codes and preconfigured hyperparameters are available via MoXing APIs. You do not have to develop your code from scratch every time, but simply refine it with your own data.
Do you want to create an AI application directly? – Jupyter Notebook is ready for you.
If you are an experienced programmer, you can deploy Jupyter Notebook as a feature of ModelArts and work within your preferred framework and libraries. Caffe, MXNet, PyTorch, Spark MLlib, TensorFlow and XGBoost-Sklearn are directly available via ModelArts. Jupyter Notebook also provides access to Conda Python and PySpark alongside several of the aforementioned solutions. You can also access your own images (e.g., on GitHub). You can upload the (container) images for this into the Software Repository for Container (SWR) in the Open Telekom Cloud and integrate them into ModelArts.
Learn more about ModelArts
Q4/2021 - General release of ModelArts (leave beta)
Why wait? Develop your AI services right now for free* !
How it works.
* During the beta phase, a limited contingent of computing resources is available to users free of charge. Minor costs are only incurred when storage space is used, such as with object storage or when working with Jupyter Notebook.
Do you have any questions or feedback about ModelArts?
We will answer any questions you have regarding testing, booking and usage – free and tailored to your needs. Feel free to give us your feedback as well! What can we do better? What works well? You can contact us via Community-Portal or via the contact form.
T-Systems International GmbH