During the AWS summit 2019 held in Milan, Julien Simon, Global Evangelist, ML/AI, AWS, introduced to the audience the possibility of creating DeepLearning models with TensorFlow and Apache MXNet using the SageMaker platform.
In this video, you can briefly understand what are TensorFlow, MXNet and SageMaker.
What is TensorFlow?
- Open source software library for ML
- Main API in Python, experimental support for other languages.
- Built-in support for many networks architectures (FC, CNN, LSTM, etc).
- Support for symbolic execution, as well as imperative execution (normal code)
- Complemented by the Keras high-level API
TensorFlow permits to Build network architecture.
85% of all TensorFlow workloads in the cloud runs on AWS
GPU instances —> If you try to scale efficiency with 256 gpu’s on the stock TensorFlow platform, your maximum scale up is 65% (training time= 30 min)
On AWS optimized TensorFlow platform the scaling efficiency reaches 90% (training time= 14 min)
Advanced features that everyday developers can get by using Amazon SageMaker
- Local mode: train on the notebook
- Script mode: use the same tensor flow code as on your local machine
- Distributed training: zero setup!
- Pipe mode: stream large datasets directly from Amazon S3
- TensorBoard: visualize the progress of your training jobs
- Keras (high level API) support.
We can use the same code used in TensorFlow, to run it on stage maker
What is Apache MXnet ?
- Open source software library for Deep learning
- Natively Implemented in C++
- Built-in support for many network architectures (FC, CNN, LSTM, etc).
- Symbolic API: python, scala, Clojure, R, Julia, Perl, Java (interference only).
- Imperative API: Gluon (python), with computer vision and natural language processing toolkits.
In conclusion Amazon SageMaker is a fully-managed service that covers the entire machine learning workflow to label and prepare your data, choose an algorithm, train the algorithm, tune and optimize it for deployment, make predictions, and take action.