Shivam Sharma
TechScalable
LEARN, CONNECT, BUILD
Ready to get started with AI and the latest technologies? Microsoft Reactor provides events, training, and community resources to help developers, entrepreneurs and startups build on AI technology and more. Join us!
LEARN, CONNECT, BUILD
Ready to get started with AI and the latest technologies? Microsoft Reactor provides events, training, and community resources to help developers, entrepreneurs and startups build on AI technology and more. Join us!
5 April, 2022 | 12:30 PM - 1:30 PM (UTC) Coordinated Universal Time
Topic: Cloud Development
Language: English
In Azure Databricks, you can perform distributed deep learning by using the open-source framework Horovod on top of TensorFlow, Keras, and PyTorch. Azure Databricks supports Uber's Horovod framework to run distributed, deep learning training jobs on Spark. In this workshop, you'll learn how to distribute deep learning model training in Azure Databricks with HorovodRunner.
The session will focus on Azure services and related products like Azure Machine Learning Service ,Azure Databricks,Spark ML,Mlflow & Horovod
What will you learn from the session :
Further Learning : https://aka.ms/HorovodandAzureDatabricks
Speaker : Shivam Sharma
Speaker BIO- Shivam is an author, cloud architect, speaker, and Co-Founder at TechScalable. Being passionate about ever evolving technology he works on Azure, GCP, Machine Learning, Kubernetes & DevOps. He is also a Microsoft Certified Trainer. He architects’ solutions on Cloud as well on-premises using wide array of platforms/technologies.
Social Handle
LinkedIn - https://www.linkedin.com/in/shivam-sharma-9828a536/
Twitter - https://twitter.com/ShivamSharma_TS
Facebook - https://www.facebook.com/TSshivamsharma/
You should be familiar with using Python to train deep learning models in Azure Databricks.
Speakers
For questions please contact us at reactor@microsoft.com