No
Yes
View More
View Less
Working...
Close
OK
Cancel
Confirm
System Message
Delete
Schedule
An unknown error has occurred and your request could not be completed. Please contact support.
Scheduled
Wait Listed
Personal Calendar
Speaking
Conference Event
Meeting
Interest
Schedule TBD
Conflict Found
This session is already scheduled at another time. Would you like to...
Loading...
Please enter a maximum of {0} characters.
Please enter a maximum of {0} words.
must be 50 characters or less.
must be 40 characters or less.
Session Summary
We were unable to load the map image.
This has not yet been assigned to a map.
Search Catalog
Reply
Replies ()
Search
New Post
Microblog
Microblog Thread
Post Reply
Post
Your session timed out.
This web page is not optimized for viewing on a mobile device. Visit this site in a desktop browser to access the full set of features.
2017 GTC San Jose

S7858 - Building low-latency, production-grade Deep Learning platforms: the unsexy journey towards real-life results (Presented by Twitter)

Session Speakers
Session Description

You'll learn how Cortex built some components of its Deep Learning platform to fit into the highly optimized Twitter Timelines ranking pipeline. After navigating through a range of specific machine learning/deep learning workflows, found in different parts of the industry, we will focus on one that does not make the headlines of Recode, but is at the core of the business: the data first deep learning pipeline. We'll see how we had to build new, custom code on CPU and GPU to even think of pushing this production, and how what matters even more than the actual algorithm is how it can be deployed, tested and debugged. We'll discuss the opportunity that such an optimized production workflow generates for a company the scale of Twitter, and the challenges ahead.


Additional Session Information
Intermediate
Talk
Deep Learning and AI
Software
25 minutes
Session Schedule