Uploaded October 2024 | Updated September 2026, 1 week ago
Datadog revolutionizes time series analysis with Toto, a groundbreaking Time Series-Optimized Transformer designed to tackle the intricate challenges of observability metrics. This session unveils how Toto harnesses advanced transformer architecture to achieve unprecedented forecasting accuracy across diverse domains.
Emaad Khwaja delves into the unique attributes that distinguish Toto in the realm of time series foundation models. He explores Toto's impressive training regimen, powered by AWS P5 instances and Ray on Kubernetes, which processed an unparalleled trillion data points—75% drawn from real-world observability scenarios. Khwaja demonstrates how this extensive training enables Toto to perform remarkably accurate zero-shot predictions on novel inputs while maintaining superior performance on general forecasting tasks. Attendees gain insights into Toto's architectural innovations and its potential to set new benchmarks in time series analysis for both observability and broader applications.
--
Interested in more?
- Watch the full Day 1 Keynote: youtu.be/jwZHJthQvXo
- Watch the full Day 2 Keynote youtu.be/Lury2ad6KG8
- Check out the Ray Summmit Breakout sessions youtube.com/playlist?list=PLzTswPQNepXntmT8jr9WaNfqQ60QwW7-U&si=qPw-_SxT9lVmbRGE
--
đź”— Connect with us:
- Subscribe to our YouTube channel: youtube.com/@anyscale
- Twitter: https://x.com/anyscalecompute
- LinkedIn: linkedin.com/company/joinanyscale
- Website: anyscale.com
Datadog revolutionizes time series analysis with Toto, a groundbreaking Time Series-Optimized Transformer designed to tackle the intricate challenges of observability metrics. This session unveils how Toto harnesses advanced transformer architecture to achieve unprecedented forecasting accuracy across diverse domains.
Emaad Khwaja delves into the unique attributes that distinguish Toto in the realm of time series foundation models. He explores Toto's impressive training regimen, powered by AWS P5 instances and Ray on Kubernetes, which processed an unparalleled trillion data points—75% drawn from real-world observability scenarios. Khwaja demonstrates how this extensive training enables Toto to perform remarkably accurate zero-shot predictions on novel inputs while maintaining superior performance on general forecasting tasks. Attendees gain insights into Toto's architectural innovations and its potential to set new benchmarks in time series analysis for both observability and broader applications.
--
Interested in more?
- Watch the full Day 1 Keynote: youtu.be/jwZHJthQvXo
- Watch the full Day 2 Keynote youtu.be/Lury2ad6KG8
- Check out the Ray Summmit Breakout sessions youtube.com/playlist?list=PLzTswPQNepXntmT8jr9WaNfqQ60QwW7-U&si=qPw-_SxT9lVmbRGE
--
đź”— Connect with us:
- Subscribe to our YouTube channel: youtube.com/@anyscale
- Twitter: https://x.com/anyscalecompute
- LinkedIn: linkedin.com/company/joinanyscale
- Website: anyscale.com
