Hello! 👋

We hope you all enjoyed some time off over the holidays and are looking forward to 2021. December was an exciting month for Determined. We held our first-ever community sync, with 34 attendees from more than 10 different organizations! We also launched our Determined AI Writers’ Program; this will give developers the chance to work closely with Determined, get paid to write about their work, and reach a wide audience of machine learning practitioners.

We couldn’t be more excited for January. The team has been busy putting the finishing touches on next year’s roadmap and we are looking forward to sharing new planned features and use cases soon. To start off the year, Determined’s Liam Li will be hosting a lunch & learn session about object detection with DETR on January 26th.

If you have any special requests or would like to chat about deep learning topics, we’d love to hear from you.

Evan, Neil, and Ameet

Lunch & Learn: Scalable Object Detection with Determined

Come join us on January 26th to learn about training modern object detection models at scale! Liam will describe Facebook’s DETR model and give attendees a hand-on walkthrough on how to train DETR, and then how to accomplish advanced tasks like hyperparameter tuning and distributed training easily using Determined. Plus, you’ll get free food! Whoever said there’s no such thing as a free lunch?

Determined is officially a TensorFlow AI Service Partner

Determined and TensorFlow share the same goal: to make deep learning dramatically easier and more accessible. We’re proud to be named a TensorFlow AI Service Partner to work alongside TensorFlow to bring world-class deep learning tooling to data scientists and engineers alike. You can read the announcement here, and see out how Determined fits into the TensorFlow partner program on their website.

Announcing: The Determined AI Writers’ Program

If you’re reading this, chances are you’ve built something amazing using machine learning. Determined wants to help you tell your story! We recently launched the Determined Writers’ Program to support data scientists creating original written content about machine learning and AI. All you need to do is use Determined for the problem you’re writing about, and you’ll work alongside somebody within Determined throughout the process. Do you already have an idea? Learn more here and fill out the form at the bottom of the article to get started!

Determined is now on the NVIDIA NGC Container Catalog

We’re excited to be featured on NVIDIA’s GPU-optimized NGC Catalog! Over the past few years, NVIDIA has helped companies power deep learning model development with state-of-the-art advancements in high-performance computing. Determined can now be accessed from NVIDIA’s NGC catalog, providing leading model training functionalities like easy distributed training and fast hyperparameter tuning to folks leveraging NVIDIA GPUs and DGX systems. You can read more about the partnership here, and download the Helm chart at the bottom of the post to get started with Determined today.

Community Sync

Determined held its first-ever community sync last week! Since our open-source release in April, our community of deep learning developers has been growing steadily, so we wanted to provide a way for people to introduce themselves, talk about problems they’re solving with deep learning, and give feedback in a more formal setting. Here’s a recap on the community sync from our Product Manager, Vishnu Mohan.

New Features

Over the past few Determined releases - i.e., between 0.13.8 and 0.13.10, we’ve:

• Added the ability to filter trial logs
• Made the WebUI mobile-device friendly
• Made various improvements and fixes to our TensorFlow Keras framework support
• Added support for priority scheduling with (optional) preemption
• Improved end-to-end TLS support to work better in a typical cloud deployment
• Improved AWS deployments, with the inclusion of our first-ever community-contributed feature!

What We’ve Been Writing About

• Data Scientists Don’t Care About Kubernetes
• Determined’s upstream contributions to Horovod: Local Gradient Aggregation
• Determined Podcast Series: TQ Chen on Efficient Model Deployment and Specialized Hardware

Work with Us

Help build the future of ML! Join our team of technical leaders in machine learning and data-driven distributed systems. Read more.