Introducing Fleets: a new open-source tool to manage cloud and on-prem clusters for AI. No more Kubernetes, KubeFlow, Slurm, etc.
Fleets enable researchers and engineers to develop AI seamlessly, without friction with Ops teams.
Introducing Fleets: a new open-source tool to manage cloud and on-prem clusters for AI. No more Kubernetes, KubeFlow, Slurm, etc.
Fleets enable researchers and engineers to develop AI seamlessly, without friction with Ops teams.
As part of our partnership and friendship with
@dstackai
, we will commit compute resources to ensuring that Fleets works flawlessly on our cluster. ❤️
Imagine creating a fleet from your on-prem GPU clusters simply by listing their hostnames and using them for all your AI needs—dev environments, tasks, services, etc. No Kubernetes, Kubeflow, Airflow, no other "flows".
We're excited to share a preview of the upcoming v0.18.7 update!
This preview is extra special with several major new features:
✨ Fleets - a brand new way to manage cloud and on-prem clusters
✨ RunPod volumes - persist data across
@runpod_io
runs using volumes
✨ dstack
So excited to see dstack reach 1250 stars! 🤩 Thanks for all the support!
If you're into containers, AI, and open-source, we’d love to have you contribute! Help us set a new standard for AI containers. FYI, our codebase is in Python.
Are you in Berlin today and interested in using containers to accelerate AI development in the cloud and on-premises? Come to the Berlin System Group meetup!
@andrey_cheptsov
will give a talk about dstack, a Kubernetes alternative for AI!
If your team heavily uses or plans on using on-prem clusters for model training and deployment, and you're open to collaborating on optimizing it, please DM us. We're creating an open-source Kubernetes alternative for AI teams and keen to collaborate!
🔥🔥🔥 Thanks to
@algo_diver
, we've got a new example on how to run
@huggingface
's Alignment Handbook distributed fine-tuning tasks with
@dstackai
(either on cloud or on-prem).
Big shout-out to
@alvarobartt
for testing and feedback too!
Excited to announce our next release! It adds initial support for Google Cloud TPUs, making it possible to run AI containers for training and deployment. Check it out:
Massive! 🚀🚀
Our new preview build lets you attach volumes to your AI workloads. For now, it's AWS-only, but GCP, Azure, RunPod, Lambda, and more are on the way!
Massive! 🚀🚀
Our new preview build lets you attach volumes to your AI workloads. For now, it's AWS-only, but GCP, Azure, RunPod, Lambda, and more are on the way!
Hey Python folks! If you love open-source, containers, infrastructure, and AI, we'd love for you to join us on our team working with . Drop us a DM if you're interested!