donnygreenberg

4 ms·

1.
Open Sourcing Kubetorch(run.house)
9 points by donnygreenberg 11mo ago|1 comments
2.
by donnygreenberg 11mo ago
Hi HN, AI/ML is a vertically integrated field. Frameworks and methods that integrate deeply down the stack win. Despite this, most open-source ML libraries and frameworks do not integrate with infrastructure whatsoever, leaving abundan
3.
by donnygreenberg 2y ago
If you want a PyTorch-like experience on your own GPUs (either static or cloud), see https://github.com/run-house/runhouse
4.
by donnygreenberg 3y ago
Would be nice if this came with scripts which could launch the examples on compatible GPUs on cloud providers (rather than trying to guess?). Would anyone else be interested in that? Considering putting it together.
5.
by donnygreenberg 3y ago
Neat idea and elegant execution. The vibe I'm getting from the README is that I'd want to reach for this as a Python dev (over Streamlit or Gradio) if I already have a FastAPI app and want a simple dashboard in front of it. Is tha
6.
by donnygreenberg 3y ago
Yes! It's natively supported in the cluster object. You can specify rh.cluster(name="my-a10", instance_type="A10:1", provider="aws"), and gcp, azure, and lambda labs are also supported, and if you leave pr
7.
Please separate orchestration and execution (Or: Packaging hell is dragging ML)(run.house)
8 points by donnygreenberg 3y ago|2 comments
8.
by donnygreenberg 3y ago
Good catch - this is actually only for functions defined inside interactive Notebook or iPython environments, and we do have an option to bypass it (which serializes the function and state), but you probably don't want it by default. A
9.
by donnygreenberg 3y ago
Yes, we work pretty closely with them and they're lovely. Everyone should try SkyPilot.
10.
by donnygreenberg 3y ago
Good question. We actually use Ray to handle a bunch of the scheduling within the compute, but largely see our role as outside the compute. Meaning, Ray provides a powerful DSL for distributed compute, while we are aggressively DSL-free so
11.
by donnygreenberg 3y ago
No, I know of CentML but don't deeply know the surface of hardware they compile for. I'm enthusiastic about projects like this and others which integrate with PyTorch 2.0. Flexible compilers make the value of being able to ship yo
12.
by donnygreenberg 3y ago
Excellent! Don't hesitate to reach out (donny at run dot house) if you want to chat about adopting our approach or using Runhouse.
13.
by donnygreenberg 3y ago
It's a great point. The funny thing is that the rest of the world just has dev, QA, and prod staging and canaries, while ML has "the 6 months it takes to translate from notebook to pipeline" or "uploading a new checkpoin
14.
by donnygreenberg 3y ago
That's a good question. I actually love the Mojo concept, but see it as very different. They're creating a portable acceleration option in Python proper, while we're trying to make it so you can easily ship around such code t
15.
by donnygreenberg 3y ago
Hi! That's awesome to hear, and very aligned with the devx we're going for. How was your system received? In fact we totally agree and are not cloudpickling the function because of the package minor version issues. We sync over th
16.
A PyTorch Approach to ML Infrastructure(run.house)
113 points by donnygreenberg 3y ago|22 comments
17.
Getting past the AI hackathon cold start problem – Remote Cluster Auto-setup(medium.com)
4 points by donnygreenberg 3y ago|1 comments
18.
by donnygreenberg 4y ago
I've been using SkyPilot for 3 months within an ML platform project and it is exactly as awesome as it sounds. The overall experience for launching and managing compute is thoughtful and ergonomic in a way that we've been conditio