Fault tolerance

This is essentially the artifact that you need to essentially distribute the model weights and to, do inference.

Um, the inference infrastructure, the checkpoint restore is mainly for fault tolerance.

So for example-

Okay

Okay, okay.

Got it.

So actually, you brought up a great point, right? Fault tolerance.

Can you talk about like, without the CRIU project, what happens today when a GPU fails? Uh, what are the consequences to the training workloads or the inferencing workloads? And then obviously I want to follow up that, follow that up with a question around after CRIU, what would that look like? So can you talk about like what happens when a GPU fails, please?

That the person is home, that the disk is in stock, that the replacement arrives before the second failure.

A one disk failure tolerance is a bet that you'll be there

and act quickly.

And the entire premise of the appliance is that it runs unattended.