Can you trigger checkpoint on demand? What are conventional abstractions people use? What layers of information is checkpointed? 🕵🏻♀️
November 14, 2025 at 4:16 PM
Can you trigger checkpoint on demand? What are conventional abstractions people use? What layers of information is checkpointed? 🕵🏻♀️
Like it’s obviously really useful for:
- speed of node rotation
- perhaps folding into scheduling preemption logic
What I don’t understand is the basics of how it’s folded into ML training.
- speed of node rotation
- perhaps folding into scheduling preemption logic
What I don’t understand is the basics of how it’s folded into ML training.
November 14, 2025 at 4:16 PM
Like it’s obviously really useful for:
- speed of node rotation
- perhaps folding into scheduling preemption logic
What I don’t understand is the basics of how it’s folded into ML training.
- speed of node rotation
- perhaps folding into scheduling preemption logic
What I don’t understand is the basics of how it’s folded into ML training.
I think also bc I don’t have like kids/health problems, worst case scenario I bum out in Atlanta for a night in a hotel room.
November 9, 2025 at 2:54 AM
I think also bc I don’t have like kids/health problems, worst case scenario I bum out in Atlanta for a night in a hotel room.
Sleeping bag for camping at the ATL airport? 🥲
November 6, 2025 at 4:38 AM
Sleeping bag for camping at the ATL airport? 🥲
I would prefer to be trapped in seattle and not trapped in Atlanta. 🙏
November 4, 2025 at 7:58 PM
I would prefer to be trapped in seattle and not trapped in Atlanta. 🙏
Also solid maybe for that. Depending on if my talk gets accepted.
October 30, 2025 at 1:29 AM
Also solid maybe for that. Depending on if my talk gets accepted.