Episode 83

Scaling Laws & Memory Math

How big models get better, and how much GPU memory they really need.

• Scaling laws intuition
• Params → memory
• Training vs inference memory
• Optimizer state

Notes for this episode are coming soon.