The Minutes of Gavin Baker on efficiency https://minutesof.com/gavin-baker/on/efficiency/ Everything Gavin Baker has said on efficiency: 3 verbatim quotes from August 2024, each with a timestamp and a link to the recording it came from. en Sun, 30 Aug 2026 16:21:43 +0000 Baker explains MFU runs at 35-40%, measuring the percentage of theoretical compute actually used for training. https://minutesof.com/q/91f8d62c-9179-4722-83b1-a80397265ed3/ https://minutesof.com/q/91f8d62c-9179-4722-83b1-a80397265ed3/ “MFU, model flops utilization, and that generally runs around 35 to 40%. And that's literally the percentage of compute, theoretical compute flops that you're actually applying to trading.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 efficiency Baker explains higher MFU allows 25% faster time to market with same GPU and power spending. https://minutesof.com/q/2ef0281e-1acf-4b2f-a09a-427746b64fa5/ https://minutesof.com/q/2ef0281e-1acf-4b2f-a09a-427746b64fa5/ “You have the same amount of GPUs and the same amount of power presumably. You could choose between faster time to market.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 efficiency Baker explains that 50% MFU versus 40% MFU enables 25% faster time to market for AI models. https://minutesof.com/q/b18ea10e-8f3d-44a4-afda-d9363c3a12ee/ https://minutesof.com/q/b18ea10e-8f3d-44a4-afda-d9363c3a12ee/ “You could choose between faster time to market. If you run a 50% MFU and your competitor's running 40 for an equivalent amount of trading flops, you could be in market 25% faster.” — Invest Like the Best Tue, 27 Aug 2024 08:00:00 +0000 efficiency