OpenAI staffer's tip: fix ugly loss curves with log-log axes and max smoothing
OpenAI's Will Depue joked that when training curves look terrible, switching both axes to log scale and maximizing smoothing 'never fails' to fix the panic.
2026-11-21 ~ 2026-11-21 · 2 related posts
- ML pro tip: make both axes logarithmic and max out smoothing, and any loss curve looks fine — willdepue · 2026-11-21
- ML pro tip: make ugly training curves look great with log-log axes and max smoothing — willdepue · 2026-11-21