chart-training-runs
A four-state training-runs card: loss against optimiser step for several concurrent experiments, each drawn as a debiased EMA over its own faint raw trace, with divergence and restart markers that shade and count the steps a rollback cost, an optional learning-rate schedule on a secondary axis, and a scan that reads every run at the same step and only calls a lead when it clears the step-to-step noise.
chart-training-runs
A four-state training-runs card: loss against optimiser step for several concurrent experiments, each drawn as a debiased EMA over its own faint raw trace, with divergence and restart markers that shade and count the steps a rollback cost, an optional learning-rate schedule on a secondary axis, and a scan that reads every run at the same step and only calls a lead when it clears the step-to-step noise.