Seed and tuning-budget audit of a two-arm comparison

Compare a baseline and a variant with paired seeds and equal tuning trials, plot expected best validation score against budget, and report intervals with a written threats-to-validity note.

The exercise happens on Self-directed project (budget-aware reporting as the method). Canonical link: https://arxiv.org/abs/1909.03004. Link check: link responded (5 Oct 2026). Opened from this app 0 times; opening is not counted as an attempt.

Supports: Evidence that you can design and defend an ML comparison, a frequent research-interview probe.