Existing anchor
13CZ
Three-run mid-bin repeat/cache control already exists.
Submission hardening / repeat controls
A bounded repeat experiment for hard/coverage and long-bin targets, designed to strengthen runtime and pLDDT interpretation without expanding the whole benchmark.
13CZ
Three-run mid-bin repeat/cache control already exists.
11BC
Two repeats planned for coverage-caveat behavior.
9K9Y
Two repeats planned for long-bin timing and confidence.
| Target | Why repeat | Existing | Planned |
|---|---|---|---|
13CZ:A | Existing mid-bin repeat/cache anchor. | 3 runs | none |
11BC:C | Hard/coverage target with only 63 aligned residues. | 1 run | 2 additional runs |
9K9Y:A | Long-bin target with strong quality and ligand-context caveat. | 1 run | 2 additional runs |
phase1a-repeat-2026-05-18-simplefold-3b-11bcc-r2phase1a-repeat-2026-05-18-simplefold-3b-11bcc-r3phase1a-repeat-2026-05-18-simplefold-3b-9k9ya-r2phase1a-repeat-2026-05-18-simplefold-3b-9k9ya-r3card.json, scores.json, predictions.json, and verdict.json.pilot/evidence/phase1a/logs/.Source document: docs/40_phase1a_repeat_controls_plan.md.