Click a node or edge for details

Verification — outgoing probability per source state (sums to 100%)

sourceout totalΣ% → targets
start91100%breakthrough 85.7% (78)   improved 9.9% (9)   stuck 4.4% (4)
improved12100%breakthrough 41.7% (5)   end_limit 25.0% (3)   improved 8.3% (1)   regressed 8.3% (1)   stuck 8.3% (1)   broke 8.3% (1)
stuck9100%stuck 44.4% (4)   end_limit 33.3% (3)   breakthrough 11.1% (1)   improved 11.1% (1)
breakthrough84100%end_submit 96.4% (81)   stay 3.6% (3)
stay7100%stay 57.1% (4)   end_submit 42.9% (3)
regressed1100%improved 100.0% (1)
broke1100%end_limit 100.0% (1)

Gap label distribution — for each label, which src→tgt edges carry it

gap labeltotal src→tgt breakdown
debug_failure2improved→end_limit 50.0% (1)   improved→broke 50.0% (1)
execute_plan1stuck→improved 100.0% (1)
explore8stuck→end_limit 25.0% (2)   start→stuck 25.0% (2)   improved→end_limit 25.0% (2)   stuck→stuck 12.5% (1)   broke→end_limit 12.5% (1)
fix_bug5improved→breakthrough 40.0% (2)   stuck→stuck 20.0% (1)   breakthrough→stay 20.0% (1)   stay→stay 20.0% (1)
implement92start→breakthrough 84.8% (78)   start→improved 9.8% (9)   start→stuck 2.2% (2)   stuck→stuck 1.1% (1)   breakthrough→stay 1.1% (1)   improved→regressed 1.1% (1)
infra_confusion3stay→end_submit 33.3% (1)   stay→stay 33.3% (1)   stuck→breakthrough 33.3% (1)
optimize_perf7stuck→stuck 14.3% (1)   improved→breakthrough 14.3% (1)   breakthrough→stay 14.3% (1)   stay→stay 14.3% (1)   improved→improved 14.3% (1)   regressed→improved 14.3% (1)   improved→stuck 14.3% (1)
revert1stay→stay 100.0% (1)
rewrite2improved→breakthrough 100.0% (2)
verify84breakthrough→end_submit 96.4% (81)   stay→end_submit 2.4% (2)   stuck→end_limit 1.2% (1)