Brad Miller: raw computed results
The raw ranking from the production similarity calculation (face-embedding centroid cosine → gender-cohort z-normalization). The web-opinion crown, curation, and fusion re-rank are removed.
- 1
85%z 4.42 - 2
84.6%z 4.39 - 3
75%z 3.96 - 4
74.4%z 3.96 - 5
70.1%z 3.91 - 6
68.1%z 3.89 - 7
59.8%z 3.80 - 8
47%z 3.60 - 9
46.4%z 3.56 - 10
46%z 3.54 - 11
45.5%z 3.50 - 12
45.3%z 3.49 - 13
45%z 3.47 - 14
44.3%z 3.42 - 15
44.1%z 3.41 - 16
44.1%z 3.41 - 17
43.9%z 3.40 - 18
43.8%z 3.39 - 19
43.2%z 3.36 - 20
43.2%z 3.35 - 21
42.8%z 3.33 - 22
42.6%z 3.32 - 23
42.5%z 3.31 - 24
42.4%z 3.30 - 25
42.3%z 3.30 - 26
42.3%z 3.30 - 27
42%z 3.27 - 28
41.8%z 3.26 - 29
41.8%z 3.26 - 30
41.2%z 3.23 - 31
41.2%z 3.23 - 32
41.2%z 3.22 - 33
40.5%z 3.18 - 34
40.5%z 3.18 - 35
40.5%z 3.18 - 36
40.1%z 3.16 - 37
39.7%z 3.13 - 38
39.2%z 3.10 - 39
39.1%z 3.09 - 40
39%z 3.08 - 41
38.9%z 3.08 - 42
38.9%z 3.08 - 43
38.8%z 3.07 - 44
38.5%z 3.05 - 45
38.1%z 3.03 - 46
38%z 3.02 - 47
37.7%z 3.00 - 48
37.7%z 3.00 - 49
37.6%z 2.99 - 50
37.6%z 2.99