Conformal Early-Rejection for Diffusion Architecture Search / results.json
Mechanism failed
1{
2 "seed": 2762,
3 "checkpoints": [
4 0.75,
5 0.5,
6 0.25
7 ],
8 "coverage_sweep": [
9 {
10 "ncal": 20,
11 "predicted_exceedance": 0.09523809523809523,
12 "observed_exceedance": 0.09381166666666667
13 },
14 {
15 "ncal": 50,
16 "predicted_exceedance": 0.09803921568627451,
17 "observed_exceedance": 0.09930388888888889
18 },
19 {
20 "ncal": 100,
21 "predicted_exceedance": 0.09900990099009901,
22 "observed_exceedance": 0.09891333333333333
23 },
24 {
25 "ncal": 500,
26 "predicted_exceedance": 0.0998003992015968,
27 "observed_exceedance": 0.09994222222222222
28 }
29 ],
30 "separability_sweep": [
31 {
32 "beta": 0.5,
33 "rejected": 164.775,
34 "evaluated": 835.225,
35 "reject_rate": 0.164775,
36 "accepted_failure": 0.39978736837966544,
37 "best_accuracy": 0.9040470728635223
38 },
39 {
40 "beta": 1.0,
41 "rejected": 169.5,
42 "evaluated": 830.5,
43 "reject_rate": 0.16949999999999998,
44 "accepted_failure": 0.3980005987006782,
45 "best_accuracy": 0.9037176757478054
46 },
47 {
48 "beta": 2.0,
49 "rejected": 161.85,
50 "evaluated": 838.15,
51 "reject_rate": 0.16185,
52 "accepted_failure": 0.4013646849001048,
53 "best_accuracy": 0.9033178961843648
54 },
55 {
56 "beta": 5.0,
57 "rejected": 163.85,
58 "evaluated": 836.15,
59 "reject_rate": 0.16385,
60 "accepted_failure": 0.403216299987184,
61 "best_accuracy": 0.9034814473176492
62 },
63 {
64 "beta": 10.0,
65 "rejected": 168.9625,
66 "evaluated": 831.0375,
67 "reject_rate": 0.1689625,
68 "accepted_failure": 0.39629988444536995,
69 "best_accuracy": 0.9036464037794417
70 }
71 ],
72 "distribution_shift_sweep": [
73 {
74 "test_shift": 0.0,
75 "rejected": 163.89,
76 "evaluated": 836.11,
77 "reject_rate": 0.16389000000000004,
78 "accepted_failure": 0.39920871093820337,
79 "best_accuracy": 0.904202619761248
80 },
81 {
82 "test_shift": -0.25,
83 "rejected": 224.9,
84 "evaluated": 775.1,
85 "reject_rate": 0.22489999999999996,
86 "accepted_failure": 0.47987205957556023,
87 "best_accuracy": 0.9024729897689305
88 },
89 {
90 "test_shift": -0.5,
91 "rejected": 303.45,
92 "evaluated": 696.55,
93 "reject_rate": 0.30345,
94 "accepted_failure": 0.5534916522516165,
95 "best_accuracy": 0.9003408879440855
96 },
97 {
98 "test_shift": -1.0,
99 "rejected": 488.76,
100 "evaluated": 511.24,
101 "reject_rate": 0.48876000000000003,
102 "accepted_failure": 0.6872725785685375,
103 "best_accuracy": 0.8956348484675797
104 }
105 ],
106 "controller_comparison": {
107 "0.05": {
108 "conformal": {
109 "rejected": 94.86,
110 "evaluated": 905.14,
111 "reject_rate": 0.09486,
112 "accepted_failure": 0.4502842914424866,
113 "best_accuracy": 0.9039759565824498
114 },
115 "uncalibrated": {
116 "rejected": 642.45,
117 "evaluated": 357.55,
118 "reject_rate": 0.6424500000000001,
119 "accepted_failure": 0.021276631784741117,
120 "best_accuracy": 0.9039759565824498
121 },
122 "none": {
123 "rejected": 0.0,
124 "evaluated": 1000.0,
125 "reject_rate": 0.0,
126 "accepted_failure": 0.5024599999999999,
127 "best_accuracy": 0.9039759565824498
128 }
129 },
130 "0.1": {
131 "conformal": {
132 "rejected": 171.25,
133 "evaluated": 828.75,
134 "reject_rate": 0.17125,
135 "accepted_failure": 0.39976413696884633,
136 "best_accuracy": 0.9039759565824498
137 },
138 "uncalibrated": {
139 "rejected": 642.45,
140 "evaluated": 357.55,
141 "reject_rate": 0.6424500000000001,
142 "accepted_failure": 0.021276631784741117,
143 "best_accuracy": 0.9039759565824498
144 },
145 "none": {
146 "rejected": 0.0,
147 "evaluated": 1000.0,
148 "reject_rate": 0.0,
149 "accepted_failure": 0.5024599999999999,
150 "best_accuracy": 0.9039759565824498
151 }
152 }
153 }
154}