Zero-Augmented Double-Scoring / results.json
Mechanism confirmed, baseline not beaten
1{
2 "predictions": {
3 "iid_topk_sweep": [
4 {
5 "M": 20,
6 "rho_aug": 0.2,
7 "K": 8,
8 "pred_mean_r": 4.0,
9 "obs_mean_r": 3.9875,
10 "pred_std_r": 1.281025230440697,
11 "obs_std_r": 1.2754778516305174,
12 "pred_eff_density": 0.2,
13 "obs_eff_density": 0.199375,
14 "pred_dummy_fraction": 0.5,
15 "obs_dummy_fraction": 0.5015625
16 },
17 {
18 "M": 20,
19 "rho_aug": 0.5,
20 "K": 20,
21 "pred_mean_r": 10.0,
22 "obs_mean_r": 9.9743,
23 "pred_std_r": 1.6012815380508714,
24 "obs_std_r": 1.5985116546337723,
25 "pred_eff_density": 0.5,
26 "obs_eff_density": 0.49871499999999996,
27 "pred_dummy_fraction": 0.5,
28 "obs_dummy_fraction": 0.501285
29 },
30 {
31 "M": 20,
32 "rho_aug": 0.8,
33 "K": 32,
34 "pred_mean_r": 16.0,
35 "obs_mean_r": 15.988,
36 "pred_std_r": 1.281025230440697,
37 "obs_std_r": 1.2890911527118631,
38 "pred_eff_density": 0.8,
39 "obs_eff_density": 0.7994,
40 "pred_dummy_fraction": 0.5,
41 "obs_dummy_fraction": 0.500375
42 },
43 {
44 "M": 50,
45 "rho_aug": 0.2,
46 "K": 20,
47 "pred_mean_r": 10.0,
48 "obs_mean_r": 10.01765,
49 "pred_std_r": 2.010075630518424,
50 "obs_std_r": 2.010705964953603,
51 "pred_eff_density": 0.2,
52 "obs_eff_density": 0.200353,
53 "pred_dummy_fraction": 0.5,
54 "obs_dummy_fraction": 0.4991175
55 },
56 {
57 "M": 50,
58 "rho_aug": 0.5,
59 "K": 50,
60 "pred_mean_r": 25.0,
61 "obs_mean_r": 25.0115,
62 "pred_std_r": 2.51259453814803,
63 "obs_std_r": 2.517313597865788,
64 "pred_eff_density": 0.5,
65 "obs_eff_density": 0.5002300000000001,
66 "pred_dummy_fraction": 0.5,
67 "obs_dummy_fraction": 0.49976999999999994
68 },
69 {
70 "M": 50,
71 "rho_aug": 0.8,
72 "K": 80,
73 "pred_mean_r": 40.0,
74 "obs_mean_r": 39.98675,
75 "pred_std_r": 2.010075630518424,
76 "obs_std_r": 2.0228876482642333,
77 "pred_eff_density": 0.8,
78 "obs_eff_density": 0.799735,
79 "pred_dummy_fraction": 0.5,
80 "obs_dummy_fraction": 0.500165625
81 },
82 {
83 "M": 100,
84 "rho_aug": 0.2,
85 "K": 40,
86 "pred_mean_r": 20.0,
87 "obs_mean_r": 20.02205,
88 "pred_std_r": 2.835524820033344,
89 "obs_std_r": 2.8377744444370485,
90 "pred_eff_density": 0.2,
91 "obs_eff_density": 0.2002205,
92 "pred_dummy_fraction": 0.5,
93 "obs_dummy_fraction": 0.49944875
94 },
95 {
96 "M": 100,
97 "rho_aug": 0.5,
98 "K": 100,
99 "pred_mean_r": 50.0,
100 "obs_mean_r": 49.99015,
101 "pred_std_r": 3.5444060250416793,
102 "obs_std_r": 3.5770033516199002,
103 "pred_eff_density": 0.5,
104 "obs_eff_density": 0.4999015,
105 "pred_dummy_fraction": 0.5,
106 "obs_dummy_fraction": 0.5000985
107 },
108 {
109 "M": 100,
110 "rho_aug": 0.8,
111 "K": 160,
112 "pred_mean_r": 80.0,
113 "obs_mean_r": 80.01715,
114 "pred_std_r": 2.835524820033344,
115 "obs_std_r": 2.8595377034583755,
116 "pred_eff_density": 0.8,
117 "obs_eff_density": 0.8001715,
118 "pred_dummy_fraction": 0.5,
119 "obs_dummy_fraction": 0.49989281249999995
120 }
121 ]
122 },
123 "representation": {
124 "M": 12,
125 "K": 9,
126 "all_masks_at_most_K_represented": true,
127 "examples_r_dummy": [
128 [
129 0,
130 9
131 ],
132 [
133 9,
134 0
135 ]
136 ]
137 },
138 "mini_experiment": {
139 "runs": [
140 {
141 "seed": 0,
142 "baseline": {
143 "acc": 0.5950000286102295,
144 "r": 6,
145 "dummy": 0,
146 "eff_density": 0.3,
147 "K": 6
148 },
149 "double": {
150 "acc": 0.6150000095367432,
151 "r": 4,
152 "dummy": 8,
153 "eff_density": 0.2,
154 "K": 12
155 }
156 },
157 {
158 "seed": 1,
159 "baseline": {
160 "acc": 0.7566666603088379,
161 "r": 6,
162 "dummy": 0,
163 "eff_density": 0.3,
164 "K": 6
165 },
166 "double": {
167 "acc": 0.7583333253860474,
168 "r": 7,
169 "dummy": 5,
170 "eff_density": 0.35,
171 "K": 12
172 }
173 },
174 {
175 "seed": 2,
176 "baseline": {
177 "acc": 0.7316666841506958,
178 "r": 6,
179 "dummy": 0,
180 "eff_density": 0.3,
181 "K": 6
182 },
183 "double": {
184 "acc": 0.7683333158493042,
185 "r": 8,
186 "dummy": 4,
187 "eff_density": 0.4,
188 "K": 12
189 }
190 },
191 {
192 "seed": 3,
193 "baseline": {
194 "acc": 0.6866666674613953,
195 "r": 6,
196 "dummy": 0,
197 "eff_density": 0.3,
198 "K": 6
199 },
200 "double": {
201 "acc": 0.7016666531562805,
202 "r": 8,
203 "dummy": 4,
204 "eff_density": 0.4,
205 "K": 12
206 }
207 },
208 {
209 "seed": 4,
210 "baseline": {
211 "acc": 0.6466666460037231,
212 "r": 6,
213 "dummy": 0,
214 "eff_density": 0.3,
215 "K": 6
216 },
217 "double": {
218 "acc": 0.6566666960716248,
219 "r": 7,
220 "dummy": 5,
221 "eff_density": 0.35,
222 "K": 12
223 }
224 }
225 ],
226 "baseline": {
227 "mean_acc": 0.6833333373069763,
228 "std_acc": 0.05808996375618053,
229 "mean_eff_density": 0.3,
230 "std_eff_density": 0.0,
231 "K": 6
232 },
233 "double": {
234 "mean_acc": 0.7,
235 "std_acc": 0.058613599525027094,
236 "mean_eff_density": 0.34,
237 "std_eff_density": 0.07348469228349534,
238 "K": 12
239 }
240 }
241}