Descent-Certified LMO Sign Switching / results.json
Failed on benchmark
1{
2 "seed": 1616,
3 "math_verification": {
4 "threshold_sweep": [
5 {
6 "tau": 0.0,
7 "predicted_post_fraction_P_rho_ge_tau": 0.659,
8 "observed_post_fraction": 0.659,
9 "post_uphill_rate": 0.346,
10 "switch_uphill_rate": 0.023916666666666666,
11 "post_mean_true_alignment": 0.23465021696378952,
12 "chosen_mean_true_alignment": 0.5225910290545672,
13 "post_mean_inner": 3.7098687082536306,
14 "switch_mean_inner": 7.275880389094258
15 },
16 {
17 "tau": 0.02,
18 "predicted_post_fraction_P_rho_ge_tau": 0.6564166666666666,
19 "observed_post_fraction": 0.6564166666666666,
20 "post_uphill_rate": 0.346,
21 "switch_uphill_rate": 0.022083333333333333,
22 "post_mean_true_alignment": 0.23465021696378952,
23 "chosen_mean_true_alignment": 0.5238502896959694,
24 "post_mean_inner": 3.7098687082536306,
25 "switch_mean_inner": 7.285496323498012
26 },
27 {
28 "tau": 0.05,
29 "predicted_post_fraction_P_rho_ge_tau": 0.6530833333333333,
30 "observed_post_fraction": 0.6530833333333333,
31 "post_uphill_rate": 0.346,
32 "switch_uphill_rate": 0.020583333333333332,
33 "post_mean_true_alignment": 0.23465021696378952,
34 "chosen_mean_true_alignment": 0.5251752698351125,
35 "post_mean_inner": 3.7098687082536306,
36 "switch_mean_inner": 7.293756898064051
37 },
38 {
39 "tau": 0.1,
40 "predicted_post_fraction_P_rho_ge_tau": 0.64675,
41 "observed_post_fraction": 0.64675,
42 "post_uphill_rate": 0.346,
43 "switch_uphill_rate": 0.018083333333333333,
44 "post_mean_true_alignment": 0.23465021696378952,
45 "chosen_mean_true_alignment": 0.5269429164262665,
46 "post_mean_inner": 3.7098687082536306,
47 "switch_mean_inner": 7.300649061591103
48 }
49 ],
50 "linear_descent_identity": {
51 "predicted_delta": 0.9385206011733538,
52 "observed_delta": 0.9385206011733537,
53 "absolute_error": 1.1102230246251565e-16,
54 "eta_sweep": [
55 {
56 "eta": 0.01,
57 "predicted_delta": 0.13407437159619337,
58 "observed_delta": 0.13407437159619337,
59 "abs_error": 0.0
60 },
61 {
62 "eta": 0.03,
63 "predicted_delta": 0.40222311478858014,
64 "observed_delta": 0.4022231147885801,
65 "abs_error": 5.551115123125783e-17
66 },
67 {
68 "eta": 0.07,
69 "predicted_delta": 0.9385206011733538,
70 "observed_delta": 0.9385206011733537,
71 "abs_error": 1.1102230246251565e-16
72 },
73 {
74 "eta": 0.15,
75 "predicted_delta": 2.0111155739429005,
76 "observed_delta": 2.0111155739429005,
77 "abs_error": 0.0
78 },
79 {
80 "eta": 0.3,
81 "predicted_delta": 4.022231147885801,
82 "observed_delta": 4.022231147885801,
83 "abs_error": 0.0
84 }
85 ]
86 },
87 "noise_sweep": [
88 {
89 "heldout_noise_std": 0.0,
90 "routing_agreement": 1.0
91 },
92 {
93 "heldout_noise_std": 0.25,
94 "routing_agreement": 0.9743333333333334
95 },
96 {
97 "heldout_noise_std": 0.55,
98 "routing_agreement": 0.9373333333333334
99 },
100 {
101 "heldout_noise_std": 1.0,
102 "routing_agreement": 0.8823333333333333
103 },
104 {
105 "heldout_noise_std": 2.0,
106 "routing_agreement": 0.765
107 }
108 ]
109 },
110 "quadratic_comparison": {
111 "post": {
112 "final_loss": 0.1444657301426662,
113 "best_loss": 0.04207063737212821,
114 "uphill_steps": 59,
115 "post_fraction": 1.0
116 },
117 "pre": {
118 "final_loss": 10.456574724429673,
119 "best_loss": 7.509928272102478,
120 "uphill_steps": 59,
121 "post_fraction": 0.0
122 },
123 "switch": {
124 "final_loss": 0.18319758786532345,
125 "best_loss": 0.040381919404750846,
126 "uphill_steps": 57,
127 "post_fraction": 0.75
128 }
129 }
130}