Support-Sparse Koopman World Model / stage2_stdout.json
Failed on benchmark
1{
2 "bench_version": 1,
3 "track": "dynamics",
4 "model": "rnn_small",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "lr": 0.006,
10 "wd": 0.0
11 },
12 "sweep": [
13 {
14 "cfg": {
15 "lr": 0.001,
16 "wd": 0.0
17 },
18 "mean": 0.41272661462426186
19 },
20 {
21 "cfg": {
22 "lr": 0.003,
23 "wd": 0.0
24 },
25 "mean": 0.0579891842789948
26 },
27 {
28 "cfg": {
29 "lr": 0.006,
30 "wd": 0.0
31 },
32 "mean": 0.050475220661610365
33 }
34 ],
35 "full": {
36 "mean": 0.050475220661610365,
37 "std": 0.015083096048661116,
38 "per_seed": [
39 0.04993843659758568,
40 0.04235526919364929,
41 0.0690467357635498,
42 0.07208486646413803,
43 0.05335729196667671,
44 0.022722482681274414,
45 0.055551957339048386,
46 0.0387447252869606
47 ],
48 "n": 8
49 }
50 },
51 "idea": {
52 "mean": 0.04891866398975253,
53 "std": 0.00862940297862839,
54 "per_seed": [
55 0.048651475459337234,
56 0.029897399246692657,
57 0.0598931610584259,
58 0.04604996740818024,
59 0.05041059851646423,
60 0.05862913280725479,
61 0.05099020525813103,
62 0.046827372163534164
63 ],
64 "n": 8
65 },
66 "comparison": {
67 "delta_mean": -0.0015565566718578339,
68 "idea_wins": 6,
69 "n_pairs": 8,
70 "per_seed_diffs": [
71 -0.0012869611382484436,
72 -0.012457869946956635,
73 -0.009153574705123901,
74 -0.026034899055957794,
75 -0.0029466934502124786,
76 0.03590665012598038,
77 -0.004561752080917358,
78 0.008082646876573563
79 ],
80 "p_value": 0.7986,
81 "mde": 0.015101068783051909,
82 "mde_rel_pct": 29.917786559647947,
83 "verdict": "no significant win",
84 "system_worked": false
85 },
86 "mechanism_signature": {
87 "one_step_latent_mse_mean": 0.03250809867555896,
88 "decoded_rollout_mse_h6": 0.305510014295578,
89 "support_change_mean": 3.440310562533947e-05,
90 "active_fraction_tau_025": 0.6012109518051147,
91 "confirmed": false
92 },
93 "idea_sweep": [
94 {
95 "cfg": {
96 "lr": 0.006,
97 "lam": 0.0005,
98 "eta": 0.002
99 },
100 "result": {
101 "mean": 0.04891866398975253,
102 "std": 0.00862940297862839,
103 "per_seed": [
104 0.048651475459337234,
105 0.029897399246692657,
106 0.0598931610584259,
107 0.04604996740818024,
108 0.05041059851646423,
109 0.05862913280725479,
110 0.05099020525813103,
111 0.046827372163534164
112 ],
113 "n": 8
114 }
115 },
116 {
117 "cfg": {
118 "lr": 0.001,
119 "lam": 0.0005,
120 "eta": 0.002
121 },
122 "result": {
123 "mean": 0.40538693591952324,
124 "std": 0.06670983818512866,
125 "per_seed": [
126 0.42158782482147217,
127 0.29316720366477966,
128 0.4900130331516266,
129 0.44477176666259766,
130 0.43406909704208374,
131 0.47257643938064575,
132 0.3660927712917328,
133 0.32081735134124756
134 ],
135 "n": 8
136 }
137 },
138 {
139 "cfg": {
140 "lr": 0.006,
141 "lam": 0.0005,
142 "eta": 0.002
143 },
144 "result": {
145 "mean": 0.04891866398975253,
146 "std": 0.00862940297862839,
147 "per_seed": [
148 0.048651475459337234,
149 0.029897399246692657,
150 0.0598931610584259,
151 0.04604996740818024,
152 0.05041059851646423,
153 0.05862913280725479,
154 0.05099020525813103,
155 0.046827372163534164
156 ],
157 "n": 8
158 }
159 }
160 ],
161 "runtime_sec": 34.89922094345093
162}