Critical Cross-Layer Weight Sharing / results.json
Mechanism confirmed, baseline not beaten
1{
2 "seed": 2716,
3 "depths": [
4 32,
5 64,
6 128,
7 256,
8 512,
9 1024,
10 2048,
11 4096,
12 8192,
13 16384,
14 32768
15 ],
16 "formula_sweeps": [
17 {
18 "gamma": 0.3,
19 "s1_slope_observed": 0.7055120547793522,
20 "s1_slope_predicted": 0.7,
21 "s2_slope_observed": 0.4259993229498566,
22 "s2_slope_predicted": 0.4,
23 "s1_at_final": 2067.9099490388685,
24 "s2_at_final": 158.0483151112958,
25 "s1_tail_slope_observed": 0.7008663230515949,
26 "s2_tail_slope_observed": 0.40906193432470234,
27 "s2_normalized_final": 2.4695049236139965,
28 "gaussian_q2_empirical_L128": 21493.061717758712,
29 "gaussian_q2_exact_L128": 21558.605545692786,
30 "gaussian_excess_per_2L": 19.213302912862446
31 },
32 {
33 "gamma": 0.5,
34 "s1_slope_observed": 0.5160477810910512,
35 "s1_slope_predicted": 0.5,
36 "s2_slope_observed": 0.14037684743002116,
37 "s2_slope_predicted": 0.0,
38 "s1_at_final": 360.58107958754226,
39 "s2_at_final": 10.974438632012166,
40 "s1_tail_slope_observed": 0.504317063981964,
41 "s2_tail_slope_observed": 0.10491115550690087,
42 "s2_normalized_final": null,
43 "gaussian_q2_empirical_L128": 18297.134124307093,
44 "gaussian_q2_exact_L128": 18419.503899776013,
45 "gaussian_excess_per_2L": 6.951187108500051
46 },
47 {
48 "gamma": 0.75,
49 "s1_slope_observed": 0.30145542420151533,
50 "s1_slope_predicted": 0.25,
51 "s2_slope_observed": 0.017733224648378922,
52 "s2_slope_predicted": 0.0,
53 "s1_at_final": 50.37629048548575,
54 "s2_at_final": 2.6013268895225026,
55 "s1_tail_slope_observed": 0.27539738774215705,
56 "s2_tail_slope_observed": 0.004570156281511213,
57 "s2_normalized_final": null,
58 "gaussian_q2_empirical_L128": 17273.024169617125,
59 "gaussian_q2_exact_L128": 17300.10180338721,
60 "gaussian_excess_per_2L": 2.5785226694812877
61 },
62 {
63 "gamma": 1.0,
64 "s1_slope_observed": 0.1403768474300212,
65 "s1_slope_predicted": 0.0,
66 "s2_slope_observed": 0.001985491711703831,
67 "s2_slope_predicted": 0.0,
68 "s1_at_final": 10.974438632012166,
69 "s2_at_final": 1.6449035497357587,
70 "s1_tail_slope_observed": 0.10491115550690087,
71 "s2_tail_slope_observed": 9.634683849950251e-05,
72 "s2_normalized_final": null,
73 "gaussian_q2_empirical_L128": 16891.13941232137,
74 "gaussian_q2_exact_L128": 16951.037846186147,
75 "gaussian_excess_per_2L": 1.2149915866646381
76 },
77 {
78 "gamma": 1.25,
79 "s1_slope_observed": 0.052641441487315116,
80 "s1_slope_predicted": 0.0,
81 "s2_slope_observed": 0.00024345211321796616,
82 "s2_slope_predicted": 0.0,
83 "s1_at_final": 4.297811181201336,
84 "s2_at_final": 1.341487144861894,
85 "s1_tail_slope_observed": 0.02574985464263504,
86 "s2_tail_slope_observed": 1.761675772098138e-06,
87 "s2_normalized_final": null,
88 "gaussian_q2_empirical_L128": 16934.23825275766,
89 "gaussian_q2_exact_L128": 16810.227495399868,
90 "gaussian_excess_per_2L": 0.6649511539057329
91 }
92 ],
93 "critical_log_checks": [
94 {
95 "gamma": 0.5,
96 "quantity": "S2",
97 "log_fit_slope": 0.14037684743002116,
98 "endpoints": [
99 4.05849519543652,
100 10.974438632012166
101 ],
102 "log_ratio_start": 1.1710341783855904,
103 "log_ratio_end": 1.055517879396281
104 },
105 {
106 "gamma": 1.0,
107 "quantity": "S1",
108 "log_fit_slope": 0.1403768474300212,
109 "endpoints": [
110 4.05849519543652,
111 10.974438632012166
112 ],
113 "log_ratio_start": 1.1710341783855904,
114 "log_ratio_end": 1.055517879396281
115 }
116 ],
117 "network_jacobian_sanity": [
118 {
119 "depth": 16,
120 "iid": {
121 "median_sv": 0.9992318380178953,
122 "max_sv": 1.3548490695423971,
123 "normalized_gradient_fourth": 1.0062354085612784
124 },
125 "powerlaw_gamma_0.3": {
126 "median_sv": 1.0124693364158148,
127 "max_sv": 2.5138964582934546,
128 "normalized_gradient_fourth": 1.0827352203586673
129 }
130 },
131 {
132 "depth": 32,
133 "iid": {
134 "median_sv": 0.9982063276567298,
135 "max_sv": 1.3644839170724725,
136 "normalized_gradient_fourth": 1.0070338422749523
137 },
138 "powerlaw_gamma_0.3": {
139 "median_sv": 0.9877556861745743,
140 "max_sv": 2.888992305970983,
141 "normalized_gradient_fourth": 1.1146772071572455
142 }
143 },
144 {
145 "depth": 64,
146 "iid": {
147 "median_sv": 0.9949002849902472,
148 "max_sv": 1.3586440681930814,
149 "normalized_gradient_fourth": 1.007257281950197
150 },
151 "powerlaw_gamma_0.3": {
152 "median_sv": 0.9839729015509657,
153 "max_sv": 3.967246737588677,
154 "normalized_gradient_fourth": 1.1691086453482293
155 }
156 },
157 {
158 "depth": 128,
159 "iid": {
160 "median_sv": 0.9938653952485531,
161 "max_sv": 1.3566726344044833,
162 "normalized_gradient_fourth": 1.007058792334494
163 },
164 "powerlaw_gamma_0.3": {
165 "median_sv": 1.0114593042006408,
166 "max_sv": 5.327694756783201,
167 "normalized_gradient_fourth": 1.2286738627508607
168 }
169 }
170 ],
171 "notes": "S1=sum_{t=1}^L t^(-gamma), S2=sum c_t^2. Slopes use finite-depth log-log regression. Gaussian check uses exact Wick formula."
172}