Phase-Blind Checkpoint Scheduling / bench_report.json
Failed on benchmark
1{
2 "bench_version": 1,
3 "track": "phase_checkpoint_regression",
4 "model": "mlp_tiny",
5 "metric_direction": "lower is better",
6 "n_seeds": 8,
7 "baseline": {
8 "best_cfg": {
9 "lr": 0.01,
10 "weight_decay": 0.001,
11 "epochs": 12,
12 "sigma": 0.05
13 },
14 "sweep": [
15 {
16 "cfg": {
17 "lr": 0.001,
18 "weight_decay": 0.0,
19 "epochs": 12,
20 "sigma": 0.0
21 },
22 "mean": 0.052177092991769314
23 },
24 {
25 "cfg": {
26 "lr": 0.001,
27 "weight_decay": 0.0001,
28 "epochs": 12,
29 "sigma": 0.02
30 },
31 "mean": 0.05208037327975035
32 },
33 {
34 "cfg": {
35 "lr": 0.001,
36 "weight_decay": 0.001,
37 "epochs": 12,
38 "sigma": 0.05
39 },
40 "mean": 0.051703399047255516
41 },
42 {
43 "cfg": {
44 "lr": 0.003,
45 "weight_decay": 0.0,
46 "epochs": 12,
47 "sigma": 0.0
48 },
49 "mean": 0.03177726315334439
50 },
51 {
52 "cfg": {
53 "lr": 0.003,
54 "weight_decay": 0.0001,
55 "epochs": 12,
56 "sigma": 0.02
57 },
58 "mean": 0.03205729741603136
59 },
60 {
61 "cfg": {
62 "lr": 0.003,
63 "weight_decay": 0.001,
64 "epochs": 12,
65 "sigma": 0.05
66 },
67 "mean": 0.03150508273392916
68 },
69 {
70 "cfg": {
71 "lr": 0.01,
72 "weight_decay": 0.0,
73 "epochs": 12,
74 "sigma": 0.0
75 },
76 "mean": 0.022574170026928186
77 },
78 {
79 "cfg": {
80 "lr": 0.01,
81 "weight_decay": 0.0001,
82 "epochs": 12,
83 "sigma": 0.02
84 },
85 "mean": 0.022592097986489534
86 },
87 {
88 "cfg": {
89 "lr": 0.01,
90 "weight_decay": 0.001,
91 "epochs": 12,
92 "sigma": 0.05
93 },
94 "mean": 0.02199754724279046
95 }
96 ],
97 "full": {
98 "mean": 0.022857969626784325,
99 "std": 0.0019890920183647087,
100 "per_seed": [
101 0.022877627983689308,
102 0.02057631127536297,
103 0.02327309176325798,
104 0.021263157948851585,
105 0.0209011472761631,
106 0.026269573718309402,
107 0.02207833342254162,
108 0.025624513626098633
109 ],
110 "n": 8
111 }
112 },
113 "idea": {
114 "mean": 0.022857969626784325,
115 "std": 0.0019890920183647087,
116 "per_seed": [
117 0.022877627983689308,
118 0.02057631127536297,
119 0.02327309176325798,
120 0.021263157948851585,
121 0.0209011472761631,
122 0.026269573718309402,
123 0.02207833342254162,
124 0.025624513626098633
125 ],
126 "n": 8
127 },
128 "comparison": {
129 "delta_mean": 0.0,
130 "idea_wins": 0,
131 "n_pairs": 8,
132 "per_seed_diffs": [
133 0.0,
134 0.0,
135 0.0,
136 0.0,
137 0.0,
138 0.0,
139 0.0,
140 0.0
141 ],
142 "p_value": 1.0,
143 "mde": 0.0,
144 "mde_rel_pct": 0.0,
145 "verdict": "no measurable effect",
146 "system_worked": false
147 },
148 "mechanism_signature": {
149 "prediction": "anonymous trained-system checkpoint phase return slope = 1",
150 "predicted_slope": 1.0,
151 "observed_slope_mean": 0.04611545590999941,
152 "priority_observed_slope_mean": 1.0049859781212493,
153 "observed_gap_drift_mean": 2.667137344314341e-18,
154 "confirmed": false,
155 "custom_track": {
156 "name": "phase_checkpoint_regression",
157 "file": "phase_track.py",
158 "domain": "training-dynamics"
159 },
160 "idea_sweep": [
161 {
162 "cfg": {
163 "lr": 0.01,
164 "weight_decay": 0.001,
165 "epochs": 12,
166 "sigma": 0.05
167 },
168 "mean": 0.02199754724279046
169 },
170 {
171 "cfg": {
172 "lr": 0.001,
173 "weight_decay": 0.001,
174 "epochs": 12,
175 "sigma": 0.05
176 },
177 "mean": 0.051703399047255516
178 },
179 {
180 "cfg": {
181 "lr": 0.01,
182 "weight_decay": 0.001,
183 "epochs": 12,
184 "sigma": 0.05
185 },
186 "mean": 0.02199754724279046
187 }
188 ],
189 "selected_idea_cfg": {
190 "lr": 0.01,
191 "weight_decay": 0.001,
192 "epochs": 12,
193 "sigma": 0.05
194 }
195 }
196}