Ablation B-MID-B1_all16
Browse files
B/MID/B1_all16/seed0/final_report.json
ADDED
|
@@ -0,0 +1,220 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"config": {
|
| 3 |
+
"group": "B",
|
| 4 |
+
"variant": "B1_all16",
|
| 5 |
+
"band": "MID",
|
| 6 |
+
"seed": 0,
|
| 7 |
+
"phase": 1,
|
| 8 |
+
"overrides": {
|
| 9 |
+
"noise_types": [
|
| 10 |
+
0,
|
| 11 |
+
1,
|
| 12 |
+
2,
|
| 13 |
+
3,
|
| 14 |
+
4,
|
| 15 |
+
5,
|
| 16 |
+
6,
|
| 17 |
+
7,
|
| 18 |
+
8,
|
| 19 |
+
9,
|
| 20 |
+
10,
|
| 21 |
+
11,
|
| 22 |
+
12,
|
| 23 |
+
13,
|
| 24 |
+
14,
|
| 25 |
+
15
|
| 26 |
+
]
|
| 27 |
+
},
|
| 28 |
+
"description": "B-MID-B1_all16"
|
| 29 |
+
},
|
| 30 |
+
"run_config": {
|
| 31 |
+
"matrix_v": 64,
|
| 32 |
+
"D": 8,
|
| 33 |
+
"patch_size": 16,
|
| 34 |
+
"hidden": 64,
|
| 35 |
+
"depth": 1,
|
| 36 |
+
"n_cross_layers": 1,
|
| 37 |
+
"n_heads": 4,
|
| 38 |
+
"max_alpha": 0.2,
|
| 39 |
+
"alpha_init": -2.0,
|
| 40 |
+
"img_size": 64,
|
| 41 |
+
"batch_size": 128,
|
| 42 |
+
"lr": 0.0001,
|
| 43 |
+
"epochs": 1,
|
| 44 |
+
"weight_decay": 0.0,
|
| 45 |
+
"use_cv_ema": true,
|
| 46 |
+
"cv_ema_alpha": 0.01,
|
| 47 |
+
"cv_alignment_epochs": 0,
|
| 48 |
+
"cv_measure_every": 50,
|
| 49 |
+
"cv_sigma_scale": 0.3,
|
| 50 |
+
"boost": 0.5,
|
| 51 |
+
"cross_attn_clip": 0.5,
|
| 52 |
+
"allowed_types": [
|
| 53 |
+
0,
|
| 54 |
+
1,
|
| 55 |
+
2,
|
| 56 |
+
3,
|
| 57 |
+
4,
|
| 58 |
+
5,
|
| 59 |
+
6,
|
| 60 |
+
7,
|
| 61 |
+
8,
|
| 62 |
+
9,
|
| 63 |
+
10,
|
| 64 |
+
11,
|
| 65 |
+
12,
|
| 66 |
+
13,
|
| 67 |
+
14,
|
| 68 |
+
15
|
| 69 |
+
],
|
| 70 |
+
"train_size": 1000000,
|
| 71 |
+
"val_size": 10000,
|
| 72 |
+
"num_workers": 2,
|
| 73 |
+
"report_every": 100,
|
| 74 |
+
"major_report_every": 10,
|
| 75 |
+
"save_every": 5,
|
| 76 |
+
"seed": 0,
|
| 77 |
+
"hf_repo": "AbstractPhil/geolip-svae-batteries",
|
| 78 |
+
"upload": false
|
| 79 |
+
},
|
| 80 |
+
"cv_ema_final": 0.3800135674409466,
|
| 81 |
+
"cv_last": 0.35093146837857997,
|
| 82 |
+
"predicted_band": "MID",
|
| 83 |
+
"expected_band": "MID",
|
| 84 |
+
"band_match": true,
|
| 85 |
+
"test_mse": 1.0088096857070923,
|
| 86 |
+
"recon_ema": 1.011015446463103,
|
| 87 |
+
"S0": 3.4625773429870605,
|
| 88 |
+
"SD": 2.159618377685547,
|
| 89 |
+
"ratio": 1.6033283300101409,
|
| 90 |
+
"erank": 7.909207344055176,
|
| 91 |
+
"observed_sphere_cv": 0.3528059752616081,
|
| 92 |
+
"uniform_sphere_cv_prediction": 0.35679185069620456,
|
| 93 |
+
"band_deviation": -0.003985875434596442,
|
| 94 |
+
"params_count": 183339,
|
| 95 |
+
"wallclock_seconds": 34.621251344680786,
|
| 96 |
+
"batches_completed": 1000,
|
| 97 |
+
"batch_limit": 1000,
|
| 98 |
+
"cv_trajectory": [
|
| 99 |
+
{
|
| 100 |
+
"batch": 0,
|
| 101 |
+
"cv": 0.3849132290425209,
|
| 102 |
+
"cv_ema": 0.3849132290425209,
|
| 103 |
+
"recon": 1.9087696075439453
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"batch": 50,
|
| 107 |
+
"cv": 0.3745136336443136,
|
| 108 |
+
"cv_ema": 0.3848092330885388,
|
| 109 |
+
"recon": 1.678182601928711
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"batch": 100,
|
| 113 |
+
"cv": 0.34874661183265016,
|
| 114 |
+
"cv_ema": 0.3844486068759799,
|
| 115 |
+
"recon": 1.4744038581848145
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"batch": 150,
|
| 119 |
+
"cv": 0.3284531917250024,
|
| 120 |
+
"cv_ema": 0.38388865272447015,
|
| 121 |
+
"recon": 1.3674907684326172
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"batch": 200,
|
| 125 |
+
"cv": 0.3400946494077929,
|
| 126 |
+
"cv_ema": 0.3834507126913034,
|
| 127 |
+
"recon": 1.0928411483764648
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"batch": 250,
|
| 131 |
+
"cv": 0.3561678506521022,
|
| 132 |
+
"cv_ema": 0.38317788407091136,
|
| 133 |
+
"recon": 1.3706424236297607
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"batch": 300,
|
| 137 |
+
"cv": 0.33828053215051107,
|
| 138 |
+
"cv_ema": 0.3827289105517073,
|
| 139 |
+
"recon": 1.0216691493988037
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"batch": 350,
|
| 143 |
+
"cv": 0.3847852656474189,
|
| 144 |
+
"cv_ema": 0.38274947410266447,
|
| 145 |
+
"recon": 1.0424699783325195
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"batch": 400,
|
| 149 |
+
"cv": 0.34372423034418637,
|
| 150 |
+
"cv_ema": 0.3823592216650797,
|
| 151 |
+
"recon": 1.0725793838500977
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"batch": 450,
|
| 155 |
+
"cv": 0.3442490202797809,
|
| 156 |
+
"cv_ema": 0.3819781196512267,
|
| 157 |
+
"recon": 1.0332896709442139
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"batch": 500,
|
| 161 |
+
"cv": 0.37312434462058813,
|
| 162 |
+
"cv_ema": 0.3818895819009203,
|
| 163 |
+
"recon": 0.9648735523223877
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"batch": 550,
|
| 167 |
+
"cv": 0.3680608454062403,
|
| 168 |
+
"cv_ema": 0.38175129453597356,
|
| 169 |
+
"recon": 1.032564640045166
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"batch": 600,
|
| 173 |
+
"cv": 0.3356815605364401,
|
| 174 |
+
"cv_ema": 0.3812905971959782,
|
| 175 |
+
"recon": 0.9076928496360779
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"batch": 650,
|
| 179 |
+
"cv": 0.3771558223567731,
|
| 180 |
+
"cv_ema": 0.38124924944758615,
|
| 181 |
+
"recon": 1.2254929542541504
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"batch": 700,
|
| 185 |
+
"cv": 0.36075011579016875,
|
| 186 |
+
"cv_ema": 0.381044258111012,
|
| 187 |
+
"recon": 1.06052827835083
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"batch": 750,
|
| 191 |
+
"cv": 0.36491354574328816,
|
| 192 |
+
"cv_ema": 0.38088295098733477,
|
| 193 |
+
"recon": 1.0862634181976318
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"batch": 800,
|
| 197 |
+
"cv": 0.35994882772489956,
|
| 198 |
+
"cv_ema": 0.38067360975471043,
|
| 199 |
+
"recon": 1.073961615562439
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"batch": 850,
|
| 203 |
+
"cv": 0.3701532342749335,
|
| 204 |
+
"cv_ema": 0.38056840599991265,
|
| 205 |
+
"recon": 1.1551573276519775
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"batch": 900,
|
| 209 |
+
"cv": 0.3544604077420659,
|
| 210 |
+
"cv_ema": 0.38030732601733414,
|
| 211 |
+
"recon": 0.9704558253288269
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"batch": 950,
|
| 215 |
+
"cv": 0.35093146837857997,
|
| 216 |
+
"cv_ema": 0.3800135674409466,
|
| 217 |
+
"recon": 1.1800693273544312
|
| 218 |
+
}
|
| 219 |
+
]
|
| 220 |
+
}
|