Ablation F-LOW-F3_silu
Browse files
F/LOW/F3_silu/seed0/final_report.json
ADDED
|
@@ -0,0 +1,203 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"config": {
|
| 3 |
+
"group": "F",
|
| 4 |
+
"variant": "F3_silu",
|
| 5 |
+
"band": "LOW",
|
| 6 |
+
"seed": 0,
|
| 7 |
+
"phase": 1,
|
| 8 |
+
"overrides": {
|
| 9 |
+
"activation": "silu"
|
| 10 |
+
},
|
| 11 |
+
"description": "F-LOW-F3_silu"
|
| 12 |
+
},
|
| 13 |
+
"run_config": {
|
| 14 |
+
"matrix_v": 64,
|
| 15 |
+
"D": 16,
|
| 16 |
+
"patch_size": 16,
|
| 17 |
+
"hidden": 64,
|
| 18 |
+
"depth": 1,
|
| 19 |
+
"n_cross_layers": 1,
|
| 20 |
+
"n_heads": 4,
|
| 21 |
+
"max_alpha": 0.2,
|
| 22 |
+
"alpha_init": -2.0,
|
| 23 |
+
"img_size": 64,
|
| 24 |
+
"batch_size": 128,
|
| 25 |
+
"lr": 0.0001,
|
| 26 |
+
"epochs": 1,
|
| 27 |
+
"weight_decay": 0.0,
|
| 28 |
+
"use_cv_ema": true,
|
| 29 |
+
"cv_ema_alpha": 0.01,
|
| 30 |
+
"cv_alignment_epochs": 0,
|
| 31 |
+
"cv_measure_every": 50,
|
| 32 |
+
"cv_sigma_scale": 0.3,
|
| 33 |
+
"boost": 0.5,
|
| 34 |
+
"cross_attn_clip": 0.5,
|
| 35 |
+
"allowed_types": [
|
| 36 |
+
0,
|
| 37 |
+
1,
|
| 38 |
+
2,
|
| 39 |
+
3,
|
| 40 |
+
4,
|
| 41 |
+
5,
|
| 42 |
+
6,
|
| 43 |
+
7,
|
| 44 |
+
8,
|
| 45 |
+
9,
|
| 46 |
+
10,
|
| 47 |
+
11,
|
| 48 |
+
12,
|
| 49 |
+
13,
|
| 50 |
+
14,
|
| 51 |
+
15
|
| 52 |
+
],
|
| 53 |
+
"train_size": 1000000,
|
| 54 |
+
"val_size": 10000,
|
| 55 |
+
"num_workers": 2,
|
| 56 |
+
"report_every": 100,
|
| 57 |
+
"major_report_every": 10,
|
| 58 |
+
"save_every": 5,
|
| 59 |
+
"seed": 0,
|
| 60 |
+
"hf_repo": "AbstractPhil/geolip-svae-batteries",
|
| 61 |
+
"upload": false
|
| 62 |
+
},
|
| 63 |
+
"cv_ema_final": 0.20382151527944323,
|
| 64 |
+
"cv_last": 0.20416849957081348,
|
| 65 |
+
"predicted_band": "LOW",
|
| 66 |
+
"expected_band": "LOW",
|
| 67 |
+
"band_match": true,
|
| 68 |
+
"test_mse": 1.0063366889953613,
|
| 69 |
+
"recon_ema": 1.0107745895523859,
|
| 70 |
+
"S0": 2.8565118312835693,
|
| 71 |
+
"SD": 1.1511564254760742,
|
| 72 |
+
"ratio": 2.4814280172982968,
|
| 73 |
+
"erank": 15.503007888793945,
|
| 74 |
+
"observed_sphere_cv": 0.19710067994169841,
|
| 75 |
+
"uniform_sphere_cv_prediction": 0.19902003258265127,
|
| 76 |
+
"band_deviation": -0.0019193526409528594,
|
| 77 |
+
"params_count": 250211,
|
| 78 |
+
"wallclock_seconds": 14.28084397315979,
|
| 79 |
+
"batches_completed": 1000,
|
| 80 |
+
"batch_limit": 1000,
|
| 81 |
+
"cv_trajectory": [
|
| 82 |
+
{
|
| 83 |
+
"batch": 0,
|
| 84 |
+
"cv": 0.20434628907067365,
|
| 85 |
+
"cv_ema": 0.20434628907067365,
|
| 86 |
+
"recon": 1.630874514579773
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"batch": 50,
|
| 90 |
+
"cv": 0.20543441937981918,
|
| 91 |
+
"cv_ema": 0.2043571703737651,
|
| 92 |
+
"recon": 1.4734901189804077
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"batch": 100,
|
| 96 |
+
"cv": 0.21939389557560077,
|
| 97 |
+
"cv_ema": 0.20450753762578347,
|
| 98 |
+
"recon": 1.3521608114242554
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"batch": 150,
|
| 102 |
+
"cv": 0.2067776366685967,
|
| 103 |
+
"cv_ema": 0.2045302386162116,
|
| 104 |
+
"recon": 1.4585273265838623
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"batch": 200,
|
| 108 |
+
"cv": 0.1977144861916051,
|
| 109 |
+
"cv_ema": 0.20446208109196554,
|
| 110 |
+
"recon": 1.1851074695587158
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"batch": 250,
|
| 114 |
+
"cv": 0.19807141723367228,
|
| 115 |
+
"cv_ema": 0.2043981744533826,
|
| 116 |
+
"recon": 1.1194274425506592
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"batch": 300,
|
| 120 |
+
"cv": 0.19455848723140648,
|
| 121 |
+
"cv_ema": 0.20429977758116283,
|
| 122 |
+
"recon": 1.0005848407745361
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"batch": 350,
|
| 126 |
+
"cv": 0.20415494995937955,
|
| 127 |
+
"cv_ema": 0.204298329304945,
|
| 128 |
+
"recon": 0.9685860872268677
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"batch": 400,
|
| 132 |
+
"cv": 0.19137346063318744,
|
| 133 |
+
"cv_ema": 0.20416908061822742,
|
| 134 |
+
"recon": 0.9788789749145508
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"batch": 450,
|
| 138 |
+
"cv": 0.20160031830585892,
|
| 139 |
+
"cv_ema": 0.20414339299510373,
|
| 140 |
+
"recon": 1.1313635110855103
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"batch": 500,
|
| 144 |
+
"cv": 0.19556119925563253,
|
| 145 |
+
"cv_ema": 0.20405757105770903,
|
| 146 |
+
"recon": 1.0095603466033936
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"batch": 550,
|
| 150 |
+
"cv": 0.2011029237401457,
|
| 151 |
+
"cv_ema": 0.2040280245845334,
|
| 152 |
+
"recon": 1.059775710105896
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"batch": 600,
|
| 156 |
+
"cv": 0.20200374899554538,
|
| 157 |
+
"cv_ema": 0.20400778182864351,
|
| 158 |
+
"recon": 1.0891534090042114
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"batch": 650,
|
| 162 |
+
"cv": 0.18572336274912737,
|
| 163 |
+
"cv_ema": 0.20382493763784834,
|
| 164 |
+
"recon": 1.019309163093567
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"batch": 700,
|
| 168 |
+
"cv": 0.20764660850186265,
|
| 169 |
+
"cv_ema": 0.2038631543464885,
|
| 170 |
+
"recon": 0.9619902968406677
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"batch": 750,
|
| 174 |
+
"cv": 0.19332388869412978,
|
| 175 |
+
"cv_ema": 0.2037577616899649,
|
| 176 |
+
"recon": 1.077964186668396
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"batch": 800,
|
| 180 |
+
"cv": 0.21176811611137902,
|
| 181 |
+
"cv_ema": 0.20383786523417904,
|
| 182 |
+
"recon": 0.9218326807022095
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"batch": 850,
|
| 186 |
+
"cv": 0.2171871501660724,
|
| 187 |
+
"cv_ema": 0.20397135808349798,
|
| 188 |
+
"recon": 0.8180744051933289
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"batch": 900,
|
| 192 |
+
"cv": 0.1886365884948191,
|
| 193 |
+
"cv_ema": 0.2038180103876112,
|
| 194 |
+
"recon": 0.8697931170463562
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"batch": 950,
|
| 198 |
+
"cv": 0.20416849957081348,
|
| 199 |
+
"cv_ema": 0.20382151527944323,
|
| 200 |
+
"recon": 0.9699296951293945
|
| 201 |
+
}
|
| 202 |
+
]
|
| 203 |
+
}
|