Upload image_aleph_64/final_report.json with huggingface_hub
Browse files- image_aleph_64/final_report.json +329 -0
image_aleph_64/final_report.json
ADDED
|
@@ -0,0 +1,329 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run_name": "image_aleph_64_tau0.1_lr0.001",
|
| 3 |
+
"config": {
|
| 4 |
+
"model_type": "aleph",
|
| 5 |
+
"V": 32,
|
| 6 |
+
"D": 4,
|
| 7 |
+
"ps": 4,
|
| 8 |
+
"hidden": 64,
|
| 9 |
+
"depth": 1,
|
| 10 |
+
"channels": 3,
|
| 11 |
+
"address": "hard",
|
| 12 |
+
"K": 64,
|
| 13 |
+
"address_tau": 0.1,
|
| 14 |
+
"address_chunk": null,
|
| 15 |
+
"codebook_init": "random",
|
| 16 |
+
"freeze_codebook": false,
|
| 17 |
+
"decode_mode": "tied",
|
| 18 |
+
"n_atoms": 64,
|
| 19 |
+
"code_tau": 1.0,
|
| 20 |
+
"readout": "linear",
|
| 21 |
+
"dec_hidden": 64,
|
| 22 |
+
"dec_depth": 1,
|
| 23 |
+
"row_norm": "sphere",
|
| 24 |
+
"n_cross": 0,
|
| 25 |
+
"n_heads": 2,
|
| 26 |
+
"smooth_mid": 8,
|
| 27 |
+
"boundary_smooth": true,
|
| 28 |
+
"activation": "gelu",
|
| 29 |
+
"init_scheme": "orthogonal"
|
| 30 |
+
},
|
| 31 |
+
"n_params": 26795,
|
| 32 |
+
"dataset": "tiny_imagenet",
|
| 33 |
+
"loss_mode": "cosine_mse",
|
| 34 |
+
"baseline_mse": 8e-05,
|
| 35 |
+
"best_test_mse": 0.014638923667371273,
|
| 36 |
+
"best_test_cos": 0.9916898846626282,
|
| 37 |
+
"history": [
|
| 38 |
+
[
|
| 39 |
+
100,
|
| 40 |
+
0.1338951736688614,
|
| 41 |
+
0.08452290445566177,
|
| 42 |
+
0.9518959522247314,
|
| 43 |
+
0.10500592291355133,
|
| 44 |
+
123.4452896118164,
|
| 45 |
+
0.21258018612861634,
|
| 46 |
+
113.29344635009765
|
| 47 |
+
],
|
| 48 |
+
[
|
| 49 |
+
200,
|
| 50 |
+
0.06699845939874649,
|
| 51 |
+
0.04208567440509796,
|
| 52 |
+
0.9765878796577454,
|
| 53 |
+
0.09970508962869644,
|
| 54 |
+
125.20772857666016,
|
| 55 |
+
0.215412175655365,
|
| 56 |
+
118.10545654296875
|
| 57 |
+
],
|
| 58 |
+
[
|
| 59 |
+
300,
|
| 60 |
+
0.047444701194763184,
|
| 61 |
+
0.030161230266094206,
|
| 62 |
+
0.9830520510673523,
|
| 63 |
+
0.09908935725688935,
|
| 64 |
+
125.13959655761718,
|
| 65 |
+
0.2190625101327896,
|
| 66 |
+
118.56011810302735
|
| 67 |
+
],
|
| 68 |
+
[
|
| 69 |
+
400,
|
| 70 |
+
0.03959755599498749,
|
| 71 |
+
0.025340546667575837,
|
| 72 |
+
0.9857062101364136,
|
| 73 |
+
0.10000364780426026,
|
| 74 |
+
124.9797607421875,
|
| 75 |
+
0.22069571614265443,
|
| 76 |
+
119.28466491699218
|
| 77 |
+
],
|
| 78 |
+
[
|
| 79 |
+
500,
|
| 80 |
+
0.03589203208684921,
|
| 81 |
+
0.02247673235833645,
|
| 82 |
+
0.9872919678688049,
|
| 83 |
+
0.10104079693555831,
|
| 84 |
+
124.96162109375,
|
| 85 |
+
0.22188216745853423,
|
| 86 |
+
120.01019592285157
|
| 87 |
+
],
|
| 88 |
+
[
|
| 89 |
+
600,
|
| 90 |
+
0.03314971551299095,
|
| 91 |
+
0.020569656789302827,
|
| 92 |
+
0.9883546829223633,
|
| 93 |
+
0.10165663361549378,
|
| 94 |
+
125.01497955322266,
|
| 95 |
+
0.2226852923631668,
|
| 96 |
+
120.4223388671875
|
| 97 |
+
],
|
| 98 |
+
[
|
| 99 |
+
700,
|
| 100 |
+
0.03013969026505947,
|
| 101 |
+
0.0191864050924778,
|
| 102 |
+
0.9891366839408875,
|
| 103 |
+
0.10278950482606888,
|
| 104 |
+
125.14998321533203,
|
| 105 |
+
0.22321296036243438,
|
| 106 |
+
120.78898468017579
|
| 107 |
+
],
|
| 108 |
+
[
|
| 109 |
+
800,
|
| 110 |
+
0.02819688618183136,
|
| 111 |
+
0.01822369061410427,
|
| 112 |
+
0.9896759390830994,
|
| 113 |
+
0.10427476167678833,
|
| 114 |
+
125.24827880859375,
|
| 115 |
+
0.22356731593608856,
|
| 116 |
+
120.98842315673828
|
| 117 |
+
],
|
| 118 |
+
[
|
| 119 |
+
900,
|
| 120 |
+
0.027659602463245392,
|
| 121 |
+
0.017599988356232645,
|
| 122 |
+
0.9900658011436463,
|
| 123 |
+
0.10600353330373764,
|
| 124 |
+
125.37111358642578,
|
| 125 |
+
0.22393932938575745,
|
| 126 |
+
121.10140228271484
|
| 127 |
+
],
|
| 128 |
+
[
|
| 129 |
+
1000,
|
| 130 |
+
0.026516078040003777,
|
| 131 |
+
0.016981299221515655,
|
| 132 |
+
0.9903982281684875,
|
| 133 |
+
0.10756459981203079,
|
| 134 |
+
125.53051300048828,
|
| 135 |
+
0.22424977421760559,
|
| 136 |
+
121.30896759033203
|
| 137 |
+
],
|
| 138 |
+
[
|
| 139 |
+
1100,
|
| 140 |
+
0.0262162946164608,
|
| 141 |
+
0.016515411809086798,
|
| 142 |
+
0.9906368732452393,
|
| 143 |
+
0.1084539532661438,
|
| 144 |
+
125.64454650878906,
|
| 145 |
+
0.2245349556207657,
|
| 146 |
+
121.44316253662109
|
| 147 |
+
],
|
| 148 |
+
[
|
| 149 |
+
1200,
|
| 150 |
+
0.025252725929021835,
|
| 151 |
+
0.016151749342679978,
|
| 152 |
+
0.9908600807189941,
|
| 153 |
+
0.10920335203409196,
|
| 154 |
+
125.7305679321289,
|
| 155 |
+
0.22477577924728392,
|
| 156 |
+
121.56105041503906
|
| 157 |
+
],
|
| 158 |
+
[
|
| 159 |
+
1300,
|
| 160 |
+
0.025090187788009644,
|
| 161 |
+
0.015863315761089326,
|
| 162 |
+
0.9910096287727356,
|
| 163 |
+
0.1098355457186699,
|
| 164 |
+
125.77598114013672,
|
| 165 |
+
0.22491119503974916,
|
| 166 |
+
121.61234588623047
|
| 167 |
+
],
|
| 168 |
+
[
|
| 169 |
+
1400,
|
| 170 |
+
0.02470283955335617,
|
| 171 |
+
0.01564104128628969,
|
| 172 |
+
0.9911393761634827,
|
| 173 |
+
0.11034584492444992,
|
| 174 |
+
125.82076721191406,
|
| 175 |
+
0.2249755382537842,
|
| 176 |
+
121.68078918457032
|
| 177 |
+
],
|
| 178 |
+
[
|
| 179 |
+
1500,
|
| 180 |
+
0.02436356619000435,
|
| 181 |
+
0.015438761748373508,
|
| 182 |
+
0.9912354111671448,
|
| 183 |
+
0.11091977655887604,
|
| 184 |
+
125.84778289794922,
|
| 185 |
+
0.22501077651977539,
|
| 186 |
+
121.69746856689453
|
| 187 |
+
],
|
| 188 |
+
[
|
| 189 |
+
1600,
|
| 190 |
+
0.024404313415288925,
|
| 191 |
+
0.015304045379161834,
|
| 192 |
+
0.9913319706916809,
|
| 193 |
+
0.11115458011627197,
|
| 194 |
+
125.89227294921875,
|
| 195 |
+
0.22516761720180511,
|
| 196 |
+
121.75006561279297
|
| 197 |
+
],
|
| 198 |
+
[
|
| 199 |
+
1700,
|
| 200 |
+
0.02390819787979126,
|
| 201 |
+
0.015172572433948516,
|
| 202 |
+
0.9913990139961243,
|
| 203 |
+
0.1111273467540741,
|
| 204 |
+
125.91994934082031,
|
| 205 |
+
0.22514898478984832,
|
| 206 |
+
121.74757385253906
|
| 207 |
+
],
|
| 208 |
+
[
|
| 209 |
+
1800,
|
| 210 |
+
0.023156777024269104,
|
| 211 |
+
0.015060418471693992,
|
| 212 |
+
0.9914594888687134,
|
| 213 |
+
0.11132417470216752,
|
| 214 |
+
125.9406005859375,
|
| 215 |
+
0.2252104848623276,
|
| 216 |
+
121.75085144042968
|
| 217 |
+
],
|
| 218 |
+
[
|
| 219 |
+
1900,
|
| 220 |
+
0.02347385138273239,
|
| 221 |
+
0.01496257595717907,
|
| 222 |
+
0.9915189385414124,
|
| 223 |
+
0.11138139814138412,
|
| 224 |
+
125.96607666015625,
|
| 225 |
+
0.225267294049263,
|
| 226 |
+
121.77407531738281
|
| 227 |
+
],
|
| 228 |
+
[
|
| 229 |
+
2000,
|
| 230 |
+
0.02340111881494522,
|
| 231 |
+
0.014874392934143544,
|
| 232 |
+
0.9915602684020997,
|
| 233 |
+
0.11145293861627578,
|
| 234 |
+
125.97136535644532,
|
| 235 |
+
0.2252195507287979,
|
| 236 |
+
121.75076446533203
|
| 237 |
+
],
|
| 238 |
+
[
|
| 239 |
+
2100,
|
| 240 |
+
0.023378636687994003,
|
| 241 |
+
0.014799886383116245,
|
| 242 |
+
0.9916011095046997,
|
| 243 |
+
0.11154815107584,
|
| 244 |
+
125.9762939453125,
|
| 245 |
+
0.22524904906749726,
|
| 246 |
+
121.7263198852539
|
| 247 |
+
],
|
| 248 |
+
[
|
| 249 |
+
2200,
|
| 250 |
+
0.02310825139284134,
|
| 251 |
+
0.014753679558634759,
|
| 252 |
+
0.9916336536407471,
|
| 253 |
+
0.11146874725818634,
|
| 254 |
+
125.98635711669922,
|
| 255 |
+
0.22525244057178498,
|
| 256 |
+
121.73275299072266
|
| 257 |
+
],
|
| 258 |
+
[
|
| 259 |
+
2300,
|
| 260 |
+
0.023118706420063972,
|
| 261 |
+
0.01471103522926569,
|
| 262 |
+
0.9916516542434692,
|
| 263 |
+
0.1114861473441124,
|
| 264 |
+
125.98931274414062,
|
| 265 |
+
0.2252629965543747,
|
| 266 |
+
121.7216567993164
|
| 267 |
+
],
|
| 268 |
+
[
|
| 269 |
+
2400,
|
| 270 |
+
0.022946298122406006,
|
| 271 |
+
0.014683219790458679,
|
| 272 |
+
0.9916661739349365,
|
| 273 |
+
0.11152514070272446,
|
| 274 |
+
125.99423675537109,
|
| 275 |
+
0.2252607136964798,
|
| 276 |
+
121.71349334716797
|
| 277 |
+
],
|
| 278 |
+
[
|
| 279 |
+
2500,
|
| 280 |
+
0.02309560962021351,
|
| 281 |
+
0.014659379236400127,
|
| 282 |
+
0.9916785836219788,
|
| 283 |
+
0.11155717521905899,
|
| 284 |
+
125.98986206054687,
|
| 285 |
+
0.2252570539712906,
|
| 286 |
+
121.69463806152343
|
| 287 |
+
],
|
| 288 |
+
[
|
| 289 |
+
2600,
|
| 290 |
+
0.02344914898276329,
|
| 291 |
+
0.014649365097284317,
|
| 292 |
+
0.991686487197876,
|
| 293 |
+
0.1115261435508728,
|
| 294 |
+
125.99332122802734,
|
| 295 |
+
0.22526324093341826,
|
| 296 |
+
121.7011489868164
|
| 297 |
+
],
|
| 298 |
+
[
|
| 299 |
+
2700,
|
| 300 |
+
0.022823315113782883,
|
| 301 |
+
0.014641879685223102,
|
| 302 |
+
0.9916890978813171,
|
| 303 |
+
0.11155008971691131,
|
| 304 |
+
125.99207305908203,
|
| 305 |
+
0.2252647489309311,
|
| 306 |
+
121.69249420166015
|
| 307 |
+
],
|
| 308 |
+
[
|
| 309 |
+
2800,
|
| 310 |
+
0.02279219776391983,
|
| 311 |
+
0.014638923667371273,
|
| 312 |
+
0.9916898846626282,
|
| 313 |
+
0.11154438704252242,
|
| 314 |
+
125.99264984130859,
|
| 315 |
+
0.22526106238365173,
|
| 316 |
+
121.69049987792968
|
| 317 |
+
]
|
| 318 |
+
],
|
| 319 |
+
"history_columns": [
|
| 320 |
+
"step",
|
| 321 |
+
"train_loss",
|
| 322 |
+
"test_mse",
|
| 323 |
+
"test_cos",
|
| 324 |
+
"test_cv",
|
| 325 |
+
"perplexity",
|
| 326 |
+
"address_margin",
|
| 327 |
+
"hard_perplexity"
|
| 328 |
+
]
|
| 329 |
+
}
|