Add v5 results
Browse files- results_v5.json +351 -0
results_v5.json
ADDED
|
@@ -0,0 +1,351 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"main_tasks": {
|
| 3 |
+
"Complex Fn (4D)": {
|
| 4 |
+
"Vanilla": {
|
| 5 |
+
"mean": 0.05014639471968015,
|
| 6 |
+
"std": 0.00929074178664744,
|
| 7 |
+
"scores": [
|
| 8 |
+
0.0560552217066288,
|
| 9 |
+
0.057355206459760666,
|
| 10 |
+
0.037028755992650986
|
| 11 |
+
],
|
| 12 |
+
"params": 4993,
|
| 13 |
+
"hidden": 48
|
| 14 |
+
},
|
| 15 |
+
"RichV1": {
|
| 16 |
+
"mean": 0.17897194375594458,
|
| 17 |
+
"std": 0.05929861375505353,
|
| 18 |
+
"scores": [
|
| 19 |
+
0.11175825446844101,
|
| 20 |
+
0.2560095191001892,
|
| 21 |
+
0.1691480576992035
|
| 22 |
+
],
|
| 23 |
+
"params": 4951,
|
| 24 |
+
"hidden": 33
|
| 25 |
+
},
|
| 26 |
+
"SinGLU": {
|
| 27 |
+
"mean": 0.012662170765300592,
|
| 28 |
+
"std": 0.009025980600835806,
|
| 29 |
+
"scores": [
|
| 30 |
+
0.02520008198916912,
|
| 31 |
+
0.008467600680887699,
|
| 32 |
+
0.0043188296258449554
|
| 33 |
+
],
|
| 34 |
+
"params": 4811,
|
| 35 |
+
"hidden": 31
|
| 36 |
+
},
|
| 37 |
+
"Hybrid": {
|
| 38 |
+
"mean": 0.07436376934250195,
|
| 39 |
+
"std": 0.020922922410045546,
|
| 40 |
+
"scores": [
|
| 41 |
+
0.05885079875588417,
|
| 42 |
+
0.1039414331316948,
|
| 43 |
+
0.06029907613992691
|
| 44 |
+
],
|
| 45 |
+
"params": 4959,
|
| 46 |
+
"hidden": 34
|
| 47 |
+
}
|
| 48 |
+
},
|
| 49 |
+
"Nested Fn (2D)": {
|
| 50 |
+
"Vanilla": {
|
| 51 |
+
"mean": 0.04167063037554423,
|
| 52 |
+
"std": 0.0013826140176434484,
|
| 53 |
+
"scores": [
|
| 54 |
+
0.04034700244665146,
|
| 55 |
+
0.04357881471514702,
|
| 56 |
+
0.04108607396483421
|
| 57 |
+
],
|
| 58 |
+
"params": 2961,
|
| 59 |
+
"hidden": 37
|
| 60 |
+
},
|
| 61 |
+
"RichV1": {
|
| 62 |
+
"mean": 0.001206373757061859,
|
| 63 |
+
"std": 0.0010383498294791104,
|
| 64 |
+
"scores": [
|
| 65 |
+
0.00035041512455791235,
|
| 66 |
+
0.0006010299548506737,
|
| 67 |
+
0.002667676191776991
|
| 68 |
+
],
|
| 69 |
+
"params": 3069,
|
| 70 |
+
"hidden": 26
|
| 71 |
+
},
|
| 72 |
+
"SinGLU": {
|
| 73 |
+
"mean": 0.000256501533537327,
|
| 74 |
+
"std": 0.00014266684270291342,
|
| 75 |
+
"scores": [
|
| 76 |
+
7.845076470403001e-05,
|
| 77 |
+
0.0002633430704008788,
|
| 78 |
+
0.0004277107655070722
|
| 79 |
+
],
|
| 80 |
+
"params": 2993,
|
| 81 |
+
"hidden": 24
|
| 82 |
+
},
|
| 83 |
+
"Hybrid": {
|
| 84 |
+
"mean": 0.0009293731806489328,
|
| 85 |
+
"std": 0.00015638037345160952,
|
| 86 |
+
"scores": [
|
| 87 |
+
0.0010697855614125729,
|
| 88 |
+
0.0007111956947483122,
|
| 89 |
+
0.0010071382857859135
|
| 90 |
+
],
|
| 91 |
+
"params": 2959,
|
| 92 |
+
"hidden": 26
|
| 93 |
+
}
|
| 94 |
+
},
|
| 95 |
+
"Spiral": {
|
| 96 |
+
"Vanilla": {
|
| 97 |
+
"mean": 0.9022222359975179,
|
| 98 |
+
"std": 0.12650085789091728,
|
| 99 |
+
"scores": [
|
| 100 |
+
0.9933333396911621,
|
| 101 |
+
0.7233333587646484,
|
| 102 |
+
0.9900000095367432
|
| 103 |
+
],
|
| 104 |
+
"params": 2999,
|
| 105 |
+
"hidden": 37
|
| 106 |
+
},
|
| 107 |
+
"RichV1": {
|
| 108 |
+
"mean": 0.998888889948527,
|
| 109 |
+
"std": 0.0015713469040821585,
|
| 110 |
+
"scores": [
|
| 111 |
+
1.0,
|
| 112 |
+
0.996666669845581,
|
| 113 |
+
1.0
|
| 114 |
+
],
|
| 115 |
+
"params": 3096,
|
| 116 |
+
"hidden": 26
|
| 117 |
+
},
|
| 118 |
+
"SinGLU": {
|
| 119 |
+
"mean": 0.4444444378217061,
|
| 120 |
+
"std": 0.021829873123329135,
|
| 121 |
+
"scores": [
|
| 122 |
+
0.4699999988079071,
|
| 123 |
+
0.4166666567325592,
|
| 124 |
+
0.4466666579246521
|
| 125 |
+
],
|
| 126 |
+
"params": 3018,
|
| 127 |
+
"hidden": 24
|
| 128 |
+
},
|
| 129 |
+
"Hybrid": {
|
| 130 |
+
"mean": 1.0,
|
| 131 |
+
"std": 0.0,
|
| 132 |
+
"scores": [
|
| 133 |
+
1.0,
|
| 134 |
+
1.0,
|
| 135 |
+
1.0
|
| 136 |
+
],
|
| 137 |
+
"params": 2986,
|
| 138 |
+
"hidden": 26
|
| 139 |
+
}
|
| 140 |
+
},
|
| 141 |
+
"Checkerboard": {
|
| 142 |
+
"Vanilla": {
|
| 143 |
+
"mean": 0.6088888843854269,
|
| 144 |
+
"std": 0.037548314598774495,
|
| 145 |
+
"scores": [
|
| 146 |
+
0.6433333158493042,
|
| 147 |
+
0.5566666722297668,
|
| 148 |
+
0.6266666650772095
|
| 149 |
+
],
|
| 150 |
+
"params": 2999,
|
| 151 |
+
"hidden": 37
|
| 152 |
+
},
|
| 153 |
+
"RichV1": {
|
| 154 |
+
"mean": 0.9044444362322489,
|
| 155 |
+
"std": 0.008748917050117025,
|
| 156 |
+
"scores": [
|
| 157 |
+
0.9166666865348816,
|
| 158 |
+
0.8966666460037231,
|
| 159 |
+
0.8999999761581421
|
| 160 |
+
],
|
| 161 |
+
"params": 3096,
|
| 162 |
+
"hidden": 26
|
| 163 |
+
},
|
| 164 |
+
"SinGLU": {
|
| 165 |
+
"mean": 0.9366666674613953,
|
| 166 |
+
"std": 0.015153520767335832,
|
| 167 |
+
"scores": [
|
| 168 |
+
0.9166666865348816,
|
| 169 |
+
0.9399999976158142,
|
| 170 |
+
0.95333331823349
|
| 171 |
+
],
|
| 172 |
+
"params": 3018,
|
| 173 |
+
"hidden": 24
|
| 174 |
+
},
|
| 175 |
+
"Hybrid": {
|
| 176 |
+
"mean": 0.8955555558204651,
|
| 177 |
+
"std": 0.013966457328736363,
|
| 178 |
+
"scores": [
|
| 179 |
+
0.8999999761581421,
|
| 180 |
+
0.8766666650772095,
|
| 181 |
+
0.9100000262260437
|
| 182 |
+
],
|
| 183 |
+
"params": 2986,
|
| 184 |
+
"hidden": 26
|
| 185 |
+
}
|
| 186 |
+
},
|
| 187 |
+
"High-Freq": {
|
| 188 |
+
"Vanilla": {
|
| 189 |
+
"mean": 1.1838690439860027,
|
| 190 |
+
"std": 0.028954010436214626,
|
| 191 |
+
"scores": [
|
| 192 |
+
1.1441768407821655,
|
| 193 |
+
1.2124271392822266,
|
| 194 |
+
1.1950031518936157
|
| 195 |
+
],
|
| 196 |
+
"params": 7999,
|
| 197 |
+
"hidden": 62
|
| 198 |
+
},
|
| 199 |
+
"RichV1": {
|
| 200 |
+
"mean": 1.6797597805658977,
|
| 201 |
+
"std": 0.18753213617719802,
|
| 202 |
+
"scores": [
|
| 203 |
+
1.5120912790298462,
|
| 204 |
+
1.9415488243103027,
|
| 205 |
+
1.585639238357544
|
| 206 |
+
],
|
| 207 |
+
"params": 7913,
|
| 208 |
+
"hidden": 43
|
| 209 |
+
},
|
| 210 |
+
"SinGLU": {
|
| 211 |
+
"mean": 1.463480571905772,
|
| 212 |
+
"std": 0.6593260909943727,
|
| 213 |
+
"scores": [
|
| 214 |
+
2.395895481109619,
|
| 215 |
+
1.0015356540679932,
|
| 216 |
+
0.9930105805397034
|
| 217 |
+
],
|
| 218 |
+
"params": 8214,
|
| 219 |
+
"hidden": 41
|
| 220 |
+
},
|
| 221 |
+
"Hybrid": {
|
| 222 |
+
"mean": 1.4304812749226887,
|
| 223 |
+
"std": 0.28564243680425555,
|
| 224 |
+
"scores": [
|
| 225 |
+
1.0885547399520874,
|
| 226 |
+
1.4151595830917358,
|
| 227 |
+
1.7877295017242432
|
| 228 |
+
],
|
| 229 |
+
"params": 7997,
|
| 230 |
+
"hidden": 44
|
| 231 |
+
}
|
| 232 |
+
},
|
| 233 |
+
"Memorization": {
|
| 234 |
+
"Vanilla": {
|
| 235 |
+
"mean": 0.023636028170585632,
|
| 236 |
+
"std": 0.005544727067146954,
|
| 237 |
+
"scores": [
|
| 238 |
+
0.028810137882828712,
|
| 239 |
+
0.01594628393650055,
|
| 240 |
+
0.026151662692427635
|
| 241 |
+
],
|
| 242 |
+
"params": 4926,
|
| 243 |
+
"hidden": 46
|
| 244 |
+
},
|
| 245 |
+
"RichV1": {
|
| 246 |
+
"mean": 5.6470338132695034e-08,
|
| 247 |
+
"std": 7.940899452098823e-08,
|
| 248 |
+
"scores": [
|
| 249 |
+
6.393849338870439e-10,
|
| 250 |
+
6.198520971674448e-13,
|
| 251 |
+
1.687710096121009e-07
|
| 252 |
+
],
|
| 253 |
+
"params": 5028,
|
| 254 |
+
"hidden": 32
|
| 255 |
+
},
|
| 256 |
+
"SinGLU": {
|
| 257 |
+
"mean": 1.4922602130836477e-08,
|
| 258 |
+
"std": 1.9137241711772547e-08,
|
| 259 |
+
"scores": [
|
| 260 |
+
2.659182207764843e-09,
|
| 261 |
+
1.6035532623970283e-10,
|
| 262 |
+
4.194826885850489e-08
|
| 263 |
+
],
|
| 264 |
+
"params": 5067,
|
| 265 |
+
"hidden": 31
|
| 266 |
+
},
|
| 267 |
+
"Hybrid": {
|
| 268 |
+
"mean": 3.557413054873967e-08,
|
| 269 |
+
"std": 4.040383113636077e-08,
|
| 270 |
+
"scores": [
|
| 271 |
+
1.3963182610154945e-08,
|
| 272 |
+
9.218822327738962e-08,
|
| 273 |
+
5.709857586744249e-10
|
| 274 |
+
],
|
| 275 |
+
"params": 4807,
|
| 276 |
+
"hidden": 32
|
| 277 |
+
}
|
| 278 |
+
}
|
| 279 |
+
},
|
| 280 |
+
"ood": {
|
| 281 |
+
"Vanilla": {
|
| 282 |
+
"id_mean": 0.19710678855578104,
|
| 283 |
+
"id_std": 0.01372835167636716,
|
| 284 |
+
"ood_mean": 1.340779185295105,
|
| 285 |
+
"ood_std": 0.07496366555450877,
|
| 286 |
+
"params": 5097,
|
| 287 |
+
"degradation": 6.80229836384182
|
| 288 |
+
},
|
| 289 |
+
"RichV1": {
|
| 290 |
+
"id_mean": 0.0030629046571751437,
|
| 291 |
+
"id_std": 0.0007370664671714653,
|
| 292 |
+
"ood_mean": 5.920439720153809,
|
| 293 |
+
"ood_std": 1.6229635781605876,
|
| 294 |
+
"params": 5101,
|
| 295 |
+
"degradation": 1932.949400264425
|
| 296 |
+
},
|
| 297 |
+
"SinGLU": {
|
| 298 |
+
"id_mean": 0.25248077015082043,
|
| 299 |
+
"id_std": 0.018526770338375,
|
| 300 |
+
"ood_mean": 6.372718969980876,
|
| 301 |
+
"ood_std": 0.8959624824918172,
|
| 302 |
+
"params": 5109,
|
| 303 |
+
"degradation": 25.24041322503138
|
| 304 |
+
},
|
| 305 |
+
"Hybrid": {
|
| 306 |
+
"id_mean": 0.0026089861057698727,
|
| 307 |
+
"id_std": 0.00042023361449633836,
|
| 308 |
+
"ood_mean": 4.857578992843628,
|
| 309 |
+
"ood_std": 1.0302407877868605,
|
| 310 |
+
"params": 4819,
|
| 311 |
+
"degradation": 1861.864646232077
|
| 312 |
+
}
|
| 313 |
+
},
|
| 314 |
+
"gradient_norms": {
|
| 315 |
+
"Vanilla": [
|
| 316 |
+
0.639647362423343,
|
| 317 |
+
0.3283495173433725,
|
| 318 |
+
0.23426363702089711,
|
| 319 |
+
0.27894396879219774,
|
| 320 |
+
0.16422125111024313
|
| 321 |
+
],
|
| 322 |
+
"RichV1": [
|
| 323 |
+
15.839934002021176,
|
| 324 |
+
6.613218660117891,
|
| 325 |
+
3.4827730708357985,
|
| 326 |
+
1.6238777871843217,
|
| 327 |
+
0.17907922274133237
|
| 328 |
+
],
|
| 329 |
+
"SinGLU": [
|
| 330 |
+
19.53187113650771,
|
| 331 |
+
14.939248550547868,
|
| 332 |
+
5.134043674832624,
|
| 333 |
+
1.3400197622788466,
|
| 334 |
+
0.4012023661781001
|
| 335 |
+
],
|
| 336 |
+
"Shared(S2)": [
|
| 337 |
+
1159.4268234957435,
|
| 338 |
+
884.1079387319159,
|
| 339 |
+
904.1129815185649,
|
| 340 |
+
714.1639373805058,
|
| 341 |
+
174.37114304372977
|
| 342 |
+
],
|
| 343 |
+
"Hybrid": [
|
| 344 |
+
13.887360774140875,
|
| 345 |
+
5.165627499528781,
|
| 346 |
+
4.734903446652124,
|
| 347 |
+
2.255145877088462,
|
| 348 |
+
0.5620669780578126
|
| 349 |
+
]
|
| 350 |
+
}
|
| 351 |
+
}
|