exp007: pmix8_dense_s1234
Browse files
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s1234/checkpoint.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e584c78bac41d0242b6ee444dd2c52e9127c0df9b4e5062c2003f4a750f483c0
|
| 3 |
+
size 28088373
|
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s1234/result.json
ADDED
|
@@ -0,0 +1,365 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"run": "pmix8_dense_s1234",
|
| 3 |
+
"tag": "pmix8_dense",
|
| 4 |
+
"seed": 1234,
|
| 5 |
+
"dev_final": 0.005977749824523926,
|
| 6 |
+
"statute_final": "uniform",
|
| 7 |
+
"params": 6922059,
|
| 8 |
+
"minutes": 9.3,
|
| 9 |
+
"bpb": 2.3438451486758853,
|
| 10 |
+
"bpb_eval": NaN,
|
| 11 |
+
"bpb_inbank": NaN,
|
| 12 |
+
"coverage": 0.8067626953125,
|
| 13 |
+
"gate_acc": 0.8631591796875,
|
| 14 |
+
"bpb_bank_cond": 2.0336265120127694,
|
| 15 |
+
"bpb_byte_cond": 2.8290389343268223,
|
| 16 |
+
"mode_spread_deg": 85.9728205170011,
|
| 17 |
+
"mix_entropy": 0.9711796087366482,
|
| 18 |
+
"write_kl": 0.12217555940151215,
|
| 19 |
+
"nn_exact": NaN,
|
| 20 |
+
"ang_err_deg": NaN,
|
| 21 |
+
"budget_deg": NaN,
|
| 22 |
+
"in_budget": NaN,
|
| 23 |
+
"conf_mean": 0.08110357820987701,
|
| 24 |
+
"kernel_invariant": 0.8547699451446533,
|
| 25 |
+
"branching_frac": 1.0,
|
| 26 |
+
"loss": 4.886106491088867,
|
| 27 |
+
"statute_trajectory": [
|
| 28 |
+
[
|
| 29 |
+
0,
|
| 30 |
+
{
|
| 31 |
+
"deviation": -0.002232074737548828,
|
| 32 |
+
"pair_fraction": 0.53125,
|
| 33 |
+
"statute": "uniform"
|
| 34 |
+
}
|
| 35 |
+
],
|
| 36 |
+
[
|
| 37 |
+
250,
|
| 38 |
+
{
|
| 39 |
+
"deviation": -0.003371119499206543,
|
| 40 |
+
"pair_fraction": 0.53125,
|
| 41 |
+
"statute": "uniform"
|
| 42 |
+
}
|
| 43 |
+
],
|
| 44 |
+
[
|
| 45 |
+
500,
|
| 46 |
+
{
|
| 47 |
+
"deviation": -0.004114508628845215,
|
| 48 |
+
"pair_fraction": 0.53125,
|
| 49 |
+
"statute": "uniform"
|
| 50 |
+
}
|
| 51 |
+
],
|
| 52 |
+
[
|
| 53 |
+
750,
|
| 54 |
+
{
|
| 55 |
+
"deviation": -0.004626154899597168,
|
| 56 |
+
"pair_fraction": 0.5625,
|
| 57 |
+
"statute": "uniform"
|
| 58 |
+
}
|
| 59 |
+
],
|
| 60 |
+
[
|
| 61 |
+
1000,
|
| 62 |
+
{
|
| 63 |
+
"deviation": -0.0050699710845947266,
|
| 64 |
+
"pair_fraction": 0.5625,
|
| 65 |
+
"statute": "uniform"
|
| 66 |
+
}
|
| 67 |
+
],
|
| 68 |
+
[
|
| 69 |
+
1250,
|
| 70 |
+
{
|
| 71 |
+
"deviation": -0.005313515663146973,
|
| 72 |
+
"pair_fraction": 0.5625,
|
| 73 |
+
"statute": "uniform"
|
| 74 |
+
}
|
| 75 |
+
],
|
| 76 |
+
[
|
| 77 |
+
1500,
|
| 78 |
+
{
|
| 79 |
+
"deviation": -0.005534172058105469,
|
| 80 |
+
"pair_fraction": 0.5625,
|
| 81 |
+
"statute": "uniform"
|
| 82 |
+
}
|
| 83 |
+
],
|
| 84 |
+
[
|
| 85 |
+
1750,
|
| 86 |
+
{
|
| 87 |
+
"deviation": -0.005790233612060547,
|
| 88 |
+
"pair_fraction": 0.5625,
|
| 89 |
+
"statute": "uniform"
|
| 90 |
+
}
|
| 91 |
+
],
|
| 92 |
+
[
|
| 93 |
+
2000,
|
| 94 |
+
{
|
| 95 |
+
"deviation": -0.00597536563873291,
|
| 96 |
+
"pair_fraction": 0.5625,
|
| 97 |
+
"statute": "uniform"
|
| 98 |
+
}
|
| 99 |
+
],
|
| 100 |
+
[
|
| 101 |
+
2250,
|
| 102 |
+
{
|
| 103 |
+
"deviation": -0.00608062744140625,
|
| 104 |
+
"pair_fraction": 0.5625,
|
| 105 |
+
"statute": "uniform"
|
| 106 |
+
}
|
| 107 |
+
],
|
| 108 |
+
[
|
| 109 |
+
2500,
|
| 110 |
+
{
|
| 111 |
+
"deviation": -0.005765676498413086,
|
| 112 |
+
"pair_fraction": 0.5625,
|
| 113 |
+
"statute": "uniform"
|
| 114 |
+
}
|
| 115 |
+
],
|
| 116 |
+
[
|
| 117 |
+
2750,
|
| 118 |
+
{
|
| 119 |
+
"deviation": -0.005376935005187988,
|
| 120 |
+
"pair_fraction": 0.53125,
|
| 121 |
+
"statute": "uniform"
|
| 122 |
+
}
|
| 123 |
+
],
|
| 124 |
+
[
|
| 125 |
+
3000,
|
| 126 |
+
{
|
| 127 |
+
"deviation": -0.004772067070007324,
|
| 128 |
+
"pair_fraction": 0.53125,
|
| 129 |
+
"statute": "uniform"
|
| 130 |
+
}
|
| 131 |
+
],
|
| 132 |
+
[
|
| 133 |
+
3250,
|
| 134 |
+
{
|
| 135 |
+
"deviation": -0.004072904586791992,
|
| 136 |
+
"pair_fraction": 0.53125,
|
| 137 |
+
"statute": "uniform"
|
| 138 |
+
}
|
| 139 |
+
],
|
| 140 |
+
[
|
| 141 |
+
3500,
|
| 142 |
+
{
|
| 143 |
+
"deviation": -0.002942323684692383,
|
| 144 |
+
"pair_fraction": 0.53125,
|
| 145 |
+
"statute": "uniform"
|
| 146 |
+
}
|
| 147 |
+
],
|
| 148 |
+
[
|
| 149 |
+
3750,
|
| 150 |
+
{
|
| 151 |
+
"deviation": -0.0017855167388916016,
|
| 152 |
+
"pair_fraction": 0.46875,
|
| 153 |
+
"statute": "uniform"
|
| 154 |
+
}
|
| 155 |
+
],
|
| 156 |
+
[
|
| 157 |
+
4000,
|
| 158 |
+
{
|
| 159 |
+
"deviation": -0.0006895065307617188,
|
| 160 |
+
"pair_fraction": 0.5625,
|
| 161 |
+
"statute": "uniform"
|
| 162 |
+
}
|
| 163 |
+
],
|
| 164 |
+
[
|
| 165 |
+
4250,
|
| 166 |
+
{
|
| 167 |
+
"deviation": 0.00024211406707763672,
|
| 168 |
+
"pair_fraction": 0.53125,
|
| 169 |
+
"statute": "uniform"
|
| 170 |
+
}
|
| 171 |
+
],
|
| 172 |
+
[
|
| 173 |
+
4500,
|
| 174 |
+
{
|
| 175 |
+
"deviation": 0.0010249614715576172,
|
| 176 |
+
"pair_fraction": 0.5,
|
| 177 |
+
"statute": "uniform"
|
| 178 |
+
}
|
| 179 |
+
],
|
| 180 |
+
[
|
| 181 |
+
4750,
|
| 182 |
+
{
|
| 183 |
+
"deviation": 0.001874089241027832,
|
| 184 |
+
"pair_fraction": 0.5,
|
| 185 |
+
"statute": "uniform"
|
| 186 |
+
}
|
| 187 |
+
],
|
| 188 |
+
[
|
| 189 |
+
5000,
|
| 190 |
+
{
|
| 191 |
+
"deviation": 0.002515554428100586,
|
| 192 |
+
"pair_fraction": 0.5,
|
| 193 |
+
"statute": "uniform"
|
| 194 |
+
}
|
| 195 |
+
],
|
| 196 |
+
[
|
| 197 |
+
5250,
|
| 198 |
+
{
|
| 199 |
+
"deviation": 0.0029686689376831055,
|
| 200 |
+
"pair_fraction": 0.53125,
|
| 201 |
+
"statute": "uniform"
|
| 202 |
+
}
|
| 203 |
+
],
|
| 204 |
+
[
|
| 205 |
+
5500,
|
| 206 |
+
{
|
| 207 |
+
"deviation": 0.0034579038619995117,
|
| 208 |
+
"pair_fraction": 0.53125,
|
| 209 |
+
"statute": "uniform"
|
| 210 |
+
}
|
| 211 |
+
],
|
| 212 |
+
[
|
| 213 |
+
5750,
|
| 214 |
+
{
|
| 215 |
+
"deviation": 0.0037964582443237305,
|
| 216 |
+
"pair_fraction": 0.53125,
|
| 217 |
+
"statute": "uniform"
|
| 218 |
+
}
|
| 219 |
+
],
|
| 220 |
+
[
|
| 221 |
+
6000,
|
| 222 |
+
{
|
| 223 |
+
"deviation": 0.004160881042480469,
|
| 224 |
+
"pair_fraction": 0.53125,
|
| 225 |
+
"statute": "uniform"
|
| 226 |
+
}
|
| 227 |
+
],
|
| 228 |
+
[
|
| 229 |
+
6250,
|
| 230 |
+
{
|
| 231 |
+
"deviation": 0.004432797431945801,
|
| 232 |
+
"pair_fraction": 0.53125,
|
| 233 |
+
"statute": "uniform"
|
| 234 |
+
}
|
| 235 |
+
],
|
| 236 |
+
[
|
| 237 |
+
6500,
|
| 238 |
+
{
|
| 239 |
+
"deviation": 0.0046842098236083984,
|
| 240 |
+
"pair_fraction": 0.53125,
|
| 241 |
+
"statute": "uniform"
|
| 242 |
+
}
|
| 243 |
+
],
|
| 244 |
+
[
|
| 245 |
+
6750,
|
| 246 |
+
{
|
| 247 |
+
"deviation": 0.004860281944274902,
|
| 248 |
+
"pair_fraction": 0.5,
|
| 249 |
+
"statute": "uniform"
|
| 250 |
+
}
|
| 251 |
+
],
|
| 252 |
+
[
|
| 253 |
+
7000,
|
| 254 |
+
{
|
| 255 |
+
"deviation": 0.005045294761657715,
|
| 256 |
+
"pair_fraction": 0.5,
|
| 257 |
+
"statute": "uniform"
|
| 258 |
+
}
|
| 259 |
+
],
|
| 260 |
+
[
|
| 261 |
+
7250,
|
| 262 |
+
{
|
| 263 |
+
"deviation": 0.005220174789428711,
|
| 264 |
+
"pair_fraction": 0.5,
|
| 265 |
+
"statute": "uniform"
|
| 266 |
+
}
|
| 267 |
+
],
|
| 268 |
+
[
|
| 269 |
+
7500,
|
| 270 |
+
{
|
| 271 |
+
"deviation": 0.0053310394287109375,
|
| 272 |
+
"pair_fraction": 0.46875,
|
| 273 |
+
"statute": "uniform"
|
| 274 |
+
}
|
| 275 |
+
],
|
| 276 |
+
[
|
| 277 |
+
7750,
|
| 278 |
+
{
|
| 279 |
+
"deviation": 0.005458354949951172,
|
| 280 |
+
"pair_fraction": 0.46875,
|
| 281 |
+
"statute": "uniform"
|
| 282 |
+
}
|
| 283 |
+
],
|
| 284 |
+
[
|
| 285 |
+
8000,
|
| 286 |
+
{
|
| 287 |
+
"deviation": 0.005547285079956055,
|
| 288 |
+
"pair_fraction": 0.46875,
|
| 289 |
+
"statute": "uniform"
|
| 290 |
+
}
|
| 291 |
+
],
|
| 292 |
+
[
|
| 293 |
+
8250,
|
| 294 |
+
{
|
| 295 |
+
"deviation": 0.0056154727935791016,
|
| 296 |
+
"pair_fraction": 0.46875,
|
| 297 |
+
"statute": "uniform"
|
| 298 |
+
}
|
| 299 |
+
],
|
| 300 |
+
[
|
| 301 |
+
8500,
|
| 302 |
+
{
|
| 303 |
+
"deviation": 0.005683422088623047,
|
| 304 |
+
"pair_fraction": 0.46875,
|
| 305 |
+
"statute": "uniform"
|
| 306 |
+
}
|
| 307 |
+
],
|
| 308 |
+
[
|
| 309 |
+
8750,
|
| 310 |
+
{
|
| 311 |
+
"deviation": 0.005743145942687988,
|
| 312 |
+
"pair_fraction": 0.46875,
|
| 313 |
+
"statute": "uniform"
|
| 314 |
+
}
|
| 315 |
+
],
|
| 316 |
+
[
|
| 317 |
+
9000,
|
| 318 |
+
{
|
| 319 |
+
"deviation": 0.005802273750305176,
|
| 320 |
+
"pair_fraction": 0.46875,
|
| 321 |
+
"statute": "uniform"
|
| 322 |
+
}
|
| 323 |
+
],
|
| 324 |
+
[
|
| 325 |
+
9250,
|
| 326 |
+
{
|
| 327 |
+
"deviation": 0.005862593650817871,
|
| 328 |
+
"pair_fraction": 0.46875,
|
| 329 |
+
"statute": "uniform"
|
| 330 |
+
}
|
| 331 |
+
],
|
| 332 |
+
[
|
| 333 |
+
9500,
|
| 334 |
+
{
|
| 335 |
+
"deviation": 0.0058956146240234375,
|
| 336 |
+
"pair_fraction": 0.46875,
|
| 337 |
+
"statute": "uniform"
|
| 338 |
+
}
|
| 339 |
+
],
|
| 340 |
+
[
|
| 341 |
+
9750,
|
| 342 |
+
{
|
| 343 |
+
"deviation": 0.005950212478637695,
|
| 344 |
+
"pair_fraction": 0.46875,
|
| 345 |
+
"statute": "uniform"
|
| 346 |
+
}
|
| 347 |
+
],
|
| 348 |
+
[
|
| 349 |
+
10000,
|
| 350 |
+
{
|
| 351 |
+
"deviation": 0.005977749824523926,
|
| 352 |
+
"pair_fraction": 0.46875,
|
| 353 |
+
"statute": "uniform"
|
| 354 |
+
}
|
| 355 |
+
]
|
| 356 |
+
],
|
| 357 |
+
"condition": {
|
| 358 |
+
"tag": "pmix8_dense",
|
| 359 |
+
"head": "hybrid",
|
| 360 |
+
"bank_scorer": "pmix",
|
| 361 |
+
"n_pointers": 8,
|
| 362 |
+
"bank_source": "atlas/bank_16384_tri.pt"
|
| 363 |
+
},
|
| 364 |
+
"steps": 10000
|
| 365 |
+
}
|
experiments/exp_007_aleph_routed_attention/runs/pmix8_dense_s1234/snapshots.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:91b2aadb89244715a5f964721538fb0918236a6efce6ab4a8774c0fc9e69ec32
|
| 3 |
+
size 57350
|