update slot resume pointers for step 5727
Browse files
slots/16/checkpoint-4632/config.json
DELETED
|
@@ -1,24 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"architectures": [
|
| 3 |
-
"TwinyForCausalLM"
|
| 4 |
-
],
|
| 5 |
-
"attention_dropout": 0.0,
|
| 6 |
-
"dtype": "float32",
|
| 7 |
-
"hidden_dropout": 0.0,
|
| 8 |
-
"hidden_size": 768,
|
| 9 |
-
"initializer_range": 0.02,
|
| 10 |
-
"intermediate_size": 3072,
|
| 11 |
-
"max_position_embeddings": 128,
|
| 12 |
-
"model_type": "twiny",
|
| 13 |
-
"neftune_alpha": 0.0,
|
| 14 |
-
"num_attention_heads": 12,
|
| 15 |
-
"num_hidden_layers": 2,
|
| 16 |
-
"num_key_value_heads": 3,
|
| 17 |
-
"qk_norm": true,
|
| 18 |
-
"rezero_init": 1.0,
|
| 19 |
-
"rms_norm_eps": 1e-06,
|
| 20 |
-
"rope_theta": 10000.0,
|
| 21 |
-
"transformers_version": "5.0.0",
|
| 22 |
-
"use_cache": false,
|
| 23 |
-
"vocab_size": 32000
|
| 24 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
slots/16/checkpoint-4632/trainer_state.json
DELETED
|
@@ -1,1658 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 1.9391936982322076,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 4632,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.00041872211245305734,
|
| 14 |
-
"grad_norm": NaN,
|
| 15 |
-
"learning_rate": 5e-05,
|
| 16 |
-
"loss": 0.6013199090957642,
|
| 17 |
-
"step": 1
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.008374442249061146,
|
| 21 |
-
"grad_norm": 1.9934037923812866,
|
| 22 |
-
"learning_rate": 4.999949009907666e-05,
|
| 23 |
-
"loss": 1.2950001766807155,
|
| 24 |
-
"step": 20
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.016748884498122292,
|
| 28 |
-
"grad_norm": 1.5410668849945068,
|
| 29 |
-
"learning_rate": 4.999758461765953e-05,
|
| 30 |
-
"loss": 0.9534212112426758,
|
| 31 |
-
"step": 40
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.025123326747183438,
|
| 35 |
-
"grad_norm": 1.7543447017669678,
|
| 36 |
-
"learning_rate": 4.999446714274127e-05,
|
| 37 |
-
"loss": 1.7281543731689453,
|
| 38 |
-
"step": 60
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.033497768996244584,
|
| 42 |
-
"grad_norm": 1.228683590888977,
|
| 43 |
-
"learning_rate": 4.999007608070043e-05,
|
| 44 |
-
"loss": 1.0517969131469727,
|
| 45 |
-
"step": 80
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.041872211245305734,
|
| 49 |
-
"grad_norm": 1.8294340372085571,
|
| 50 |
-
"learning_rate": 4.9984078259584456e-05,
|
| 51 |
-
"loss": 0.6183329105377198,
|
| 52 |
-
"step": 100
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.050246653494366876,
|
| 56 |
-
"grad_norm": 2.365013837814331,
|
| 57 |
-
"learning_rate": 4.997666984880032e-05,
|
| 58 |
-
"loss": 0.6331160545349122,
|
| 59 |
-
"step": 120
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.058621095743428026,
|
| 63 |
-
"grad_norm": 1.479344367980957,
|
| 64 |
-
"learning_rate": 4.9968325679939295e-05,
|
| 65 |
-
"loss": 0.7190958023071289,
|
| 66 |
-
"step": 140
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.06699553799248917,
|
| 70 |
-
"grad_norm": 1.2224725484848022,
|
| 71 |
-
"learning_rate": 4.995816789488419e-05,
|
| 72 |
-
"loss": 0.9246182441711426,
|
| 73 |
-
"step": 160
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.07536998024155032,
|
| 77 |
-
"grad_norm": 1.1624501943588257,
|
| 78 |
-
"learning_rate": 4.994660098304831e-05,
|
| 79 |
-
"loss": 0.801685905456543,
|
| 80 |
-
"step": 180
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.08374442249061147,
|
| 84 |
-
"grad_norm": 1.4690343141555786,
|
| 85 |
-
"learning_rate": 4.993362559749423e-05,
|
| 86 |
-
"loss": 0.6664809226989746,
|
| 87 |
-
"step": 200
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.09211886473967261,
|
| 91 |
-
"grad_norm": 1.4142041206359863,
|
| 92 |
-
"learning_rate": 4.991999504877833e-05,
|
| 93 |
-
"loss": 0.9447656631469726,
|
| 94 |
-
"step": 220
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.10049330698873375,
|
| 98 |
-
"grad_norm": 1.426574468612671,
|
| 99 |
-
"learning_rate": 4.9904275318961756e-05,
|
| 100 |
-
"loss": 0.5283678054809571,
|
| 101 |
-
"step": 240
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.10886774923779491,
|
| 105 |
-
"grad_norm": 1.2339314222335815,
|
| 106 |
-
"learning_rate": 4.988803917569756e-05,
|
| 107 |
-
"loss": 0.9661962509155273,
|
| 108 |
-
"step": 260
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.11724219148685605,
|
| 112 |
-
"grad_norm": 1.0341933965682983,
|
| 113 |
-
"learning_rate": 4.986957847640217e-05,
|
| 114 |
-
"loss": 0.5136320114135742,
|
| 115 |
-
"step": 280
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.1256166337359172,
|
| 119 |
-
"grad_norm": 1.5590699911117554,
|
| 120 |
-
"learning_rate": 4.984971365204418e-05,
|
| 121 |
-
"loss": 0.566891860961914,
|
| 122 |
-
"step": 300
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.13399107598497834,
|
| 126 |
-
"grad_norm": 2.064842700958252,
|
| 127 |
-
"learning_rate": 4.982844582418252e-05,
|
| 128 |
-
"loss": 0.6304716110229492,
|
| 129 |
-
"step": 320
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.14236551823403948,
|
| 133 |
-
"grad_norm": 1.4325364828109741,
|
| 134 |
-
"learning_rate": 4.980577619358906e-05,
|
| 135 |
-
"loss": 0.7040385723114013,
|
| 136 |
-
"step": 340
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.15073996048310065,
|
| 140 |
-
"grad_norm": 1.0557610988616943,
|
| 141 |
-
"learning_rate": 4.978170604018079e-05,
|
| 142 |
-
"loss": 0.6057440757751464,
|
| 143 |
-
"step": 360
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.1591144027321618,
|
| 147 |
-
"grad_norm": 1.2590022087097168,
|
| 148 |
-
"learning_rate": 4.9756236722947616e-05,
|
| 149 |
-
"loss": 0.4318533420562744,
|
| 150 |
-
"step": 380
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.16748884498122293,
|
| 154 |
-
"grad_norm": 1.700644850730896,
|
| 155 |
-
"learning_rate": 4.972936967987556e-05,
|
| 156 |
-
"loss": 0.8144928932189941,
|
| 157 |
-
"step": 400
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.17586328723028408,
|
| 161 |
-
"grad_norm": 0.6783745884895325,
|
| 162 |
-
"learning_rate": 4.970110642786564e-05,
|
| 163 |
-
"loss": 0.5173389911651611,
|
| 164 |
-
"step": 420
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.18423772947934522,
|
| 168 |
-
"grad_norm": 0.6910356879234314,
|
| 169 |
-
"learning_rate": 4.967144856264818e-05,
|
| 170 |
-
"loss": 0.4949141502380371,
|
| 171 |
-
"step": 440
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.19261217172840636,
|
| 175 |
-
"grad_norm": 0.8224261999130249,
|
| 176 |
-
"learning_rate": 4.9640397758692715e-05,
|
| 177 |
-
"loss": 0.4392205238342285,
|
| 178 |
-
"step": 460
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.2009866139774675,
|
| 182 |
-
"grad_norm": 1.6215633153915405,
|
| 183 |
-
"learning_rate": 4.960795576911349e-05,
|
| 184 |
-
"loss": 0.4062031745910645,
|
| 185 |
-
"step": 480
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.20936105622652867,
|
| 189 |
-
"grad_norm": 0.8354963660240173,
|
| 190 |
-
"learning_rate": 4.9574124425570436e-05,
|
| 191 |
-
"loss": 0.34378366470336913,
|
| 192 |
-
"step": 500
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.21773549847558982,
|
| 196 |
-
"grad_norm": 1.0993494987487793,
|
| 197 |
-
"learning_rate": 4.9538905638165805e-05,
|
| 198 |
-
"loss": 0.5852083206176758,
|
| 199 |
-
"step": 520
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.22610994072465096,
|
| 203 |
-
"grad_norm": 2.658684015274048,
|
| 204 |
-
"learning_rate": 4.950230139533626e-05,
|
| 205 |
-
"loss": 0.4075888156890869,
|
| 206 |
-
"step": 540
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.2344843829737121,
|
| 210 |
-
"grad_norm": 0.9528117179870605,
|
| 211 |
-
"learning_rate": 4.946431376374067e-05,
|
| 212 |
-
"loss": 0.3786322832107544,
|
| 213 |
-
"step": 560
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.24285882522277324,
|
| 217 |
-
"grad_norm": 2.995469093322754,
|
| 218 |
-
"learning_rate": 4.9424944888143395e-05,
|
| 219 |
-
"loss": 0.5390625,
|
| 220 |
-
"step": 580
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.2512332674718344,
|
| 224 |
-
"grad_norm": 0.8218050599098206,
|
| 225 |
-
"learning_rate": 4.9384196991293205e-05,
|
| 226 |
-
"loss": 0.39401955604553224,
|
| 227 |
-
"step": 600
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.25960770972089553,
|
| 231 |
-
"grad_norm": 1.5551509857177734,
|
| 232 |
-
"learning_rate": 4.9342072373797774e-05,
|
| 233 |
-
"loss": 0.47962727546691897,
|
| 234 |
-
"step": 620
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.2679821519699567,
|
| 238 |
-
"grad_norm": 1.12935209274292,
|
| 239 |
-
"learning_rate": 4.9298573413993806e-05,
|
| 240 |
-
"loss": 0.42098431587219237,
|
| 241 |
-
"step": 640
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.2763565942190178,
|
| 245 |
-
"grad_norm": 2.1346328258514404,
|
| 246 |
-
"learning_rate": 4.925370256781273e-05,
|
| 247 |
-
"loss": 0.4414808750152588,
|
| 248 |
-
"step": 660
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.28473103646807896,
|
| 252 |
-
"grad_norm": 1.285367727279663,
|
| 253 |
-
"learning_rate": 4.920746236864208e-05,
|
| 254 |
-
"loss": 0.4640543460845947,
|
| 255 |
-
"step": 680
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.29310547871714016,
|
| 259 |
-
"grad_norm": 0.7970554232597351,
|
| 260 |
-
"learning_rate": 4.915985542718241e-05,
|
| 261 |
-
"loss": 0.3569347381591797,
|
| 262 |
-
"step": 700
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.3014799209662013,
|
| 266 |
-
"grad_norm": 0.6530137658119202,
|
| 267 |
-
"learning_rate": 4.9110884431299935e-05,
|
| 268 |
-
"loss": 0.41652464866638184,
|
| 269 |
-
"step": 720
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.30985436321526244,
|
| 273 |
-
"grad_norm": 0.956561267375946,
|
| 274 |
-
"learning_rate": 4.906055214587476e-05,
|
| 275 |
-
"loss": 0.3489442110061646,
|
| 276 |
-
"step": 740
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.3182288054643236,
|
| 280 |
-
"grad_norm": 1.9765814542770386,
|
| 281 |
-
"learning_rate": 4.9008861412644756e-05,
|
| 282 |
-
"loss": 0.40424065589904784,
|
| 283 |
-
"step": 760
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.3266032477133847,
|
| 287 |
-
"grad_norm": 1.658229112625122,
|
| 288 |
-
"learning_rate": 4.895581515004517e-05,
|
| 289 |
-
"loss": 0.31643118858337405,
|
| 290 |
-
"step": 780
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.33497768996244587,
|
| 294 |
-
"grad_norm": 2.2325282096862793,
|
| 295 |
-
"learning_rate": 4.89014163530438e-05,
|
| 296 |
-
"loss": 0.3643105745315552,
|
| 297 |
-
"step": 800
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.343352132211507,
|
| 301 |
-
"grad_norm": 0.4895799160003662,
|
| 302 |
-
"learning_rate": 4.8845668092971934e-05,
|
| 303 |
-
"loss": 0.3214404821395874,
|
| 304 |
-
"step": 820
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.35172657446056815,
|
| 308 |
-
"grad_norm": 1.4924691915512085,
|
| 309 |
-
"learning_rate": 4.87885735173509e-05,
|
| 310 |
-
"loss": 0.4828330993652344,
|
| 311 |
-
"step": 840
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.3601010167096293,
|
| 315 |
-
"grad_norm": 0.6766088604927063,
|
| 316 |
-
"learning_rate": 4.873013584971442e-05,
|
| 317 |
-
"loss": 0.3910247325897217,
|
| 318 |
-
"step": 860
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.36847545895869044,
|
| 322 |
-
"grad_norm": 0.48699629306793213,
|
| 323 |
-
"learning_rate": 4.867035838942657e-05,
|
| 324 |
-
"loss": 0.2918583631515503,
|
| 325 |
-
"step": 880
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.3768499012077516,
|
| 329 |
-
"grad_norm": 1.2841662168502808,
|
| 330 |
-
"learning_rate": 4.86092445114955e-05,
|
| 331 |
-
"loss": 0.2856093406677246,
|
| 332 |
-
"step": 900
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.3852243434568127,
|
| 336 |
-
"grad_norm": 0.21915394067764282,
|
| 337 |
-
"learning_rate": 4.8546797666382896e-05,
|
| 338 |
-
"loss": 0.33952271938323975,
|
| 339 |
-
"step": 920
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.39359878570587387,
|
| 343 |
-
"grad_norm": 0.6552733778953552,
|
| 344 |
-
"learning_rate": 4.8483021379809154e-05,
|
| 345 |
-
"loss": 0.3813821315765381,
|
| 346 |
-
"step": 940
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.401973227954935,
|
| 350 |
-
"grad_norm": 0.4852622449398041,
|
| 351 |
-
"learning_rate": 4.8417919252554354e-05,
|
| 352 |
-
"loss": 0.3341342926025391,
|
| 353 |
-
"step": 960
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.4103476702039962,
|
| 357 |
-
"grad_norm": 1.5116126537322998,
|
| 358 |
-
"learning_rate": 4.8351494960254916e-05,
|
| 359 |
-
"loss": 0.29020323753356936,
|
| 360 |
-
"step": 980
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.41872211245305735,
|
| 364 |
-
"grad_norm": 0.4594159424304962,
|
| 365 |
-
"learning_rate": 4.8283752253196094e-05,
|
| 366 |
-
"loss": 0.25537676811218263,
|
| 367 |
-
"step": 1000
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.4270965547021185,
|
| 371 |
-
"grad_norm": 0.8960456252098083,
|
| 372 |
-
"learning_rate": 4.8214694956100245e-05,
|
| 373 |
-
"loss": 0.26440722942352296,
|
| 374 |
-
"step": 1020
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.43547099695117963,
|
| 378 |
-
"grad_norm": 0.6404821872711182,
|
| 379 |
-
"learning_rate": 4.814432696791089e-05,
|
| 380 |
-
"loss": 0.2976436376571655,
|
| 381 |
-
"step": 1040
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.4438454392002408,
|
| 385 |
-
"grad_norm": 1.3145599365234375,
|
| 386 |
-
"learning_rate": 4.807265226157255e-05,
|
| 387 |
-
"loss": 0.5187122344970703,
|
| 388 |
-
"step": 1060
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.4522198814493019,
|
| 392 |
-
"grad_norm": 0.7199428081512451,
|
| 393 |
-
"learning_rate": 4.7999674883806465e-05,
|
| 394 |
-
"loss": 0.3027026653289795,
|
| 395 |
-
"step": 1080
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.46059432369836306,
|
| 399 |
-
"grad_norm": 0.2537328600883484,
|
| 400 |
-
"learning_rate": 4.792539895488213e-05,
|
| 401 |
-
"loss": 0.44816179275512696,
|
| 402 |
-
"step": 1100
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.4689687659474242,
|
| 406 |
-
"grad_norm": 0.6229515075683594,
|
| 407 |
-
"learning_rate": 4.784982866838462e-05,
|
| 408 |
-
"loss": 0.2838144063949585,
|
| 409 |
-
"step": 1120
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.47734320819648535,
|
| 413 |
-
"grad_norm": 0.32296815514564514,
|
| 414 |
-
"learning_rate": 4.777296829097785e-05,
|
| 415 |
-
"loss": 0.4447695255279541,
|
| 416 |
-
"step": 1140
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.4857176504455465,
|
| 420 |
-
"grad_norm": 0.8773243427276611,
|
| 421 |
-
"learning_rate": 4.769482216216369e-05,
|
| 422 |
-
"loss": 0.3049264192581177,
|
| 423 |
-
"step": 1160
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.49409209269460763,
|
| 427 |
-
"grad_norm": 6.263951301574707,
|
| 428 |
-
"learning_rate": 4.761539469403695e-05,
|
| 429 |
-
"loss": 0.2543496131896973,
|
| 430 |
-
"step": 1180
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.5024665349436688,
|
| 434 |
-
"grad_norm": 0.31666627526283264,
|
| 435 |
-
"learning_rate": 4.7534690371036274e-05,
|
| 436 |
-
"loss": 0.22089865207672119,
|
| 437 |
-
"step": 1200
|
| 438 |
-
},
|
| 439 |
-
{
|
| 440 |
-
"epoch": 0.5108409771927299,
|
| 441 |
-
"grad_norm": 1.305454134941101,
|
| 442 |
-
"learning_rate": 4.7452713749690934e-05,
|
| 443 |
-
"loss": 0.30597648620605467,
|
| 444 |
-
"step": 1220
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.5192154194417911,
|
| 448 |
-
"grad_norm": 0.7708957195281982,
|
| 449 |
-
"learning_rate": 4.73694694583636e-05,
|
| 450 |
-
"loss": 0.31495089530944825,
|
| 451 |
-
"step": 1240
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.5275898616908522,
|
| 455 |
-
"grad_norm": 1.2214106321334839,
|
| 456 |
-
"learning_rate": 4.728496219698902e-05,
|
| 457 |
-
"loss": 0.2793067455291748,
|
| 458 |
-
"step": 1260
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.5359643039399133,
|
| 462 |
-
"grad_norm": 0.6414151191711426,
|
| 463 |
-
"learning_rate": 4.719919673680863e-05,
|
| 464 |
-
"loss": 0.24640004634857177,
|
| 465 |
-
"step": 1280
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 0.5443387461889745,
|
| 469 |
-
"grad_norm": 0.4211215376853943,
|
| 470 |
-
"learning_rate": 4.711217792010122e-05,
|
| 471 |
-
"loss": 0.21954472064971925,
|
| 472 |
-
"step": 1300
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.5527131884380356,
|
| 476 |
-
"grad_norm": 1.1475203037261963,
|
| 477 |
-
"learning_rate": 4.70239106599095e-05,
|
| 478 |
-
"loss": 0.29560739994049073,
|
| 479 |
-
"step": 1320
|
| 480 |
-
},
|
| 481 |
-
{
|
| 482 |
-
"epoch": 0.5610876306870968,
|
| 483 |
-
"grad_norm": 1.3786718845367432,
|
| 484 |
-
"learning_rate": 4.6934399939762746e-05,
|
| 485 |
-
"loss": 0.2873764276504517,
|
| 486 |
-
"step": 1340
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 0.5694620729361579,
|
| 490 |
-
"grad_norm": 1.0504395961761475,
|
| 491 |
-
"learning_rate": 4.684365081339541e-05,
|
| 492 |
-
"loss": 0.35107598304748533,
|
| 493 |
-
"step": 1360
|
| 494 |
-
},
|
| 495 |
-
{
|
| 496 |
-
"epoch": 0.5778365151852191,
|
| 497 |
-
"grad_norm": 2.374439001083374,
|
| 498 |
-
"learning_rate": 4.675166840446179e-05,
|
| 499 |
-
"loss": 0.33103017807006835,
|
| 500 |
-
"step": 1380
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.5862109574342803,
|
| 504 |
-
"grad_norm": 0.30569833517074585,
|
| 505 |
-
"learning_rate": 4.6658457906246754e-05,
|
| 506 |
-
"loss": 0.15970700979232788,
|
| 507 |
-
"step": 1400
|
| 508 |
-
},
|
| 509 |
-
{
|
| 510 |
-
"epoch": 0.5945853996833415,
|
| 511 |
-
"grad_norm": 2.7184505462646484,
|
| 512 |
-
"learning_rate": 4.656402458137255e-05,
|
| 513 |
-
"loss": 0.18424086570739745,
|
| 514 |
-
"step": 1420
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.6029598419324026,
|
| 518 |
-
"grad_norm": 0.9072023630142212,
|
| 519 |
-
"learning_rate": 4.646837376150164e-05,
|
| 520 |
-
"loss": 0.2588572263717651,
|
| 521 |
-
"step": 1440
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 0.6113342841814637,
|
| 525 |
-
"grad_norm": 0.4133734405040741,
|
| 526 |
-
"learning_rate": 4.637151084703571e-05,
|
| 527 |
-
"loss": 0.20942835807800292,
|
| 528 |
-
"step": 1460
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 0.6197087264305249,
|
| 532 |
-
"grad_norm": 0.8726677298545837,
|
| 533 |
-
"learning_rate": 4.6273441306810724e-05,
|
| 534 |
-
"loss": 0.36864147186279295,
|
| 535 |
-
"step": 1480
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"epoch": 0.628083168679586,
|
| 539 |
-
"grad_norm": 0.9570859670639038,
|
| 540 |
-
"learning_rate": 4.617417067778825e-05,
|
| 541 |
-
"loss": 0.35473387241363524,
|
| 542 |
-
"step": 1500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.6364576109286472,
|
| 546 |
-
"grad_norm": 1.6990162134170532,
|
| 547 |
-
"learning_rate": 4.6073704564742734e-05,
|
| 548 |
-
"loss": 0.28413536548614504,
|
| 549 |
-
"step": 1520
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.6448320531777083,
|
| 553 |
-
"grad_norm": 0.5367034673690796,
|
| 554 |
-
"learning_rate": 4.5972048639945105e-05,
|
| 555 |
-
"loss": 0.23524901866912842,
|
| 556 |
-
"step": 1540
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 0.6532064954267695,
|
| 560 |
-
"grad_norm": 0.1774810403585434,
|
| 561 |
-
"learning_rate": 4.5869208642842536e-05,
|
| 562 |
-
"loss": 0.16743130683898927,
|
| 563 |
-
"step": 1560
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 0.6615809376758306,
|
| 567 |
-
"grad_norm": 0.4111934304237366,
|
| 568 |
-
"learning_rate": 4.5765190379734384e-05,
|
| 569 |
-
"loss": 0.2366396427154541,
|
| 570 |
-
"step": 1580
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.6699553799248917,
|
| 574 |
-
"grad_norm": 1.3504371643066406,
|
| 575 |
-
"learning_rate": 4.565999972344436e-05,
|
| 576 |
-
"loss": 0.17410001754760743,
|
| 577 |
-
"step": 1600
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 0.6783298221739529,
|
| 581 |
-
"grad_norm": 0.26619571447372437,
|
| 582 |
-
"learning_rate": 4.5553642612988954e-05,
|
| 583 |
-
"loss": 0.21413214206695558,
|
| 584 |
-
"step": 1620
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.686704264423014,
|
| 588 |
-
"grad_norm": 1.2675085067749023,
|
| 589 |
-
"learning_rate": 4.544612505324213e-05,
|
| 590 |
-
"loss": 0.22462894916534423,
|
| 591 |
-
"step": 1640
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 0.6950787066720752,
|
| 595 |
-
"grad_norm": 0.2958528995513916,
|
| 596 |
-
"learning_rate": 4.533745311459631e-05,
|
| 597 |
-
"loss": 0.17030240297317506,
|
| 598 |
-
"step": 1660
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.7034531489211363,
|
| 602 |
-
"grad_norm": 0.29442816972732544,
|
| 603 |
-
"learning_rate": 4.5227632932619604e-05,
|
| 604 |
-
"loss": 0.30537502765655516,
|
| 605 |
-
"step": 1680
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 0.7118275911701974,
|
| 609 |
-
"grad_norm": 1.2775323390960693,
|
| 610 |
-
"learning_rate": 4.5116670707709396e-05,
|
| 611 |
-
"loss": 0.21231284141540527,
|
| 612 |
-
"step": 1700
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.7202020334192586,
|
| 616 |
-
"grad_norm": 0.6058893203735352,
|
| 617 |
-
"learning_rate": 4.5004572704742345e-05,
|
| 618 |
-
"loss": 0.2254181146621704,
|
| 619 |
-
"step": 1720
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"epoch": 0.7285764756683197,
|
| 623 |
-
"grad_norm": 0.67788165807724,
|
| 624 |
-
"learning_rate": 4.489134525272057e-05,
|
| 625 |
-
"loss": 0.2267812728881836,
|
| 626 |
-
"step": 1740
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.7369509179173809,
|
| 630 |
-
"grad_norm": 0.5695115923881531,
|
| 631 |
-
"learning_rate": 4.477699474441441e-05,
|
| 632 |
-
"loss": 0.2287515878677368,
|
| 633 |
-
"step": 1760
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 0.745325360166442,
|
| 637 |
-
"grad_norm": 1.9975258111953735,
|
| 638 |
-
"learning_rate": 4.466152763600142e-05,
|
| 639 |
-
"loss": 0.15261965990066528,
|
| 640 |
-
"step": 1780
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.7536998024155032,
|
| 644 |
-
"grad_norm": 0.8055058717727661,
|
| 645 |
-
"learning_rate": 4.454495044670189e-05,
|
| 646 |
-
"loss": 0.2953073740005493,
|
| 647 |
-
"step": 1800
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.7620742446645643,
|
| 651 |
-
"grad_norm": 0.3822990357875824,
|
| 652 |
-
"learning_rate": 4.4427269758410806e-05,
|
| 653 |
-
"loss": 0.22958216667175294,
|
| 654 |
-
"step": 1820
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.7704486869136254,
|
| 658 |
-
"grad_norm": 1.03629469871521,
|
| 659 |
-
"learning_rate": 4.4308492215326145e-05,
|
| 660 |
-
"loss": 0.2106247663497925,
|
| 661 |
-
"step": 1840
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.7788231291626866,
|
| 665 |
-
"grad_norm": 0.15963371098041534,
|
| 666 |
-
"learning_rate": 4.418862452357387e-05,
|
| 667 |
-
"loss": 0.14056196212768554,
|
| 668 |
-
"step": 1860
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.7871975714117477,
|
| 672 |
-
"grad_norm": 0.46281635761260986,
|
| 673 |
-
"learning_rate": 4.4067673450829184e-05,
|
| 674 |
-
"loss": 0.2616506814956665,
|
| 675 |
-
"step": 1880
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.7955720136608089,
|
| 679 |
-
"grad_norm": 0.8718962669372559,
|
| 680 |
-
"learning_rate": 4.394564582593452e-05,
|
| 681 |
-
"loss": 0.20136590003967286,
|
| 682 |
-
"step": 1900
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.80394645590987,
|
| 686 |
-
"grad_norm": 0.16621536016464233,
|
| 687 |
-
"learning_rate": 4.382254853851396e-05,
|
| 688 |
-
"loss": 0.1678437829017639,
|
| 689 |
-
"step": 1920
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.8123208981589312,
|
| 693 |
-
"grad_norm": 0.31210511922836304,
|
| 694 |
-
"learning_rate": 4.3698388538584214e-05,
|
| 695 |
-
"loss": 0.1877954602241516,
|
| 696 |
-
"step": 1940
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 0.8206953404079924,
|
| 700 |
-
"grad_norm": 0.22836421430110931,
|
| 701 |
-
"learning_rate": 4.357317283616228e-05,
|
| 702 |
-
"loss": 0.1965322971343994,
|
| 703 |
-
"step": 1960
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.8290697826570536,
|
| 707 |
-
"grad_norm": 1.5269449949264526,
|
| 708 |
-
"learning_rate": 4.3446908500869624e-05,
|
| 709 |
-
"loss": 0.14575521945953368,
|
| 710 |
-
"step": 1980
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.8374442249061147,
|
| 714 |
-
"grad_norm": 0.541411817073822,
|
| 715 |
-
"learning_rate": 4.3319602661533054e-05,
|
| 716 |
-
"loss": 0.1215706706047058,
|
| 717 |
-
"step": 2000
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.8458186671551758,
|
| 721 |
-
"grad_norm": 1.0127756595611572,
|
| 722 |
-
"learning_rate": 4.319126250578223e-05,
|
| 723 |
-
"loss": 0.1414382576942444,
|
| 724 |
-
"step": 2020
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.854193109404237,
|
| 728 |
-
"grad_norm": 0.496611624956131,
|
| 729 |
-
"learning_rate": 4.306189527964382e-05,
|
| 730 |
-
"loss": 0.1537245512008667,
|
| 731 |
-
"step": 2040
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.8625675516532981,
|
| 735 |
-
"grad_norm": 0.17574390769004822,
|
| 736 |
-
"learning_rate": 4.2931508287132435e-05,
|
| 737 |
-
"loss": 0.14648995399475098,
|
| 738 |
-
"step": 2060
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.8709419939023593,
|
| 742 |
-
"grad_norm": 1.1143279075622559,
|
| 743 |
-
"learning_rate": 4.2800108889838244e-05,
|
| 744 |
-
"loss": 0.1767507791519165,
|
| 745 |
-
"step": 2080
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.8793164361514204,
|
| 749 |
-
"grad_norm": 0.40195536613464355,
|
| 750 |
-
"learning_rate": 4.26677045065113e-05,
|
| 751 |
-
"loss": 0.12700635194778442,
|
| 752 |
-
"step": 2100
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.8876908784004816,
|
| 756 |
-
"grad_norm": 0.5011544823646545,
|
| 757 |
-
"learning_rate": 4.253430261264272e-05,
|
| 758 |
-
"loss": 0.1776489496231079,
|
| 759 |
-
"step": 2120
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.8960653206495427,
|
| 763 |
-
"grad_norm": 0.9425447583198547,
|
| 764 |
-
"learning_rate": 4.239991074004262e-05,
|
| 765 |
-
"loss": 0.17866064310073854,
|
| 766 |
-
"step": 2140
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.9044397628986038,
|
| 770 |
-
"grad_norm": 1.5652395486831665,
|
| 771 |
-
"learning_rate": 4.226453647641484e-05,
|
| 772 |
-
"loss": 0.18456958532333373,
|
| 773 |
-
"step": 2160
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.912814205147665,
|
| 777 |
-
"grad_norm": 0.9656435251235962,
|
| 778 |
-
"learning_rate": 4.2128187464928604e-05,
|
| 779 |
-
"loss": 0.20541505813598632,
|
| 780 |
-
"step": 2180
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.9211886473967261,
|
| 784 |
-
"grad_norm": 0.20453469455242157,
|
| 785 |
-
"learning_rate": 4.1990871403786934e-05,
|
| 786 |
-
"loss": 0.14855676889419556,
|
| 787 |
-
"step": 2200
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.9295630896457873,
|
| 791 |
-
"grad_norm": 0.19641469419002533,
|
| 792 |
-
"learning_rate": 4.185259604579203e-05,
|
| 793 |
-
"loss": 0.09722235202789306,
|
| 794 |
-
"step": 2220
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.9379375318948484,
|
| 798 |
-
"grad_norm": 0.34978342056274414,
|
| 799 |
-
"learning_rate": 4.171336919790759e-05,
|
| 800 |
-
"loss": 0.1498737931251526,
|
| 801 |
-
"step": 2240
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.9463119741439096,
|
| 805 |
-
"grad_norm": 1.2353084087371826,
|
| 806 |
-
"learning_rate": 4.1573198720817955e-05,
|
| 807 |
-
"loss": 0.17144311666488649,
|
| 808 |
-
"step": 2260
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.9546864163929707,
|
| 812 |
-
"grad_norm": 1.176924228668213,
|
| 813 |
-
"learning_rate": 4.143209252848433e-05,
|
| 814 |
-
"loss": 0.20422227382659913,
|
| 815 |
-
"step": 2280
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.9630608586420318,
|
| 819 |
-
"grad_norm": 0.27958446741104126,
|
| 820 |
-
"learning_rate": 4.1290058587698054e-05,
|
| 821 |
-
"loss": 0.1507509708404541,
|
| 822 |
-
"step": 2300
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.971435300891093,
|
| 826 |
-
"grad_norm": 1.0945500135421753,
|
| 827 |
-
"learning_rate": 4.114710491763063e-05,
|
| 828 |
-
"loss": 0.21894502639770508,
|
| 829 |
-
"step": 2320
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.9798097431401541,
|
| 833 |
-
"grad_norm": 0.3587752878665924,
|
| 834 |
-
"learning_rate": 4.100323958938112e-05,
|
| 835 |
-
"loss": 0.09666566252708435,
|
| 836 |
-
"step": 2340
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.9881841853892153,
|
| 840 |
-
"grad_norm": 1.3135223388671875,
|
| 841 |
-
"learning_rate": 4.085847072552036e-05,
|
| 842 |
-
"loss": 0.16662789583206178,
|
| 843 |
-
"step": 2360
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 0.9965586276382764,
|
| 847 |
-
"grad_norm": 0.44114890694618225,
|
| 848 |
-
"learning_rate": 4.0712806499632397e-05,
|
| 849 |
-
"loss": 0.1317327857017517,
|
| 850 |
-
"step": 2380
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 1.0046059432369836,
|
| 854 |
-
"grad_norm": 0.4595654308795929,
|
| 855 |
-
"learning_rate": 4.0566255135853014e-05,
|
| 856 |
-
"loss": 0.16484848260879517,
|
| 857 |
-
"step": 2400
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 1.0129803854860449,
|
| 861 |
-
"grad_norm": 0.2524283528327942,
|
| 862 |
-
"learning_rate": 4.04188249084054e-05,
|
| 863 |
-
"loss": 0.09629667997360229,
|
| 864 |
-
"step": 2420
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 1.0213548277351059,
|
| 868 |
-
"grad_norm": 0.640849769115448,
|
| 869 |
-
"learning_rate": 4.0270524141133e-05,
|
| 870 |
-
"loss": 0.08449938893318176,
|
| 871 |
-
"step": 2440
|
| 872 |
-
},
|
| 873 |
-
{
|
| 874 |
-
"epoch": 1.0297292699841671,
|
| 875 |
-
"grad_norm": 0.9351454377174377,
|
| 876 |
-
"learning_rate": 4.01213612070295e-05,
|
| 877 |
-
"loss": 0.13373035192489624,
|
| 878 |
-
"step": 2460
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 1.0381037122332282,
|
| 882 |
-
"grad_norm": 0.1494932770729065,
|
| 883 |
-
"learning_rate": 3.997134452776617e-05,
|
| 884 |
-
"loss": 0.09310936331748962,
|
| 885 |
-
"step": 2480
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 1.0464781544822894,
|
| 889 |
-
"grad_norm": 0.7994132041931152,
|
| 890 |
-
"learning_rate": 3.982048257321631e-05,
|
| 891 |
-
"loss": 0.11848243474960327,
|
| 892 |
-
"step": 2500
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 1.0548525967313505,
|
| 896 |
-
"grad_norm": 1.789854884147644,
|
| 897 |
-
"learning_rate": 3.966878386097711e-05,
|
| 898 |
-
"loss": 0.08053499460220337,
|
| 899 |
-
"step": 2520
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 1.0632270389804117,
|
| 903 |
-
"grad_norm": 0.09759683907032013,
|
| 904 |
-
"learning_rate": 3.95162569558887e-05,
|
| 905 |
-
"loss": 0.13048415184020995,
|
| 906 |
-
"step": 2540
|
| 907 |
-
},
|
| 908 |
-
{
|
| 909 |
-
"epoch": 1.0716014812294727,
|
| 910 |
-
"grad_norm": 1.745640754699707,
|
| 911 |
-
"learning_rate": 3.936291046955061e-05,
|
| 912 |
-
"loss": 0.06222975254058838,
|
| 913 |
-
"step": 2560
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 1.079975923478534,
|
| 917 |
-
"grad_norm": 0.9128367304801941,
|
| 918 |
-
"learning_rate": 3.920875305983553e-05,
|
| 919 |
-
"loss": 0.07491644620895385,
|
| 920 |
-
"step": 2580
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 1.088350365727595,
|
| 924 |
-
"grad_norm": 1.008280873298645,
|
| 925 |
-
"learning_rate": 3.905379343040054e-05,
|
| 926 |
-
"loss": 0.14277656078338624,
|
| 927 |
-
"step": 2600
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 1.0967248079766563,
|
| 931 |
-
"grad_norm": 0.48391154408454895,
|
| 932 |
-
"learning_rate": 3.889804033019567e-05,
|
| 933 |
-
"loss": 0.0745099127292633,
|
| 934 |
-
"step": 2620
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 1.1050992502257173,
|
| 938 |
-
"grad_norm": 0.2227552831172943,
|
| 939 |
-
"learning_rate": 3.8741502552969956e-05,
|
| 940 |
-
"loss": 0.09417952299118042,
|
| 941 |
-
"step": 2640
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 1.1134736924747786,
|
| 945 |
-
"grad_norm": 0.16306816041469574,
|
| 946 |
-
"learning_rate": 3.8584188936774925e-05,
|
| 947 |
-
"loss": 0.12839992046356202,
|
| 948 |
-
"step": 2660
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 1.1218481347238396,
|
| 952 |
-
"grad_norm": 0.07621652632951736,
|
| 953 |
-
"learning_rate": 3.842610836346563e-05,
|
| 954 |
-
"loss": 0.10739836692810059,
|
| 955 |
-
"step": 2680
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 1.1302225769729008,
|
| 959 |
-
"grad_norm": 0.5850180983543396,
|
| 960 |
-
"learning_rate": 3.8267269758199175e-05,
|
| 961 |
-
"loss": 0.07980061173439026,
|
| 962 |
-
"step": 2700
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 1.1385970192219619,
|
| 966 |
-
"grad_norm": 2.3212532997131348,
|
| 967 |
-
"learning_rate": 3.8107682088930794e-05,
|
| 968 |
-
"loss": 0.08940492272377014,
|
| 969 |
-
"step": 2720
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 1.1469714614710231,
|
| 973 |
-
"grad_norm": 0.10123835504055023,
|
| 974 |
-
"learning_rate": 3.7947354365907536e-05,
|
| 975 |
-
"loss": 0.09007878303527832,
|
| 976 |
-
"step": 2740
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 1.1553459037200842,
|
| 980 |
-
"grad_norm": 0.9358357191085815,
|
| 981 |
-
"learning_rate": 3.7786295641159544e-05,
|
| 982 |
-
"loss": 0.08590203523635864,
|
| 983 |
-
"step": 2760
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 1.1637203459691454,
|
| 987 |
-
"grad_norm": 0.22766231000423431,
|
| 988 |
-
"learning_rate": 3.7624515007989004e-05,
|
| 989 |
-
"loss": 0.12319990396499633,
|
| 990 |
-
"step": 2780
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 1.1720947882182067,
|
| 994 |
-
"grad_norm": 1.0723813772201538,
|
| 995 |
-
"learning_rate": 3.746202160045669e-05,
|
| 996 |
-
"loss": 0.07759063839912414,
|
| 997 |
-
"step": 2800
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 1.1804692304672677,
|
| 1001 |
-
"grad_norm": 0.2642482817173004,
|
| 1002 |
-
"learning_rate": 3.729882459286632e-05,
|
| 1003 |
-
"loss": 0.07289831042289734,
|
| 1004 |
-
"step": 2820
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 1.1888436727163287,
|
| 1008 |
-
"grad_norm": 0.11382978409528732,
|
| 1009 |
-
"learning_rate": 3.713493319924655e-05,
|
| 1010 |
-
"loss": 0.08417699337005616,
|
| 1011 |
-
"step": 2840
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 1.19721811496539,
|
| 1015 |
-
"grad_norm": 0.6055602431297302,
|
| 1016 |
-
"learning_rate": 3.697035667283072e-05,
|
| 1017 |
-
"loss": 0.0856675386428833,
|
| 1018 |
-
"step": 2860
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 1.2055925572144512,
|
| 1022 |
-
"grad_norm": 0.12562307715415955,
|
| 1023 |
-
"learning_rate": 3.68051043055345e-05,
|
| 1024 |
-
"loss": 0.10780172348022461,
|
| 1025 |
-
"step": 2880
|
| 1026 |
-
},
|
| 1027 |
-
{
|
| 1028 |
-
"epoch": 1.2139669994635123,
|
| 1029 |
-
"grad_norm": 0.06781217455863953,
|
| 1030 |
-
"learning_rate": 3.66391854274312e-05,
|
| 1031 |
-
"loss": 0.055737143754959105,
|
| 1032 |
-
"step": 2900
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 1.2223414417125735,
|
| 1036 |
-
"grad_norm": 0.23094801604747772,
|
| 1037 |
-
"learning_rate": 3.647260940622504e-05,
|
| 1038 |
-
"loss": 0.08827725648880005,
|
| 1039 |
-
"step": 2920
|
| 1040 |
-
},
|
| 1041 |
-
{
|
| 1042 |
-
"epoch": 1.2307158839616346,
|
| 1043 |
-
"grad_norm": 0.26578766107559204,
|
| 1044 |
-
"learning_rate": 3.6305385646722224e-05,
|
| 1045 |
-
"loss": 0.10403858423233033,
|
| 1046 |
-
"step": 2940
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 1.2390903262106958,
|
| 1050 |
-
"grad_norm": 1.2067272663116455,
|
| 1051 |
-
"learning_rate": 3.613752359029998e-05,
|
| 1052 |
-
"loss": 0.09574756622314454,
|
| 1053 |
-
"step": 2960
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 1.2474647684597568,
|
| 1057 |
-
"grad_norm": 0.28369495272636414,
|
| 1058 |
-
"learning_rate": 3.5969032714373495e-05,
|
| 1059 |
-
"loss": 0.12156195640563965,
|
| 1060 |
-
"step": 2980
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 1.255839210708818,
|
| 1064 |
-
"grad_norm": 0.6278151273727417,
|
| 1065 |
-
"learning_rate": 3.57999225318608e-05,
|
| 1066 |
-
"loss": 0.09732474088668823,
|
| 1067 |
-
"step": 3000
|
| 1068 |
-
},
|
| 1069 |
-
{
|
| 1070 |
-
"epoch": 1.2642136529578791,
|
| 1071 |
-
"grad_norm": 0.7716904282569885,
|
| 1072 |
-
"learning_rate": 3.563020259064569e-05,
|
| 1073 |
-
"loss": 0.10135951042175292,
|
| 1074 |
-
"step": 3020
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 1.2725880952069404,
|
| 1078 |
-
"grad_norm": 0.07069838047027588,
|
| 1079 |
-
"learning_rate": 3.5459882473038695e-05,
|
| 1080 |
-
"loss": 0.2272268295288086,
|
| 1081 |
-
"step": 3040
|
| 1082 |
-
},
|
| 1083 |
-
{
|
| 1084 |
-
"epoch": 1.2809625374560014,
|
| 1085 |
-
"grad_norm": 0.5034748911857605,
|
| 1086 |
-
"learning_rate": 3.528897179523598e-05,
|
| 1087 |
-
"loss": 0.07878611087799073,
|
| 1088 |
-
"step": 3060
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 1.2893369797050627,
|
| 1092 |
-
"grad_norm": 0.1114577129483223,
|
| 1093 |
-
"learning_rate": 3.5117480206776506e-05,
|
| 1094 |
-
"loss": 0.07403420805931091,
|
| 1095 |
-
"step": 3080
|
| 1096 |
-
},
|
| 1097 |
-
{
|
| 1098 |
-
"epoch": 1.2977114219541237,
|
| 1099 |
-
"grad_norm": 0.1419895440340042,
|
| 1100 |
-
"learning_rate": 3.494541738999717e-05,
|
| 1101 |
-
"loss": 0.06619678139686584,
|
| 1102 |
-
"step": 3100
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 1.306085864203185,
|
| 1106 |
-
"grad_norm": 0.3242378234863281,
|
| 1107 |
-
"learning_rate": 3.4772793059486176e-05,
|
| 1108 |
-
"loss": 0.07819657921791076,
|
| 1109 |
-
"step": 3120
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"epoch": 1.314460306452246,
|
| 1113 |
-
"grad_norm": 0.13734346628189087,
|
| 1114 |
-
"learning_rate": 3.4599616961534506e-05,
|
| 1115 |
-
"loss": 0.07510924339294434,
|
| 1116 |
-
"step": 3140
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 1.3228347487013072,
|
| 1120 |
-
"grad_norm": 0.9328349232673645,
|
| 1121 |
-
"learning_rate": 3.44258988735857e-05,
|
| 1122 |
-
"loss": 0.07773483395576478,
|
| 1123 |
-
"step": 3160
|
| 1124 |
-
},
|
| 1125 |
-
{
|
| 1126 |
-
"epoch": 1.3312091909503683,
|
| 1127 |
-
"grad_norm": 0.14395828545093536,
|
| 1128 |
-
"learning_rate": 3.4251648603683824e-05,
|
| 1129 |
-
"loss": 0.07977863550186157,
|
| 1130 |
-
"step": 3180
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.3395836331994295,
|
| 1134 |
-
"grad_norm": 0.4411904513835907,
|
| 1135 |
-
"learning_rate": 3.407687598991965e-05,
|
| 1136 |
-
"loss": 0.07483948469161987,
|
| 1137 |
-
"step": 3200
|
| 1138 |
-
},
|
| 1139 |
-
{
|
| 1140 |
-
"epoch": 1.3479580754484906,
|
| 1141 |
-
"grad_norm": 0.2287372499704361,
|
| 1142 |
-
"learning_rate": 3.390159089987529e-05,
|
| 1143 |
-
"loss": 0.08884957432746887,
|
| 1144 |
-
"step": 3220
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 1.3563325176975518,
|
| 1148 |
-
"grad_norm": 0.7106556296348572,
|
| 1149 |
-
"learning_rate": 3.3725803230067006e-05,
|
| 1150 |
-
"loss": 0.10559600591659546,
|
| 1151 |
-
"step": 3240
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 1.3647069599466128,
|
| 1155 |
-
"grad_norm": 0.7979939579963684,
|
| 1156 |
-
"learning_rate": 3.354952290538649e-05,
|
| 1157 |
-
"loss": 0.11070560216903687,
|
| 1158 |
-
"step": 3260
|
| 1159 |
-
},
|
| 1160 |
-
{
|
| 1161 |
-
"epoch": 1.373081402195674,
|
| 1162 |
-
"grad_norm": 0.2569740414619446,
|
| 1163 |
-
"learning_rate": 3.337275987854052e-05,
|
| 1164 |
-
"loss": 0.0596321702003479,
|
| 1165 |
-
"step": 3280
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 1.3814558444447353,
|
| 1169 |
-
"grad_norm": 0.16822633147239685,
|
| 1170 |
-
"learning_rate": 3.3195524129488986e-05,
|
| 1171 |
-
"loss": 0.08175230622291565,
|
| 1172 |
-
"step": 3300
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.3898302866937964,
|
| 1176 |
-
"grad_norm": 0.1559642255306244,
|
| 1177 |
-
"learning_rate": 3.3017825664881496e-05,
|
| 1178 |
-
"loss": 0.07060136198997498,
|
| 1179 |
-
"step": 3320
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 1.3982047289428574,
|
| 1183 |
-
"grad_norm": 0.565130352973938,
|
| 1184 |
-
"learning_rate": 3.283967451749233e-05,
|
| 1185 |
-
"loss": 0.06781386733055114,
|
| 1186 |
-
"step": 3340
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 1.4065791711919187,
|
| 1190 |
-
"grad_norm": 0.05673038586974144,
|
| 1191 |
-
"learning_rate": 3.266108074565406e-05,
|
| 1192 |
-
"loss": 0.043686714768409726,
|
| 1193 |
-
"step": 3360
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 1.41495361344098,
|
| 1197 |
-
"grad_norm": 0.5792385935783386,
|
| 1198 |
-
"learning_rate": 3.2482054432689616e-05,
|
| 1199 |
-
"loss": 0.05318898558616638,
|
| 1200 |
-
"step": 3380
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 1.423328055690041,
|
| 1204 |
-
"grad_norm": 0.8359180688858032,
|
| 1205 |
-
"learning_rate": 3.2302605686343016e-05,
|
| 1206 |
-
"loss": 0.07649694681167603,
|
| 1207 |
-
"step": 3400
|
| 1208 |
-
},
|
| 1209 |
-
{
|
| 1210 |
-
"epoch": 1.431702497939102,
|
| 1211 |
-
"grad_norm": 0.11994684487581253,
|
| 1212 |
-
"learning_rate": 3.212274463820866e-05,
|
| 1213 |
-
"loss": 0.043499928712844846,
|
| 1214 |
-
"step": 3420
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.4400769401881632,
|
| 1218 |
-
"grad_norm": 0.3480674624443054,
|
| 1219 |
-
"learning_rate": 3.194248144315933e-05,
|
| 1220 |
-
"loss": 0.07629216313362122,
|
| 1221 |
-
"step": 3440
|
| 1222 |
-
},
|
| 1223 |
-
{
|
| 1224 |
-
"epoch": 1.4484513824372245,
|
| 1225 |
-
"grad_norm": 0.349407821893692,
|
| 1226 |
-
"learning_rate": 3.176182627877284e-05,
|
| 1227 |
-
"loss": 0.06937903761863709,
|
| 1228 |
-
"step": 3460
|
| 1229 |
-
},
|
| 1230 |
-
{
|
| 1231 |
-
"epoch": 1.4568258246862855,
|
| 1232 |
-
"grad_norm": 0.32339081168174744,
|
| 1233 |
-
"learning_rate": 3.15807893447574e-05,
|
| 1234 |
-
"loss": 0.0664923369884491,
|
| 1235 |
-
"step": 3480
|
| 1236 |
-
},
|
| 1237 |
-
{
|
| 1238 |
-
"epoch": 1.4652002669353466,
|
| 1239 |
-
"grad_norm": 1.2784419059753418,
|
| 1240 |
-
"learning_rate": 3.1399380862375785e-05,
|
| 1241 |
-
"loss": 0.043991288542747496,
|
| 1242 |
-
"step": 3500
|
| 1243 |
-
},
|
| 1244 |
-
{
|
| 1245 |
-
"epoch": 1.4735747091844078,
|
| 1246 |
-
"grad_norm": 0.9719874262809753,
|
| 1247 |
-
"learning_rate": 3.121761107386817e-05,
|
| 1248 |
-
"loss": 0.0807355523109436,
|
| 1249 |
-
"step": 3520
|
| 1250 |
-
},
|
| 1251 |
-
{
|
| 1252 |
-
"epoch": 1.481949151433469,
|
| 1253 |
-
"grad_norm": 0.06417154520750046,
|
| 1254 |
-
"learning_rate": 3.103549024187398e-05,
|
| 1255 |
-
"loss": 0.06301149725914001,
|
| 1256 |
-
"step": 3540
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.49032359368253,
|
| 1260 |
-
"grad_norm": 0.5472319722175598,
|
| 1261 |
-
"learning_rate": 3.085302864885235e-05,
|
| 1262 |
-
"loss": 0.13920137882232667,
|
| 1263 |
-
"step": 3560
|
| 1264 |
-
},
|
| 1265 |
-
{
|
| 1266 |
-
"epoch": 1.4986980359315913,
|
| 1267 |
-
"grad_norm": 0.4602014720439911,
|
| 1268 |
-
"learning_rate": 3.067023659650163e-05,
|
| 1269 |
-
"loss": 0.10979901552200318,
|
| 1270 |
-
"step": 3580
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.5070724781806524,
|
| 1274 |
-
"grad_norm": 0.48343533277511597,
|
| 1275 |
-
"learning_rate": 3.048712440517777e-05,
|
| 1276 |
-
"loss": 0.056929266452789305,
|
| 1277 |
-
"step": 3600
|
| 1278 |
-
},
|
| 1279 |
-
{
|
| 1280 |
-
"epoch": 1.5154469204297136,
|
| 1281 |
-
"grad_norm": 0.5795832276344299,
|
| 1282 |
-
"learning_rate": 3.030370241331159e-05,
|
| 1283 |
-
"loss": 0.06327124834060668,
|
| 1284 |
-
"step": 3620
|
| 1285 |
-
},
|
| 1286 |
-
{
|
| 1287 |
-
"epoch": 1.5238213626787747,
|
| 1288 |
-
"grad_norm": 0.07871934771537781,
|
| 1289 |
-
"learning_rate": 3.0119980976825156e-05,
|
| 1290 |
-
"loss": 0.05837821960449219,
|
| 1291 |
-
"step": 3640
|
| 1292 |
-
},
|
| 1293 |
-
{
|
| 1294 |
-
"epoch": 1.532195804927836,
|
| 1295 |
-
"grad_norm": 0.38906294107437134,
|
| 1296 |
-
"learning_rate": 2.9935970468546992e-05,
|
| 1297 |
-
"loss": 0.05847613215446472,
|
| 1298 |
-
"step": 3660
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.540570247176897,
|
| 1302 |
-
"grad_norm": 1.1948798894882202,
|
| 1303 |
-
"learning_rate": 2.9751681277626507e-05,
|
| 1304 |
-
"loss": 0.06106034517288208,
|
| 1305 |
-
"step": 3680
|
| 1306 |
-
},
|
| 1307 |
-
{
|
| 1308 |
-
"epoch": 1.5489446894259582,
|
| 1309 |
-
"grad_norm": 0.06581711769104004,
|
| 1310 |
-
"learning_rate": 2.956712380894742e-05,
|
| 1311 |
-
"loss": 0.06192805767059326,
|
| 1312 |
-
"step": 3700
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 1.5573191316750195,
|
| 1316 |
-
"grad_norm": 0.11205727607011795,
|
| 1317 |
-
"learning_rate": 2.9382308482540237e-05,
|
| 1318 |
-
"loss": 0.05574634075164795,
|
| 1319 |
-
"step": 3720
|
| 1320 |
-
},
|
| 1321 |
-
{
|
| 1322 |
-
"epoch": 1.5656935739240805,
|
| 1323 |
-
"grad_norm": 3.2353687286376953,
|
| 1324 |
-
"learning_rate": 2.919724573299406e-05,
|
| 1325 |
-
"loss": 0.07629644870758057,
|
| 1326 |
-
"step": 3740
|
| 1327 |
-
},
|
| 1328 |
-
{
|
| 1329 |
-
"epoch": 1.5740680161731415,
|
| 1330 |
-
"grad_norm": 0.07098077982664108,
|
| 1331 |
-
"learning_rate": 2.9011946008867342e-05,
|
| 1332 |
-
"loss": 0.030209916830062866,
|
| 1333 |
-
"step": 3760
|
| 1334 |
-
},
|
| 1335 |
-
{
|
| 1336 |
-
"epoch": 1.5824424584222028,
|
| 1337 |
-
"grad_norm": 0.28404733538627625,
|
| 1338 |
-
"learning_rate": 2.882641977209803e-05,
|
| 1339 |
-
"loss": 0.05649316906929016,
|
| 1340 |
-
"step": 3780
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.590816900671264,
|
| 1344 |
-
"grad_norm": 0.46935683488845825,
|
| 1345 |
-
"learning_rate": 2.8640677497412877e-05,
|
| 1346 |
-
"loss": 0.05797313451766968,
|
| 1347 |
-
"step": 3800
|
| 1348 |
-
},
|
| 1349 |
-
{
|
| 1350 |
-
"epoch": 1.599191342920325,
|
| 1351 |
-
"grad_norm": 0.5720381736755371,
|
| 1352 |
-
"learning_rate": 2.8454729671736013e-05,
|
| 1353 |
-
"loss": 0.05965529084205627,
|
| 1354 |
-
"step": 3820
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 1.607565785169386,
|
| 1358 |
-
"grad_norm": 0.1056889146566391,
|
| 1359 |
-
"learning_rate": 2.826858679359693e-05,
|
| 1360 |
-
"loss": 0.06270216703414917,
|
| 1361 |
-
"step": 3840
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 1.6159402274184473,
|
| 1365 |
-
"grad_norm": 0.17195488512516022,
|
| 1366 |
-
"learning_rate": 2.8082259372537635e-05,
|
| 1367 |
-
"loss": 0.08613481521606445,
|
| 1368 |
-
"step": 3860
|
| 1369 |
-
},
|
| 1370 |
-
{
|
| 1371 |
-
"epoch": 1.6243146696675086,
|
| 1372 |
-
"grad_norm": 0.21867765486240387,
|
| 1373 |
-
"learning_rate": 2.789575792851939e-05,
|
| 1374 |
-
"loss": 0.05220889449119568,
|
| 1375 |
-
"step": 3880
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 1.6326891119165696,
|
| 1379 |
-
"grad_norm": 0.05280565097928047,
|
| 1380 |
-
"learning_rate": 2.770909299132869e-05,
|
| 1381 |
-
"loss": 0.05668648481369019,
|
| 1382 |
-
"step": 3900
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.6410635541656307,
|
| 1386 |
-
"grad_norm": 1.2573331594467163,
|
| 1387 |
-
"learning_rate": 2.7522275099982797e-05,
|
| 1388 |
-
"loss": 0.09791213274002075,
|
| 1389 |
-
"step": 3920
|
| 1390 |
-
},
|
| 1391 |
-
{
|
| 1392 |
-
"epoch": 1.649437996414692,
|
| 1393 |
-
"grad_norm": 0.24661803245544434,
|
| 1394 |
-
"learning_rate": 2.7335314802134676e-05,
|
| 1395 |
-
"loss": 0.05671970248222351,
|
| 1396 |
-
"step": 3940
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 1.6578124386637532,
|
| 1400 |
-
"grad_norm": 0.11609286069869995,
|
| 1401 |
-
"learning_rate": 2.7148222653477505e-05,
|
| 1402 |
-
"loss": 0.052801978588104245,
|
| 1403 |
-
"step": 3960
|
| 1404 |
-
},
|
| 1405 |
-
{
|
| 1406 |
-
"epoch": 1.6661868809128142,
|
| 1407 |
-
"grad_norm": 0.0745048075914383,
|
| 1408 |
-
"learning_rate": 2.696100921714871e-05,
|
| 1409 |
-
"loss": 0.05944955348968506,
|
| 1410 |
-
"step": 3980
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"epoch": 1.6745613231618752,
|
| 1414 |
-
"grad_norm": 0.6762334108352661,
|
| 1415 |
-
"learning_rate": 2.677368506313354e-05,
|
| 1416 |
-
"loss": 0.04590807855129242,
|
| 1417 |
-
"step": 4000
|
| 1418 |
-
},
|
| 1419 |
-
{
|
| 1420 |
-
"epoch": 1.6829357654109365,
|
| 1421 |
-
"grad_norm": 0.0797283723950386,
|
| 1422 |
-
"learning_rate": 2.6586260767668353e-05,
|
| 1423 |
-
"loss": 0.05602290630340576,
|
| 1424 |
-
"step": 4020
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.6913102076599977,
|
| 1428 |
-
"grad_norm": 0.0805838331580162,
|
| 1429 |
-
"learning_rate": 2.6398746912643395e-05,
|
| 1430 |
-
"loss": 0.07643119692802429,
|
| 1431 |
-
"step": 4040
|
| 1432 |
-
},
|
| 1433 |
-
{
|
| 1434 |
-
"epoch": 1.6996846499090588,
|
| 1435 |
-
"grad_norm": 0.5817922353744507,
|
| 1436 |
-
"learning_rate": 2.621115408500545e-05,
|
| 1437 |
-
"loss": 0.06817017793655396,
|
| 1438 |
-
"step": 4060
|
| 1439 |
-
},
|
| 1440 |
-
{
|
| 1441 |
-
"epoch": 1.7080590921581198,
|
| 1442 |
-
"grad_norm": 0.13826222717761993,
|
| 1443 |
-
"learning_rate": 2.602349287616004e-05,
|
| 1444 |
-
"loss": 0.10364378690719604,
|
| 1445 |
-
"step": 4080
|
| 1446 |
-
},
|
| 1447 |
-
{
|
| 1448 |
-
"epoch": 1.716433534407181,
|
| 1449 |
-
"grad_norm": 0.7005400061607361,
|
| 1450 |
-
"learning_rate": 2.5835773881373456e-05,
|
| 1451 |
-
"loss": 0.0438168853521347,
|
| 1452 |
-
"step": 4100
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 1.7248079766562423,
|
| 1456 |
-
"grad_norm": 0.213337242603302,
|
| 1457 |
-
"learning_rate": 2.5648007699174566e-05,
|
| 1458 |
-
"loss": 0.07051296830177307,
|
| 1459 |
-
"step": 4120
|
| 1460 |
-
},
|
| 1461 |
-
{
|
| 1462 |
-
"epoch": 1.7331824189053033,
|
| 1463 |
-
"grad_norm": 0.4417899549007416,
|
| 1464 |
-
"learning_rate": 2.546020493075641e-05,
|
| 1465 |
-
"loss": 0.06489490866661071,
|
| 1466 |
-
"step": 4140
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.7415568611543644,
|
| 1470 |
-
"grad_norm": 1.6167367696762085,
|
| 1471 |
-
"learning_rate": 2.527237617937767e-05,
|
| 1472 |
-
"loss": 0.03677745163440704,
|
| 1473 |
-
"step": 4160
|
| 1474 |
-
},
|
| 1475 |
-
{
|
| 1476 |
-
"epoch": 1.7499313034034256,
|
| 1477 |
-
"grad_norm": 0.834720253944397,
|
| 1478 |
-
"learning_rate": 2.5084532049764027e-05,
|
| 1479 |
-
"loss": 0.05091303586959839,
|
| 1480 |
-
"step": 4180
|
| 1481 |
-
},
|
| 1482 |
-
{
|
| 1483 |
-
"epoch": 1.7583057456524869,
|
| 1484 |
-
"grad_norm": 0.5853901505470276,
|
| 1485 |
-
"learning_rate": 2.489668314750939e-05,
|
| 1486 |
-
"loss": 0.07355493307113647,
|
| 1487 |
-
"step": 4200
|
| 1488 |
-
},
|
| 1489 |
-
{
|
| 1490 |
-
"epoch": 1.7666801879015481,
|
| 1491 |
-
"grad_norm": 0.44475457072257996,
|
| 1492 |
-
"learning_rate": 2.4708840078477134e-05,
|
| 1493 |
-
"loss": 0.059920096397399904,
|
| 1494 |
-
"step": 4220
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 1.7750546301506092,
|
| 1498 |
-
"grad_norm": 0.16852276027202606,
|
| 1499 |
-
"learning_rate": 2.4521013448201314e-05,
|
| 1500 |
-
"loss": 0.060352939367294314,
|
| 1501 |
-
"step": 4240
|
| 1502 |
-
},
|
| 1503 |
-
{
|
| 1504 |
-
"epoch": 1.7834290723996702,
|
| 1505 |
-
"grad_norm": 0.15743663907051086,
|
| 1506 |
-
"learning_rate": 2.433321386128784e-05,
|
| 1507 |
-
"loss": 0.09045843482017517,
|
| 1508 |
-
"step": 4260
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.7918035146487314,
|
| 1512 |
-
"grad_norm": 0.4637702405452728,
|
| 1513 |
-
"learning_rate": 2.414545192081575e-05,
|
| 1514 |
-
"loss": 0.08557018041610717,
|
| 1515 |
-
"step": 4280
|
| 1516 |
-
},
|
| 1517 |
-
{
|
| 1518 |
-
"epoch": 1.8001779568977927,
|
| 1519 |
-
"grad_norm": 0.05946366488933563,
|
| 1520 |
-
"learning_rate": 2.395773822773863e-05,
|
| 1521 |
-
"loss": 0.07715467214584351,
|
| 1522 |
-
"step": 4300
|
| 1523 |
-
},
|
| 1524 |
-
{
|
| 1525 |
-
"epoch": 1.8085523991468537,
|
| 1526 |
-
"grad_norm": 0.1889827400445938,
|
| 1527 |
-
"learning_rate": 2.3770083380286004e-05,
|
| 1528 |
-
"loss": 0.05025853514671326,
|
| 1529 |
-
"step": 4320
|
| 1530 |
-
},
|
| 1531 |
-
{
|
| 1532 |
-
"epoch": 1.8169268413959148,
|
| 1533 |
-
"grad_norm": 0.7031142711639404,
|
| 1534 |
-
"learning_rate": 2.3582497973364996e-05,
|
| 1535 |
-
"loss": 0.02963052988052368,
|
| 1536 |
-
"step": 4340
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 1.825301283644976,
|
| 1540 |
-
"grad_norm": 1.4223092794418335,
|
| 1541 |
-
"learning_rate": 2.339499259796216e-05,
|
| 1542 |
-
"loss": 0.06973366737365723,
|
| 1543 |
-
"step": 4360
|
| 1544 |
-
},
|
| 1545 |
-
{
|
| 1546 |
-
"epoch": 1.8336757258940373,
|
| 1547 |
-
"grad_norm": 0.1039121076464653,
|
| 1548 |
-
"learning_rate": 2.320757784054552e-05,
|
| 1549 |
-
"loss": 0.032325556874275206,
|
| 1550 |
-
"step": 4380
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.8420501681430983,
|
| 1554 |
-
"grad_norm": 0.31871646642684937,
|
| 1555 |
-
"learning_rate": 2.302026428246682e-05,
|
| 1556 |
-
"loss": 0.0332093209028244,
|
| 1557 |
-
"step": 4400
|
| 1558 |
-
},
|
| 1559 |
-
{
|
| 1560 |
-
"epoch": 1.8504246103921593,
|
| 1561 |
-
"grad_norm": 0.22291108965873718,
|
| 1562 |
-
"learning_rate": 2.2833062499364153e-05,
|
| 1563 |
-
"loss": 0.037931859493255615,
|
| 1564 |
-
"step": 4420
|
| 1565 |
-
},
|
| 1566 |
-
{
|
| 1567 |
-
"epoch": 1.8587990526412206,
|
| 1568 |
-
"grad_norm": 0.5042901635169983,
|
| 1569 |
-
"learning_rate": 2.2645983060564868e-05,
|
| 1570 |
-
"loss": 0.04829805791378021,
|
| 1571 |
-
"step": 4440
|
| 1572 |
-
},
|
| 1573 |
-
{
|
| 1574 |
-
"epoch": 1.8671734948902818,
|
| 1575 |
-
"grad_norm": 1.4926162958145142,
|
| 1576 |
-
"learning_rate": 2.245903652848877e-05,
|
| 1577 |
-
"loss": 0.04460406899452209,
|
| 1578 |
-
"step": 4460
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 1.8755479371393429,
|
| 1582 |
-
"grad_norm": 0.2670135498046875,
|
| 1583 |
-
"learning_rate": 2.2272233458051837e-05,
|
| 1584 |
-
"loss": 0.045635414123535153,
|
| 1585 |
-
"step": 4480
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"epoch": 1.883922379388404,
|
| 1589 |
-
"grad_norm": 2.4627184867858887,
|
| 1590 |
-
"learning_rate": 2.2085584396070245e-05,
|
| 1591 |
-
"loss": 0.06845098137855529,
|
| 1592 |
-
"step": 4500
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.8922968216374652,
|
| 1596 |
-
"grad_norm": 0.06812036037445068,
|
| 1597 |
-
"learning_rate": 2.1899099880664965e-05,
|
| 1598 |
-
"loss": 0.05972928404808044,
|
| 1599 |
-
"step": 4520
|
| 1600 |
-
},
|
| 1601 |
-
{
|
| 1602 |
-
"epoch": 1.9006712638865264,
|
| 1603 |
-
"grad_norm": 0.04887200891971588,
|
| 1604 |
-
"learning_rate": 2.17127904406667e-05,
|
| 1605 |
-
"loss": 0.11951442956924438,
|
| 1606 |
-
"step": 4540
|
| 1607 |
-
},
|
| 1608 |
-
{
|
| 1609 |
-
"epoch": 1.9090457061355874,
|
| 1610 |
-
"grad_norm": 0.2047152817249298,
|
| 1611 |
-
"learning_rate": 2.15266665950215e-05,
|
| 1612 |
-
"loss": 0.023097112774848938,
|
| 1613 |
-
"step": 4560
|
| 1614 |
-
},
|
| 1615 |
-
{
|
| 1616 |
-
"epoch": 1.9174201483846485,
|
| 1617 |
-
"grad_norm": 0.1777394413948059,
|
| 1618 |
-
"learning_rate": 2.1340738852196834e-05,
|
| 1619 |
-
"loss": 0.04754082262516022,
|
| 1620 |
-
"step": 4580
|
| 1621 |
-
},
|
| 1622 |
-
{
|
| 1623 |
-
"epoch": 1.9257945906337097,
|
| 1624 |
-
"grad_norm": 0.21407704055309296,
|
| 1625 |
-
"learning_rate": 2.11550177095883e-05,
|
| 1626 |
-
"loss": 0.0496823787689209,
|
| 1627 |
-
"step": 4600
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"epoch": 1.934169032882771,
|
| 1631 |
-
"grad_norm": 0.06560799479484558,
|
| 1632 |
-
"learning_rate": 2.096951365292693e-05,
|
| 1633 |
-
"loss": 0.04696822464466095,
|
| 1634 |
-
"step": 4620
|
| 1635 |
-
}
|
| 1636 |
-
],
|
| 1637 |
-
"logging_steps": 20,
|
| 1638 |
-
"max_steps": 8362,
|
| 1639 |
-
"num_input_tokens_seen": 0,
|
| 1640 |
-
"num_train_epochs": 4,
|
| 1641 |
-
"save_steps": 1000000000,
|
| 1642 |
-
"stateful_callbacks": {
|
| 1643 |
-
"TrainerControl": {
|
| 1644 |
-
"args": {
|
| 1645 |
-
"should_epoch_stop": false,
|
| 1646 |
-
"should_evaluate": false,
|
| 1647 |
-
"should_log": false,
|
| 1648 |
-
"should_save": true,
|
| 1649 |
-
"should_training_stop": false
|
| 1650 |
-
},
|
| 1651 |
-
"attributes": {}
|
| 1652 |
-
}
|
| 1653 |
-
},
|
| 1654 |
-
"total_flos": 1.463868956604672e+16,
|
| 1655 |
-
"train_batch_size": 1,
|
| 1656 |
-
"trial_name": null,
|
| 1657 |
-
"trial_params": null
|
| 1658 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
slots/16/latest.json
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
{"worker_id": "slot:16", "checkpoint": "slots/16/checkpoint-
|
|
|
|
| 1 |
+
{"worker_id": "slot:16", "checkpoint": "slots/16/checkpoint-5727", "step": 5727, "updated_at": 1776440093}
|