prune 4 old checkpoints
Browse files- slots/1/checkpoint-6452/config.json +0 -24
- slots/1/checkpoint-6452/trainer_state.json +0 -2295
- slots/13/checkpoint-8035/config.json +0 -24
- slots/13/checkpoint-8035/trainer_state.json +0 -2848
- slots/19/checkpoint-6078/config.json +0 -24
- slots/19/checkpoint-6078/trainer_state.json +0 -2162
- slots/7/checkpoint-6622/config.json +0 -24
- slots/7/checkpoint-6622/trainer_state.json +0 -2358
slots/1/checkpoint-6452/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/1/checkpoint-6452/trainer_state.json
DELETED
|
@@ -1,2295 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 2.0257514941558634,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 6452,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.0003140426116568692,
|
| 14 |
-
"grad_norm": 11.236513137817383,
|
| 15 |
-
"learning_rate": 5e-05,
|
| 16 |
-
"loss": 14.929291725158691,
|
| 17 |
-
"step": 1
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.0062808522331373835,
|
| 21 |
-
"grad_norm": 8.230694770812988,
|
| 22 |
-
"learning_rate": 4.9999363065476e-05,
|
| 23 |
-
"loss": 13.227896439401727,
|
| 24 |
-
"step": 20
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.012561704466274767,
|
| 28 |
-
"grad_norm": 15.541277885437012,
|
| 29 |
-
"learning_rate": 4.999745229435885e-05,
|
| 30 |
-
"loss": 11.76095199584961,
|
| 31 |
-
"step": 40
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.018842556699412153,
|
| 35 |
-
"grad_norm": 19.88605308532715,
|
| 36 |
-
"learning_rate": 4.999406489735261e-05,
|
| 37 |
-
"loss": 10.630082702636718,
|
| 38 |
-
"step": 60
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.025123408932549534,
|
| 42 |
-
"grad_norm": 21.772388458251953,
|
| 43 |
-
"learning_rate": 4.998953978539926e-05,
|
| 44 |
-
"loss": 10.612107086181641,
|
| 45 |
-
"step": 80
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.03140426116568692,
|
| 49 |
-
"grad_norm": 24.13600730895996,
|
| 50 |
-
"learning_rate": 4.998374133739565e-05,
|
| 51 |
-
"loss": 8.763066864013672,
|
| 52 |
-
"step": 100
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.037685113398824306,
|
| 56 |
-
"grad_norm": 33.56028747558594,
|
| 57 |
-
"learning_rate": 4.997666984880032e-05,
|
| 58 |
-
"loss": 11.96340103149414,
|
| 59 |
-
"step": 120
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.04396596563196169,
|
| 63 |
-
"grad_norm": 51.29957962036133,
|
| 64 |
-
"learning_rate": 4.996785126662353e-05,
|
| 65 |
-
"loss": 7.662068939208984,
|
| 66 |
-
"step": 140
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.05024681786509907,
|
| 70 |
-
"grad_norm": 34.75215530395508,
|
| 71 |
-
"learning_rate": 4.9957623010947197e-05,
|
| 72 |
-
"loss": 7.163829040527344,
|
| 73 |
-
"step": 160
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.056527670098236456,
|
| 77 |
-
"grad_norm": 48.993743896484375,
|
| 78 |
-
"learning_rate": 4.994598565925399e-05,
|
| 79 |
-
"loss": 6.738508605957032,
|
| 80 |
-
"step": 180
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.06280852233137384,
|
| 84 |
-
"grad_norm": 37.503150939941406,
|
| 85 |
-
"learning_rate": 4.9932939868583476e-05,
|
| 86 |
-
"loss": 5.927839660644532,
|
| 87 |
-
"step": 200
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.06908937456451122,
|
| 91 |
-
"grad_norm": 34.306514739990234,
|
| 92 |
-
"learning_rate": 4.991848637549505e-05,
|
| 93 |
-
"loss": 5.548626327514649,
|
| 94 |
-
"step": 220
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.07537022679764861,
|
| 98 |
-
"grad_norm": 25.989212036132812,
|
| 99 |
-
"learning_rate": 4.990345241504926e-05,
|
| 100 |
-
"loss": 5.030820083618164,
|
| 101 |
-
"step": 240
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.08165107903078599,
|
| 105 |
-
"grad_norm": 23.606063842773438,
|
| 106 |
-
"learning_rate": 4.98862563217225e-05,
|
| 107 |
-
"loss": 4.923490524291992,
|
| 108 |
-
"step": 260
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.08793193126392337,
|
| 112 |
-
"grad_norm": 36.173194885253906,
|
| 113 |
-
"learning_rate": 4.9867655161709595e-05,
|
| 114 |
-
"loss": 4.673915100097656,
|
| 115 |
-
"step": 280
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.09421278349706076,
|
| 119 |
-
"grad_norm": 18.624378204345703,
|
| 120 |
-
"learning_rate": 4.984764998522357e-05,
|
| 121 |
-
"loss": 3.8839893341064453,
|
| 122 |
-
"step": 300
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.10049363573019814,
|
| 126 |
-
"grad_norm": 22.662349700927734,
|
| 127 |
-
"learning_rate": 4.9826241921747555e-05,
|
| 128 |
-
"loss": 3.063496971130371,
|
| 129 |
-
"step": 320
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.10677448796333552,
|
| 133 |
-
"grad_norm": 25.49854278564453,
|
| 134 |
-
"learning_rate": 4.9803432179971055e-05,
|
| 135 |
-
"loss": 3.0309188842773436,
|
| 136 |
-
"step": 340
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.11305534019647291,
|
| 140 |
-
"grad_norm": 15.640332221984863,
|
| 141 |
-
"learning_rate": 4.9779222047721693e-05,
|
| 142 |
-
"loss": 2.6161094665527345,
|
| 143 |
-
"step": 360
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.1193361924296103,
|
| 147 |
-
"grad_norm": 13.35746955871582,
|
| 148 |
-
"learning_rate": 4.9753612891892514e-05,
|
| 149 |
-
"loss": 2.3498945236206055,
|
| 150 |
-
"step": 380
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.12561704466274767,
|
| 154 |
-
"grad_norm": 10.242664337158203,
|
| 155 |
-
"learning_rate": 4.9726606158364765e-05,
|
| 156 |
-
"loss": 2.069006156921387,
|
| 157 |
-
"step": 400
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.13189789689588505,
|
| 161 |
-
"grad_norm": 8.599258422851562,
|
| 162 |
-
"learning_rate": 4.969965664306835e-05,
|
| 163 |
-
"loss": 1.7923540115356444,
|
| 164 |
-
"step": 420
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.13817874912902245,
|
| 168 |
-
"grad_norm": 5.654482364654541,
|
| 169 |
-
"learning_rate": 4.966992909049121e-05,
|
| 170 |
-
"loss": 1.7063348770141602,
|
| 171 |
-
"step": 440
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.14445960136215982,
|
| 175 |
-
"grad_norm": 9.39411449432373,
|
| 176 |
-
"learning_rate": 4.9638808684964765e-05,
|
| 177 |
-
"loss": 1.4251695632934571,
|
| 178 |
-
"step": 460
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.15074045359529722,
|
| 182 |
-
"grad_norm": 4.354184627532959,
|
| 183 |
-
"learning_rate": 4.960629718353295e-05,
|
| 184 |
-
"loss": 1.2388976097106934,
|
| 185 |
-
"step": 480
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.1570213058284346,
|
| 189 |
-
"grad_norm": 4.586623191833496,
|
| 190 |
-
"learning_rate": 4.9572396421780296e-05,
|
| 191 |
-
"loss": 1.2111310958862305,
|
| 192 |
-
"step": 500
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.16330215806157197,
|
| 196 |
-
"grad_norm": 3.979524850845337,
|
| 197 |
-
"learning_rate": 4.9537108313728365e-05,
|
| 198 |
-
"loss": 1.4075250625610352,
|
| 199 |
-
"step": 520
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.16958301029470937,
|
| 203 |
-
"grad_norm": 3.687380790710449,
|
| 204 |
-
"learning_rate": 4.9500434851727635e-05,
|
| 205 |
-
"loss": 0.9275534629821778,
|
| 206 |
-
"step": 540
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.17586386252784675,
|
| 210 |
-
"grad_norm": 2.872728109359741,
|
| 211 |
-
"learning_rate": 4.946431376374067e-05,
|
| 212 |
-
"loss": 1.418515110015869,
|
| 213 |
-
"step": 560
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.18214471476098412,
|
| 217 |
-
"grad_norm": 2.1888179779052734,
|
| 218 |
-
"learning_rate": 4.9424944888143395e-05,
|
| 219 |
-
"loss": 0.7708720684051513,
|
| 220 |
-
"step": 580
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.18842556699412152,
|
| 224 |
-
"grad_norm": 2.96533465385437,
|
| 225 |
-
"learning_rate": 4.9384196991293205e-05,
|
| 226 |
-
"loss": 0.6360139369964599,
|
| 227 |
-
"step": 600
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.1947064192272589,
|
| 231 |
-
"grad_norm": 3.389042615890503,
|
| 232 |
-
"learning_rate": 4.9342072373797774e-05,
|
| 233 |
-
"loss": 1.2306670188903808,
|
| 234 |
-
"step": 620
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.20098727146039627,
|
| 238 |
-
"grad_norm": 3.054168939590454,
|
| 239 |
-
"learning_rate": 4.9298573413993806e-05,
|
| 240 |
-
"loss": 0.684608793258667,
|
| 241 |
-
"step": 640
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.20726812369353367,
|
| 245 |
-
"grad_norm": 2.2155184745788574,
|
| 246 |
-
"learning_rate": 4.925370256781273e-05,
|
| 247 |
-
"loss": 0.6700978755950928,
|
| 248 |
-
"step": 660
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.21354897592667105,
|
| 252 |
-
"grad_norm": 1.223708987236023,
|
| 253 |
-
"learning_rate": 4.920746236864208e-05,
|
| 254 |
-
"loss": 0.707273292541504,
|
| 255 |
-
"step": 680
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.21982982815980842,
|
| 259 |
-
"grad_norm": 1.1626676321029663,
|
| 260 |
-
"learning_rate": 4.915985542718241e-05,
|
| 261 |
-
"loss": 0.6534492492675781,
|
| 262 |
-
"step": 700
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.22611068039294582,
|
| 266 |
-
"grad_norm": 1.1415362358093262,
|
| 267 |
-
"learning_rate": 4.9110884431299935e-05,
|
| 268 |
-
"loss": 0.45153131484985354,
|
| 269 |
-
"step": 720
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.2323915326260832,
|
| 273 |
-
"grad_norm": 1.0272735357284546,
|
| 274 |
-
"learning_rate": 4.906055214587476e-05,
|
| 275 |
-
"loss": 0.46799488067626954,
|
| 276 |
-
"step": 740
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.2386723848592206,
|
| 280 |
-
"grad_norm": 1.2896318435668945,
|
| 281 |
-
"learning_rate": 4.9008861412644756e-05,
|
| 282 |
-
"loss": 0.4087998390197754,
|
| 283 |
-
"step": 760
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.24495323709235797,
|
| 287 |
-
"grad_norm": 0.873556911945343,
|
| 288 |
-
"learning_rate": 4.895581515004517e-05,
|
| 289 |
-
"loss": 0.3405194044113159,
|
| 290 |
-
"step": 780
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.25123408932549535,
|
| 294 |
-
"grad_norm": 1.2751230001449585,
|
| 295 |
-
"learning_rate": 4.89014163530438e-05,
|
| 296 |
-
"loss": 0.401216459274292,
|
| 297 |
-
"step": 800
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.25751494155863275,
|
| 301 |
-
"grad_norm": 0.8582271337509155,
|
| 302 |
-
"learning_rate": 4.8845668092971934e-05,
|
| 303 |
-
"loss": 0.36631605625152586,
|
| 304 |
-
"step": 820
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.2637957937917701,
|
| 308 |
-
"grad_norm": 0.7967858910560608,
|
| 309 |
-
"learning_rate": 4.87885735173509e-05,
|
| 310 |
-
"loss": 0.33444316387176515,
|
| 311 |
-
"step": 840
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.2700766460249075,
|
| 315 |
-
"grad_norm": 0.7824229001998901,
|
| 316 |
-
"learning_rate": 4.873013584971442e-05,
|
| 317 |
-
"loss": 0.3244049549102783,
|
| 318 |
-
"step": 860
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.2763574982580449,
|
| 322 |
-
"grad_norm": 0.774782121181488,
|
| 323 |
-
"learning_rate": 4.867035838942657e-05,
|
| 324 |
-
"loss": 0.6262166023254394,
|
| 325 |
-
"step": 880
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.2826383504911823,
|
| 329 |
-
"grad_norm": 0.6737077832221985,
|
| 330 |
-
"learning_rate": 4.86092445114955e-05,
|
| 331 |
-
"loss": 0.28058154582977296,
|
| 332 |
-
"step": 900
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.28891920272431965,
|
| 336 |
-
"grad_norm": 1.0165168046951294,
|
| 337 |
-
"learning_rate": 4.8546797666382896e-05,
|
| 338 |
-
"loss": 0.3938837289810181,
|
| 339 |
-
"step": 920
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.29520005495745705,
|
| 343 |
-
"grad_norm": 1.093992829322815,
|
| 344 |
-
"learning_rate": 4.8483021379809154e-05,
|
| 345 |
-
"loss": 0.3142520904541016,
|
| 346 |
-
"step": 940
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.30148090719059445,
|
| 350 |
-
"grad_norm": 0.6402747631072998,
|
| 351 |
-
"learning_rate": 4.8417919252554354e-05,
|
| 352 |
-
"loss": 0.29062786102294924,
|
| 353 |
-
"step": 960
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.3077617594237318,
|
| 357 |
-
"grad_norm": 1.9970321655273438,
|
| 358 |
-
"learning_rate": 4.8351494960254916e-05,
|
| 359 |
-
"loss": 0.31338260173797605,
|
| 360 |
-
"step": 980
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.3140426116568692,
|
| 364 |
-
"grad_norm": 1.1727873086929321,
|
| 365 |
-
"learning_rate": 4.8283752253196094e-05,
|
| 366 |
-
"loss": 0.2333289623260498,
|
| 367 |
-
"step": 1000
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.3203234638900066,
|
| 371 |
-
"grad_norm": 0.7795476913452148,
|
| 372 |
-
"learning_rate": 4.8214694956100245e-05,
|
| 373 |
-
"loss": 0.2488691806793213,
|
| 374 |
-
"step": 1020
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.32660431612314395,
|
| 378 |
-
"grad_norm": 1.1675890684127808,
|
| 379 |
-
"learning_rate": 4.814432696791089e-05,
|
| 380 |
-
"loss": 0.22975556850433348,
|
| 381 |
-
"step": 1040
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.33288516835628135,
|
| 385 |
-
"grad_norm": 0.5809735059738159,
|
| 386 |
-
"learning_rate": 4.807265226157255e-05,
|
| 387 |
-
"loss": 0.2238070011138916,
|
| 388 |
-
"step": 1060
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.33916602058941875,
|
| 392 |
-
"grad_norm": 0.6172874569892883,
|
| 393 |
-
"learning_rate": 4.7999674883806465e-05,
|
| 394 |
-
"loss": 0.22853121757507325,
|
| 395 |
-
"step": 1080
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.3454468728225561,
|
| 399 |
-
"grad_norm": 0.9794221520423889,
|
| 400 |
-
"learning_rate": 4.792539895488213e-05,
|
| 401 |
-
"loss": 0.2538387060165405,
|
| 402 |
-
"step": 1100
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.3517277250556935,
|
| 406 |
-
"grad_norm": 0.6366945505142212,
|
| 407 |
-
"learning_rate": 4.784982866838462e-05,
|
| 408 |
-
"loss": 0.207783842086792,
|
| 409 |
-
"step": 1120
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.3580085772888309,
|
| 413 |
-
"grad_norm": 0.5059712529182434,
|
| 414 |
-
"learning_rate": 4.777296829097785e-05,
|
| 415 |
-
"loss": 0.2022031545639038,
|
| 416 |
-
"step": 1140
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.36428942952196824,
|
| 420 |
-
"grad_norm": 0.42327556014060974,
|
| 421 |
-
"learning_rate": 4.769482216216369e-05,
|
| 422 |
-
"loss": 0.19924311637878417,
|
| 423 |
-
"step": 1160
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.37057028175510565,
|
| 427 |
-
"grad_norm": 1.0261644124984741,
|
| 428 |
-
"learning_rate": 4.761539469403695e-05,
|
| 429 |
-
"loss": 0.1949863314628601,
|
| 430 |
-
"step": 1180
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.37685113398824305,
|
| 434 |
-
"grad_norm": 0.78790283203125,
|
| 435 |
-
"learning_rate": 4.7534690371036274e-05,
|
| 436 |
-
"loss": 0.1740255832672119,
|
| 437 |
-
"step": 1200
|
| 438 |
-
},
|
| 439 |
-
{
|
| 440 |
-
"epoch": 0.3831319862213804,
|
| 441 |
-
"grad_norm": 0.49199551343917847,
|
| 442 |
-
"learning_rate": 4.7452713749690934e-05,
|
| 443 |
-
"loss": 0.16614840030670167,
|
| 444 |
-
"step": 1220
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.3894128384545178,
|
| 448 |
-
"grad_norm": 0.5265403985977173,
|
| 449 |
-
"learning_rate": 4.73694694583636e-05,
|
| 450 |
-
"loss": 0.17591538429260253,
|
| 451 |
-
"step": 1240
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.3956936906876552,
|
| 455 |
-
"grad_norm": 0.5221320390701294,
|
| 456 |
-
"learning_rate": 4.728496219698902e-05,
|
| 457 |
-
"loss": 0.18058695793151855,
|
| 458 |
-
"step": 1260
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.40197454292079254,
|
| 462 |
-
"grad_norm": 1.2317312955856323,
|
| 463 |
-
"learning_rate": 4.719919673680863e-05,
|
| 464 |
-
"loss": 0.12510637044906617,
|
| 465 |
-
"step": 1280
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 0.40825539515392995,
|
| 469 |
-
"grad_norm": 0.21159659326076508,
|
| 470 |
-
"learning_rate": 4.711217792010122e-05,
|
| 471 |
-
"loss": 0.3407147884368896,
|
| 472 |
-
"step": 1300
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.41453624738706735,
|
| 476 |
-
"grad_norm": 1.4082627296447754,
|
| 477 |
-
"learning_rate": 4.70239106599095e-05,
|
| 478 |
-
"loss": 0.16848219633102418,
|
| 479 |
-
"step": 1320
|
| 480 |
-
},
|
| 481 |
-
{
|
| 482 |
-
"epoch": 0.4208170996202047,
|
| 483 |
-
"grad_norm": 0.35336634516716003,
|
| 484 |
-
"learning_rate": 4.6934399939762746e-05,
|
| 485 |
-
"loss": 0.13055137395858765,
|
| 486 |
-
"step": 1340
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 0.4270979518533421,
|
| 490 |
-
"grad_norm": 0.8723010420799255,
|
| 491 |
-
"learning_rate": 4.684365081339541e-05,
|
| 492 |
-
"loss": 0.1356166958808899,
|
| 493 |
-
"step": 1360
|
| 494 |
-
},
|
| 495 |
-
{
|
| 496 |
-
"epoch": 0.4333788040864795,
|
| 497 |
-
"grad_norm": 0.49977484345436096,
|
| 498 |
-
"learning_rate": 4.675166840446179e-05,
|
| 499 |
-
"loss": 0.17129049301147461,
|
| 500 |
-
"step": 1380
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.43965965631961684,
|
| 504 |
-
"grad_norm": 0.8447715640068054,
|
| 505 |
-
"learning_rate": 4.6658457906246754e-05,
|
| 506 |
-
"loss": 0.16764267683029174,
|
| 507 |
-
"step": 1400
|
| 508 |
-
},
|
| 509 |
-
{
|
| 510 |
-
"epoch": 0.44594050855275424,
|
| 511 |
-
"grad_norm": 0.491344153881073,
|
| 512 |
-
"learning_rate": 4.656402458137255e-05,
|
| 513 |
-
"loss": 0.2734020709991455,
|
| 514 |
-
"step": 1420
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.45222136078589165,
|
| 518 |
-
"grad_norm": 0.7019853591918945,
|
| 519 |
-
"learning_rate": 4.646837376150164e-05,
|
| 520 |
-
"loss": 0.15193712711334229,
|
| 521 |
-
"step": 1440
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 0.458502213019029,
|
| 525 |
-
"grad_norm": 0.5096240639686584,
|
| 526 |
-
"learning_rate": 4.637151084703571e-05,
|
| 527 |
-
"loss": 0.31453657150268555,
|
| 528 |
-
"step": 1460
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 0.4647830652521664,
|
| 532 |
-
"grad_norm": 0.457865446805954,
|
| 533 |
-
"learning_rate": 4.6273441306810724e-05,
|
| 534 |
-
"loss": 0.14785597324371338,
|
| 535 |
-
"step": 1480
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"epoch": 0.4710639174853038,
|
| 539 |
-
"grad_norm": 0.5041202306747437,
|
| 540 |
-
"learning_rate": 4.617417067778825e-05,
|
| 541 |
-
"loss": 0.13907097578048705,
|
| 542 |
-
"step": 1500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.4773447697184412,
|
| 546 |
-
"grad_norm": 3.153047800064087,
|
| 547 |
-
"learning_rate": 4.6073704564742734e-05,
|
| 548 |
-
"loss": 0.1594400644302368,
|
| 549 |
-
"step": 1520
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.48362562195157854,
|
| 553 |
-
"grad_norm": 6.409017562866211,
|
| 554 |
-
"learning_rate": 4.5972048639945105e-05,
|
| 555 |
-
"loss": 0.1708819031715393,
|
| 556 |
-
"step": 1540
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 0.48990647418471595,
|
| 560 |
-
"grad_norm": 0.5795732736587524,
|
| 561 |
-
"learning_rate": 4.5869208642842536e-05,
|
| 562 |
-
"loss": 0.5376199245452881,
|
| 563 |
-
"step": 1560
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 0.49618732641785335,
|
| 567 |
-
"grad_norm": 1.402137279510498,
|
| 568 |
-
"learning_rate": 4.5765190379734384e-05,
|
| 569 |
-
"loss": 0.13104203939437867,
|
| 570 |
-
"step": 1580
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.5024681786509907,
|
| 574 |
-
"grad_norm": 1.01236093044281,
|
| 575 |
-
"learning_rate": 4.565999972344436e-05,
|
| 576 |
-
"loss": 0.14823287725448608,
|
| 577 |
-
"step": 1600
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 0.508749030884128,
|
| 581 |
-
"grad_norm": 0.47942015528678894,
|
| 582 |
-
"learning_rate": 4.5553642612988954e-05,
|
| 583 |
-
"loss": 0.13399872779846192,
|
| 584 |
-
"step": 1620
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.5150298831172655,
|
| 588 |
-
"grad_norm": 0.5736086964607239,
|
| 589 |
-
"learning_rate": 4.544612505324213e-05,
|
| 590 |
-
"loss": 0.09690673351287842,
|
| 591 |
-
"step": 1640
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 0.5213107353504028,
|
| 595 |
-
"grad_norm": 0.4799337387084961,
|
| 596 |
-
"learning_rate": 4.533745311459631e-05,
|
| 597 |
-
"loss": 0.12333678007125855,
|
| 598 |
-
"step": 1660
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.5275915875835402,
|
| 602 |
-
"grad_norm": 0.49539658427238464,
|
| 603 |
-
"learning_rate": 4.5227632932619604e-05,
|
| 604 |
-
"loss": 0.1237714409828186,
|
| 605 |
-
"step": 1680
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 0.5338724398166776,
|
| 609 |
-
"grad_norm": 3.5184292793273926,
|
| 610 |
-
"learning_rate": 4.5116670707709396e-05,
|
| 611 |
-
"loss": 0.12902543544769288,
|
| 612 |
-
"step": 1700
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.540153292049815,
|
| 616 |
-
"grad_norm": 0.543497622013092,
|
| 617 |
-
"learning_rate": 4.5004572704742345e-05,
|
| 618 |
-
"loss": 0.11819875240325928,
|
| 619 |
-
"step": 1720
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"epoch": 0.5464341442829523,
|
| 623 |
-
"grad_norm": 0.47288012504577637,
|
| 624 |
-
"learning_rate": 4.489134525272057e-05,
|
| 625 |
-
"loss": 0.09024000763893128,
|
| 626 |
-
"step": 1740
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.5527149965160898,
|
| 630 |
-
"grad_norm": 0.5600817203521729,
|
| 631 |
-
"learning_rate": 4.477699474441441e-05,
|
| 632 |
-
"loss": 0.10022807121276855,
|
| 633 |
-
"step": 1760
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 0.5589958487492271,
|
| 637 |
-
"grad_norm": 0.38338902592658997,
|
| 638 |
-
"learning_rate": 4.466152763600142e-05,
|
| 639 |
-
"loss": 0.1357239603996277,
|
| 640 |
-
"step": 1780
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.5652767009823646,
|
| 644 |
-
"grad_norm": 0.12702441215515137,
|
| 645 |
-
"learning_rate": 4.454495044670189e-05,
|
| 646 |
-
"loss": 0.36903703212738037,
|
| 647 |
-
"step": 1800
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.571557553215502,
|
| 651 |
-
"grad_norm": 0.32971400022506714,
|
| 652 |
-
"learning_rate": 4.4427269758410806e-05,
|
| 653 |
-
"loss": 0.09078516364097595,
|
| 654 |
-
"step": 1820
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.5778384054486393,
|
| 658 |
-
"grad_norm": 0.5972228050231934,
|
| 659 |
-
"learning_rate": 4.4308492215326145e-05,
|
| 660 |
-
"loss": 0.13515870571136473,
|
| 661 |
-
"step": 1840
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.5841192576817767,
|
| 665 |
-
"grad_norm": 0.4297616183757782,
|
| 666 |
-
"learning_rate": 4.418862452357387e-05,
|
| 667 |
-
"loss": 0.16306908130645753,
|
| 668 |
-
"step": 1860
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.5904001099149141,
|
| 672 |
-
"grad_norm": 0.3069750666618347,
|
| 673 |
-
"learning_rate": 4.4067673450829184e-05,
|
| 674 |
-
"loss": 0.11015008687973023,
|
| 675 |
-
"step": 1880
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.5966809621480514,
|
| 679 |
-
"grad_norm": 0.2897976040840149,
|
| 680 |
-
"learning_rate": 4.394564582593452e-05,
|
| 681 |
-
"loss": 0.10074495077133179,
|
| 682 |
-
"step": 1900
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.6029618143811889,
|
| 686 |
-
"grad_norm": 0.2866128981113434,
|
| 687 |
-
"learning_rate": 4.382254853851396e-05,
|
| 688 |
-
"loss": 0.10123426914215088,
|
| 689 |
-
"step": 1920
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.6092426666143262,
|
| 693 |
-
"grad_norm": 0.611520528793335,
|
| 694 |
-
"learning_rate": 4.3698388538584214e-05,
|
| 695 |
-
"loss": 0.1677260398864746,
|
| 696 |
-
"step": 1940
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 0.6155235188474636,
|
| 700 |
-
"grad_norm": 0.6373476982116699,
|
| 701 |
-
"learning_rate": 4.357317283616228e-05,
|
| 702 |
-
"loss": 0.07637218236923218,
|
| 703 |
-
"step": 1960
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.621804371080601,
|
| 707 |
-
"grad_norm": 0.24775440990924835,
|
| 708 |
-
"learning_rate": 4.3446908500869624e-05,
|
| 709 |
-
"loss": 0.07576732635498047,
|
| 710 |
-
"step": 1980
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.6280852233137384,
|
| 714 |
-
"grad_norm": 0.36698204278945923,
|
| 715 |
-
"learning_rate": 4.3319602661533054e-05,
|
| 716 |
-
"loss": 0.08246396780014038,
|
| 717 |
-
"step": 2000
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.6343660755468757,
|
| 721 |
-
"grad_norm": 0.4456537365913391,
|
| 722 |
-
"learning_rate": 4.319126250578223e-05,
|
| 723 |
-
"loss": 0.08348861336708069,
|
| 724 |
-
"step": 2020
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.6406469277800132,
|
| 728 |
-
"grad_norm": 0.12586534023284912,
|
| 729 |
-
"learning_rate": 4.306189527964382e-05,
|
| 730 |
-
"loss": 0.11425427198410035,
|
| 731 |
-
"step": 2040
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.6469277800131505,
|
| 735 |
-
"grad_norm": 0.42404818534851074,
|
| 736 |
-
"learning_rate": 4.2931508287132435e-05,
|
| 737 |
-
"loss": 0.43116607666015627,
|
| 738 |
-
"step": 2060
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.6532086322462879,
|
| 742 |
-
"grad_norm": 0.322374165058136,
|
| 743 |
-
"learning_rate": 4.2800108889838244e-05,
|
| 744 |
-
"loss": 0.07275817394256592,
|
| 745 |
-
"step": 2080
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.6594894844794253,
|
| 749 |
-
"grad_norm": 0.24643594026565552,
|
| 750 |
-
"learning_rate": 4.26677045065113e-05,
|
| 751 |
-
"loss": 0.0800868034362793,
|
| 752 |
-
"step": 2100
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.6657703367125627,
|
| 756 |
-
"grad_norm": 0.32612672448158264,
|
| 757 |
-
"learning_rate": 4.253430261264272e-05,
|
| 758 |
-
"loss": 0.11347514390945435,
|
| 759 |
-
"step": 2120
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.6720511889457,
|
| 763 |
-
"grad_norm": 0.9363917708396912,
|
| 764 |
-
"learning_rate": 4.239991074004262e-05,
|
| 765 |
-
"loss": 0.14267417192459106,
|
| 766 |
-
"step": 2140
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.6783320411788375,
|
| 770 |
-
"grad_norm": 5.144097328186035,
|
| 771 |
-
"learning_rate": 4.226453647641484e-05,
|
| 772 |
-
"loss": 0.2039745807647705,
|
| 773 |
-
"step": 2160
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.6846128934119748,
|
| 777 |
-
"grad_norm": 0.275455117225647,
|
| 778 |
-
"learning_rate": 4.2128187464928604e-05,
|
| 779 |
-
"loss": 0.12193784713745118,
|
| 780 |
-
"step": 2180
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.6908937456451122,
|
| 784 |
-
"grad_norm": 0.23535878956317902,
|
| 785 |
-
"learning_rate": 4.1990871403786934e-05,
|
| 786 |
-
"loss": 0.08345263600349426,
|
| 787 |
-
"step": 2200
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.6971745978782496,
|
| 791 |
-
"grad_norm": 0.24043764173984528,
|
| 792 |
-
"learning_rate": 4.185259604579203e-05,
|
| 793 |
-
"loss": 0.06981859803199768,
|
| 794 |
-
"step": 2220
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.703455450111387,
|
| 798 |
-
"grad_norm": 0.6327846050262451,
|
| 799 |
-
"learning_rate": 4.171336919790759e-05,
|
| 800 |
-
"loss": 0.1702319025993347,
|
| 801 |
-
"step": 2240
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.7097363023445243,
|
| 805 |
-
"grad_norm": 0.059428345412015915,
|
| 806 |
-
"learning_rate": 4.1573198720817955e-05,
|
| 807 |
-
"loss": 0.07783352732658386,
|
| 808 |
-
"step": 2260
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.7160171545776618,
|
| 812 |
-
"grad_norm": 0.4627244770526886,
|
| 813 |
-
"learning_rate": 4.143209252848433e-05,
|
| 814 |
-
"loss": 0.07844600677490235,
|
| 815 |
-
"step": 2280
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.7222980068107991,
|
| 819 |
-
"grad_norm": 0.7847932577133179,
|
| 820 |
-
"learning_rate": 4.1290058587698054e-05,
|
| 821 |
-
"loss": 0.09436154365539551,
|
| 822 |
-
"step": 2300
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.7285788590439365,
|
| 826 |
-
"grad_norm": 0.6646613478660583,
|
| 827 |
-
"learning_rate": 4.114710491763063e-05,
|
| 828 |
-
"loss": 0.10847301483154297,
|
| 829 |
-
"step": 2320
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.734859711277074,
|
| 833 |
-
"grad_norm": 0.13904166221618652,
|
| 834 |
-
"learning_rate": 4.100323958938112e-05,
|
| 835 |
-
"loss": 0.10292520523071289,
|
| 836 |
-
"step": 2340
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.7411405635102113,
|
| 840 |
-
"grad_norm": 0.17684051394462585,
|
| 841 |
-
"learning_rate": 4.085847072552036e-05,
|
| 842 |
-
"loss": 0.10409547090530395,
|
| 843 |
-
"step": 2360
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 0.7474214157433486,
|
| 847 |
-
"grad_norm": 2.33097767829895,
|
| 848 |
-
"learning_rate": 4.0712806499632397e-05,
|
| 849 |
-
"loss": 0.07961801886558532,
|
| 850 |
-
"step": 2380
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 0.7537022679764861,
|
| 854 |
-
"grad_norm": 0.476825475692749,
|
| 855 |
-
"learning_rate": 4.0566255135853014e-05,
|
| 856 |
-
"loss": 0.07112305760383605,
|
| 857 |
-
"step": 2400
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 0.7599831202096234,
|
| 861 |
-
"grad_norm": 0.5014712810516357,
|
| 862 |
-
"learning_rate": 4.04188249084054e-05,
|
| 863 |
-
"loss": 0.1152498722076416,
|
| 864 |
-
"step": 2420
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 0.7662639724427608,
|
| 868 |
-
"grad_norm": 0.4747202694416046,
|
| 869 |
-
"learning_rate": 4.0270524141133e-05,
|
| 870 |
-
"loss": 0.12496150732040405,
|
| 871 |
-
"step": 2440
|
| 872 |
-
},
|
| 873 |
-
{
|
| 874 |
-
"epoch": 0.7725448246758982,
|
| 875 |
-
"grad_norm": 0.3705868721008301,
|
| 876 |
-
"learning_rate": 4.01213612070295e-05,
|
| 877 |
-
"loss": 0.09812939167022705,
|
| 878 |
-
"step": 2460
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 0.7788256769090356,
|
| 882 |
-
"grad_norm": 0.18966151773929596,
|
| 883 |
-
"learning_rate": 3.997134452776617e-05,
|
| 884 |
-
"loss": 0.09260130524635315,
|
| 885 |
-
"step": 2480
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 0.7851065291421729,
|
| 889 |
-
"grad_norm": 0.4178507328033447,
|
| 890 |
-
"learning_rate": 3.982048257321631e-05,
|
| 891 |
-
"loss": 0.08796253800392151,
|
| 892 |
-
"step": 2500
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 0.7913873813753104,
|
| 896 |
-
"grad_norm": 4.684873580932617,
|
| 897 |
-
"learning_rate": 3.966878386097711e-05,
|
| 898 |
-
"loss": 0.09675108194351197,
|
| 899 |
-
"step": 2520
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 0.7976682336084477,
|
| 903 |
-
"grad_norm": 0.7546767592430115,
|
| 904 |
-
"learning_rate": 3.95162569558887e-05,
|
| 905 |
-
"loss": 0.061058801412582395,
|
| 906 |
-
"step": 2540
|
| 907 |
-
},
|
| 908 |
-
{
|
| 909 |
-
"epoch": 0.8039490858415851,
|
| 910 |
-
"grad_norm": 1.2984436750411987,
|
| 911 |
-
"learning_rate": 3.936291046955061e-05,
|
| 912 |
-
"loss": 0.0732529878616333,
|
| 913 |
-
"step": 2560
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 0.8102299380747225,
|
| 917 |
-
"grad_norm": 0.08046652376651764,
|
| 918 |
-
"learning_rate": 3.920875305983553e-05,
|
| 919 |
-
"loss": 0.056511855125427245,
|
| 920 |
-
"step": 2580
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.8165107903078599,
|
| 924 |
-
"grad_norm": 0.24541246891021729,
|
| 925 |
-
"learning_rate": 3.905379343040054e-05,
|
| 926 |
-
"loss": 0.06492850780487061,
|
| 927 |
-
"step": 2600
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 0.8227916425409972,
|
| 931 |
-
"grad_norm": 0.34249547123908997,
|
| 932 |
-
"learning_rate": 3.889804033019567e-05,
|
| 933 |
-
"loss": 0.06372414827346802,
|
| 934 |
-
"step": 2620
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.8290724947741347,
|
| 938 |
-
"grad_norm": 0.6350385546684265,
|
| 939 |
-
"learning_rate": 3.8741502552969956e-05,
|
| 940 |
-
"loss": 0.07334259748458863,
|
| 941 |
-
"step": 2640
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 0.835353347007272,
|
| 945 |
-
"grad_norm": 0.5676931738853455,
|
| 946 |
-
"learning_rate": 3.8584188936774925e-05,
|
| 947 |
-
"loss": 0.051751017570495605,
|
| 948 |
-
"step": 2660
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.8416341992404094,
|
| 952 |
-
"grad_norm": 1.5374674797058105,
|
| 953 |
-
"learning_rate": 3.842610836346563e-05,
|
| 954 |
-
"loss": 0.13946666717529296,
|
| 955 |
-
"step": 2680
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 0.8479150514735468,
|
| 959 |
-
"grad_norm": 0.4549136459827423,
|
| 960 |
-
"learning_rate": 3.8267269758199175e-05,
|
| 961 |
-
"loss": 0.06678192019462585,
|
| 962 |
-
"step": 2700
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.8541959037066842,
|
| 966 |
-
"grad_norm": 0.4328550696372986,
|
| 967 |
-
"learning_rate": 3.8107682088930794e-05,
|
| 968 |
-
"loss": 0.0756498098373413,
|
| 969 |
-
"step": 2720
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.8604767559398215,
|
| 973 |
-
"grad_norm": 1.7214152812957764,
|
| 974 |
-
"learning_rate": 3.7947354365907536e-05,
|
| 975 |
-
"loss": 0.07417964935302734,
|
| 976 |
-
"step": 2740
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 0.866757608172959,
|
| 980 |
-
"grad_norm": 0.3970869481563568,
|
| 981 |
-
"learning_rate": 3.7786295641159544e-05,
|
| 982 |
-
"loss": 0.055640166997909545,
|
| 983 |
-
"step": 2760
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 0.8730384604060963,
|
| 987 |
-
"grad_norm": 0.16465847194194794,
|
| 988 |
-
"learning_rate": 3.7624515007989004e-05,
|
| 989 |
-
"loss": 0.057208091020584106,
|
| 990 |
-
"step": 2780
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.8793193126392337,
|
| 994 |
-
"grad_norm": 1.0308940410614014,
|
| 995 |
-
"learning_rate": 3.746202160045669e-05,
|
| 996 |
-
"loss": 0.1124604344367981,
|
| 997 |
-
"step": 2800
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 0.8856001648723711,
|
| 1001 |
-
"grad_norm": 1.095229983329773,
|
| 1002 |
-
"learning_rate": 3.729882459286632e-05,
|
| 1003 |
-
"loss": 0.06749147176742554,
|
| 1004 |
-
"step": 2820
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 0.8918810171055085,
|
| 1008 |
-
"grad_norm": 0.7268246412277222,
|
| 1009 |
-
"learning_rate": 3.713493319924655e-05,
|
| 1010 |
-
"loss": 0.05816763639450073,
|
| 1011 |
-
"step": 2840
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 0.8981618693386458,
|
| 1015 |
-
"grad_norm": 0.06477148085832596,
|
| 1016 |
-
"learning_rate": 3.697035667283072e-05,
|
| 1017 |
-
"loss": 0.07365052700042725,
|
| 1018 |
-
"step": 2860
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 0.9044427215717833,
|
| 1022 |
-
"grad_norm": 0.14640307426452637,
|
| 1023 |
-
"learning_rate": 3.68051043055345e-05,
|
| 1024 |
-
"loss": 0.10839139223098755,
|
| 1025 |
-
"step": 2880
|
| 1026 |
-
},
|
| 1027 |
-
{
|
| 1028 |
-
"epoch": 0.9107235738049206,
|
| 1029 |
-
"grad_norm": 0.6694417595863342,
|
| 1030 |
-
"learning_rate": 3.66391854274312e-05,
|
| 1031 |
-
"loss": 0.0827883541584015,
|
| 1032 |
-
"step": 2900
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 0.917004426038058,
|
| 1036 |
-
"grad_norm": 0.7503764629364014,
|
| 1037 |
-
"learning_rate": 3.647260940622504e-05,
|
| 1038 |
-
"loss": 0.05839287042617798,
|
| 1039 |
-
"step": 2920
|
| 1040 |
-
},
|
| 1041 |
-
{
|
| 1042 |
-
"epoch": 0.9232852782711954,
|
| 1043 |
-
"grad_norm": 0.6776005029678345,
|
| 1044 |
-
"learning_rate": 3.6305385646722224e-05,
|
| 1045 |
-
"loss": 0.08498733639717101,
|
| 1046 |
-
"step": 2940
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 0.9295661305043328,
|
| 1050 |
-
"grad_norm": 0.1537044644355774,
|
| 1051 |
-
"learning_rate": 3.613752359029998e-05,
|
| 1052 |
-
"loss": 0.1408205509185791,
|
| 1053 |
-
"step": 2960
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 0.9358469827374702,
|
| 1057 |
-
"grad_norm": 0.18748170137405396,
|
| 1058 |
-
"learning_rate": 3.5969032714373495e-05,
|
| 1059 |
-
"loss": 0.07079569697380066,
|
| 1060 |
-
"step": 2980
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 0.9421278349706076,
|
| 1064 |
-
"grad_norm": 1.4993492364883423,
|
| 1065 |
-
"learning_rate": 3.57999225318608e-05,
|
| 1066 |
-
"loss": 0.07452060580253601,
|
| 1067 |
-
"step": 3000
|
| 1068 |
-
},
|
| 1069 |
-
{
|
| 1070 |
-
"epoch": 0.9484086872037449,
|
| 1071 |
-
"grad_norm": 2.759636163711548,
|
| 1072 |
-
"learning_rate": 3.5638702921686935e-05,
|
| 1073 |
-
"loss": 0.30155444145202637,
|
| 1074 |
-
"step": 3020
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.9546895394368824,
|
| 1078 |
-
"grad_norm": 0.35174673795700073,
|
| 1079 |
-
"learning_rate": 3.546841258479434e-05,
|
| 1080 |
-
"loss": 0.07043100595474243,
|
| 1081 |
-
"step": 3040
|
| 1082 |
-
},
|
| 1083 |
-
{
|
| 1084 |
-
"epoch": 0.9609703916700197,
|
| 1085 |
-
"grad_norm": 0.1061045303940773,
|
| 1086 |
-
"learning_rate": 3.529753120609982e-05,
|
| 1087 |
-
"loss": 0.06520496010780334,
|
| 1088 |
-
"step": 3060
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 0.9672512439031571,
|
| 1092 |
-
"grad_norm": 0.32999858260154724,
|
| 1093 |
-
"learning_rate": 3.51260684334881e-05,
|
| 1094 |
-
"loss": 0.11166098117828369,
|
| 1095 |
-
"step": 3080
|
| 1096 |
-
},
|
| 1097 |
-
{
|
| 1098 |
-
"epoch": 0.9735320961362945,
|
| 1099 |
-
"grad_norm": 0.527561366558075,
|
| 1100 |
-
"learning_rate": 3.4954033947669163e-05,
|
| 1101 |
-
"loss": 0.05884957313537598,
|
| 1102 |
-
"step": 3100
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 0.9798129483694319,
|
| 1106 |
-
"grad_norm": 0.21894922852516174,
|
| 1107 |
-
"learning_rate": 3.478143746163164e-05,
|
| 1108 |
-
"loss": 0.04760840237140655,
|
| 1109 |
-
"step": 3120
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"epoch": 0.9860938006025692,
|
| 1113 |
-
"grad_norm": 0.39011839032173157,
|
| 1114 |
-
"learning_rate": 3.460828872009444e-05,
|
| 1115 |
-
"loss": 0.07473071217536927,
|
| 1116 |
-
"step": 3140
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 0.9923746528357067,
|
| 1120 |
-
"grad_norm": 0.24014221131801605,
|
| 1121 |
-
"learning_rate": 3.443459749895658e-05,
|
| 1122 |
-
"loss": 0.05981050133705139,
|
| 1123 |
-
"step": 3160
|
| 1124 |
-
},
|
| 1125 |
-
{
|
| 1126 |
-
"epoch": 0.998655505068844,
|
| 1127 |
-
"grad_norm": 0.4369584918022156,
|
| 1128 |
-
"learning_rate": 3.4260373604745205e-05,
|
| 1129 |
-
"loss": 0.06494853496551514,
|
| 1130 |
-
"step": 3180
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.004710639174853,
|
| 1134 |
-
"grad_norm": 0.11435407400131226,
|
| 1135 |
-
"learning_rate": 3.4085626874061974e-05,
|
| 1136 |
-
"loss": 0.037298232316970825,
|
| 1137 |
-
"step": 3200
|
| 1138 |
-
},
|
| 1139 |
-
{
|
| 1140 |
-
"epoch": 1.0109914914079905,
|
| 1141 |
-
"grad_norm": 0.10405384004116058,
|
| 1142 |
-
"learning_rate": 3.391036717302761e-05,
|
| 1143 |
-
"loss": 0.03180496394634247,
|
| 1144 |
-
"step": 3220
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 1.0172723436411277,
|
| 1148 |
-
"grad_norm": 0.2586648166179657,
|
| 1149 |
-
"learning_rate": 3.373460439672495e-05,
|
| 1150 |
-
"loss": 0.04955384433269501,
|
| 1151 |
-
"step": 3240
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 1.0235531958742652,
|
| 1155 |
-
"grad_norm": 0.08358602225780487,
|
| 1156 |
-
"learning_rate": 3.35583484686402e-05,
|
| 1157 |
-
"loss": 0.05318580269813537,
|
| 1158 |
-
"step": 3260
|
| 1159 |
-
},
|
| 1160 |
-
{
|
| 1161 |
-
"epoch": 1.0298340481074026,
|
| 1162 |
-
"grad_norm": 0.07553794234991074,
|
| 1163 |
-
"learning_rate": 3.338160934010271e-05,
|
| 1164 |
-
"loss": 0.08674532771110535,
|
| 1165 |
-
"step": 3280
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 1.0361149003405399,
|
| 1169 |
-
"grad_norm": 0.2271793782711029,
|
| 1170 |
-
"learning_rate": 3.32043969897231e-05,
|
| 1171 |
-
"loss": 0.04139872789382935,
|
| 1172 |
-
"step": 3300
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.0423957525736773,
|
| 1176 |
-
"grad_norm": 0.25413981080055237,
|
| 1177 |
-
"learning_rate": 3.302672142282988e-05,
|
| 1178 |
-
"loss": 0.035733970999717715,
|
| 1179 |
-
"step": 3320
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 1.0486766048068148,
|
| 1183 |
-
"grad_norm": 0.09163467586040497,
|
| 1184 |
-
"learning_rate": 3.284859267090456e-05,
|
| 1185 |
-
"loss": 0.08379928469657898,
|
| 1186 |
-
"step": 3340
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 1.054957457039952,
|
| 1190 |
-
"grad_norm": 0.07256153970956802,
|
| 1191 |
-
"learning_rate": 3.267002079101525e-05,
|
| 1192 |
-
"loss": 0.03688819110393524,
|
| 1193 |
-
"step": 3360
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 1.0612383092730895,
|
| 1197 |
-
"grad_norm": 0.305923193693161,
|
| 1198 |
-
"learning_rate": 3.2491015865248884e-05,
|
| 1199 |
-
"loss": 0.049816560745239255,
|
| 1200 |
-
"step": 3380
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 1.067519161506227,
|
| 1204 |
-
"grad_norm": 0.046878550201654434,
|
| 1205 |
-
"learning_rate": 3.231158800014197e-05,
|
| 1206 |
-
"loss": 0.03804933726787567,
|
| 1207 |
-
"step": 3400
|
| 1208 |
-
},
|
| 1209 |
-
{
|
| 1210 |
-
"epoch": 1.0738000137393642,
|
| 1211 |
-
"grad_norm": 0.6410548686981201,
|
| 1212 |
-
"learning_rate": 3.213174732610995e-05,
|
| 1213 |
-
"loss": 0.0333125501871109,
|
| 1214 |
-
"step": 3420
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.0800808659725016,
|
| 1218 |
-
"grad_norm": 0.06022074818611145,
|
| 1219 |
-
"learning_rate": 3.195150399687531e-05,
|
| 1220 |
-
"loss": 0.028510311245918275,
|
| 1221 |
-
"step": 3440
|
| 1222 |
-
},
|
| 1223 |
-
{
|
| 1224 |
-
"epoch": 1.086361718205639,
|
| 1225 |
-
"grad_norm": 0.41786953806877136,
|
| 1226 |
-
"learning_rate": 3.177086818889423e-05,
|
| 1227 |
-
"loss": 0.046647512912750246,
|
| 1228 |
-
"step": 3460
|
| 1229 |
-
},
|
| 1230 |
-
{
|
| 1231 |
-
"epoch": 1.0926425704387763,
|
| 1232 |
-
"grad_norm": 0.1293671578168869,
|
| 1233 |
-
"learning_rate": 3.1589850100782085e-05,
|
| 1234 |
-
"loss": 0.03643852770328522,
|
| 1235 |
-
"step": 3480
|
| 1236 |
-
},
|
| 1237 |
-
{
|
| 1238 |
-
"epoch": 1.0989234226719138,
|
| 1239 |
-
"grad_norm": 0.05348878353834152,
|
| 1240 |
-
"learning_rate": 3.1408459952737594e-05,
|
| 1241 |
-
"loss": 0.03161140382289886,
|
| 1242 |
-
"step": 3500
|
| 1243 |
-
},
|
| 1244 |
-
{
|
| 1245 |
-
"epoch": 1.1052042749050512,
|
| 1246 |
-
"grad_norm": 0.29638731479644775,
|
| 1247 |
-
"learning_rate": 3.122670798596583e-05,
|
| 1248 |
-
"loss": 0.04662284553050995,
|
| 1249 |
-
"step": 3520
|
| 1250 |
-
},
|
| 1251 |
-
{
|
| 1252 |
-
"epoch": 1.1114851271381885,
|
| 1253 |
-
"grad_norm": 0.11425400525331497,
|
| 1254 |
-
"learning_rate": 3.1044604462099965e-05,
|
| 1255 |
-
"loss": 0.04306844174861908,
|
| 1256 |
-
"step": 3540
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.117765979371326,
|
| 1260 |
-
"grad_norm": 13.269792556762695,
|
| 1261 |
-
"learning_rate": 3.0862159662621933e-05,
|
| 1262 |
-
"loss": 0.10026944875717163,
|
| 1263 |
-
"step": 3560
|
| 1264 |
-
},
|
| 1265 |
-
{
|
| 1266 |
-
"epoch": 1.1240468316044634,
|
| 1267 |
-
"grad_norm": 0.09814783930778503,
|
| 1268 |
-
"learning_rate": 3.067938388828195e-05,
|
| 1269 |
-
"loss": 0.03279039561748505,
|
| 1270 |
-
"step": 3580
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.1303276838376006,
|
| 1274 |
-
"grad_norm": 2.8222439289093018,
|
| 1275 |
-
"learning_rate": 3.049628745851689e-05,
|
| 1276 |
-
"loss": 0.03593765795230865,
|
| 1277 |
-
"step": 3600
|
| 1278 |
-
},
|
| 1279 |
-
{
|
| 1280 |
-
"epoch": 1.136608536070738,
|
| 1281 |
-
"grad_norm": 0.14487864077091217,
|
| 1282 |
-
"learning_rate": 3.031288071086772e-05,
|
| 1283 |
-
"loss": 0.0726437509059906,
|
| 1284 |
-
"step": 3620
|
| 1285 |
-
},
|
| 1286 |
-
{
|
| 1287 |
-
"epoch": 1.1428893883038755,
|
| 1288 |
-
"grad_norm": 1.1210272312164307,
|
| 1289 |
-
"learning_rate": 3.0129174000395776e-05,
|
| 1290 |
-
"loss": 0.03795652091503143,
|
| 1291 |
-
"step": 3640
|
| 1292 |
-
},
|
| 1293 |
-
{
|
| 1294 |
-
"epoch": 1.1491702405370128,
|
| 1295 |
-
"grad_norm": 0.42077013850212097,
|
| 1296 |
-
"learning_rate": 2.99451776990982e-05,
|
| 1297 |
-
"loss": 0.04205833375453949,
|
| 1298 |
-
"step": 3660
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.1554510927701502,
|
| 1302 |
-
"grad_norm": 0.08231453597545624,
|
| 1303 |
-
"learning_rate": 2.976090219532226e-05,
|
| 1304 |
-
"loss": 0.04487053751945495,
|
| 1305 |
-
"step": 3680
|
| 1306 |
-
},
|
| 1307 |
-
{
|
| 1308 |
-
"epoch": 1.1617319450032877,
|
| 1309 |
-
"grad_norm": 0.1877751648426056,
|
| 1310 |
-
"learning_rate": 2.9576357893178885e-05,
|
| 1311 |
-
"loss": 0.034676852822303775,
|
| 1312 |
-
"step": 3700
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 1.168012797236425,
|
| 1316 |
-
"grad_norm": 0.6968119144439697,
|
| 1317 |
-
"learning_rate": 2.939155521195526e-05,
|
| 1318 |
-
"loss": 0.03220001757144928,
|
| 1319 |
-
"step": 3720
|
| 1320 |
-
},
|
| 1321 |
-
{
|
| 1322 |
-
"epoch": 1.1742936494695624,
|
| 1323 |
-
"grad_norm": 0.06414724141359329,
|
| 1324 |
-
"learning_rate": 2.9206504585526493e-05,
|
| 1325 |
-
"loss": 0.041881483793258664,
|
| 1326 |
-
"step": 3740
|
| 1327 |
-
},
|
| 1328 |
-
{
|
| 1329 |
-
"epoch": 1.1805745017026998,
|
| 1330 |
-
"grad_norm": 0.10654035210609436,
|
| 1331 |
-
"learning_rate": 2.9021216461766604e-05,
|
| 1332 |
-
"loss": 0.03662925064563751,
|
| 1333 |
-
"step": 3760
|
| 1334 |
-
},
|
| 1335 |
-
{
|
| 1336 |
-
"epoch": 1.186855353935837,
|
| 1337 |
-
"grad_norm": 0.04183540865778923,
|
| 1338 |
-
"learning_rate": 2.8835701301958574e-05,
|
| 1339 |
-
"loss": 0.04763528108596802,
|
| 1340 |
-
"step": 3780
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.1931362061689745,
|
| 1344 |
-
"grad_norm": 2.18105411529541,
|
| 1345 |
-
"learning_rate": 2.864996958020375e-05,
|
| 1346 |
-
"loss": 0.06307411193847656,
|
| 1347 |
-
"step": 3800
|
| 1348 |
-
},
|
| 1349 |
-
{
|
| 1350 |
-
"epoch": 1.199417058402112,
|
| 1351 |
-
"grad_norm": 0.3592161536216736,
|
| 1352 |
-
"learning_rate": 2.8464031782830474e-05,
|
| 1353 |
-
"loss": 0.051524901390075685,
|
| 1354 |
-
"step": 3820
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 1.2056979106352492,
|
| 1358 |
-
"grad_norm": 0.11931532621383667,
|
| 1359 |
-
"learning_rate": 2.827789840780201e-05,
|
| 1360 |
-
"loss": 0.035766962170600894,
|
| 1361 |
-
"step": 3840
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 1.2119787628683867,
|
| 1365 |
-
"grad_norm": 0.06202741712331772,
|
| 1366 |
-
"learning_rate": 2.8091579964123854e-05,
|
| 1367 |
-
"loss": 0.04704097211360932,
|
| 1368 |
-
"step": 3860
|
| 1369 |
-
},
|
| 1370 |
-
{
|
| 1371 |
-
"epoch": 1.2182596151015241,
|
| 1372 |
-
"grad_norm": 0.04776179790496826,
|
| 1373 |
-
"learning_rate": 2.7905086971250393e-05,
|
| 1374 |
-
"loss": 0.028332751989364625,
|
| 1375 |
-
"step": 3880
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 1.2245404673346614,
|
| 1379 |
-
"grad_norm": 0.2371724396944046,
|
| 1380 |
-
"learning_rate": 2.771842995849098e-05,
|
| 1381 |
-
"loss": 0.027332887053489685,
|
| 1382 |
-
"step": 3900
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.2308213195677988,
|
| 1386 |
-
"grad_norm": 0.05050333961844444,
|
| 1387 |
-
"learning_rate": 2.753161946441546e-05,
|
| 1388 |
-
"loss": 0.02915753722190857,
|
| 1389 |
-
"step": 3920
|
| 1390 |
-
},
|
| 1391 |
-
{
|
| 1392 |
-
"epoch": 1.2371021718009363,
|
| 1393 |
-
"grad_norm": 0.08952230215072632,
|
| 1394 |
-
"learning_rate": 2.734466603625915e-05,
|
| 1395 |
-
"loss": 0.025665390491485595,
|
| 1396 |
-
"step": 3940
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 1.2433830240340735,
|
| 1400 |
-
"grad_norm": 0.20825526118278503,
|
| 1401 |
-
"learning_rate": 2.7157580229327385e-05,
|
| 1402 |
-
"loss": 0.2276303768157959,
|
| 1403 |
-
"step": 3960
|
| 1404 |
-
},
|
| 1405 |
-
{
|
| 1406 |
-
"epoch": 1.249663876267211,
|
| 1407 |
-
"grad_norm": 1.8631341457366943,
|
| 1408 |
-
"learning_rate": 2.6970372606399525e-05,
|
| 1409 |
-
"loss": 0.03601144254207611,
|
| 1410 |
-
"step": 3980
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"epoch": 1.2559447285003484,
|
| 1414 |
-
"grad_norm": 0.21078747510910034,
|
| 1415 |
-
"learning_rate": 2.6783053737132607e-05,
|
| 1416 |
-
"loss": 0.08526483774185181,
|
| 1417 |
-
"step": 4000
|
| 1418 |
-
},
|
| 1419 |
-
{
|
| 1420 |
-
"epoch": 1.2622255807334857,
|
| 1421 |
-
"grad_norm": 0.051980458199977875,
|
| 1422 |
-
"learning_rate": 2.6595634197464596e-05,
|
| 1423 |
-
"loss": 0.022251833975315095,
|
| 1424 |
-
"step": 4020
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.2685064329666231,
|
| 1428 |
-
"grad_norm": 0.3019862771034241,
|
| 1429 |
-
"learning_rate": 2.6408124569017257e-05,
|
| 1430 |
-
"loss": 0.05009118318557739,
|
| 1431 |
-
"step": 4040
|
| 1432 |
-
},
|
| 1433 |
-
{
|
| 1434 |
-
"epoch": 1.2747872851997606,
|
| 1435 |
-
"grad_norm": 0.6371468305587769,
|
| 1436 |
-
"learning_rate": 2.622053543849872e-05,
|
| 1437 |
-
"loss": 0.040181088447570804,
|
| 1438 |
-
"step": 4060
|
| 1439 |
-
},
|
| 1440 |
-
{
|
| 1441 |
-
"epoch": 1.2810681374328978,
|
| 1442 |
-
"grad_norm": 1.1055502891540527,
|
| 1443 |
-
"learning_rate": 2.6032877397105783e-05,
|
| 1444 |
-
"loss": 0.02929992079734802,
|
| 1445 |
-
"step": 4080
|
| 1446 |
-
},
|
| 1447 |
-
{
|
| 1448 |
-
"epoch": 1.2873489896660353,
|
| 1449 |
-
"grad_norm": 0.26957738399505615,
|
| 1450 |
-
"learning_rate": 2.58451610399259e-05,
|
| 1451 |
-
"loss": 0.020054836571216584,
|
| 1452 |
-
"step": 4100
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 1.2936298418991727,
|
| 1456 |
-
"grad_norm": 0.07279884070158005,
|
| 1457 |
-
"learning_rate": 2.5657396965339003e-05,
|
| 1458 |
-
"loss": 0.039886316657066344,
|
| 1459 |
-
"step": 4120
|
| 1460 |
-
},
|
| 1461 |
-
{
|
| 1462 |
-
"epoch": 1.29991069413231,
|
| 1463 |
-
"grad_norm": 0.09953224658966064,
|
| 1464 |
-
"learning_rate": 2.5469595774419158e-05,
|
| 1465 |
-
"loss": 0.03994681239128113,
|
| 1466 |
-
"step": 4140
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.3061915463654474,
|
| 1470 |
-
"grad_norm": 0.14353635907173157,
|
| 1471 |
-
"learning_rate": 2.5281768070335988e-05,
|
| 1472 |
-
"loss": 0.03711488544940948,
|
| 1473 |
-
"step": 4160
|
| 1474 |
-
},
|
| 1475 |
-
{
|
| 1476 |
-
"epoch": 1.3124723985985849,
|
| 1477 |
-
"grad_norm": 0.27422836422920227,
|
| 1478 |
-
"learning_rate": 2.5093924457756018e-05,
|
| 1479 |
-
"loss": 0.023467497527599336,
|
| 1480 |
-
"step": 4180
|
| 1481 |
-
},
|
| 1482 |
-
{
|
| 1483 |
-
"epoch": 1.3187532508317221,
|
| 1484 |
-
"grad_norm": 0.0728367418050766,
|
| 1485 |
-
"learning_rate": 2.490607554224399e-05,
|
| 1486 |
-
"loss": 0.018936164677143097,
|
| 1487 |
-
"step": 4200
|
| 1488 |
-
},
|
| 1489 |
-
{
|
| 1490 |
-
"epoch": 1.3250341030648596,
|
| 1491 |
-
"grad_norm": 0.24521099030971527,
|
| 1492 |
-
"learning_rate": 2.4718231929664025e-05,
|
| 1493 |
-
"loss": 0.03874744176864624,
|
| 1494 |
-
"step": 4220
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 1.331314955297997,
|
| 1498 |
-
"grad_norm": 0.13463203608989716,
|
| 1499 |
-
"learning_rate": 2.453040422558084e-05,
|
| 1500 |
-
"loss": 0.03414709270000458,
|
| 1501 |
-
"step": 4240
|
| 1502 |
-
},
|
| 1503 |
-
{
|
| 1504 |
-
"epoch": 1.3375958075311343,
|
| 1505 |
-
"grad_norm": 0.12021703273057938,
|
| 1506 |
-
"learning_rate": 2.4342603034661003e-05,
|
| 1507 |
-
"loss": 0.028204512596130372,
|
| 1508 |
-
"step": 4260
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.3438766597642717,
|
| 1512 |
-
"grad_norm": 0.09300895035266876,
|
| 1513 |
-
"learning_rate": 2.4154838960074117e-05,
|
| 1514 |
-
"loss": 0.03560083508491516,
|
| 1515 |
-
"step": 4280
|
| 1516 |
-
},
|
| 1517 |
-
{
|
| 1518 |
-
"epoch": 1.3501575119974092,
|
| 1519 |
-
"grad_norm": 0.07003216445446014,
|
| 1520 |
-
"learning_rate": 2.3967122602894223e-05,
|
| 1521 |
-
"loss": 0.04477728307247162,
|
| 1522 |
-
"step": 4300
|
| 1523 |
-
},
|
| 1524 |
-
{
|
| 1525 |
-
"epoch": 1.3564383642305464,
|
| 1526 |
-
"grad_norm": 0.15798458456993103,
|
| 1527 |
-
"learning_rate": 2.377946456150128e-05,
|
| 1528 |
-
"loss": 0.05230234265327453,
|
| 1529 |
-
"step": 4320
|
| 1530 |
-
},
|
| 1531 |
-
{
|
| 1532 |
-
"epoch": 1.3627192164636839,
|
| 1533 |
-
"grad_norm": 0.05332406610250473,
|
| 1534 |
-
"learning_rate": 2.3591875430982752e-05,
|
| 1535 |
-
"loss": 0.029861843585968016,
|
| 1536 |
-
"step": 4340
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 1.3690000686968213,
|
| 1540 |
-
"grad_norm": 0.05614598095417023,
|
| 1541 |
-
"learning_rate": 2.3404365802535413e-05,
|
| 1542 |
-
"loss": 0.03531986773014069,
|
| 1543 |
-
"step": 4360
|
| 1544 |
-
},
|
| 1545 |
-
{
|
| 1546 |
-
"epoch": 1.3752809209299586,
|
| 1547 |
-
"grad_norm": 0.33550721406936646,
|
| 1548 |
-
"learning_rate": 2.3216946262867396e-05,
|
| 1549 |
-
"loss": 0.031084492802619934,
|
| 1550 |
-
"step": 4380
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.381561773163096,
|
| 1554 |
-
"grad_norm": 0.3401692807674408,
|
| 1555 |
-
"learning_rate": 2.3029627393600484e-05,
|
| 1556 |
-
"loss": 0.02833726108074188,
|
| 1557 |
-
"step": 4400
|
| 1558 |
-
},
|
| 1559 |
-
{
|
| 1560 |
-
"epoch": 1.3878426253962335,
|
| 1561 |
-
"grad_norm": 0.15662887692451477,
|
| 1562 |
-
"learning_rate": 2.2842419770672624e-05,
|
| 1563 |
-
"loss": 0.03976942598819733,
|
| 1564 |
-
"step": 4420
|
| 1565 |
-
},
|
| 1566 |
-
{
|
| 1567 |
-
"epoch": 1.3941234776293707,
|
| 1568 |
-
"grad_norm": 0.0530722439289093,
|
| 1569 |
-
"learning_rate": 2.265533396374085e-05,
|
| 1570 |
-
"loss": 0.03581877648830414,
|
| 1571 |
-
"step": 4440
|
| 1572 |
-
},
|
| 1573 |
-
{
|
| 1574 |
-
"epoch": 1.4004043298625082,
|
| 1575 |
-
"grad_norm": 0.09495271742343903,
|
| 1576 |
-
"learning_rate": 2.246838053558455e-05,
|
| 1577 |
-
"loss": 0.02298358678817749,
|
| 1578 |
-
"step": 4460
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 1.4066851820956456,
|
| 1582 |
-
"grad_norm": 0.04610360413789749,
|
| 1583 |
-
"learning_rate": 2.2281570041509024e-05,
|
| 1584 |
-
"loss": 0.04205411374568939,
|
| 1585 |
-
"step": 4480
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"epoch": 1.4129660343287829,
|
| 1589 |
-
"grad_norm": 0.07970083504915237,
|
| 1590 |
-
"learning_rate": 2.209491302874961e-05,
|
| 1591 |
-
"loss": 0.028789132833480835,
|
| 1592 |
-
"step": 4500
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.4192468865619203,
|
| 1596 |
-
"grad_norm": 1.6510498523712158,
|
| 1597 |
-
"learning_rate": 2.1908420035876148e-05,
|
| 1598 |
-
"loss": 0.030281484127044678,
|
| 1599 |
-
"step": 4520
|
| 1600 |
-
},
|
| 1601 |
-
{
|
| 1602 |
-
"epoch": 1.4255277387950578,
|
| 1603 |
-
"grad_norm": 0.061110883951187134,
|
| 1604 |
-
"learning_rate": 2.1722101592197995e-05,
|
| 1605 |
-
"loss": 0.026464346051216125,
|
| 1606 |
-
"step": 4540
|
| 1607 |
-
},
|
| 1608 |
-
{
|
| 1609 |
-
"epoch": 1.431808591028195,
|
| 1610 |
-
"grad_norm": 0.10222575813531876,
|
| 1611 |
-
"learning_rate": 2.1535968217169535e-05,
|
| 1612 |
-
"loss": 0.024321816861629486,
|
| 1613 |
-
"step": 4560
|
| 1614 |
-
},
|
| 1615 |
-
{
|
| 1616 |
-
"epoch": 1.4380894432613325,
|
| 1617 |
-
"grad_norm": 0.13632528483867645,
|
| 1618 |
-
"learning_rate": 2.135003041979625e-05,
|
| 1619 |
-
"loss": 0.022314466536045074,
|
| 1620 |
-
"step": 4580
|
| 1621 |
-
},
|
| 1622 |
-
{
|
| 1623 |
-
"epoch": 1.44437029549447,
|
| 1624 |
-
"grad_norm": 0.3248070478439331,
|
| 1625 |
-
"learning_rate": 2.1164298698041432e-05,
|
| 1626 |
-
"loss": 0.05798411965370178,
|
| 1627 |
-
"step": 4600
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"epoch": 1.4506511477276072,
|
| 1631 |
-
"grad_norm": 0.12926240265369415,
|
| 1632 |
-
"learning_rate": 2.0978783538233405e-05,
|
| 1633 |
-
"loss": 0.02899252772331238,
|
| 1634 |
-
"step": 4620
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 1.4569319999607446,
|
| 1638 |
-
"grad_norm": 0.23640146851539612,
|
| 1639 |
-
"learning_rate": 2.0793495414473506e-05,
|
| 1640 |
-
"loss": 0.020399539172649382,
|
| 1641 |
-
"step": 4640
|
| 1642 |
-
},
|
| 1643 |
-
{
|
| 1644 |
-
"epoch": 1.463212852193882,
|
| 1645 |
-
"grad_norm": 0.07195625454187393,
|
| 1646 |
-
"learning_rate": 2.0608444788044744e-05,
|
| 1647 |
-
"loss": 0.02832261621952057,
|
| 1648 |
-
"step": 4660
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 1.4694937044270193,
|
| 1652 |
-
"grad_norm": 0.05917816609144211,
|
| 1653 |
-
"learning_rate": 2.0423642106821118e-05,
|
| 1654 |
-
"loss": 0.017405569553375244,
|
| 1655 |
-
"step": 4680
|
| 1656 |
-
},
|
| 1657 |
-
{
|
| 1658 |
-
"epoch": 1.4757745566601568,
|
| 1659 |
-
"grad_norm": 0.050064459443092346,
|
| 1660 |
-
"learning_rate": 2.0239097804677755e-05,
|
| 1661 |
-
"loss": 0.03790850043296814,
|
| 1662 |
-
"step": 4700
|
| 1663 |
-
},
|
| 1664 |
-
{
|
| 1665 |
-
"epoch": 1.4820554088932942,
|
| 1666 |
-
"grad_norm": 0.06861293315887451,
|
| 1667 |
-
"learning_rate": 2.005482230090181e-05,
|
| 1668 |
-
"loss": 0.02318151295185089,
|
| 1669 |
-
"step": 4720
|
| 1670 |
-
},
|
| 1671 |
-
{
|
| 1672 |
-
"epoch": 1.4883362611264315,
|
| 1673 |
-
"grad_norm": 0.5692355036735535,
|
| 1674 |
-
"learning_rate": 1.987082599960423e-05,
|
| 1675 |
-
"loss": 0.05805472135543823,
|
| 1676 |
-
"step": 4740
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.494617113359569,
|
| 1680 |
-
"grad_norm": 0.10575691610574722,
|
| 1681 |
-
"learning_rate": 1.9687119289132286e-05,
|
| 1682 |
-
"loss": 0.025386866927146912,
|
| 1683 |
-
"step": 4760
|
| 1684 |
-
},
|
| 1685 |
-
{
|
| 1686 |
-
"epoch": 1.5008979655927064,
|
| 1687 |
-
"grad_norm": 0.07561904937028885,
|
| 1688 |
-
"learning_rate": 1.9503712541483105e-05,
|
| 1689 |
-
"loss": 0.02556038498878479,
|
| 1690 |
-
"step": 4780
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 1.5071788178258436,
|
| 1694 |
-
"grad_norm": 0.06342057138681412,
|
| 1695 |
-
"learning_rate": 1.9320616111718055e-05,
|
| 1696 |
-
"loss": 0.026987478137016296,
|
| 1697 |
-
"step": 4800
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 1.513459670058981,
|
| 1701 |
-
"grad_norm": 0.22881099581718445,
|
| 1702 |
-
"learning_rate": 1.913784033737807e-05,
|
| 1703 |
-
"loss": 0.023304201662540436,
|
| 1704 |
-
"step": 4820
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 1.5197405222921185,
|
| 1708 |
-
"grad_norm": 0.34815022349357605,
|
| 1709 |
-
"learning_rate": 1.8955395537900045e-05,
|
| 1710 |
-
"loss": 0.023628510534763336,
|
| 1711 |
-
"step": 4840
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 1.5260213745252558,
|
| 1715 |
-
"grad_norm": 0.07004115730524063,
|
| 1716 |
-
"learning_rate": 1.877329201403417e-05,
|
| 1717 |
-
"loss": 0.01844399720430374,
|
| 1718 |
-
"step": 4860
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 1.5323022267583932,
|
| 1722 |
-
"grad_norm": 0.5382964015007019,
|
| 1723 |
-
"learning_rate": 1.8591540047262408e-05,
|
| 1724 |
-
"loss": 0.022429858148097993,
|
| 1725 |
-
"step": 4880
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 1.5385830789915307,
|
| 1729 |
-
"grad_norm": 0.10155873745679855,
|
| 1730 |
-
"learning_rate": 1.841014989921792e-05,
|
| 1731 |
-
"loss": 0.019565099477767946,
|
| 1732 |
-
"step": 4900
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 1.544863931224668,
|
| 1736 |
-
"grad_norm": 0.45096078515052795,
|
| 1737 |
-
"learning_rate": 1.8229131811105773e-05,
|
| 1738 |
-
"loss": 0.02386317253112793,
|
| 1739 |
-
"step": 4920
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 1.5511447834578054,
|
| 1743 |
-
"grad_norm": 1.6493111848831177,
|
| 1744 |
-
"learning_rate": 1.8048496003124692e-05,
|
| 1745 |
-
"loss": 0.022837825119495392,
|
| 1746 |
-
"step": 4940
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 1.5574256356909428,
|
| 1750 |
-
"grad_norm": 0.2704576849937439,
|
| 1751 |
-
"learning_rate": 1.786825267389005e-05,
|
| 1752 |
-
"loss": 0.02158147543668747,
|
| 1753 |
-
"step": 4960
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 1.56370648792408,
|
| 1757 |
-
"grad_norm": 0.14172030985355377,
|
| 1758 |
-
"learning_rate": 1.768841199985804e-05,
|
| 1759 |
-
"loss": 0.02343100756406784,
|
| 1760 |
-
"step": 4980
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 1.5699873401572177,
|
| 1764 |
-
"grad_norm": 0.3839413523674011,
|
| 1765 |
-
"learning_rate": 1.7508984134751115e-05,
|
| 1766 |
-
"loss": 0.0211731418967247,
|
| 1767 |
-
"step": 5000
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 1.576268192390355,
|
| 1771 |
-
"grad_norm": 0.17986074090003967,
|
| 1772 |
-
"learning_rate": 1.7329979208984757e-05,
|
| 1773 |
-
"loss": 0.050633370876312256,
|
| 1774 |
-
"step": 5020
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 1.5825490446234922,
|
| 1778 |
-
"grad_norm": 0.11192335188388824,
|
| 1779 |
-
"learning_rate": 1.715140732909545e-05,
|
| 1780 |
-
"loss": 0.019000351428985596,
|
| 1781 |
-
"step": 5040
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 1.58882989685663,
|
| 1785 |
-
"grad_norm": 0.06556563079357147,
|
| 1786 |
-
"learning_rate": 1.697327857717012e-05,
|
| 1787 |
-
"loss": 0.031407642364501956,
|
| 1788 |
-
"step": 5060
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 1.5951107490897671,
|
| 1792 |
-
"grad_norm": 0.5898942351341248,
|
| 1793 |
-
"learning_rate": 1.67956030102769e-05,
|
| 1794 |
-
"loss": 0.03214563131332397,
|
| 1795 |
-
"step": 5080
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 1.6013916013229044,
|
| 1799 |
-
"grad_norm": 0.11632583290338516,
|
| 1800 |
-
"learning_rate": 1.6618390659897297e-05,
|
| 1801 |
-
"loss": 0.12849061489105223,
|
| 1802 |
-
"step": 5100
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 1.607672453556042,
|
| 1806 |
-
"grad_norm": 0.26994606852531433,
|
| 1807 |
-
"learning_rate": 1.64416515313598e-05,
|
| 1808 |
-
"loss": 0.01800369322299957,
|
| 1809 |
-
"step": 5120
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 1.6139533057891793,
|
| 1813 |
-
"grad_norm": 0.3220551908016205,
|
| 1814 |
-
"learning_rate": 1.626539560327505e-05,
|
| 1815 |
-
"loss": 0.022022050619125367,
|
| 1816 |
-
"step": 5140
|
| 1817 |
-
},
|
| 1818 |
-
{
|
| 1819 |
-
"epoch": 1.6202341580223165,
|
| 1820 |
-
"grad_norm": 0.16517284512519836,
|
| 1821 |
-
"learning_rate": 1.6089632826972395e-05,
|
| 1822 |
-
"loss": 0.015372823178768157,
|
| 1823 |
-
"step": 5160
|
| 1824 |
-
},
|
| 1825 |
-
{
|
| 1826 |
-
"epoch": 1.6265150102554542,
|
| 1827 |
-
"grad_norm": 0.03851303830742836,
|
| 1828 |
-
"learning_rate": 1.591437312593804e-05,
|
| 1829 |
-
"loss": 0.0164533868432045,
|
| 1830 |
-
"step": 5180
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 1.6327958624885914,
|
| 1834 |
-
"grad_norm": 0.3491153419017792,
|
| 1835 |
-
"learning_rate": 1.5739626395254798e-05,
|
| 1836 |
-
"loss": 0.031122979521751405,
|
| 1837 |
-
"step": 5200
|
| 1838 |
-
},
|
| 1839 |
-
{
|
| 1840 |
-
"epoch": 1.6390767147217287,
|
| 1841 |
-
"grad_norm": 0.06409186869859695,
|
| 1842 |
-
"learning_rate": 1.556540250104343e-05,
|
| 1843 |
-
"loss": 0.0328720360994339,
|
| 1844 |
-
"step": 5220
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 1.6453575669548663,
|
| 1848 |
-
"grad_norm": 0.04858129099011421,
|
| 1849 |
-
"learning_rate": 1.539171127990557e-05,
|
| 1850 |
-
"loss": 0.05794671773910522,
|
| 1851 |
-
"step": 5240
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"epoch": 1.6516384191880036,
|
| 1855 |
-
"grad_norm": 0.262850821018219,
|
| 1856 |
-
"learning_rate": 1.5218562538368364e-05,
|
| 1857 |
-
"loss": 0.014299227297306061,
|
| 1858 |
-
"step": 5260
|
| 1859 |
-
},
|
| 1860 |
-
{
|
| 1861 |
-
"epoch": 1.6579192714211408,
|
| 1862 |
-
"grad_norm": 0.12737710773944855,
|
| 1863 |
-
"learning_rate": 1.504596605233084e-05,
|
| 1864 |
-
"loss": 0.02150038182735443,
|
| 1865 |
-
"step": 5280
|
| 1866 |
-
},
|
| 1867 |
-
{
|
| 1868 |
-
"epoch": 1.6642001236542785,
|
| 1869 |
-
"grad_norm": 0.06830194592475891,
|
| 1870 |
-
"learning_rate": 1.4873931566511901e-05,
|
| 1871 |
-
"loss": 0.01579444855451584,
|
| 1872 |
-
"step": 5300
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 1.6704809758874157,
|
| 1876 |
-
"grad_norm": 0.08396535366773605,
|
| 1877 |
-
"learning_rate": 1.4702468793900188e-05,
|
| 1878 |
-
"loss": 0.027909812331199647,
|
| 1879 |
-
"step": 5320
|
| 1880 |
-
},
|
| 1881 |
-
{
|
| 1882 |
-
"epoch": 1.676761828120553,
|
| 1883 |
-
"grad_norm": 0.26403504610061646,
|
| 1884 |
-
"learning_rate": 1.4531587415205667e-05,
|
| 1885 |
-
"loss": 0.02287571579217911,
|
| 1886 |
-
"step": 5340
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 1.6830426803536906,
|
| 1890 |
-
"grad_norm": 0.43412601947784424,
|
| 1891 |
-
"learning_rate": 1.436129707831308e-05,
|
| 1892 |
-
"loss": 0.017245300114154816,
|
| 1893 |
-
"step": 5360
|
| 1894 |
-
},
|
| 1895 |
-
{
|
| 1896 |
-
"epoch": 1.6893235325868279,
|
| 1897 |
-
"grad_norm": 0.18354500830173492,
|
| 1898 |
-
"learning_rate": 1.4191607397737206e-05,
|
| 1899 |
-
"loss": 0.031501269340515135,
|
| 1900 |
-
"step": 5380
|
| 1901 |
-
},
|
| 1902 |
-
{
|
| 1903 |
-
"epoch": 1.6956043848199651,
|
| 1904 |
-
"grad_norm": 0.04084882140159607,
|
| 1905 |
-
"learning_rate": 1.402252795408004e-05,
|
| 1906 |
-
"loss": 0.036852356791496274,
|
| 1907 |
-
"step": 5400
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 1.7018852370531028,
|
| 1911 |
-
"grad_norm": 0.06115623563528061,
|
| 1912 |
-
"learning_rate": 1.3854068293489909e-05,
|
| 1913 |
-
"loss": 0.028507155179977418,
|
| 1914 |
-
"step": 5420
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 1.70816608928624,
|
| 1918 |
-
"grad_norm": 0.32743072509765625,
|
| 1919 |
-
"learning_rate": 1.368623792712243e-05,
|
| 1920 |
-
"loss": 0.014020594954490661,
|
| 1921 |
-
"step": 5440
|
| 1922 |
-
},
|
| 1923 |
-
{
|
| 1924 |
-
"epoch": 1.7144469415193773,
|
| 1925 |
-
"grad_norm": 0.06222687289118767,
|
| 1926 |
-
"learning_rate": 1.3519046330603584e-05,
|
| 1927 |
-
"loss": 0.05104716420173645,
|
| 1928 |
-
"step": 5460
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 1.720727793752515,
|
| 1932 |
-
"grad_norm": 0.06253579258918762,
|
| 1933 |
-
"learning_rate": 1.3352502943494672e-05,
|
| 1934 |
-
"loss": 0.026476505398750304,
|
| 1935 |
-
"step": 5480
|
| 1936 |
-
},
|
| 1937 |
-
{
|
| 1938 |
-
"epoch": 1.7270086459856522,
|
| 1939 |
-
"grad_norm": 0.4662589430809021,
|
| 1940 |
-
"learning_rate": 1.318661716875942e-05,
|
| 1941 |
-
"loss": 0.01999303698539734,
|
| 1942 |
-
"step": 5500
|
| 1943 |
-
},
|
| 1944 |
-
{
|
| 1945 |
-
"epoch": 1.7332894982187894,
|
| 1946 |
-
"grad_norm": 0.1247578039765358,
|
| 1947 |
-
"learning_rate": 1.3021398372233035e-05,
|
| 1948 |
-
"loss": 0.02069215327501297,
|
| 1949 |
-
"step": 5520
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 1.739570350451927,
|
| 1953 |
-
"grad_norm": 0.3788617253303528,
|
| 1954 |
-
"learning_rate": 1.2856855882093432e-05,
|
| 1955 |
-
"loss": 0.018105532228946685,
|
| 1956 |
-
"step": 5540
|
| 1957 |
-
},
|
| 1958 |
-
{
|
| 1959 |
-
"epoch": 1.7458512026850643,
|
| 1960 |
-
"grad_norm": 1.8883055448532104,
|
| 1961 |
-
"learning_rate": 1.2692998988334612e-05,
|
| 1962 |
-
"loss": 0.04176854789257049,
|
| 1963 |
-
"step": 5560
|
| 1964 |
-
},
|
| 1965 |
-
{
|
| 1966 |
-
"epoch": 1.7521320549182016,
|
| 1967 |
-
"grad_norm": 0.21989472210407257,
|
| 1968 |
-
"learning_rate": 1.252983694224208e-05,
|
| 1969 |
-
"loss": 0.020176875591278075,
|
| 1970 |
-
"step": 5580
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 1.7584129071513392,
|
| 1974 |
-
"grad_norm": 0.16766223311424255,
|
| 1975 |
-
"learning_rate": 1.236737895587056e-05,
|
| 1976 |
-
"loss": 0.020288291573524474,
|
| 1977 |
-
"step": 5600
|
| 1978 |
-
},
|
| 1979 |
-
{
|
| 1980 |
-
"epoch": 1.7646937593844765,
|
| 1981 |
-
"grad_norm": 0.09795085340738297,
|
| 1982 |
-
"learning_rate": 1.2205634201523922e-05,
|
| 1983 |
-
"loss": 0.015002159774303437,
|
| 1984 |
-
"step": 5620
|
| 1985 |
-
},
|
| 1986 |
-
{
|
| 1987 |
-
"epoch": 1.7709746116176137,
|
| 1988 |
-
"grad_norm": 0.029697278514504433,
|
| 1989 |
-
"learning_rate": 1.2044611811237242e-05,
|
| 1990 |
-
"loss": 0.023318004608154298,
|
| 1991 |
-
"step": 5640
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 1.7772554638507514,
|
| 1995 |
-
"grad_norm": 0.3376557528972626,
|
| 1996 |
-
"learning_rate": 1.1884320876261268e-05,
|
| 1997 |
-
"loss": 0.0196805402636528,
|
| 1998 |
-
"step": 5660
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 1.7835363160838886,
|
| 2002 |
-
"grad_norm": 0.06531202793121338,
|
| 2003 |
-
"learning_rate": 1.17247704465491e-05,
|
| 2004 |
-
"loss": 0.020646215975284578,
|
| 2005 |
-
"step": 5680
|
| 2006 |
-
},
|
| 2007 |
-
{
|
| 2008 |
-
"epoch": 1.7898171683170259,
|
| 2009 |
-
"grad_norm": 0.36572521924972534,
|
| 2010 |
-
"learning_rate": 1.1565969530245277e-05,
|
| 2011 |
-
"loss": 0.025085961818695067,
|
| 2012 |
-
"step": 5700
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 1.7960980205501635,
|
| 2016 |
-
"grad_norm": 0.1814069002866745,
|
| 2017 |
-
"learning_rate": 1.140792709317713e-05,
|
| 2018 |
-
"loss": 0.030590158700942994,
|
| 2019 |
-
"step": 5720
|
| 2020 |
-
},
|
| 2021 |
-
{
|
| 2022 |
-
"epoch": 1.8023788727833008,
|
| 2023 |
-
"grad_norm": 0.24981825053691864,
|
| 2024 |
-
"learning_rate": 1.12506520583486e-05,
|
| 2025 |
-
"loss": 0.0215068444609642,
|
| 2026 |
-
"step": 5740
|
| 2027 |
-
},
|
| 2028 |
-
{
|
| 2029 |
-
"epoch": 1.808659725016438,
|
| 2030 |
-
"grad_norm": 0.35276567935943604,
|
| 2031 |
-
"learning_rate": 1.1094153305436453e-05,
|
| 2032 |
-
"loss": 0.01926688849925995,
|
| 2033 |
-
"step": 5760
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 1.8149405772495757,
|
| 2037 |
-
"grad_norm": 0.02407790906727314,
|
| 2038 |
-
"learning_rate": 1.0938439670288928e-05,
|
| 2039 |
-
"loss": 0.013058166205883025,
|
| 2040 |
-
"step": 5780
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 1.821221429482713,
|
| 2044 |
-
"grad_norm": 0.08507027477025986,
|
| 2045 |
-
"learning_rate": 1.0783519944426874e-05,
|
| 2046 |
-
"loss": 0.01877378821372986,
|
| 2047 |
-
"step": 5800
|
| 2048 |
-
},
|
| 2049 |
-
{
|
| 2050 |
-
"epoch": 1.8275022817158502,
|
| 2051 |
-
"grad_norm": 0.18674074113368988,
|
| 2052 |
-
"learning_rate": 1.0629402874547378e-05,
|
| 2053 |
-
"loss": 0.018952280282974243,
|
| 2054 |
-
"step": 5820
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 1.8337831339489878,
|
| 2058 |
-
"grad_norm": 0.057133324444293976,
|
| 2059 |
-
"learning_rate": 1.047609716202996e-05,
|
| 2060 |
-
"loss": 0.030013075470924376,
|
| 2061 |
-
"step": 5840
|
| 2062 |
-
},
|
| 2063 |
-
{
|
| 2064 |
-
"epoch": 1.840063986182125,
|
| 2065 |
-
"grad_norm": 0.10284405946731567,
|
| 2066 |
-
"learning_rate": 1.0323611462445246e-05,
|
| 2067 |
-
"loss": 0.020608581602573395,
|
| 2068 |
-
"step": 5860
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 1.8463448384152623,
|
| 2072 |
-
"grad_norm": 0.25261375308036804,
|
| 2073 |
-
"learning_rate": 1.0171954385066303e-05,
|
| 2074 |
-
"loss": 0.02777992784976959,
|
| 2075 |
-
"step": 5880
|
| 2076 |
-
},
|
| 2077 |
-
{
|
| 2078 |
-
"epoch": 1.8526256906484,
|
| 2079 |
-
"grad_norm": 0.38947853446006775,
|
| 2080 |
-
"learning_rate": 1.0021134492382606e-05,
|
| 2081 |
-
"loss": 0.02200171649456024,
|
| 2082 |
-
"step": 5900
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 1.8589065428815372,
|
| 2086 |
-
"grad_norm": 0.1173999160528183,
|
| 2087 |
-
"learning_rate": 9.87116029961652e-06,
|
| 2088 |
-
"loss": 0.019530104100704195,
|
| 2089 |
-
"step": 5920
|
| 2090 |
-
},
|
| 2091 |
-
{
|
| 2092 |
-
"epoch": 1.8651873951146745,
|
| 2093 |
-
"grad_norm": 2.3180384635925293,
|
| 2094 |
-
"learning_rate": 9.722040274242594e-06,
|
| 2095 |
-
"loss": 0.028142544627189636,
|
| 2096 |
-
"step": 5940
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 1.8714682473478121,
|
| 2100 |
-
"grad_norm": 0.4677383601665497,
|
| 2101 |
-
"learning_rate": 9.573782835509471e-06,
|
| 2102 |
-
"loss": 0.017200836539268495,
|
| 2103 |
-
"step": 5960
|
| 2104 |
-
},
|
| 2105 |
-
{
|
| 2106 |
-
"epoch": 1.8777490995809494,
|
| 2107 |
-
"grad_norm": 0.19456204771995544,
|
| 2108 |
-
"learning_rate": 9.426396353964565e-06,
|
| 2109 |
-
"loss": 0.015525393187999725,
|
| 2110 |
-
"step": 5980
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 1.8840299518140866,
|
| 2114 |
-
"grad_norm": 0.561476469039917,
|
| 2115 |
-
"learning_rate": 9.279889150981421e-06,
|
| 2116 |
-
"loss": 0.04386698603630066,
|
| 2117 |
-
"step": 6000
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 1.8903108040472243,
|
| 2121 |
-
"grad_norm": 0.0613991916179657,
|
| 2122 |
-
"learning_rate": 9.134269498289922e-06,
|
| 2123 |
-
"loss": 0.031869342923164366,
|
| 2124 |
-
"step": 6020
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 1.8965916562803615,
|
| 2128 |
-
"grad_norm": 0.047774072736501694,
|
| 2129 |
-
"learning_rate": 8.989545617509283e-06,
|
| 2130 |
-
"loss": 0.02780761420726776,
|
| 2131 |
-
"step": 6040
|
| 2132 |
-
},
|
| 2133 |
-
{
|
| 2134 |
-
"epoch": 1.9028725085134988,
|
| 2135 |
-
"grad_norm": 0.33418071269989014,
|
| 2136 |
-
"learning_rate": 8.845725679683822e-06,
|
| 2137 |
-
"loss": 0.04635312259197235,
|
| 2138 |
-
"step": 6060
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 1.9091533607466364,
|
| 2142 |
-
"grad_norm": 0.0721767470240593,
|
| 2143 |
-
"learning_rate": 8.702817804821664e-06,
|
| 2144 |
-
"loss": 0.01725383847951889,
|
| 2145 |
-
"step": 6080
|
| 2146 |
-
},
|
| 2147 |
-
{
|
| 2148 |
-
"epoch": 1.9154342129797737,
|
| 2149 |
-
"grad_norm": 0.04245443269610405,
|
| 2150 |
-
"learning_rate": 8.560830061436265e-06,
|
| 2151 |
-
"loss": 0.02144094407558441,
|
| 2152 |
-
"step": 6100
|
| 2153 |
-
},
|
| 2154 |
-
{
|
| 2155 |
-
"epoch": 1.921715065212911,
|
| 2156 |
-
"grad_norm": 0.1140967458486557,
|
| 2157 |
-
"learning_rate": 8.419770466090896e-06,
|
| 2158 |
-
"loss": 0.014450745284557342,
|
| 2159 |
-
"step": 6120
|
| 2160 |
-
},
|
| 2161 |
-
{
|
| 2162 |
-
"epoch": 1.9279959174460486,
|
| 2163 |
-
"grad_norm": 0.06213388964533806,
|
| 2164 |
-
"learning_rate": 8.27964698294599e-06,
|
| 2165 |
-
"loss": 0.023716664314270018,
|
| 2166 |
-
"step": 6140
|
| 2167 |
-
},
|
| 2168 |
-
{
|
| 2169 |
-
"epoch": 1.9342767696791858,
|
| 2170 |
-
"grad_norm": 0.2430906891822815,
|
| 2171 |
-
"learning_rate": 8.140467523309527e-06,
|
| 2172 |
-
"loss": 0.011777687072753906,
|
| 2173 |
-
"step": 6160
|
| 2174 |
-
},
|
| 2175 |
-
{
|
| 2176 |
-
"epoch": 1.9405576219123233,
|
| 2177 |
-
"grad_norm": 0.044944941997528076,
|
| 2178 |
-
"learning_rate": 8.002239945190345e-06,
|
| 2179 |
-
"loss": 0.01790996491909027,
|
| 2180 |
-
"step": 6180
|
| 2181 |
-
},
|
| 2182 |
-
{
|
| 2183 |
-
"epoch": 1.9468384741454607,
|
| 2184 |
-
"grad_norm": 0.032080262899398804,
|
| 2185 |
-
"learning_rate": 7.864972052854485e-06,
|
| 2186 |
-
"loss": 0.03710399568080902,
|
| 2187 |
-
"step": 6200
|
| 2188 |
-
},
|
| 2189 |
-
{
|
| 2190 |
-
"epoch": 1.953119326378598,
|
| 2191 |
-
"grad_norm": 0.4400108754634857,
|
| 2192 |
-
"learning_rate": 7.72867159638456e-06,
|
| 2193 |
-
"loss": 0.022929078340530394,
|
| 2194 |
-
"step": 6220
|
| 2195 |
-
},
|
| 2196 |
-
{
|
| 2197 |
-
"epoch": 1.9594001786117354,
|
| 2198 |
-
"grad_norm": 0.04446094483137131,
|
| 2199 |
-
"learning_rate": 7.593346271242216e-06,
|
| 2200 |
-
"loss": 0.01109272986650467,
|
| 2201 |
-
"step": 6240
|
| 2202 |
-
},
|
| 2203 |
-
{
|
| 2204 |
-
"epoch": 1.965681030844873,
|
| 2205 |
-
"grad_norm": 0.37326884269714355,
|
| 2206 |
-
"learning_rate": 7.459003717833607e-06,
|
| 2207 |
-
"loss": 0.04479179680347443,
|
| 2208 |
-
"step": 6260
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 1.9719618830780101,
|
| 2212 |
-
"grad_norm": 0.03807030990719795,
|
| 2213 |
-
"learning_rate": 7.325651521078053e-06,
|
| 2214 |
-
"loss": 0.020693680644035338,
|
| 2215 |
-
"step": 6280
|
| 2216 |
-
},
|
| 2217 |
-
{
|
| 2218 |
-
"epoch": 1.9782427353111476,
|
| 2219 |
-
"grad_norm": 0.07757691293954849,
|
| 2220 |
-
"learning_rate": 7.193297209979777e-06,
|
| 2221 |
-
"loss": 0.015228751301765441,
|
| 2222 |
-
"step": 6300
|
| 2223 |
-
},
|
| 2224 |
-
{
|
| 2225 |
-
"epoch": 1.984523587544285,
|
| 2226 |
-
"grad_norm": 0.06824570149183273,
|
| 2227 |
-
"learning_rate": 7.061948257202858e-06,
|
| 2228 |
-
"loss": 0.01840972900390625,
|
| 2229 |
-
"step": 6320
|
| 2230 |
-
},
|
| 2231 |
-
{
|
| 2232 |
-
"epoch": 1.9908044397774223,
|
| 2233 |
-
"grad_norm": 0.1820923089981079,
|
| 2234 |
-
"learning_rate": 6.931612078649283e-06,
|
| 2235 |
-
"loss": 0.02284184694290161,
|
| 2236 |
-
"step": 6340
|
| 2237 |
-
},
|
| 2238 |
-
{
|
| 2239 |
-
"epoch": 1.9970852920105597,
|
| 2240 |
-
"grad_norm": 0.0430208295583725,
|
| 2241 |
-
"learning_rate": 6.8022960330402616e-06,
|
| 2242 |
-
"loss": 0.016644990444183348,
|
| 2243 |
-
"step": 6360
|
| 2244 |
-
},
|
| 2245 |
-
{
|
| 2246 |
-
"epoch": 2.003140426116569,
|
| 2247 |
-
"grad_norm": 0.050990670919418335,
|
| 2248 |
-
"learning_rate": 6.674007421500786e-06,
|
| 2249 |
-
"loss": 0.03776961266994476,
|
| 2250 |
-
"step": 6380
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 2.009421278349706,
|
| 2254 |
-
"grad_norm": 0.13200590014457703,
|
| 2255 |
-
"learning_rate": 6.546753487147375e-06,
|
| 2256 |
-
"loss": 0.026686957478523253,
|
| 2257 |
-
"step": 6400
|
| 2258 |
-
},
|
| 2259 |
-
{
|
| 2260 |
-
"epoch": 2.0157021305828433,
|
| 2261 |
-
"grad_norm": 0.08269485086202621,
|
| 2262 |
-
"learning_rate": 6.42054141467914e-06,
|
| 2263 |
-
"loss": 0.01832299828529358,
|
| 2264 |
-
"step": 6420
|
| 2265 |
-
},
|
| 2266 |
-
{
|
| 2267 |
-
"epoch": 2.021982982815981,
|
| 2268 |
-
"grad_norm": 0.04629537835717201,
|
| 2269 |
-
"learning_rate": 6.2953783299721486e-06,
|
| 2270 |
-
"loss": 0.01113806888461113,
|
| 2271 |
-
"step": 6440
|
| 2272 |
-
}
|
| 2273 |
-
],
|
| 2274 |
-
"logging_steps": 20,
|
| 2275 |
-
"max_steps": 8362,
|
| 2276 |
-
"num_input_tokens_seen": 0,
|
| 2277 |
-
"num_train_epochs": 3,
|
| 2278 |
-
"save_steps": 1000000000,
|
| 2279 |
-
"stateful_callbacks": {
|
| 2280 |
-
"TrainerControl": {
|
| 2281 |
-
"args": {
|
| 2282 |
-
"should_epoch_stop": false,
|
| 2283 |
-
"should_evaluate": false,
|
| 2284 |
-
"should_log": false,
|
| 2285 |
-
"should_save": true,
|
| 2286 |
-
"should_training_stop": false
|
| 2287 |
-
},
|
| 2288 |
-
"attributes": {}
|
| 2289 |
-
}
|
| 2290 |
-
},
|
| 2291 |
-
"total_flos": 2.038940224187904e+16,
|
| 2292 |
-
"train_batch_size": 1,
|
| 2293 |
-
"trial_name": null,
|
| 2294 |
-
"trial_params": null
|
| 2295 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
slots/13/checkpoint-8035/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/13/checkpoint-8035/trainer_state.json
DELETED
|
@@ -1,2848 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 2.5228809484086874,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 8035,
|
| 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 |
-
"epoch": 1.942543475131832,
|
| 1638 |
-
"grad_norm": 0.33941805362701416,
|
| 1639 |
-
"learning_rate": 2.0784237155687184e-05,
|
| 1640 |
-
"loss": 0.05176846385002136,
|
| 1641 |
-
"step": 4640
|
| 1642 |
-
},
|
| 1643 |
-
{
|
| 1644 |
-
"epoch": 1.950917917380893,
|
| 1645 |
-
"grad_norm": 0.32326143980026245,
|
| 1646 |
-
"learning_rate": 2.0599198678495653e-05,
|
| 1647 |
-
"loss": 0.04779183864593506,
|
| 1648 |
-
"step": 4660
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 1.9592923596299543,
|
| 1652 |
-
"grad_norm": 0.10640216618776321,
|
| 1653 |
-
"learning_rate": 2.041440866854038e-05,
|
| 1654 |
-
"loss": 0.03176500201225281,
|
| 1655 |
-
"step": 4680
|
| 1656 |
-
},
|
| 1657 |
-
{
|
| 1658 |
-
"epoch": 1.9676668018790155,
|
| 1659 |
-
"grad_norm": 0.14410381019115448,
|
| 1660 |
-
"learning_rate": 2.0229877558981075e-05,
|
| 1661 |
-
"loss": 0.10440460443496705,
|
| 1662 |
-
"step": 4700
|
| 1663 |
-
},
|
| 1664 |
-
{
|
| 1665 |
-
"epoch": 1.9760412441280766,
|
| 1666 |
-
"grad_norm": 0.4526270031929016,
|
| 1667 |
-
"learning_rate": 2.0045615768360078e-05,
|
| 1668 |
-
"loss": 0.03654649257659912,
|
| 1669 |
-
"step": 4720
|
| 1670 |
-
},
|
| 1671 |
-
{
|
| 1672 |
-
"epoch": 1.9844156863771376,
|
| 1673 |
-
"grad_norm": 0.09464205056428909,
|
| 1674 |
-
"learning_rate": 1.9861633700014065e-05,
|
| 1675 |
-
"loss": 0.041169199347496035,
|
| 1676 |
-
"step": 4740
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.9927901286261989,
|
| 1680 |
-
"grad_norm": 0.7187637090682983,
|
| 1681 |
-
"learning_rate": 1.9677941741486735e-05,
|
| 1682 |
-
"loss": 0.07295008301734925,
|
| 1683 |
-
"step": 4760
|
| 1684 |
-
},
|
| 1685 |
-
{
|
| 1686 |
-
"epoch": 2.000837444224906,
|
| 1687 |
-
"grad_norm": 0.14082716405391693,
|
| 1688 |
-
"learning_rate": 1.949455026394233e-05,
|
| 1689 |
-
"loss": 0.04572053253650665,
|
| 1690 |
-
"step": 4780
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 2.009211886473967,
|
| 1694 |
-
"grad_norm": 2.1813251972198486,
|
| 1695 |
-
"learning_rate": 1.931146962158006e-05,
|
| 1696 |
-
"loss": 0.06506224870681762,
|
| 1697 |
-
"step": 4800
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 2.0175863287230285,
|
| 1701 |
-
"grad_norm": 0.5109630227088928,
|
| 1702 |
-
"learning_rate": 1.912871015104952e-05,
|
| 1703 |
-
"loss": 0.038307076692581175,
|
| 1704 |
-
"step": 4820
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 2.0259607709720897,
|
| 1708 |
-
"grad_norm": 0.5181107521057129,
|
| 1709 |
-
"learning_rate": 1.894628217086709e-05,
|
| 1710 |
-
"loss": 0.04958427548408508,
|
| 1711 |
-
"step": 4840
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 2.0343352132211505,
|
| 1715 |
-
"grad_norm": 0.15236905217170715,
|
| 1716 |
-
"learning_rate": 1.876419598083338e-05,
|
| 1717 |
-
"loss": 0.03732606470584869,
|
| 1718 |
-
"step": 4860
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 2.0427096554702118,
|
| 1722 |
-
"grad_norm": 0.27413368225097656,
|
| 1723 |
-
"learning_rate": 1.8582461861451654e-05,
|
| 1724 |
-
"loss": 0.02974282205104828,
|
| 1725 |
-
"step": 4880
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 2.051084097719273,
|
| 1729 |
-
"grad_norm": 0.30715715885162354,
|
| 1730 |
-
"learning_rate": 1.840109007334743e-05,
|
| 1731 |
-
"loss": 0.03197464346885681,
|
| 1732 |
-
"step": 4900
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 2.0594585399683343,
|
| 1736 |
-
"grad_norm": 0.15932349860668182,
|
| 1737 |
-
"learning_rate": 1.8220090856689192e-05,
|
| 1738 |
-
"loss": 0.020184148848056794,
|
| 1739 |
-
"step": 4920
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 2.067832982217395,
|
| 1743 |
-
"grad_norm": 0.2691262364387512,
|
| 1744 |
-
"learning_rate": 1.8039474430610193e-05,
|
| 1745 |
-
"loss": 0.034737563133239745,
|
| 1746 |
-
"step": 4940
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 2.0762074244664563,
|
| 1750 |
-
"grad_norm": 0.5201924443244934,
|
| 1751 |
-
"learning_rate": 1.7859250992631497e-05,
|
| 1752 |
-
"loss": 0.02623576819896698,
|
| 1753 |
-
"step": 4960
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 2.0845818667155176,
|
| 1757 |
-
"grad_norm": 0.2914315462112427,
|
| 1758 |
-
"learning_rate": 1.7679430718086243e-05,
|
| 1759 |
-
"loss": 0.06491096019744873,
|
| 1760 |
-
"step": 4980
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 2.092956308964579,
|
| 1764 |
-
"grad_norm": 0.1011788621544838,
|
| 1765 |
-
"learning_rate": 1.7500023759545165e-05,
|
| 1766 |
-
"loss": 0.02444133162498474,
|
| 1767 |
-
"step": 5000
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 2.1013307512136397,
|
| 1771 |
-
"grad_norm": 0.4195410907268524,
|
| 1772 |
-
"learning_rate": 1.7321040246243335e-05,
|
| 1773 |
-
"loss": 0.03377276659011841,
|
| 1774 |
-
"step": 5020
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 2.109705193462701,
|
| 1778 |
-
"grad_norm": 0.2820679843425751,
|
| 1779 |
-
"learning_rate": 1.7142490283508324e-05,
|
| 1780 |
-
"loss": 0.038240480422973636,
|
| 1781 |
-
"step": 5040
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 2.118079635711762,
|
| 1785 |
-
"grad_norm": 0.1285504251718521,
|
| 1786 |
-
"learning_rate": 1.696438395218963e-05,
|
| 1787 |
-
"loss": 0.04785881340503693,
|
| 1788 |
-
"step": 5060
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 2.1264540779608234,
|
| 1792 |
-
"grad_norm": 0.07012035697698593,
|
| 1793 |
-
"learning_rate": 1.6786731308089513e-05,
|
| 1794 |
-
"loss": 0.035026317834854125,
|
| 1795 |
-
"step": 5080
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 2.1348285202098847,
|
| 1799 |
-
"grad_norm": 0.1849839687347412,
|
| 1800 |
-
"learning_rate": 1.6609542381395266e-05,
|
| 1801 |
-
"loss": 0.019530372321605684,
|
| 1802 |
-
"step": 5100
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 2.1432029624589455,
|
| 1806 |
-
"grad_norm": 0.19696128368377686,
|
| 1807 |
-
"learning_rate": 1.64328271761129e-05,
|
| 1808 |
-
"loss": 0.031582942605018614,
|
| 1809 |
-
"step": 5120
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 2.1515774047080067,
|
| 1813 |
-
"grad_norm": 0.575410783290863,
|
| 1814 |
-
"learning_rate": 1.625659566950236e-05,
|
| 1815 |
-
"loss": 0.06284295320510865,
|
| 1816 |
-
"step": 5140
|
| 1817 |
-
},
|
| 1818 |
-
{
|
| 1819 |
-
"epoch": 2.159951846957068,
|
| 1820 |
-
"grad_norm": 0.26818394660949707,
|
| 1821 |
-
"learning_rate": 1.6080857811514148e-05,
|
| 1822 |
-
"loss": 0.033028793334960935,
|
| 1823 |
-
"step": 5160
|
| 1824 |
-
},
|
| 1825 |
-
{
|
| 1826 |
-
"epoch": 2.1683262892061292,
|
| 1827 |
-
"grad_norm": 0.4050503671169281,
|
| 1828 |
-
"learning_rate": 1.5905623524227616e-05,
|
| 1829 |
-
"loss": 0.027020111680030823,
|
| 1830 |
-
"step": 5180
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 2.17670073145519,
|
| 1834 |
-
"grad_norm": 0.9115064144134521,
|
| 1835 |
-
"learning_rate": 1.573090270129072e-05,
|
| 1836 |
-
"loss": 0.037383803725242616,
|
| 1837 |
-
"step": 5200
|
| 1838 |
-
},
|
| 1839 |
-
{
|
| 1840 |
-
"epoch": 2.1850751737042513,
|
| 1841 |
-
"grad_norm": 0.055088263005018234,
|
| 1842 |
-
"learning_rate": 1.555670520736149e-05,
|
| 1843 |
-
"loss": 0.04163878262042999,
|
| 1844 |
-
"step": 5220
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 2.1934496159533126,
|
| 1848 |
-
"grad_norm": 0.08490361273288727,
|
| 1849 |
-
"learning_rate": 1.5383040877550997e-05,
|
| 1850 |
-
"loss": 0.027175077795982362,
|
| 1851 |
-
"step": 5240
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"epoch": 2.201824058202374,
|
| 1855 |
-
"grad_norm": 0.05155947431921959,
|
| 1856 |
-
"learning_rate": 1.5209919516868129e-05,
|
| 1857 |
-
"loss": 0.017401106655597687,
|
| 1858 |
-
"step": 5260
|
| 1859 |
-
},
|
| 1860 |
-
{
|
| 1861 |
-
"epoch": 2.2101985004514346,
|
| 1862 |
-
"grad_norm": 0.3288068473339081,
|
| 1863 |
-
"learning_rate": 1.5037350899665992e-05,
|
| 1864 |
-
"loss": 0.0302372545003891,
|
| 1865 |
-
"step": 5280
|
| 1866 |
-
},
|
| 1867 |
-
{
|
| 1868 |
-
"epoch": 2.218572942700496,
|
| 1869 |
-
"grad_norm": 0.16062849760055542,
|
| 1870 |
-
"learning_rate": 1.4865344769090034e-05,
|
| 1871 |
-
"loss": 0.039090073108673094,
|
| 1872 |
-
"step": 5300
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 2.226947384949557,
|
| 1876 |
-
"grad_norm": 0.39235061407089233,
|
| 1877 |
-
"learning_rate": 1.4693910836527956e-05,
|
| 1878 |
-
"loss": 0.04479800760746002,
|
| 1879 |
-
"step": 5320
|
| 1880 |
-
},
|
| 1881 |
-
{
|
| 1882 |
-
"epoch": 2.2353218271986184,
|
| 1883 |
-
"grad_norm": 0.08488383889198303,
|
| 1884 |
-
"learning_rate": 1.4523058781061454e-05,
|
| 1885 |
-
"loss": 0.03554574549198151,
|
| 1886 |
-
"step": 5340
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 2.243696269447679,
|
| 1890 |
-
"grad_norm": 0.1608286052942276,
|
| 1891 |
-
"learning_rate": 1.435279824891968e-05,
|
| 1892 |
-
"loss": 0.04044436812400818,
|
| 1893 |
-
"step": 5360
|
| 1894 |
-
},
|
| 1895 |
-
{
|
| 1896 |
-
"epoch": 2.2520707116967404,
|
| 1897 |
-
"grad_norm": 0.10187438875436783,
|
| 1898 |
-
"learning_rate": 1.4183138852934641e-05,
|
| 1899 |
-
"loss": 0.041433611512184144,
|
| 1900 |
-
"step": 5380
|
| 1901 |
-
},
|
| 1902 |
-
{
|
| 1903 |
-
"epoch": 2.2604451539458017,
|
| 1904 |
-
"grad_norm": 1.007249116897583,
|
| 1905 |
-
"learning_rate": 1.4014090171998511e-05,
|
| 1906 |
-
"loss": 0.03556722104549408,
|
| 1907 |
-
"step": 5400
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 2.268819596194863,
|
| 1911 |
-
"grad_norm": 0.1751636415719986,
|
| 1912 |
-
"learning_rate": 1.3845661750522733e-05,
|
| 1913 |
-
"loss": 0.02698090374469757,
|
| 1914 |
-
"step": 5420
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 2.2771940384439238,
|
| 1918 |
-
"grad_norm": 0.5745561122894287,
|
| 1919 |
-
"learning_rate": 1.3677863097899196e-05,
|
| 1920 |
-
"loss": 0.027302804589271545,
|
| 1921 |
-
"step": 5440
|
| 1922 |
-
},
|
| 1923 |
-
{
|
| 1924 |
-
"epoch": 2.285568480692985,
|
| 1925 |
-
"grad_norm": 0.18385080993175507,
|
| 1926 |
-
"learning_rate": 1.3510703687963317e-05,
|
| 1927 |
-
"loss": 0.036330342292785645,
|
| 1928 |
-
"step": 5460
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 2.2939429229420463,
|
| 1932 |
-
"grad_norm": 0.8466984629631042,
|
| 1933 |
-
"learning_rate": 1.334419295845919e-05,
|
| 1934 |
-
"loss": 0.04368994832038879,
|
| 1935 |
-
"step": 5480
|
| 1936 |
-
},
|
| 1937 |
-
{
|
| 1938 |
-
"epoch": 2.3023173651911075,
|
| 1939 |
-
"grad_norm": 0.5879298448562622,
|
| 1940 |
-
"learning_rate": 1.3178340310506681e-05,
|
| 1941 |
-
"loss": 0.0322884202003479,
|
| 1942 |
-
"step": 5500
|
| 1943 |
-
},
|
| 1944 |
-
{
|
| 1945 |
-
"epoch": 2.3106918074401683,
|
| 1946 |
-
"grad_norm": 0.1511489599943161,
|
| 1947 |
-
"learning_rate": 1.3013155108070685e-05,
|
| 1948 |
-
"loss": 0.05695425868034363,
|
| 1949 |
-
"step": 5520
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 2.3190662496892296,
|
| 1953 |
-
"grad_norm": 0.059570491313934326,
|
| 1954 |
-
"learning_rate": 1.2848646677432416e-05,
|
| 1955 |
-
"loss": 0.02256210893392563,
|
| 1956 |
-
"step": 5540
|
| 1957 |
-
},
|
| 1958 |
-
{
|
| 1959 |
-
"epoch": 2.327440691938291,
|
| 1960 |
-
"grad_norm": 0.0460665300488472,
|
| 1961 |
-
"learning_rate": 1.2684824306662871e-05,
|
| 1962 |
-
"loss": 0.016327956318855287,
|
| 1963 |
-
"step": 5560
|
| 1964 |
-
},
|
| 1965 |
-
{
|
| 1966 |
-
"epoch": 2.335815134187352,
|
| 1967 |
-
"grad_norm": 0.05761329084634781,
|
| 1968 |
-
"learning_rate": 1.2521697245098413e-05,
|
| 1969 |
-
"loss": 0.030600738525390626,
|
| 1970 |
-
"step": 5580
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 2.3441895764364133,
|
| 1974 |
-
"grad_norm": 0.07144748419523239,
|
| 1975 |
-
"learning_rate": 1.2359274702818564e-05,
|
| 1976 |
-
"loss": 0.0311413437128067,
|
| 1977 |
-
"step": 5600
|
| 1978 |
-
},
|
| 1979 |
-
{
|
| 1980 |
-
"epoch": 2.352564018685474,
|
| 1981 |
-
"grad_norm": 0.05444992706179619,
|
| 1982 |
-
"learning_rate": 1.2197565850126023e-05,
|
| 1983 |
-
"loss": 0.026527315378189087,
|
| 1984 |
-
"step": 5620
|
| 1985 |
-
},
|
| 1986 |
-
{
|
| 1987 |
-
"epoch": 2.3609384609345354,
|
| 1988 |
-
"grad_norm": 0.39611950516700745,
|
| 1989 |
-
"learning_rate": 1.2036579817028884e-05,
|
| 1990 |
-
"loss": 0.035798245668411256,
|
| 1991 |
-
"step": 5640
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 2.3693129031835967,
|
| 1995 |
-
"grad_norm": 0.03898654505610466,
|
| 1996 |
-
"learning_rate": 1.1876325692725171e-05,
|
| 1997 |
-
"loss": 0.03876761496067047,
|
| 1998 |
-
"step": 5660
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 2.3776873454326575,
|
| 2002 |
-
"grad_norm": 0.06425919383764267,
|
| 2003 |
-
"learning_rate": 1.1716812525089702e-05,
|
| 2004 |
-
"loss": 0.025335514545440675,
|
| 2005 |
-
"step": 5680
|
| 2006 |
-
},
|
| 2007 |
-
{
|
| 2008 |
-
"epoch": 2.3860617876817187,
|
| 2009 |
-
"grad_norm": 0.07012301683425903,
|
| 2010 |
-
"learning_rate": 1.1558049320163186e-05,
|
| 2011 |
-
"loss": 0.0288514643907547,
|
| 2012 |
-
"step": 5700
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 2.39443622993078,
|
| 2016 |
-
"grad_norm": 0.12998658418655396,
|
| 2017 |
-
"learning_rate": 1.1400045041643795e-05,
|
| 2018 |
-
"loss": 0.04393172860145569,
|
| 2019 |
-
"step": 5720
|
| 2020 |
-
},
|
| 2021 |
-
{
|
| 2022 |
-
"epoch": 2.4028106721798412,
|
| 2023 |
-
"grad_norm": 0.4555356204509735,
|
| 2024 |
-
"learning_rate": 1.1242808610381051e-05,
|
| 2025 |
-
"loss": 0.034080561995506284,
|
| 2026 |
-
"step": 5740
|
| 2027 |
-
},
|
| 2028 |
-
{
|
| 2029 |
-
"epoch": 2.4111851144289025,
|
| 2030 |
-
"grad_norm": 0.6439165472984314,
|
| 2031 |
-
"learning_rate": 1.108634890387219e-05,
|
| 2032 |
-
"loss": 0.019711557030677795,
|
| 2033 |
-
"step": 5760
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 2.4195595566779633,
|
| 2037 |
-
"grad_norm": 0.04977191612124443,
|
| 2038 |
-
"learning_rate": 1.0930674755760908e-05,
|
| 2039 |
-
"loss": 0.044436204433441165,
|
| 2040 |
-
"step": 5780
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 2.4279339989270246,
|
| 2044 |
-
"grad_norm": 0.04331052675843239,
|
| 2045 |
-
"learning_rate": 1.0775794955338625e-05,
|
| 2046 |
-
"loss": 0.038460010290145875,
|
| 2047 |
-
"step": 5800
|
| 2048 |
-
},
|
| 2049 |
-
{
|
| 2050 |
-
"epoch": 2.436308441176086,
|
| 2051 |
-
"grad_norm": 0.5490692853927612,
|
| 2052 |
-
"learning_rate": 1.0621718247048278e-05,
|
| 2053 |
-
"loss": 0.030916082859039306,
|
| 2054 |
-
"step": 5820
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 2.444682883425147,
|
| 2058 |
-
"grad_norm": 0.17242658138275146,
|
| 2059 |
-
"learning_rate": 1.0468453329990563e-05,
|
| 2060 |
-
"loss": 0.025243097543716432,
|
| 2061 |
-
"step": 5840
|
| 2062 |
-
},
|
| 2063 |
-
{
|
| 2064 |
-
"epoch": 2.453057325674208,
|
| 2065 |
-
"grad_norm": 0.3855954706668854,
|
| 2066 |
-
"learning_rate": 1.0316008857432827e-05,
|
| 2067 |
-
"loss": 0.030227524042129517,
|
| 2068 |
-
"step": 5860
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 2.461431767923269,
|
| 2072 |
-
"grad_norm": 0.564544677734375,
|
| 2073 |
-
"learning_rate": 1.0164393436320477e-05,
|
| 2074 |
-
"loss": 0.03143407106399536,
|
| 2075 |
-
"step": 5880
|
| 2076 |
-
},
|
| 2077 |
-
{
|
| 2078 |
-
"epoch": 2.4698062101723304,
|
| 2079 |
-
"grad_norm": 0.31516560912132263,
|
| 2080 |
-
"learning_rate": 1.0013615626791087e-05,
|
| 2081 |
-
"loss": 0.035627859830856326,
|
| 2082 |
-
"step": 5900
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 2.4781806524213916,
|
| 2086 |
-
"grad_norm": 0.06647393107414246,
|
| 2087 |
-
"learning_rate": 9.863683941691033e-06,
|
| 2088 |
-
"loss": 0.03705155551433563,
|
| 2089 |
-
"step": 5920
|
| 2090 |
-
},
|
| 2091 |
-
{
|
| 2092 |
-
"epoch": 2.4865550946704524,
|
| 2093 |
-
"grad_norm": 0.2980024814605713,
|
| 2094 |
-
"learning_rate": 9.714606846094892e-06,
|
| 2095 |
-
"loss": 0.05262235403060913,
|
| 2096 |
-
"step": 5940
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 2.4949295369195137,
|
| 2100 |
-
"grad_norm": 0.05878954753279686,
|
| 2101 |
-
"learning_rate": 9.566392756827533e-06,
|
| 2102 |
-
"loss": 0.019335083663463593,
|
| 2103 |
-
"step": 5960
|
| 2104 |
-
},
|
| 2105 |
-
{
|
| 2106 |
-
"epoch": 2.503303979168575,
|
| 2107 |
-
"grad_norm": 0.5564544796943665,
|
| 2108 |
-
"learning_rate": 9.419050041988859e-06,
|
| 2109 |
-
"loss": 0.028229591250419617,
|
| 2110 |
-
"step": 5980
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 2.511678421417636,
|
| 2114 |
-
"grad_norm": 0.08901913464069366,
|
| 2115 |
-
"learning_rate": 9.272587020481371e-06,
|
| 2116 |
-
"loss": 0.03342890441417694,
|
| 2117 |
-
"step": 6000
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 2.520052863666697,
|
| 2121 |
-
"grad_norm": 0.8243499994277954,
|
| 2122 |
-
"learning_rate": 9.127011961540491e-06,
|
| 2123 |
-
"loss": 0.021094033122062684,
|
| 2124 |
-
"step": 6020
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 2.5284273059157583,
|
| 2128 |
-
"grad_norm": 0.17781858146190643,
|
| 2129 |
-
"learning_rate": 8.982333084267683e-06,
|
| 2130 |
-
"loss": 0.029926052689552306,
|
| 2131 |
-
"step": 6040
|
| 2132 |
-
},
|
| 2133 |
-
{
|
| 2134 |
-
"epoch": 2.5368017481648195,
|
| 2135 |
-
"grad_norm": 0.04101826995611191,
|
| 2136 |
-
"learning_rate": 8.838558557166391e-06,
|
| 2137 |
-
"loss": 0.04929870367050171,
|
| 2138 |
-
"step": 6060
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 2.5451761904138808,
|
| 2142 |
-
"grad_norm": 0.310251384973526,
|
| 2143 |
-
"learning_rate": 8.695696497680864e-06,
|
| 2144 |
-
"loss": 0.027747470140457153,
|
| 2145 |
-
"step": 6080
|
| 2146 |
-
},
|
| 2147 |
-
{
|
| 2148 |
-
"epoch": 2.553550632662942,
|
| 2149 |
-
"grad_norm": 0.724092960357666,
|
| 2150 |
-
"learning_rate": 8.553754971737864e-06,
|
| 2151 |
-
"loss": 0.03592235445976257,
|
| 2152 |
-
"step": 6100
|
| 2153 |
-
},
|
| 2154 |
-
{
|
| 2155 |
-
"epoch": 2.561925074912003,
|
| 2156 |
-
"grad_norm": 0.21852202713489532,
|
| 2157 |
-
"learning_rate": 8.412741993291226e-06,
|
| 2158 |
-
"loss": 0.03389221727848053,
|
| 2159 |
-
"step": 6120
|
| 2160 |
-
},
|
| 2161 |
-
{
|
| 2162 |
-
"epoch": 2.570299517161064,
|
| 2163 |
-
"grad_norm": 0.3016412854194641,
|
| 2164 |
-
"learning_rate": 8.272665523869416e-06,
|
| 2165 |
-
"loss": 0.04466528296470642,
|
| 2166 |
-
"step": 6140
|
| 2167 |
-
},
|
| 2168 |
-
{
|
| 2169 |
-
"epoch": 2.5786739594101253,
|
| 2170 |
-
"grad_norm": 0.3840382695198059,
|
| 2171 |
-
"learning_rate": 8.13353347212605e-06,
|
| 2172 |
-
"loss": 0.036823111772537234,
|
| 2173 |
-
"step": 6160
|
| 2174 |
-
},
|
| 2175 |
-
{
|
| 2176 |
-
"epoch": 2.587048401659186,
|
| 2177 |
-
"grad_norm": 0.052707478404045105,
|
| 2178 |
-
"learning_rate": 7.995353693393337e-06,
|
| 2179 |
-
"loss": 0.030457329750061036,
|
| 2180 |
-
"step": 6180
|
| 2181 |
-
},
|
| 2182 |
-
{
|
| 2183 |
-
"epoch": 2.5954228439082474,
|
| 2184 |
-
"grad_norm": 0.861706554889679,
|
| 2185 |
-
"learning_rate": 7.85813398923858e-06,
|
| 2186 |
-
"loss": 0.0386508047580719,
|
| 2187 |
-
"step": 6200
|
| 2188 |
-
},
|
| 2189 |
-
{
|
| 2190 |
-
"epoch": 2.6037972861573087,
|
| 2191 |
-
"grad_norm": 0.036389291286468506,
|
| 2192 |
-
"learning_rate": 7.721882107023711e-06,
|
| 2193 |
-
"loss": 0.022716432809829712,
|
| 2194 |
-
"step": 6220
|
| 2195 |
-
},
|
| 2196 |
-
{
|
| 2197 |
-
"epoch": 2.61217172840637,
|
| 2198 |
-
"grad_norm": 0.5068594813346863,
|
| 2199 |
-
"learning_rate": 7.5866057394679e-06,
|
| 2200 |
-
"loss": 0.037295091152191165,
|
| 2201 |
-
"step": 6240
|
| 2202 |
-
},
|
| 2203 |
-
{
|
| 2204 |
-
"epoch": 2.620546170655431,
|
| 2205 |
-
"grad_norm": 0.03255036100745201,
|
| 2206 |
-
"learning_rate": 7.452312524213173e-06,
|
| 2207 |
-
"loss": 0.023309530317783357,
|
| 2208 |
-
"step": 6260
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 2.628920612904492,
|
| 2212 |
-
"grad_norm": 0.37486711144447327,
|
| 2213 |
-
"learning_rate": 7.319010043393235e-06,
|
| 2214 |
-
"loss": 0.02240334004163742,
|
| 2215 |
-
"step": 6280
|
| 2216 |
-
},
|
| 2217 |
-
{
|
| 2218 |
-
"epoch": 2.6372950551535532,
|
| 2219 |
-
"grad_norm": 0.17838098108768463,
|
| 2220 |
-
"learning_rate": 7.186705823205394e-06,
|
| 2221 |
-
"loss": 0.021342363953590394,
|
| 2222 |
-
"step": 6300
|
| 2223 |
-
},
|
| 2224 |
-
{
|
| 2225 |
-
"epoch": 2.6456694974026145,
|
| 2226 |
-
"grad_norm": 0.0714629665017128,
|
| 2227 |
-
"learning_rate": 7.055407333485594e-06,
|
| 2228 |
-
"loss": 0.027359399199485778,
|
| 2229 |
-
"step": 6320
|
| 2230 |
-
},
|
| 2231 |
-
{
|
| 2232 |
-
"epoch": 2.6540439396516753,
|
| 2233 |
-
"grad_norm": 0.49168136715888977,
|
| 2234 |
-
"learning_rate": 6.925121987286709e-06,
|
| 2235 |
-
"loss": 0.12207809686660767,
|
| 2236 |
-
"step": 6340
|
| 2237 |
-
},
|
| 2238 |
-
{
|
| 2239 |
-
"epoch": 2.6624183819007365,
|
| 2240 |
-
"grad_norm": 0.13302330672740936,
|
| 2241 |
-
"learning_rate": 6.7958571404599766e-06,
|
| 2242 |
-
"loss": 0.02340977340936661,
|
| 2243 |
-
"step": 6360
|
| 2244 |
-
},
|
| 2245 |
-
{
|
| 2246 |
-
"epoch": 2.670792824149798,
|
| 2247 |
-
"grad_norm": 0.22196047008037567,
|
| 2248 |
-
"learning_rate": 6.667620091239732e-06,
|
| 2249 |
-
"loss": 0.02150045782327652,
|
| 2250 |
-
"step": 6380
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 2.679167266398859,
|
| 2254 |
-
"grad_norm": 0.05339694768190384,
|
| 2255 |
-
"learning_rate": 6.540418079831301e-06,
|
| 2256 |
-
"loss": 0.04855234920978546,
|
| 2257 |
-
"step": 6400
|
| 2258 |
-
},
|
| 2259 |
-
{
|
| 2260 |
-
"epoch": 2.6875417086479203,
|
| 2261 |
-
"grad_norm": 0.04468238353729248,
|
| 2262 |
-
"learning_rate": 6.414258288002259e-06,
|
| 2263 |
-
"loss": 0.01810908317565918,
|
| 2264 |
-
"step": 6420
|
| 2265 |
-
},
|
| 2266 |
-
{
|
| 2267 |
-
"epoch": 2.695916150896981,
|
| 2268 |
-
"grad_norm": 0.17243818938732147,
|
| 2269 |
-
"learning_rate": 6.2891478386769315e-06,
|
| 2270 |
-
"loss": 0.02477678209543228,
|
| 2271 |
-
"step": 6440
|
| 2272 |
-
},
|
| 2273 |
-
{
|
| 2274 |
-
"epoch": 2.7042905931460424,
|
| 2275 |
-
"grad_norm": 0.11534826457500458,
|
| 2276 |
-
"learning_rate": 6.165093795534254e-06,
|
| 2277 |
-
"loss": 0.026881667971611022,
|
| 2278 |
-
"step": 6460
|
| 2279 |
-
},
|
| 2280 |
-
{
|
| 2281 |
-
"epoch": 2.7126650353951036,
|
| 2282 |
-
"grad_norm": 0.36479777097702026,
|
| 2283 |
-
"learning_rate": 6.042103162608937e-06,
|
| 2284 |
-
"loss": 0.021916355192661285,
|
| 2285 |
-
"step": 6480
|
| 2286 |
-
},
|
| 2287 |
-
{
|
| 2288 |
-
"epoch": 2.721039477644165,
|
| 2289 |
-
"grad_norm": 0.3529409170150757,
|
| 2290 |
-
"learning_rate": 5.920182883896053e-06,
|
| 2291 |
-
"loss": 0.022671876847743987,
|
| 2292 |
-
"step": 6500
|
| 2293 |
-
},
|
| 2294 |
-
{
|
| 2295 |
-
"epoch": 2.7294139198932257,
|
| 2296 |
-
"grad_norm": 0.24025386571884155,
|
| 2297 |
-
"learning_rate": 5.7993398429589506e-06,
|
| 2298 |
-
"loss": 0.05065078735351562,
|
| 2299 |
-
"step": 6520
|
| 2300 |
-
},
|
| 2301 |
-
{
|
| 2302 |
-
"epoch": 2.737788362142287,
|
| 2303 |
-
"grad_norm": 0.13825002312660217,
|
| 2304 |
-
"learning_rate": 5.679580862540629e-06,
|
| 2305 |
-
"loss": 0.06285998821258545,
|
| 2306 |
-
"step": 6540
|
| 2307 |
-
},
|
| 2308 |
-
{
|
| 2309 |
-
"epoch": 2.746162804391348,
|
| 2310 |
-
"grad_norm": 0.09770816564559937,
|
| 2311 |
-
"learning_rate": 5.560912704178514e-06,
|
| 2312 |
-
"loss": 0.1244073748588562,
|
| 2313 |
-
"step": 6560
|
| 2314 |
-
},
|
| 2315 |
-
{
|
| 2316 |
-
"epoch": 2.7545372466404094,
|
| 2317 |
-
"grad_norm": 0.3843393623828888,
|
| 2318 |
-
"learning_rate": 5.443342067822743e-06,
|
| 2319 |
-
"loss": 0.018991626799106598,
|
| 2320 |
-
"step": 6580
|
| 2321 |
-
},
|
| 2322 |
-
{
|
| 2323 |
-
"epoch": 2.7629116888894707,
|
| 2324 |
-
"grad_norm": 0.34221217036247253,
|
| 2325 |
-
"learning_rate": 5.326875591457833e-06,
|
| 2326 |
-
"loss": 0.025901442766189574,
|
| 2327 |
-
"step": 6600
|
| 2328 |
-
},
|
| 2329 |
-
{
|
| 2330 |
-
"epoch": 2.7712861311385315,
|
| 2331 |
-
"grad_norm": 0.07205979526042938,
|
| 2332 |
-
"learning_rate": 5.2115198507279365e-06,
|
| 2333 |
-
"loss": 0.02461903393268585,
|
| 2334 |
-
"step": 6620
|
| 2335 |
-
},
|
| 2336 |
-
{
|
| 2337 |
-
"epoch": 2.7796605733875928,
|
| 2338 |
-
"grad_norm": 0.16574513912200928,
|
| 2339 |
-
"learning_rate": 5.097281358565595e-06,
|
| 2340 |
-
"loss": 0.030345502495765685,
|
| 2341 |
-
"step": 6640
|
| 2342 |
-
},
|
| 2343 |
-
{
|
| 2344 |
-
"epoch": 2.788035015636654,
|
| 2345 |
-
"grad_norm": 0.6843248605728149,
|
| 2346 |
-
"learning_rate": 4.984166564823992e-06,
|
| 2347 |
-
"loss": 0.03047039806842804,
|
| 2348 |
-
"step": 6660
|
| 2349 |
-
},
|
| 2350 |
-
{
|
| 2351 |
-
"epoch": 2.796409457885715,
|
| 2352 |
-
"grad_norm": 0.08679454028606415,
|
| 2353 |
-
"learning_rate": 4.872181855912811e-06,
|
| 2354 |
-
"loss": 0.03248770534992218,
|
| 2355 |
-
"step": 6680
|
| 2356 |
-
},
|
| 2357 |
-
{
|
| 2358 |
-
"epoch": 2.804783900134776,
|
| 2359 |
-
"grad_norm": 0.038589492440223694,
|
| 2360 |
-
"learning_rate": 4.761333554437653e-06,
|
| 2361 |
-
"loss": 0.026063334941864014,
|
| 2362 |
-
"step": 6700
|
| 2363 |
-
},
|
| 2364 |
-
{
|
| 2365 |
-
"epoch": 2.8131583423838373,
|
| 2366 |
-
"grad_norm": 0.6347565650939941,
|
| 2367 |
-
"learning_rate": 4.651627918843099e-06,
|
| 2368 |
-
"loss": 0.02364843189716339,
|
| 2369 |
-
"step": 6720
|
| 2370 |
-
},
|
| 2371 |
-
{
|
| 2372 |
-
"epoch": 2.8215327846328986,
|
| 2373 |
-
"grad_norm": 0.03580786660313606,
|
| 2374 |
-
"learning_rate": 4.543071143059311e-06,
|
| 2375 |
-
"loss": 0.02118242233991623,
|
| 2376 |
-
"step": 6740
|
| 2377 |
-
},
|
| 2378 |
-
{
|
| 2379 |
-
"epoch": 2.82990722688196,
|
| 2380 |
-
"grad_norm": 0.6818614602088928,
|
| 2381 |
-
"learning_rate": 4.435669356152352e-06,
|
| 2382 |
-
"loss": 0.026198378205299376,
|
| 2383 |
-
"step": 6760
|
| 2384 |
-
},
|
| 2385 |
-
{
|
| 2386 |
-
"epoch": 2.8382816691310206,
|
| 2387 |
-
"grad_norm": 0.18815983831882477,
|
| 2388 |
-
"learning_rate": 4.3294286219781595e-06,
|
| 2389 |
-
"loss": 0.029299169778823853,
|
| 2390 |
-
"step": 6780
|
| 2391 |
-
},
|
| 2392 |
-
{
|
| 2393 |
-
"epoch": 2.846656111380082,
|
| 2394 |
-
"grad_norm": 0.49268460273742676,
|
| 2395 |
-
"learning_rate": 4.224354938840144e-06,
|
| 2396 |
-
"loss": 0.03095126748085022,
|
| 2397 |
-
"step": 6800
|
| 2398 |
-
},
|
| 2399 |
-
{
|
| 2400 |
-
"epoch": 2.855030553629143,
|
| 2401 |
-
"grad_norm": 0.15540994703769684,
|
| 2402 |
-
"learning_rate": 4.120454239150548e-06,
|
| 2403 |
-
"loss": 0.024085646867752074,
|
| 2404 |
-
"step": 6820
|
| 2405 |
-
},
|
| 2406 |
-
{
|
| 2407 |
-
"epoch": 2.863404995878204,
|
| 2408 |
-
"grad_norm": 0.03290439397096634,
|
| 2409 |
-
"learning_rate": 4.017732389095511e-06,
|
| 2410 |
-
"loss": 0.02222718745470047,
|
| 2411 |
-
"step": 6840
|
| 2412 |
-
},
|
| 2413 |
-
{
|
| 2414 |
-
"epoch": 2.871779438127265,
|
| 2415 |
-
"grad_norm": 0.25523442029953003,
|
| 2416 |
-
"learning_rate": 3.916195188303853e-06,
|
| 2417 |
-
"loss": 0.03877650797367096,
|
| 2418 |
-
"step": 6860
|
| 2419 |
-
},
|
| 2420 |
-
{
|
| 2421 |
-
"epoch": 2.8801538803763265,
|
| 2422 |
-
"grad_norm": 0.06997464597225189,
|
| 2423 |
-
"learning_rate": 3.81584836951964e-06,
|
| 2424 |
-
"loss": 0.04235384464263916,
|
| 2425 |
-
"step": 6880
|
| 2426 |
-
},
|
| 2427 |
-
{
|
| 2428 |
-
"epoch": 2.8885283226253877,
|
| 2429 |
-
"grad_norm": 0.13868676126003265,
|
| 2430 |
-
"learning_rate": 3.716697598278507e-06,
|
| 2431 |
-
"loss": 0.031541892886161806,
|
| 2432 |
-
"step": 6900
|
| 2433 |
-
},
|
| 2434 |
-
{
|
| 2435 |
-
"epoch": 2.896902764874449,
|
| 2436 |
-
"grad_norm": 0.023875195533037186,
|
| 2437 |
-
"learning_rate": 3.6187484725878047e-06,
|
| 2438 |
-
"loss": 0.01966340243816376,
|
| 2439 |
-
"step": 6920
|
| 2440 |
-
},
|
| 2441 |
-
{
|
| 2442 |
-
"epoch": 2.90527720712351,
|
| 2443 |
-
"grad_norm": 0.06577973067760468,
|
| 2444 |
-
"learning_rate": 3.5220065226105105e-06,
|
| 2445 |
-
"loss": 0.02548326551914215,
|
| 2446 |
-
"step": 6940
|
| 2447 |
-
},
|
| 2448 |
-
{
|
| 2449 |
-
"epoch": 2.913651649372571,
|
| 2450 |
-
"grad_norm": 0.11466669291257858,
|
| 2451 |
-
"learning_rate": 3.426477210353013e-06,
|
| 2452 |
-
"loss": 0.026891687512397768,
|
| 2453 |
-
"step": 6960
|
| 2454 |
-
},
|
| 2455 |
-
{
|
| 2456 |
-
"epoch": 2.9220260916216323,
|
| 2457 |
-
"grad_norm": 0.11031544208526611,
|
| 2458 |
-
"learning_rate": 3.332165929356726e-06,
|
| 2459 |
-
"loss": 0.020203272998332977,
|
| 2460 |
-
"step": 6980
|
| 2461 |
-
},
|
| 2462 |
-
{
|
| 2463 |
-
"epoch": 2.930400533870693,
|
| 2464 |
-
"grad_norm": 0.6550182700157166,
|
| 2465 |
-
"learning_rate": 3.2390780043935864e-06,
|
| 2466 |
-
"loss": 0.04080580472946167,
|
| 2467 |
-
"step": 7000
|
| 2468 |
-
},
|
| 2469 |
-
{
|
| 2470 |
-
"epoch": 2.9387749761197544,
|
| 2471 |
-
"grad_norm": 0.06129005551338196,
|
| 2472 |
-
"learning_rate": 3.147218691165396e-06,
|
| 2473 |
-
"loss": 0.01906572878360748,
|
| 2474 |
-
"step": 7020
|
| 2475 |
-
},
|
| 2476 |
-
{
|
| 2477 |
-
"epoch": 2.9471494183688156,
|
| 2478 |
-
"grad_norm": 0.37061557173728943,
|
| 2479 |
-
"learning_rate": 3.0565931760070917e-06,
|
| 2480 |
-
"loss": 0.02230120003223419,
|
| 2481 |
-
"step": 7040
|
| 2482 |
-
},
|
| 2483 |
-
{
|
| 2484 |
-
"epoch": 2.955523860617877,
|
| 2485 |
-
"grad_norm": 0.11101196706295013,
|
| 2486 |
-
"learning_rate": 2.967206575593956e-06,
|
| 2487 |
-
"loss": 0.0184899240732193,
|
| 2488 |
-
"step": 7060
|
| 2489 |
-
},
|
| 2490 |
-
{
|
| 2491 |
-
"epoch": 2.963898302866938,
|
| 2492 |
-
"grad_norm": 0.06029270961880684,
|
| 2493 |
-
"learning_rate": 2.87906393665269e-06,
|
| 2494 |
-
"loss": 0.030630800127983093,
|
| 2495 |
-
"step": 7080
|
| 2496 |
-
},
|
| 2497 |
-
{
|
| 2498 |
-
"epoch": 2.9722727451159994,
|
| 2499 |
-
"grad_norm": 0.1466781049966812,
|
| 2500 |
-
"learning_rate": 2.792170235676497e-06,
|
| 2501 |
-
"loss": 0.030768883228302003,
|
| 2502 |
-
"step": 7100
|
| 2503 |
-
},
|
| 2504 |
-
{
|
| 2505 |
-
"epoch": 2.98064718736506,
|
| 2506 |
-
"grad_norm": 0.6590607762336731,
|
| 2507 |
-
"learning_rate": 2.7065303786441314e-06,
|
| 2508 |
-
"loss": 0.025333571434020995,
|
| 2509 |
-
"step": 7120
|
| 2510 |
-
},
|
| 2511 |
-
{
|
| 2512 |
-
"epoch": 2.9890216296141214,
|
| 2513 |
-
"grad_norm": 0.04609564691781998,
|
| 2514 |
-
"learning_rate": 2.622149200742871e-06,
|
| 2515 |
-
"loss": 0.028621068596839903,
|
| 2516 |
-
"step": 7140
|
| 2517 |
-
},
|
| 2518 |
-
{
|
| 2519 |
-
"epoch": 2.9973960718631827,
|
| 2520 |
-
"grad_norm": 0.05696053057909012,
|
| 2521 |
-
"learning_rate": 2.539031466095551e-06,
|
| 2522 |
-
"loss": 0.030104705691337587,
|
| 2523 |
-
"step": 7160
|
| 2524 |
-
},
|
| 2525 |
-
{
|
| 2526 |
-
"epoch": 3.0054433874618898,
|
| 2527 |
-
"grad_norm": 0.08858000487089157,
|
| 2528 |
-
"learning_rate": 2.457181867491573e-06,
|
| 2529 |
-
"loss": 0.014444483816623688,
|
| 2530 |
-
"step": 7180
|
| 2531 |
-
},
|
| 2532 |
-
{
|
| 2533 |
-
"epoch": 3.013817829710951,
|
| 2534 |
-
"grad_norm": 0.8151156902313232,
|
| 2535 |
-
"learning_rate": 2.376605026121964e-06,
|
| 2536 |
-
"loss": 0.030318111181259155,
|
| 2537 |
-
"step": 7200
|
| 2538 |
-
},
|
| 2539 |
-
{
|
| 2540 |
-
"epoch": 3.022192271960012,
|
| 2541 |
-
"grad_norm": 0.08565035462379456,
|
| 2542 |
-
"learning_rate": 2.297305491318447e-06,
|
| 2543 |
-
"loss": 0.01848241835832596,
|
| 2544 |
-
"step": 7220
|
| 2545 |
-
},
|
| 2546 |
-
{
|
| 2547 |
-
"epoch": 3.030566714209073,
|
| 2548 |
-
"grad_norm": 0.02643604762852192,
|
| 2549 |
-
"learning_rate": 2.219287740296605e-06,
|
| 2550 |
-
"loss": 0.023431843519210814,
|
| 2551 |
-
"step": 7240
|
| 2552 |
-
},
|
| 2553 |
-
{
|
| 2554 |
-
"epoch": 3.0389411564581343,
|
| 2555 |
-
"grad_norm": 0.038074638694524765,
|
| 2556 |
-
"learning_rate": 2.142556177903096e-06,
|
| 2557 |
-
"loss": 0.02755042314529419,
|
| 2558 |
-
"step": 7260
|
| 2559 |
-
},
|
| 2560 |
-
{
|
| 2561 |
-
"epoch": 3.0473155987071956,
|
| 2562 |
-
"grad_norm": 0.03524810075759888,
|
| 2563 |
-
"learning_rate": 2.067115136366948e-06,
|
| 2564 |
-
"loss": 0.03158809542655945,
|
| 2565 |
-
"step": 7280
|
| 2566 |
-
},
|
| 2567 |
-
{
|
| 2568 |
-
"epoch": 3.0556900409562564,
|
| 2569 |
-
"grad_norm": 0.0943976640701294,
|
| 2570 |
-
"learning_rate": 1.9929688750549734e-06,
|
| 2571 |
-
"loss": 0.016621415317058564,
|
| 2572 |
-
"step": 7300
|
| 2573 |
-
},
|
| 2574 |
-
{
|
| 2575 |
-
"epoch": 3.0640644832053177,
|
| 2576 |
-
"grad_norm": 0.026402320712804794,
|
| 2577 |
-
"learning_rate": 1.9201215802312794e-06,
|
| 2578 |
-
"loss": 0.029634922742843628,
|
| 2579 |
-
"step": 7320
|
| 2580 |
-
},
|
| 2581 |
-
{
|
| 2582 |
-
"epoch": 2.3046213333071632,
|
| 2583 |
-
"grad_norm": 20.23296546936035,
|
| 2584 |
-
"learning_rate": 1.8485773648209192e-06,
|
| 2585 |
-
"loss": 15.542088558799342,
|
| 2586 |
-
"step": 7340
|
| 2587 |
-
},
|
| 2588 |
-
{
|
| 2589 |
-
"epoch": 2.3109021855403005,
|
| 2590 |
-
"grad_norm": 23.0955867767334,
|
| 2591 |
-
"learning_rate": 1.7783402681776833e-06,
|
| 2592 |
-
"loss": 15.532046508789062,
|
| 2593 |
-
"step": 7360
|
| 2594 |
-
},
|
| 2595 |
-
{
|
| 2596 |
-
"epoch": 2.3171830377734377,
|
| 2597 |
-
"grad_norm": 13.154098510742188,
|
| 2598 |
-
"learning_rate": 1.7094142558560177e-06,
|
| 2599 |
-
"loss": 13.967564392089844,
|
| 2600 |
-
"step": 7380
|
| 2601 |
-
},
|
| 2602 |
-
{
|
| 2603 |
-
"epoch": 2.3234638900065754,
|
| 2604 |
-
"grad_norm": 18.517717361450195,
|
| 2605 |
-
"learning_rate": 1.6418032193871612e-06,
|
| 2606 |
-
"loss": 14.239532470703125,
|
| 2607 |
-
"step": 7400
|
| 2608 |
-
},
|
| 2609 |
-
{
|
| 2610 |
-
"epoch": 2.3297447422397126,
|
| 2611 |
-
"grad_norm": 19.73468780517578,
|
| 2612 |
-
"learning_rate": 1.5755109760594105e-06,
|
| 2613 |
-
"loss": 14.224758911132813,
|
| 2614 |
-
"step": 7420
|
| 2615 |
-
},
|
| 2616 |
-
{
|
| 2617 |
-
"epoch": 2.33602559447285,
|
| 2618 |
-
"grad_norm": 23.925220489501953,
|
| 2619 |
-
"learning_rate": 1.510541268702595e-06,
|
| 2620 |
-
"loss": 17.3900634765625,
|
| 2621 |
-
"step": 7440
|
| 2622 |
-
},
|
| 2623 |
-
{
|
| 2624 |
-
"epoch": 2.3423064467059875,
|
| 2625 |
-
"grad_norm": 17.185338973999023,
|
| 2626 |
-
"learning_rate": 1.4468977654767736e-06,
|
| 2627 |
-
"loss": 13.145829772949218,
|
| 2628 |
-
"step": 7460
|
| 2629 |
-
},
|
| 2630 |
-
{
|
| 2631 |
-
"epoch": 2.3485872989391248,
|
| 2632 |
-
"grad_norm": 13.490628242492676,
|
| 2633 |
-
"learning_rate": 1.3845840596651256e-06,
|
| 2634 |
-
"loss": 11.708284759521485,
|
| 2635 |
-
"step": 7480
|
| 2636 |
-
},
|
| 2637 |
-
{
|
| 2638 |
-
"epoch": 2.354868151172262,
|
| 2639 |
-
"grad_norm": 17.3670654296875,
|
| 2640 |
-
"learning_rate": 1.3236036694710712e-06,
|
| 2641 |
-
"loss": 12.966706848144531,
|
| 2642 |
-
"step": 7500
|
| 2643 |
-
},
|
| 2644 |
-
{
|
| 2645 |
-
"epoch": 2.3611490034053997,
|
| 2646 |
-
"grad_norm": 19.007699966430664,
|
| 2647 |
-
"learning_rate": 1.2639600378196336e-06,
|
| 2648 |
-
"loss": 12.327762603759766,
|
| 2649 |
-
"step": 7520
|
| 2650 |
-
},
|
| 2651 |
-
{
|
| 2652 |
-
"epoch": 2.367429855638537,
|
| 2653 |
-
"grad_norm": 15.086852073669434,
|
| 2654 |
-
"learning_rate": 1.2056565321630742e-06,
|
| 2655 |
-
"loss": 11.243121337890624,
|
| 2656 |
-
"step": 7540
|
| 2657 |
-
},
|
| 2658 |
-
{
|
| 2659 |
-
"epoch": 2.373710707871674,
|
| 2660 |
-
"grad_norm": 14.045313835144043,
|
| 2661 |
-
"learning_rate": 1.1486964442907338e-06,
|
| 2662 |
-
"loss": 12.811164855957031,
|
| 2663 |
-
"step": 7560
|
| 2664 |
-
},
|
| 2665 |
-
{
|
| 2666 |
-
"epoch": 2.379991560104812,
|
| 2667 |
-
"grad_norm": 22.401371002197266,
|
| 2668 |
-
"learning_rate": 1.093082990143207e-06,
|
| 2669 |
-
"loss": 11.368438720703125,
|
| 2670 |
-
"step": 7580
|
| 2671 |
-
},
|
| 2672 |
-
{
|
| 2673 |
-
"epoch": 2.386272412337949,
|
| 2674 |
-
"grad_norm": 16.42538833618164,
|
| 2675 |
-
"learning_rate": 1.0388193096307619e-06,
|
| 2676 |
-
"loss": 11.5912353515625,
|
| 2677 |
-
"step": 7600
|
| 2678 |
-
},
|
| 2679 |
-
{
|
| 2680 |
-
"epoch": 2.3925532645710863,
|
| 2681 |
-
"grad_norm": 17.476543426513672,
|
| 2682 |
-
"learning_rate": 9.85908466456062e-07,
|
| 2683 |
-
"loss": 11.786304473876953,
|
| 2684 |
-
"step": 7620
|
| 2685 |
-
},
|
| 2686 |
-
{
|
| 2687 |
-
"epoch": 2.398834116804224,
|
| 2688 |
-
"grad_norm": 19.577699661254883,
|
| 2689 |
-
"learning_rate": 9.343534479411886e-07,
|
| 2690 |
-
"loss": 12.563404083251953,
|
| 2691 |
-
"step": 7640
|
| 2692 |
-
},
|
| 2693 |
-
{
|
| 2694 |
-
"epoch": 2.405114969037361,
|
| 2695 |
-
"grad_norm": 17.37641143798828,
|
| 2696 |
-
"learning_rate": 8.841571648589814e-07,
|
| 2697 |
-
"loss": 12.625107574462891,
|
| 2698 |
-
"step": 7660
|
| 2699 |
-
},
|
| 2700 |
-
{
|
| 2701 |
-
"epoch": 2.4113958212704985,
|
| 2702 |
-
"grad_norm": 15.331337928771973,
|
| 2703 |
-
"learning_rate": 8.377318067968348e-07,
|
| 2704 |
-
"loss": 12.805096435546876,
|
| 2705 |
-
"step": 7680
|
| 2706 |
-
},
|
| 2707 |
-
{
|
| 2708 |
-
"epoch": 2.417676673503636,
|
| 2709 |
-
"grad_norm": 14.103033065795898,
|
| 2710 |
-
"learning_rate": 7.901931392556267e-07,
|
| 2711 |
-
"loss": 10.87573471069336,
|
| 2712 |
-
"step": 7700
|
| 2713 |
-
},
|
| 2714 |
-
{
|
| 2715 |
-
"epoch": 2.4239575257367734,
|
| 2716 |
-
"grad_norm": 12.514013290405273,
|
| 2717 |
-
"learning_rate": 7.440213463723894e-07,
|
| 2718 |
-
"loss": 10.322096252441407,
|
| 2719 |
-
"step": 7720
|
| 2720 |
-
},
|
| 2721 |
-
{
|
| 2722 |
-
"epoch": 2.4302383779699106,
|
| 2723 |
-
"grad_norm": 17.600311279296875,
|
| 2724 |
-
"learning_rate": 6.992190349855016e-07,
|
| 2725 |
-
"loss": 13.445755004882812,
|
| 2726 |
-
"step": 7740
|
| 2727 |
-
},
|
| 2728 |
-
{
|
| 2729 |
-
"epoch": 2.4365192302030483,
|
| 2730 |
-
"grad_norm": 17.075410842895508,
|
| 2731 |
-
"learning_rate": 6.557887346130387e-07,
|
| 2732 |
-
"loss": 11.708379364013672,
|
| 2733 |
-
"step": 7760
|
| 2734 |
-
},
|
| 2735 |
-
{
|
| 2736 |
-
"epoch": 2.4428000824361855,
|
| 2737 |
-
"grad_norm": 14.414114952087402,
|
| 2738 |
-
"learning_rate": 6.137328973099599e-07,
|
| 2739 |
-
"loss": 11.876699829101563,
|
| 2740 |
-
"step": 7780
|
| 2741 |
-
},
|
| 2742 |
-
{
|
| 2743 |
-
"epoch": 2.4490809346693228,
|
| 2744 |
-
"grad_norm": 20.820781707763672,
|
| 2745 |
-
"learning_rate": 5.730538975296656e-07,
|
| 2746 |
-
"loss": 12.301487731933594,
|
| 2747 |
-
"step": 7800
|
| 2748 |
-
},
|
| 2749 |
-
{
|
| 2750 |
-
"epoch": 2.4553617869024604,
|
| 2751 |
-
"grad_norm": 17.03597068786621,
|
| 2752 |
-
"learning_rate": 5.337540319899331e-07,
|
| 2753 |
-
"loss": 12.67370376586914,
|
| 2754 |
-
"step": 7820
|
| 2755 |
-
},
|
| 2756 |
-
{
|
| 2757 |
-
"epoch": 2.4616426391355977,
|
| 2758 |
-
"grad_norm": 18.336877822875977,
|
| 2759 |
-
"learning_rate": 4.958355195432613e-07,
|
| 2760 |
-
"loss": 11.884120178222656,
|
| 2761 |
-
"step": 7840
|
| 2762 |
-
},
|
| 2763 |
-
{
|
| 2764 |
-
"epoch": 2.467923491368735,
|
| 2765 |
-
"grad_norm": 22.29871940612793,
|
| 2766 |
-
"learning_rate": 4.593005010515711e-07,
|
| 2767 |
-
"loss": 11.572581481933593,
|
| 2768 |
-
"step": 7860
|
| 2769 |
-
},
|
| 2770 |
-
{
|
| 2771 |
-
"epoch": 2.4742043436018726,
|
| 2772 |
-
"grad_norm": 16.737030029296875,
|
| 2773 |
-
"learning_rate": 4.241510392653525e-07,
|
| 2774 |
-
"loss": 10.467595672607422,
|
| 2775 |
-
"step": 7880
|
| 2776 |
-
},
|
| 2777 |
-
{
|
| 2778 |
-
"epoch": 2.48048519583501,
|
| 2779 |
-
"grad_norm": 13.895743370056152,
|
| 2780 |
-
"learning_rate": 3.9038911870719063e-07,
|
| 2781 |
-
"loss": 11.676488494873047,
|
| 2782 |
-
"step": 7900
|
| 2783 |
-
},
|
| 2784 |
-
{
|
| 2785 |
-
"epoch": 2.486766048068147,
|
| 2786 |
-
"grad_norm": 19.80741310119629,
|
| 2787 |
-
"learning_rate": 3.580166455597278e-07,
|
| 2788 |
-
"loss": 10.244381713867188,
|
| 2789 |
-
"step": 7920
|
| 2790 |
-
},
|
| 2791 |
-
{
|
| 2792 |
-
"epoch": 2.4930469003012847,
|
| 2793 |
-
"grad_norm": 13.476153373718262,
|
| 2794 |
-
"learning_rate": 3.270354475580412e-07,
|
| 2795 |
-
"loss": 11.219804382324218,
|
| 2796 |
-
"step": 7940
|
| 2797 |
-
},
|
| 2798 |
-
{
|
| 2799 |
-
"epoch": 2.499327752534422,
|
| 2800 |
-
"grad_norm": 11.395315170288086,
|
| 2801 |
-
"learning_rate": 2.9744727388644477e-07,
|
| 2802 |
-
"loss": 10.30084686279297,
|
| 2803 |
-
"step": 7960
|
| 2804 |
-
},
|
| 2805 |
-
{
|
| 2806 |
-
"epoch": 2.505608604767559,
|
| 2807 |
-
"grad_norm": 20.2418270111084,
|
| 2808 |
-
"learning_rate": 2.692537950797325e-07,
|
| 2809 |
-
"loss": 10.656898498535156,
|
| 2810 |
-
"step": 7980
|
| 2811 |
-
},
|
| 2812 |
-
{
|
| 2813 |
-
"epoch": 2.511889457000697,
|
| 2814 |
-
"grad_norm": 16.177560806274414,
|
| 2815 |
-
"learning_rate": 2.424566029288672e-07,
|
| 2816 |
-
"loss": 10.936573791503907,
|
| 2817 |
-
"step": 8000
|
| 2818 |
-
},
|
| 2819 |
-
{
|
| 2820 |
-
"epoch": 2.518170309233834,
|
| 2821 |
-
"grad_norm": 14.45263957977295,
|
| 2822 |
-
"learning_rate": 2.1705721039110005e-07,
|
| 2823 |
-
"loss": 11.340428924560547,
|
| 2824 |
-
"step": 8020
|
| 2825 |
-
}
|
| 2826 |
-
],
|
| 2827 |
-
"logging_steps": 20,
|
| 2828 |
-
"max_steps": 8362,
|
| 2829 |
-
"num_input_tokens_seen": 0,
|
| 2830 |
-
"num_train_epochs": 3,
|
| 2831 |
-
"save_steps": 1000000000,
|
| 2832 |
-
"stateful_callbacks": {
|
| 2833 |
-
"TrainerControl": {
|
| 2834 |
-
"args": {
|
| 2835 |
-
"should_epoch_stop": false,
|
| 2836 |
-
"should_evaluate": false,
|
| 2837 |
-
"should_log": false,
|
| 2838 |
-
"should_save": true,
|
| 2839 |
-
"should_training_stop": false
|
| 2840 |
-
},
|
| 2841 |
-
"attributes": {}
|
| 2842 |
-
}
|
| 2843 |
-
},
|
| 2844 |
-
"total_flos": 2.53901980411776e+16,
|
| 2845 |
-
"train_batch_size": 1,
|
| 2846 |
-
"trial_name": null,
|
| 2847 |
-
"trial_params": null
|
| 2848 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
slots/19/checkpoint-6078/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/19/checkpoint-6078/trainer_state.json
DELETED
|
@@ -1,2162 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 1.9085252755233224,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 6078,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.0003140426116568692,
|
| 14 |
-
"grad_norm": 11.236513137817383,
|
| 15 |
-
"learning_rate": 5e-05,
|
| 16 |
-
"loss": 14.929291725158691,
|
| 17 |
-
"step": 1
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.0062808522331373835,
|
| 21 |
-
"grad_norm": 8.230694770812988,
|
| 22 |
-
"learning_rate": 4.9999363065476e-05,
|
| 23 |
-
"loss": 13.227896439401727,
|
| 24 |
-
"step": 20
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.012561704466274767,
|
| 28 |
-
"grad_norm": 15.541277885437012,
|
| 29 |
-
"learning_rate": 4.999745229435885e-05,
|
| 30 |
-
"loss": 11.76095199584961,
|
| 31 |
-
"step": 40
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.018842556699412153,
|
| 35 |
-
"grad_norm": 19.88605308532715,
|
| 36 |
-
"learning_rate": 4.999406489735261e-05,
|
| 37 |
-
"loss": 10.630082702636718,
|
| 38 |
-
"step": 60
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.025123408932549534,
|
| 42 |
-
"grad_norm": 21.772388458251953,
|
| 43 |
-
"learning_rate": 4.998953978539926e-05,
|
| 44 |
-
"loss": 10.612107086181641,
|
| 45 |
-
"step": 80
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.03140426116568692,
|
| 49 |
-
"grad_norm": 24.13600730895996,
|
| 50 |
-
"learning_rate": 4.998374133739565e-05,
|
| 51 |
-
"loss": 8.763066864013672,
|
| 52 |
-
"step": 100
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.037685113398824306,
|
| 56 |
-
"grad_norm": 33.56028747558594,
|
| 57 |
-
"learning_rate": 4.997666984880032e-05,
|
| 58 |
-
"loss": 11.96340103149414,
|
| 59 |
-
"step": 120
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.04396596563196169,
|
| 63 |
-
"grad_norm": 51.29957962036133,
|
| 64 |
-
"learning_rate": 4.996785126662353e-05,
|
| 65 |
-
"loss": 7.662068939208984,
|
| 66 |
-
"step": 140
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.05024681786509907,
|
| 70 |
-
"grad_norm": 34.75215530395508,
|
| 71 |
-
"learning_rate": 4.9957623010947197e-05,
|
| 72 |
-
"loss": 7.163829040527344,
|
| 73 |
-
"step": 160
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.056527670098236456,
|
| 77 |
-
"grad_norm": 48.993743896484375,
|
| 78 |
-
"learning_rate": 4.994598565925399e-05,
|
| 79 |
-
"loss": 6.738508605957032,
|
| 80 |
-
"step": 180
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.06280852233137384,
|
| 84 |
-
"grad_norm": 37.503150939941406,
|
| 85 |
-
"learning_rate": 4.9932939868583476e-05,
|
| 86 |
-
"loss": 5.927839660644532,
|
| 87 |
-
"step": 200
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.06908937456451122,
|
| 91 |
-
"grad_norm": 34.306514739990234,
|
| 92 |
-
"learning_rate": 4.991848637549505e-05,
|
| 93 |
-
"loss": 5.548626327514649,
|
| 94 |
-
"step": 220
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.07537022679764861,
|
| 98 |
-
"grad_norm": 25.989212036132812,
|
| 99 |
-
"learning_rate": 4.990345241504926e-05,
|
| 100 |
-
"loss": 5.030820083618164,
|
| 101 |
-
"step": 240
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.08165107903078599,
|
| 105 |
-
"grad_norm": 23.606063842773438,
|
| 106 |
-
"learning_rate": 4.98862563217225e-05,
|
| 107 |
-
"loss": 4.923490524291992,
|
| 108 |
-
"step": 260
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.08793193126392337,
|
| 112 |
-
"grad_norm": 36.173194885253906,
|
| 113 |
-
"learning_rate": 4.9867655161709595e-05,
|
| 114 |
-
"loss": 4.673915100097656,
|
| 115 |
-
"step": 280
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.09421278349706076,
|
| 119 |
-
"grad_norm": 18.624378204345703,
|
| 120 |
-
"learning_rate": 4.984764998522357e-05,
|
| 121 |
-
"loss": 3.8839893341064453,
|
| 122 |
-
"step": 300
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.10049363573019814,
|
| 126 |
-
"grad_norm": 22.662349700927734,
|
| 127 |
-
"learning_rate": 4.9826241921747555e-05,
|
| 128 |
-
"loss": 3.063496971130371,
|
| 129 |
-
"step": 320
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.10677448796333552,
|
| 133 |
-
"grad_norm": 25.49854278564453,
|
| 134 |
-
"learning_rate": 4.9803432179971055e-05,
|
| 135 |
-
"loss": 3.0309188842773436,
|
| 136 |
-
"step": 340
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.11305534019647291,
|
| 140 |
-
"grad_norm": 15.640332221984863,
|
| 141 |
-
"learning_rate": 4.9779222047721693e-05,
|
| 142 |
-
"loss": 2.6161094665527345,
|
| 143 |
-
"step": 360
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.1193361924296103,
|
| 147 |
-
"grad_norm": 13.35746955871582,
|
| 148 |
-
"learning_rate": 4.9753612891892514e-05,
|
| 149 |
-
"loss": 2.3498945236206055,
|
| 150 |
-
"step": 380
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.12561704466274767,
|
| 154 |
-
"grad_norm": 10.242664337158203,
|
| 155 |
-
"learning_rate": 4.9726606158364765e-05,
|
| 156 |
-
"loss": 2.069006156921387,
|
| 157 |
-
"step": 400
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.13189789689588505,
|
| 161 |
-
"grad_norm": 8.599258422851562,
|
| 162 |
-
"learning_rate": 4.969965664306835e-05,
|
| 163 |
-
"loss": 1.7923540115356444,
|
| 164 |
-
"step": 420
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.13817874912902245,
|
| 168 |
-
"grad_norm": 5.654482364654541,
|
| 169 |
-
"learning_rate": 4.966992909049121e-05,
|
| 170 |
-
"loss": 1.7063348770141602,
|
| 171 |
-
"step": 440
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.14445960136215982,
|
| 175 |
-
"grad_norm": 9.39411449432373,
|
| 176 |
-
"learning_rate": 4.9638808684964765e-05,
|
| 177 |
-
"loss": 1.4251695632934571,
|
| 178 |
-
"step": 460
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.15074045359529722,
|
| 182 |
-
"grad_norm": 4.354184627532959,
|
| 183 |
-
"learning_rate": 4.960629718353295e-05,
|
| 184 |
-
"loss": 1.2388976097106934,
|
| 185 |
-
"step": 480
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.1570213058284346,
|
| 189 |
-
"grad_norm": 4.586623191833496,
|
| 190 |
-
"learning_rate": 4.9572396421780296e-05,
|
| 191 |
-
"loss": 1.2111310958862305,
|
| 192 |
-
"step": 500
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.16330215806157197,
|
| 196 |
-
"grad_norm": 3.979524850845337,
|
| 197 |
-
"learning_rate": 4.9537108313728365e-05,
|
| 198 |
-
"loss": 1.4075250625610352,
|
| 199 |
-
"step": 520
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.16958301029470937,
|
| 203 |
-
"grad_norm": 3.687380790710449,
|
| 204 |
-
"learning_rate": 4.9500434851727635e-05,
|
| 205 |
-
"loss": 0.9275534629821778,
|
| 206 |
-
"step": 540
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.17586386252784675,
|
| 210 |
-
"grad_norm": 2.872728109359741,
|
| 211 |
-
"learning_rate": 4.946431376374067e-05,
|
| 212 |
-
"loss": 1.418515110015869,
|
| 213 |
-
"step": 560
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.18214471476098412,
|
| 217 |
-
"grad_norm": 2.1888179779052734,
|
| 218 |
-
"learning_rate": 4.9424944888143395e-05,
|
| 219 |
-
"loss": 0.7708720684051513,
|
| 220 |
-
"step": 580
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.18842556699412152,
|
| 224 |
-
"grad_norm": 2.96533465385437,
|
| 225 |
-
"learning_rate": 4.9384196991293205e-05,
|
| 226 |
-
"loss": 0.6360139369964599,
|
| 227 |
-
"step": 600
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.1947064192272589,
|
| 231 |
-
"grad_norm": 3.389042615890503,
|
| 232 |
-
"learning_rate": 4.9342072373797774e-05,
|
| 233 |
-
"loss": 1.2306670188903808,
|
| 234 |
-
"step": 620
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.20098727146039627,
|
| 238 |
-
"grad_norm": 3.054168939590454,
|
| 239 |
-
"learning_rate": 4.9298573413993806e-05,
|
| 240 |
-
"loss": 0.684608793258667,
|
| 241 |
-
"step": 640
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.20726812369353367,
|
| 245 |
-
"grad_norm": 2.2155184745788574,
|
| 246 |
-
"learning_rate": 4.925370256781273e-05,
|
| 247 |
-
"loss": 0.6700978755950928,
|
| 248 |
-
"step": 660
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.21354897592667105,
|
| 252 |
-
"grad_norm": 1.223708987236023,
|
| 253 |
-
"learning_rate": 4.920746236864208e-05,
|
| 254 |
-
"loss": 0.707273292541504,
|
| 255 |
-
"step": 680
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.21982982815980842,
|
| 259 |
-
"grad_norm": 1.1626676321029663,
|
| 260 |
-
"learning_rate": 4.915985542718241e-05,
|
| 261 |
-
"loss": 0.6534492492675781,
|
| 262 |
-
"step": 700
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.22611068039294582,
|
| 266 |
-
"grad_norm": 1.1415362358093262,
|
| 267 |
-
"learning_rate": 4.9110884431299935e-05,
|
| 268 |
-
"loss": 0.45153131484985354,
|
| 269 |
-
"step": 720
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.2323915326260832,
|
| 273 |
-
"grad_norm": 1.0272735357284546,
|
| 274 |
-
"learning_rate": 4.906055214587476e-05,
|
| 275 |
-
"loss": 0.46799488067626954,
|
| 276 |
-
"step": 740
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.2386723848592206,
|
| 280 |
-
"grad_norm": 1.2896318435668945,
|
| 281 |
-
"learning_rate": 4.9008861412644756e-05,
|
| 282 |
-
"loss": 0.4087998390197754,
|
| 283 |
-
"step": 760
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.24495323709235797,
|
| 287 |
-
"grad_norm": 0.873556911945343,
|
| 288 |
-
"learning_rate": 4.895581515004517e-05,
|
| 289 |
-
"loss": 0.3405194044113159,
|
| 290 |
-
"step": 780
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.25123408932549535,
|
| 294 |
-
"grad_norm": 1.2751230001449585,
|
| 295 |
-
"learning_rate": 4.89014163530438e-05,
|
| 296 |
-
"loss": 0.401216459274292,
|
| 297 |
-
"step": 800
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.25751494155863275,
|
| 301 |
-
"grad_norm": 0.8582271337509155,
|
| 302 |
-
"learning_rate": 4.8845668092971934e-05,
|
| 303 |
-
"loss": 0.36631605625152586,
|
| 304 |
-
"step": 820
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.2637957937917701,
|
| 308 |
-
"grad_norm": 0.7967858910560608,
|
| 309 |
-
"learning_rate": 4.87885735173509e-05,
|
| 310 |
-
"loss": 0.33444316387176515,
|
| 311 |
-
"step": 840
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.2700766460249075,
|
| 315 |
-
"grad_norm": 0.7824229001998901,
|
| 316 |
-
"learning_rate": 4.873013584971442e-05,
|
| 317 |
-
"loss": 0.3244049549102783,
|
| 318 |
-
"step": 860
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.2763574982580449,
|
| 322 |
-
"grad_norm": 0.774782121181488,
|
| 323 |
-
"learning_rate": 4.867035838942657e-05,
|
| 324 |
-
"loss": 0.6262166023254394,
|
| 325 |
-
"step": 880
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.2826383504911823,
|
| 329 |
-
"grad_norm": 0.6737077832221985,
|
| 330 |
-
"learning_rate": 4.86092445114955e-05,
|
| 331 |
-
"loss": 0.28058154582977296,
|
| 332 |
-
"step": 900
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.28891920272431965,
|
| 336 |
-
"grad_norm": 1.0165168046951294,
|
| 337 |
-
"learning_rate": 4.8546797666382896e-05,
|
| 338 |
-
"loss": 0.3938837289810181,
|
| 339 |
-
"step": 920
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.29520005495745705,
|
| 343 |
-
"grad_norm": 1.093992829322815,
|
| 344 |
-
"learning_rate": 4.8483021379809154e-05,
|
| 345 |
-
"loss": 0.3142520904541016,
|
| 346 |
-
"step": 940
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.30148090719059445,
|
| 350 |
-
"grad_norm": 0.6402747631072998,
|
| 351 |
-
"learning_rate": 4.8417919252554354e-05,
|
| 352 |
-
"loss": 0.29062786102294924,
|
| 353 |
-
"step": 960
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.3077617594237318,
|
| 357 |
-
"grad_norm": 1.9970321655273438,
|
| 358 |
-
"learning_rate": 4.8351494960254916e-05,
|
| 359 |
-
"loss": 0.31338260173797605,
|
| 360 |
-
"step": 980
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.3140426116568692,
|
| 364 |
-
"grad_norm": 1.1727873086929321,
|
| 365 |
-
"learning_rate": 4.8283752253196094e-05,
|
| 366 |
-
"loss": 0.2333289623260498,
|
| 367 |
-
"step": 1000
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.3203234638900066,
|
| 371 |
-
"grad_norm": 0.7795476913452148,
|
| 372 |
-
"learning_rate": 4.8214694956100245e-05,
|
| 373 |
-
"loss": 0.2488691806793213,
|
| 374 |
-
"step": 1020
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.32660431612314395,
|
| 378 |
-
"grad_norm": 1.1675890684127808,
|
| 379 |
-
"learning_rate": 4.814432696791089e-05,
|
| 380 |
-
"loss": 0.22975556850433348,
|
| 381 |
-
"step": 1040
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.33288516835628135,
|
| 385 |
-
"grad_norm": 0.5809735059738159,
|
| 386 |
-
"learning_rate": 4.807265226157255e-05,
|
| 387 |
-
"loss": 0.2238070011138916,
|
| 388 |
-
"step": 1060
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.33916602058941875,
|
| 392 |
-
"grad_norm": 0.6172874569892883,
|
| 393 |
-
"learning_rate": 4.7999674883806465e-05,
|
| 394 |
-
"loss": 0.22853121757507325,
|
| 395 |
-
"step": 1080
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.3454468728225561,
|
| 399 |
-
"grad_norm": 0.9794221520423889,
|
| 400 |
-
"learning_rate": 4.792539895488213e-05,
|
| 401 |
-
"loss": 0.2538387060165405,
|
| 402 |
-
"step": 1100
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.3517277250556935,
|
| 406 |
-
"grad_norm": 0.6366945505142212,
|
| 407 |
-
"learning_rate": 4.784982866838462e-05,
|
| 408 |
-
"loss": 0.207783842086792,
|
| 409 |
-
"step": 1120
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.3580085772888309,
|
| 413 |
-
"grad_norm": 0.5059712529182434,
|
| 414 |
-
"learning_rate": 4.777296829097785e-05,
|
| 415 |
-
"loss": 0.2022031545639038,
|
| 416 |
-
"step": 1140
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.36428942952196824,
|
| 420 |
-
"grad_norm": 0.42327556014060974,
|
| 421 |
-
"learning_rate": 4.769482216216369e-05,
|
| 422 |
-
"loss": 0.19924311637878417,
|
| 423 |
-
"step": 1160
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.37057028175510565,
|
| 427 |
-
"grad_norm": 1.0261644124984741,
|
| 428 |
-
"learning_rate": 4.761539469403695e-05,
|
| 429 |
-
"loss": 0.1949863314628601,
|
| 430 |
-
"step": 1180
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.37685113398824305,
|
| 434 |
-
"grad_norm": 0.78790283203125,
|
| 435 |
-
"learning_rate": 4.7534690371036274e-05,
|
| 436 |
-
"loss": 0.1740255832672119,
|
| 437 |
-
"step": 1200
|
| 438 |
-
},
|
| 439 |
-
{
|
| 440 |
-
"epoch": 0.3831319862213804,
|
| 441 |
-
"grad_norm": 0.49199551343917847,
|
| 442 |
-
"learning_rate": 4.7452713749690934e-05,
|
| 443 |
-
"loss": 0.16614840030670167,
|
| 444 |
-
"step": 1220
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.3894128384545178,
|
| 448 |
-
"grad_norm": 0.5265403985977173,
|
| 449 |
-
"learning_rate": 4.73694694583636e-05,
|
| 450 |
-
"loss": 0.17591538429260253,
|
| 451 |
-
"step": 1240
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.3956936906876552,
|
| 455 |
-
"grad_norm": 0.5221320390701294,
|
| 456 |
-
"learning_rate": 4.728496219698902e-05,
|
| 457 |
-
"loss": 0.18058695793151855,
|
| 458 |
-
"step": 1260
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.40197454292079254,
|
| 462 |
-
"grad_norm": 1.2317312955856323,
|
| 463 |
-
"learning_rate": 4.719919673680863e-05,
|
| 464 |
-
"loss": 0.12510637044906617,
|
| 465 |
-
"step": 1280
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 0.40825539515392995,
|
| 469 |
-
"grad_norm": 0.21159659326076508,
|
| 470 |
-
"learning_rate": 4.711217792010122e-05,
|
| 471 |
-
"loss": 0.3407147884368896,
|
| 472 |
-
"step": 1300
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.41453624738706735,
|
| 476 |
-
"grad_norm": 1.4082627296447754,
|
| 477 |
-
"learning_rate": 4.70239106599095e-05,
|
| 478 |
-
"loss": 0.16848219633102418,
|
| 479 |
-
"step": 1320
|
| 480 |
-
},
|
| 481 |
-
{
|
| 482 |
-
"epoch": 0.4208170996202047,
|
| 483 |
-
"grad_norm": 0.35336634516716003,
|
| 484 |
-
"learning_rate": 4.6934399939762746e-05,
|
| 485 |
-
"loss": 0.13055137395858765,
|
| 486 |
-
"step": 1340
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 0.4270979518533421,
|
| 490 |
-
"grad_norm": 0.8723010420799255,
|
| 491 |
-
"learning_rate": 4.684365081339541e-05,
|
| 492 |
-
"loss": 0.1356166958808899,
|
| 493 |
-
"step": 1360
|
| 494 |
-
},
|
| 495 |
-
{
|
| 496 |
-
"epoch": 0.4333788040864795,
|
| 497 |
-
"grad_norm": 0.49977484345436096,
|
| 498 |
-
"learning_rate": 4.675166840446179e-05,
|
| 499 |
-
"loss": 0.17129049301147461,
|
| 500 |
-
"step": 1380
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.43965965631961684,
|
| 504 |
-
"grad_norm": 0.8447715640068054,
|
| 505 |
-
"learning_rate": 4.6658457906246754e-05,
|
| 506 |
-
"loss": 0.16764267683029174,
|
| 507 |
-
"step": 1400
|
| 508 |
-
},
|
| 509 |
-
{
|
| 510 |
-
"epoch": 0.44594050855275424,
|
| 511 |
-
"grad_norm": 0.491344153881073,
|
| 512 |
-
"learning_rate": 4.656402458137255e-05,
|
| 513 |
-
"loss": 0.2734020709991455,
|
| 514 |
-
"step": 1420
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.45222136078589165,
|
| 518 |
-
"grad_norm": 0.7019853591918945,
|
| 519 |
-
"learning_rate": 4.646837376150164e-05,
|
| 520 |
-
"loss": 0.15193712711334229,
|
| 521 |
-
"step": 1440
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 0.458502213019029,
|
| 525 |
-
"grad_norm": 0.5096240639686584,
|
| 526 |
-
"learning_rate": 4.637151084703571e-05,
|
| 527 |
-
"loss": 0.31453657150268555,
|
| 528 |
-
"step": 1460
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 0.4647830652521664,
|
| 532 |
-
"grad_norm": 0.457865446805954,
|
| 533 |
-
"learning_rate": 4.6273441306810724e-05,
|
| 534 |
-
"loss": 0.14785597324371338,
|
| 535 |
-
"step": 1480
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"epoch": 0.4710639174853038,
|
| 539 |
-
"grad_norm": 0.5041202306747437,
|
| 540 |
-
"learning_rate": 4.617417067778825e-05,
|
| 541 |
-
"loss": 0.13907097578048705,
|
| 542 |
-
"step": 1500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.4773447697184412,
|
| 546 |
-
"grad_norm": 3.153047800064087,
|
| 547 |
-
"learning_rate": 4.6073704564742734e-05,
|
| 548 |
-
"loss": 0.1594400644302368,
|
| 549 |
-
"step": 1520
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.48362562195157854,
|
| 553 |
-
"grad_norm": 6.409017562866211,
|
| 554 |
-
"learning_rate": 4.5972048639945105e-05,
|
| 555 |
-
"loss": 0.1708819031715393,
|
| 556 |
-
"step": 1540
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 0.48990647418471595,
|
| 560 |
-
"grad_norm": 0.5795732736587524,
|
| 561 |
-
"learning_rate": 4.5869208642842536e-05,
|
| 562 |
-
"loss": 0.5376199245452881,
|
| 563 |
-
"step": 1560
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 0.49618732641785335,
|
| 567 |
-
"grad_norm": 1.402137279510498,
|
| 568 |
-
"learning_rate": 4.5765190379734384e-05,
|
| 569 |
-
"loss": 0.13104203939437867,
|
| 570 |
-
"step": 1580
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.5024681786509907,
|
| 574 |
-
"grad_norm": 1.01236093044281,
|
| 575 |
-
"learning_rate": 4.565999972344436e-05,
|
| 576 |
-
"loss": 0.14823287725448608,
|
| 577 |
-
"step": 1600
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 0.508749030884128,
|
| 581 |
-
"grad_norm": 0.47942015528678894,
|
| 582 |
-
"learning_rate": 4.5553642612988954e-05,
|
| 583 |
-
"loss": 0.13399872779846192,
|
| 584 |
-
"step": 1620
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.5150298831172655,
|
| 588 |
-
"grad_norm": 0.5736086964607239,
|
| 589 |
-
"learning_rate": 4.544612505324213e-05,
|
| 590 |
-
"loss": 0.09690673351287842,
|
| 591 |
-
"step": 1640
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 0.5213107353504028,
|
| 595 |
-
"grad_norm": 0.4799337387084961,
|
| 596 |
-
"learning_rate": 4.533745311459631e-05,
|
| 597 |
-
"loss": 0.12333678007125855,
|
| 598 |
-
"step": 1660
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.5275915875835402,
|
| 602 |
-
"grad_norm": 0.49539658427238464,
|
| 603 |
-
"learning_rate": 4.5227632932619604e-05,
|
| 604 |
-
"loss": 0.1237714409828186,
|
| 605 |
-
"step": 1680
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 0.5338724398166776,
|
| 609 |
-
"grad_norm": 3.5184292793273926,
|
| 610 |
-
"learning_rate": 4.5116670707709396e-05,
|
| 611 |
-
"loss": 0.12902543544769288,
|
| 612 |
-
"step": 1700
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.540153292049815,
|
| 616 |
-
"grad_norm": 0.543497622013092,
|
| 617 |
-
"learning_rate": 4.5004572704742345e-05,
|
| 618 |
-
"loss": 0.11819875240325928,
|
| 619 |
-
"step": 1720
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"epoch": 0.5464341442829523,
|
| 623 |
-
"grad_norm": 0.47288012504577637,
|
| 624 |
-
"learning_rate": 4.489134525272057e-05,
|
| 625 |
-
"loss": 0.09024000763893128,
|
| 626 |
-
"step": 1740
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.5527149965160898,
|
| 630 |
-
"grad_norm": 0.5600817203521729,
|
| 631 |
-
"learning_rate": 4.477699474441441e-05,
|
| 632 |
-
"loss": 0.10022807121276855,
|
| 633 |
-
"step": 1760
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 0.5589958487492271,
|
| 637 |
-
"grad_norm": 0.38338902592658997,
|
| 638 |
-
"learning_rate": 4.466152763600142e-05,
|
| 639 |
-
"loss": 0.1357239603996277,
|
| 640 |
-
"step": 1780
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.5652767009823646,
|
| 644 |
-
"grad_norm": 0.12702441215515137,
|
| 645 |
-
"learning_rate": 4.454495044670189e-05,
|
| 646 |
-
"loss": 0.36903703212738037,
|
| 647 |
-
"step": 1800
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.571557553215502,
|
| 651 |
-
"grad_norm": 0.32971400022506714,
|
| 652 |
-
"learning_rate": 4.4427269758410806e-05,
|
| 653 |
-
"loss": 0.09078516364097595,
|
| 654 |
-
"step": 1820
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.5778384054486393,
|
| 658 |
-
"grad_norm": 0.5972228050231934,
|
| 659 |
-
"learning_rate": 4.4308492215326145e-05,
|
| 660 |
-
"loss": 0.13515870571136473,
|
| 661 |
-
"step": 1840
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.5841192576817767,
|
| 665 |
-
"grad_norm": 0.4297616183757782,
|
| 666 |
-
"learning_rate": 4.418862452357387e-05,
|
| 667 |
-
"loss": 0.16306908130645753,
|
| 668 |
-
"step": 1860
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.5904001099149141,
|
| 672 |
-
"grad_norm": 0.3069750666618347,
|
| 673 |
-
"learning_rate": 4.4067673450829184e-05,
|
| 674 |
-
"loss": 0.11015008687973023,
|
| 675 |
-
"step": 1880
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.5966809621480514,
|
| 679 |
-
"grad_norm": 0.2897976040840149,
|
| 680 |
-
"learning_rate": 4.394564582593452e-05,
|
| 681 |
-
"loss": 0.10074495077133179,
|
| 682 |
-
"step": 1900
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.6029618143811889,
|
| 686 |
-
"grad_norm": 0.2866128981113434,
|
| 687 |
-
"learning_rate": 4.382254853851396e-05,
|
| 688 |
-
"loss": 0.10123426914215088,
|
| 689 |
-
"step": 1920
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.6092426666143262,
|
| 693 |
-
"grad_norm": 0.611520528793335,
|
| 694 |
-
"learning_rate": 4.3698388538584214e-05,
|
| 695 |
-
"loss": 0.1677260398864746,
|
| 696 |
-
"step": 1940
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 0.6155235188474636,
|
| 700 |
-
"grad_norm": 0.6373476982116699,
|
| 701 |
-
"learning_rate": 4.357317283616228e-05,
|
| 702 |
-
"loss": 0.07637218236923218,
|
| 703 |
-
"step": 1960
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.621804371080601,
|
| 707 |
-
"grad_norm": 0.24775440990924835,
|
| 708 |
-
"learning_rate": 4.3446908500869624e-05,
|
| 709 |
-
"loss": 0.07576732635498047,
|
| 710 |
-
"step": 1980
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.6280852233137384,
|
| 714 |
-
"grad_norm": 0.36698204278945923,
|
| 715 |
-
"learning_rate": 4.3319602661533054e-05,
|
| 716 |
-
"loss": 0.08246396780014038,
|
| 717 |
-
"step": 2000
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.6343660755468757,
|
| 721 |
-
"grad_norm": 0.4456537365913391,
|
| 722 |
-
"learning_rate": 4.319126250578223e-05,
|
| 723 |
-
"loss": 0.08348861336708069,
|
| 724 |
-
"step": 2020
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.6406469277800132,
|
| 728 |
-
"grad_norm": 0.12586534023284912,
|
| 729 |
-
"learning_rate": 4.306189527964382e-05,
|
| 730 |
-
"loss": 0.11425427198410035,
|
| 731 |
-
"step": 2040
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.6469277800131505,
|
| 735 |
-
"grad_norm": 0.42404818534851074,
|
| 736 |
-
"learning_rate": 4.2931508287132435e-05,
|
| 737 |
-
"loss": 0.43116607666015627,
|
| 738 |
-
"step": 2060
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.6532086322462879,
|
| 742 |
-
"grad_norm": 0.322374165058136,
|
| 743 |
-
"learning_rate": 4.2800108889838244e-05,
|
| 744 |
-
"loss": 0.07275817394256592,
|
| 745 |
-
"step": 2080
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.6594894844794253,
|
| 749 |
-
"grad_norm": 0.24643594026565552,
|
| 750 |
-
"learning_rate": 4.26677045065113e-05,
|
| 751 |
-
"loss": 0.0800868034362793,
|
| 752 |
-
"step": 2100
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.6657703367125627,
|
| 756 |
-
"grad_norm": 0.32612672448158264,
|
| 757 |
-
"learning_rate": 4.253430261264272e-05,
|
| 758 |
-
"loss": 0.11347514390945435,
|
| 759 |
-
"step": 2120
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.6720511889457,
|
| 763 |
-
"grad_norm": 0.9363917708396912,
|
| 764 |
-
"learning_rate": 4.239991074004262e-05,
|
| 765 |
-
"loss": 0.14267417192459106,
|
| 766 |
-
"step": 2140
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.6783320411788375,
|
| 770 |
-
"grad_norm": 5.144097328186035,
|
| 771 |
-
"learning_rate": 4.226453647641484e-05,
|
| 772 |
-
"loss": 0.2039745807647705,
|
| 773 |
-
"step": 2160
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.6846128934119748,
|
| 777 |
-
"grad_norm": 0.275455117225647,
|
| 778 |
-
"learning_rate": 4.2128187464928604e-05,
|
| 779 |
-
"loss": 0.12193784713745118,
|
| 780 |
-
"step": 2180
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.6908937456451122,
|
| 784 |
-
"grad_norm": 0.23535878956317902,
|
| 785 |
-
"learning_rate": 4.1990871403786934e-05,
|
| 786 |
-
"loss": 0.08345263600349426,
|
| 787 |
-
"step": 2200
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.6971745978782496,
|
| 791 |
-
"grad_norm": 0.24043764173984528,
|
| 792 |
-
"learning_rate": 4.185259604579203e-05,
|
| 793 |
-
"loss": 0.06981859803199768,
|
| 794 |
-
"step": 2220
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.703455450111387,
|
| 798 |
-
"grad_norm": 0.6327846050262451,
|
| 799 |
-
"learning_rate": 4.171336919790759e-05,
|
| 800 |
-
"loss": 0.1702319025993347,
|
| 801 |
-
"step": 2240
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.7097363023445243,
|
| 805 |
-
"grad_norm": 0.059428345412015915,
|
| 806 |
-
"learning_rate": 4.1573198720817955e-05,
|
| 807 |
-
"loss": 0.07783352732658386,
|
| 808 |
-
"step": 2260
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.7160171545776618,
|
| 812 |
-
"grad_norm": 0.4627244770526886,
|
| 813 |
-
"learning_rate": 4.143209252848433e-05,
|
| 814 |
-
"loss": 0.07844600677490235,
|
| 815 |
-
"step": 2280
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.7222980068107991,
|
| 819 |
-
"grad_norm": 0.7847932577133179,
|
| 820 |
-
"learning_rate": 4.1290058587698054e-05,
|
| 821 |
-
"loss": 0.09436154365539551,
|
| 822 |
-
"step": 2300
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.7285788590439365,
|
| 826 |
-
"grad_norm": 0.6646613478660583,
|
| 827 |
-
"learning_rate": 4.114710491763063e-05,
|
| 828 |
-
"loss": 0.10847301483154297,
|
| 829 |
-
"step": 2320
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.734859711277074,
|
| 833 |
-
"grad_norm": 0.13904166221618652,
|
| 834 |
-
"learning_rate": 4.100323958938112e-05,
|
| 835 |
-
"loss": 0.10292520523071289,
|
| 836 |
-
"step": 2340
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.7411405635102113,
|
| 840 |
-
"grad_norm": 0.17684051394462585,
|
| 841 |
-
"learning_rate": 4.085847072552036e-05,
|
| 842 |
-
"loss": 0.10409547090530395,
|
| 843 |
-
"step": 2360
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 0.7474214157433486,
|
| 847 |
-
"grad_norm": 2.33097767829895,
|
| 848 |
-
"learning_rate": 4.0712806499632397e-05,
|
| 849 |
-
"loss": 0.07961801886558532,
|
| 850 |
-
"step": 2380
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 0.7537022679764861,
|
| 854 |
-
"grad_norm": 0.476825475692749,
|
| 855 |
-
"learning_rate": 4.0566255135853014e-05,
|
| 856 |
-
"loss": 0.07112305760383605,
|
| 857 |
-
"step": 2400
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 0.7599831202096234,
|
| 861 |
-
"grad_norm": 0.5014712810516357,
|
| 862 |
-
"learning_rate": 4.04188249084054e-05,
|
| 863 |
-
"loss": 0.1152498722076416,
|
| 864 |
-
"step": 2420
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 0.7662639724427608,
|
| 868 |
-
"grad_norm": 0.4747202694416046,
|
| 869 |
-
"learning_rate": 4.0270524141133e-05,
|
| 870 |
-
"loss": 0.12496150732040405,
|
| 871 |
-
"step": 2440
|
| 872 |
-
},
|
| 873 |
-
{
|
| 874 |
-
"epoch": 0.7725448246758982,
|
| 875 |
-
"grad_norm": 0.3705868721008301,
|
| 876 |
-
"learning_rate": 4.01213612070295e-05,
|
| 877 |
-
"loss": 0.09812939167022705,
|
| 878 |
-
"step": 2460
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 0.7788256769090356,
|
| 882 |
-
"grad_norm": 0.18966151773929596,
|
| 883 |
-
"learning_rate": 3.997134452776617e-05,
|
| 884 |
-
"loss": 0.09260130524635315,
|
| 885 |
-
"step": 2480
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 0.7851065291421729,
|
| 889 |
-
"grad_norm": 0.4178507328033447,
|
| 890 |
-
"learning_rate": 3.982048257321631e-05,
|
| 891 |
-
"loss": 0.08796253800392151,
|
| 892 |
-
"step": 2500
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 0.7913873813753104,
|
| 896 |
-
"grad_norm": 4.684873580932617,
|
| 897 |
-
"learning_rate": 3.966878386097711e-05,
|
| 898 |
-
"loss": 0.09675108194351197,
|
| 899 |
-
"step": 2520
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 0.7976682336084477,
|
| 903 |
-
"grad_norm": 0.7546767592430115,
|
| 904 |
-
"learning_rate": 3.95162569558887e-05,
|
| 905 |
-
"loss": 0.061058801412582395,
|
| 906 |
-
"step": 2540
|
| 907 |
-
},
|
| 908 |
-
{
|
| 909 |
-
"epoch": 0.8039490858415851,
|
| 910 |
-
"grad_norm": 1.2984436750411987,
|
| 911 |
-
"learning_rate": 3.936291046955061e-05,
|
| 912 |
-
"loss": 0.0732529878616333,
|
| 913 |
-
"step": 2560
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 0.8102299380747225,
|
| 917 |
-
"grad_norm": 0.08046652376651764,
|
| 918 |
-
"learning_rate": 3.920875305983553e-05,
|
| 919 |
-
"loss": 0.056511855125427245,
|
| 920 |
-
"step": 2580
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.8165107903078599,
|
| 924 |
-
"grad_norm": 0.24541246891021729,
|
| 925 |
-
"learning_rate": 3.905379343040054e-05,
|
| 926 |
-
"loss": 0.06492850780487061,
|
| 927 |
-
"step": 2600
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 0.8227916425409972,
|
| 931 |
-
"grad_norm": 0.34249547123908997,
|
| 932 |
-
"learning_rate": 3.889804033019567e-05,
|
| 933 |
-
"loss": 0.06372414827346802,
|
| 934 |
-
"step": 2620
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.8290724947741347,
|
| 938 |
-
"grad_norm": 0.6350385546684265,
|
| 939 |
-
"learning_rate": 3.8741502552969956e-05,
|
| 940 |
-
"loss": 0.07334259748458863,
|
| 941 |
-
"step": 2640
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 0.835353347007272,
|
| 945 |
-
"grad_norm": 0.5676931738853455,
|
| 946 |
-
"learning_rate": 3.8584188936774925e-05,
|
| 947 |
-
"loss": 0.051751017570495605,
|
| 948 |
-
"step": 2660
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.8416341992404094,
|
| 952 |
-
"grad_norm": 1.5374674797058105,
|
| 953 |
-
"learning_rate": 3.842610836346563e-05,
|
| 954 |
-
"loss": 0.13946666717529296,
|
| 955 |
-
"step": 2680
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 0.8479150514735468,
|
| 959 |
-
"grad_norm": 0.4549136459827423,
|
| 960 |
-
"learning_rate": 3.8267269758199175e-05,
|
| 961 |
-
"loss": 0.06678192019462585,
|
| 962 |
-
"step": 2700
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.8541959037066842,
|
| 966 |
-
"grad_norm": 0.4328550696372986,
|
| 967 |
-
"learning_rate": 3.8107682088930794e-05,
|
| 968 |
-
"loss": 0.0756498098373413,
|
| 969 |
-
"step": 2720
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.8604767559398215,
|
| 973 |
-
"grad_norm": 1.7214152812957764,
|
| 974 |
-
"learning_rate": 3.7947354365907536e-05,
|
| 975 |
-
"loss": 0.07417964935302734,
|
| 976 |
-
"step": 2740
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 0.866757608172959,
|
| 980 |
-
"grad_norm": 0.3970869481563568,
|
| 981 |
-
"learning_rate": 3.7786295641159544e-05,
|
| 982 |
-
"loss": 0.055640166997909545,
|
| 983 |
-
"step": 2760
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 0.8730384604060963,
|
| 987 |
-
"grad_norm": 0.16465847194194794,
|
| 988 |
-
"learning_rate": 3.7624515007989004e-05,
|
| 989 |
-
"loss": 0.057208091020584106,
|
| 990 |
-
"step": 2780
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.8793193126392337,
|
| 994 |
-
"grad_norm": 1.0308940410614014,
|
| 995 |
-
"learning_rate": 3.746202160045669e-05,
|
| 996 |
-
"loss": 0.1124604344367981,
|
| 997 |
-
"step": 2800
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 0.8856001648723711,
|
| 1001 |
-
"grad_norm": 1.095229983329773,
|
| 1002 |
-
"learning_rate": 3.729882459286632e-05,
|
| 1003 |
-
"loss": 0.06749147176742554,
|
| 1004 |
-
"step": 2820
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 0.8918810171055085,
|
| 1008 |
-
"grad_norm": 0.7268246412277222,
|
| 1009 |
-
"learning_rate": 3.713493319924655e-05,
|
| 1010 |
-
"loss": 0.05816763639450073,
|
| 1011 |
-
"step": 2840
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 0.8981618693386458,
|
| 1015 |
-
"grad_norm": 0.06477148085832596,
|
| 1016 |
-
"learning_rate": 3.697035667283072e-05,
|
| 1017 |
-
"loss": 0.07365052700042725,
|
| 1018 |
-
"step": 2860
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 0.9044427215717833,
|
| 1022 |
-
"grad_norm": 0.14640307426452637,
|
| 1023 |
-
"learning_rate": 3.68051043055345e-05,
|
| 1024 |
-
"loss": 0.10839139223098755,
|
| 1025 |
-
"step": 2880
|
| 1026 |
-
},
|
| 1027 |
-
{
|
| 1028 |
-
"epoch": 0.9107235738049206,
|
| 1029 |
-
"grad_norm": 0.6694417595863342,
|
| 1030 |
-
"learning_rate": 3.66391854274312e-05,
|
| 1031 |
-
"loss": 0.0827883541584015,
|
| 1032 |
-
"step": 2900
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 0.917004426038058,
|
| 1036 |
-
"grad_norm": 0.7503764629364014,
|
| 1037 |
-
"learning_rate": 3.647260940622504e-05,
|
| 1038 |
-
"loss": 0.05839287042617798,
|
| 1039 |
-
"step": 2920
|
| 1040 |
-
},
|
| 1041 |
-
{
|
| 1042 |
-
"epoch": 0.9232852782711954,
|
| 1043 |
-
"grad_norm": 0.6776005029678345,
|
| 1044 |
-
"learning_rate": 3.6305385646722224e-05,
|
| 1045 |
-
"loss": 0.08498733639717101,
|
| 1046 |
-
"step": 2940
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 0.9295661305043328,
|
| 1050 |
-
"grad_norm": 0.1537044644355774,
|
| 1051 |
-
"learning_rate": 3.613752359029998e-05,
|
| 1052 |
-
"loss": 0.1408205509185791,
|
| 1053 |
-
"step": 2960
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 0.9358469827374702,
|
| 1057 |
-
"grad_norm": 0.18748170137405396,
|
| 1058 |
-
"learning_rate": 3.5969032714373495e-05,
|
| 1059 |
-
"loss": 0.07079569697380066,
|
| 1060 |
-
"step": 2980
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 0.9421278349706076,
|
| 1064 |
-
"grad_norm": 1.4993492364883423,
|
| 1065 |
-
"learning_rate": 3.57999225318608e-05,
|
| 1066 |
-
"loss": 0.07452060580253601,
|
| 1067 |
-
"step": 3000
|
| 1068 |
-
},
|
| 1069 |
-
{
|
| 1070 |
-
"epoch": 0.9484086872037449,
|
| 1071 |
-
"grad_norm": 2.759636163711548,
|
| 1072 |
-
"learning_rate": 3.5638702921686935e-05,
|
| 1073 |
-
"loss": 0.30155444145202637,
|
| 1074 |
-
"step": 3020
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.9546895394368824,
|
| 1078 |
-
"grad_norm": 0.35174673795700073,
|
| 1079 |
-
"learning_rate": 3.546841258479434e-05,
|
| 1080 |
-
"loss": 0.07043100595474243,
|
| 1081 |
-
"step": 3040
|
| 1082 |
-
},
|
| 1083 |
-
{
|
| 1084 |
-
"epoch": 0.9609703916700197,
|
| 1085 |
-
"grad_norm": 0.1061045303940773,
|
| 1086 |
-
"learning_rate": 3.529753120609982e-05,
|
| 1087 |
-
"loss": 0.06520496010780334,
|
| 1088 |
-
"step": 3060
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 0.9672512439031571,
|
| 1092 |
-
"grad_norm": 0.32999858260154724,
|
| 1093 |
-
"learning_rate": 3.51260684334881e-05,
|
| 1094 |
-
"loss": 0.11166098117828369,
|
| 1095 |
-
"step": 3080
|
| 1096 |
-
},
|
| 1097 |
-
{
|
| 1098 |
-
"epoch": 0.9735320961362945,
|
| 1099 |
-
"grad_norm": 0.527561366558075,
|
| 1100 |
-
"learning_rate": 3.4954033947669163e-05,
|
| 1101 |
-
"loss": 0.05884957313537598,
|
| 1102 |
-
"step": 3100
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 0.9798129483694319,
|
| 1106 |
-
"grad_norm": 0.21894922852516174,
|
| 1107 |
-
"learning_rate": 3.478143746163164e-05,
|
| 1108 |
-
"loss": 0.04760840237140655,
|
| 1109 |
-
"step": 3120
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"epoch": 0.9860938006025692,
|
| 1113 |
-
"grad_norm": 0.39011839032173157,
|
| 1114 |
-
"learning_rate": 3.460828872009444e-05,
|
| 1115 |
-
"loss": 0.07473071217536927,
|
| 1116 |
-
"step": 3140
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 0.9923746528357067,
|
| 1120 |
-
"grad_norm": 0.24014221131801605,
|
| 1121 |
-
"learning_rate": 3.443459749895658e-05,
|
| 1122 |
-
"loss": 0.05981050133705139,
|
| 1123 |
-
"step": 3160
|
| 1124 |
-
},
|
| 1125 |
-
{
|
| 1126 |
-
"epoch": 0.998655505068844,
|
| 1127 |
-
"grad_norm": 0.4369584918022156,
|
| 1128 |
-
"learning_rate": 3.4260373604745205e-05,
|
| 1129 |
-
"loss": 0.06494853496551514,
|
| 1130 |
-
"step": 3180
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.004710639174853,
|
| 1134 |
-
"grad_norm": 0.11435407400131226,
|
| 1135 |
-
"learning_rate": 3.4085626874061974e-05,
|
| 1136 |
-
"loss": 0.037298232316970825,
|
| 1137 |
-
"step": 3200
|
| 1138 |
-
},
|
| 1139 |
-
{
|
| 1140 |
-
"epoch": 1.0109914914079905,
|
| 1141 |
-
"grad_norm": 0.10405384004116058,
|
| 1142 |
-
"learning_rate": 3.391036717302761e-05,
|
| 1143 |
-
"loss": 0.03180496394634247,
|
| 1144 |
-
"step": 3220
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 1.0172723436411277,
|
| 1148 |
-
"grad_norm": 0.2586648166179657,
|
| 1149 |
-
"learning_rate": 3.373460439672495e-05,
|
| 1150 |
-
"loss": 0.04955384433269501,
|
| 1151 |
-
"step": 3240
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 1.0235531958742652,
|
| 1155 |
-
"grad_norm": 0.08358602225780487,
|
| 1156 |
-
"learning_rate": 3.35583484686402e-05,
|
| 1157 |
-
"loss": 0.05318580269813537,
|
| 1158 |
-
"step": 3260
|
| 1159 |
-
},
|
| 1160 |
-
{
|
| 1161 |
-
"epoch": 1.0298340481074026,
|
| 1162 |
-
"grad_norm": 0.07553794234991074,
|
| 1163 |
-
"learning_rate": 3.338160934010271e-05,
|
| 1164 |
-
"loss": 0.08674532771110535,
|
| 1165 |
-
"step": 3280
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 1.0361149003405399,
|
| 1169 |
-
"grad_norm": 0.2271793782711029,
|
| 1170 |
-
"learning_rate": 3.32043969897231e-05,
|
| 1171 |
-
"loss": 0.04139872789382935,
|
| 1172 |
-
"step": 3300
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.0423957525736773,
|
| 1176 |
-
"grad_norm": 0.25413981080055237,
|
| 1177 |
-
"learning_rate": 3.302672142282988e-05,
|
| 1178 |
-
"loss": 0.035733970999717715,
|
| 1179 |
-
"step": 3320
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 1.0486766048068148,
|
| 1183 |
-
"grad_norm": 0.09163467586040497,
|
| 1184 |
-
"learning_rate": 3.284859267090456e-05,
|
| 1185 |
-
"loss": 0.08379928469657898,
|
| 1186 |
-
"step": 3340
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 1.054957457039952,
|
| 1190 |
-
"grad_norm": 0.07256153970956802,
|
| 1191 |
-
"learning_rate": 3.267002079101525e-05,
|
| 1192 |
-
"loss": 0.03688819110393524,
|
| 1193 |
-
"step": 3360
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 1.0612383092730895,
|
| 1197 |
-
"grad_norm": 0.305923193693161,
|
| 1198 |
-
"learning_rate": 3.2491015865248884e-05,
|
| 1199 |
-
"loss": 0.049816560745239255,
|
| 1200 |
-
"step": 3380
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 1.067519161506227,
|
| 1204 |
-
"grad_norm": 0.046878550201654434,
|
| 1205 |
-
"learning_rate": 3.231158800014197e-05,
|
| 1206 |
-
"loss": 0.03804933726787567,
|
| 1207 |
-
"step": 3400
|
| 1208 |
-
},
|
| 1209 |
-
{
|
| 1210 |
-
"epoch": 1.0738000137393642,
|
| 1211 |
-
"grad_norm": 0.6410548686981201,
|
| 1212 |
-
"learning_rate": 3.213174732610995e-05,
|
| 1213 |
-
"loss": 0.0333125501871109,
|
| 1214 |
-
"step": 3420
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.0800808659725016,
|
| 1218 |
-
"grad_norm": 0.06022074818611145,
|
| 1219 |
-
"learning_rate": 3.195150399687531e-05,
|
| 1220 |
-
"loss": 0.028510311245918275,
|
| 1221 |
-
"step": 3440
|
| 1222 |
-
},
|
| 1223 |
-
{
|
| 1224 |
-
"epoch": 1.086361718205639,
|
| 1225 |
-
"grad_norm": 0.41786953806877136,
|
| 1226 |
-
"learning_rate": 3.177086818889423e-05,
|
| 1227 |
-
"loss": 0.046647512912750246,
|
| 1228 |
-
"step": 3460
|
| 1229 |
-
},
|
| 1230 |
-
{
|
| 1231 |
-
"epoch": 1.0926425704387763,
|
| 1232 |
-
"grad_norm": 0.1293671578168869,
|
| 1233 |
-
"learning_rate": 3.1589850100782085e-05,
|
| 1234 |
-
"loss": 0.03643852770328522,
|
| 1235 |
-
"step": 3480
|
| 1236 |
-
},
|
| 1237 |
-
{
|
| 1238 |
-
"epoch": 1.0989234226719138,
|
| 1239 |
-
"grad_norm": 0.05348878353834152,
|
| 1240 |
-
"learning_rate": 3.1408459952737594e-05,
|
| 1241 |
-
"loss": 0.03161140382289886,
|
| 1242 |
-
"step": 3500
|
| 1243 |
-
},
|
| 1244 |
-
{
|
| 1245 |
-
"epoch": 1.1052042749050512,
|
| 1246 |
-
"grad_norm": 0.29638731479644775,
|
| 1247 |
-
"learning_rate": 3.122670798596583e-05,
|
| 1248 |
-
"loss": 0.04662284553050995,
|
| 1249 |
-
"step": 3520
|
| 1250 |
-
},
|
| 1251 |
-
{
|
| 1252 |
-
"epoch": 1.1114851271381885,
|
| 1253 |
-
"grad_norm": 0.11425400525331497,
|
| 1254 |
-
"learning_rate": 3.1044604462099965e-05,
|
| 1255 |
-
"loss": 0.04306844174861908,
|
| 1256 |
-
"step": 3540
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.117765979371326,
|
| 1260 |
-
"grad_norm": 13.269792556762695,
|
| 1261 |
-
"learning_rate": 3.0862159662621933e-05,
|
| 1262 |
-
"loss": 0.10026944875717163,
|
| 1263 |
-
"step": 3560
|
| 1264 |
-
},
|
| 1265 |
-
{
|
| 1266 |
-
"epoch": 1.1240468316044634,
|
| 1267 |
-
"grad_norm": 0.09814783930778503,
|
| 1268 |
-
"learning_rate": 3.067938388828195e-05,
|
| 1269 |
-
"loss": 0.03279039561748505,
|
| 1270 |
-
"step": 3580
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.1303276838376006,
|
| 1274 |
-
"grad_norm": 2.8222439289093018,
|
| 1275 |
-
"learning_rate": 3.049628745851689e-05,
|
| 1276 |
-
"loss": 0.03593765795230865,
|
| 1277 |
-
"step": 3600
|
| 1278 |
-
},
|
| 1279 |
-
{
|
| 1280 |
-
"epoch": 1.136608536070738,
|
| 1281 |
-
"grad_norm": 0.14487864077091217,
|
| 1282 |
-
"learning_rate": 3.031288071086772e-05,
|
| 1283 |
-
"loss": 0.0726437509059906,
|
| 1284 |
-
"step": 3620
|
| 1285 |
-
},
|
| 1286 |
-
{
|
| 1287 |
-
"epoch": 1.1428893883038755,
|
| 1288 |
-
"grad_norm": 1.1210272312164307,
|
| 1289 |
-
"learning_rate": 3.0129174000395776e-05,
|
| 1290 |
-
"loss": 0.03795652091503143,
|
| 1291 |
-
"step": 3640
|
| 1292 |
-
},
|
| 1293 |
-
{
|
| 1294 |
-
"epoch": 1.1491702405370128,
|
| 1295 |
-
"grad_norm": 0.42077013850212097,
|
| 1296 |
-
"learning_rate": 2.99451776990982e-05,
|
| 1297 |
-
"loss": 0.04205833375453949,
|
| 1298 |
-
"step": 3660
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.1554510927701502,
|
| 1302 |
-
"grad_norm": 0.08231453597545624,
|
| 1303 |
-
"learning_rate": 2.976090219532226e-05,
|
| 1304 |
-
"loss": 0.04487053751945495,
|
| 1305 |
-
"step": 3680
|
| 1306 |
-
},
|
| 1307 |
-
{
|
| 1308 |
-
"epoch": 1.1617319450032877,
|
| 1309 |
-
"grad_norm": 0.1877751648426056,
|
| 1310 |
-
"learning_rate": 2.9576357893178885e-05,
|
| 1311 |
-
"loss": 0.034676852822303775,
|
| 1312 |
-
"step": 3700
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 1.168012797236425,
|
| 1316 |
-
"grad_norm": 0.6968119144439697,
|
| 1317 |
-
"learning_rate": 2.939155521195526e-05,
|
| 1318 |
-
"loss": 0.03220001757144928,
|
| 1319 |
-
"step": 3720
|
| 1320 |
-
},
|
| 1321 |
-
{
|
| 1322 |
-
"epoch": 1.1742936494695624,
|
| 1323 |
-
"grad_norm": 0.06414724141359329,
|
| 1324 |
-
"learning_rate": 2.9206504585526493e-05,
|
| 1325 |
-
"loss": 0.041881483793258664,
|
| 1326 |
-
"step": 3740
|
| 1327 |
-
},
|
| 1328 |
-
{
|
| 1329 |
-
"epoch": 1.1805745017026998,
|
| 1330 |
-
"grad_norm": 0.10654035210609436,
|
| 1331 |
-
"learning_rate": 2.9021216461766604e-05,
|
| 1332 |
-
"loss": 0.03662925064563751,
|
| 1333 |
-
"step": 3760
|
| 1334 |
-
},
|
| 1335 |
-
{
|
| 1336 |
-
"epoch": 1.186855353935837,
|
| 1337 |
-
"grad_norm": 0.04183540865778923,
|
| 1338 |
-
"learning_rate": 2.8835701301958574e-05,
|
| 1339 |
-
"loss": 0.04763528108596802,
|
| 1340 |
-
"step": 3780
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.1931362061689745,
|
| 1344 |
-
"grad_norm": 2.18105411529541,
|
| 1345 |
-
"learning_rate": 2.864996958020375e-05,
|
| 1346 |
-
"loss": 0.06307411193847656,
|
| 1347 |
-
"step": 3800
|
| 1348 |
-
},
|
| 1349 |
-
{
|
| 1350 |
-
"epoch": 1.199417058402112,
|
| 1351 |
-
"grad_norm": 0.3592161536216736,
|
| 1352 |
-
"learning_rate": 2.8464031782830474e-05,
|
| 1353 |
-
"loss": 0.051524901390075685,
|
| 1354 |
-
"step": 3820
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 1.2056979106352492,
|
| 1358 |
-
"grad_norm": 0.11931532621383667,
|
| 1359 |
-
"learning_rate": 2.827789840780201e-05,
|
| 1360 |
-
"loss": 0.035766962170600894,
|
| 1361 |
-
"step": 3840
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 1.2119787628683867,
|
| 1365 |
-
"grad_norm": 0.06202741712331772,
|
| 1366 |
-
"learning_rate": 2.8091579964123854e-05,
|
| 1367 |
-
"loss": 0.04704097211360932,
|
| 1368 |
-
"step": 3860
|
| 1369 |
-
},
|
| 1370 |
-
{
|
| 1371 |
-
"epoch": 1.2182596151015241,
|
| 1372 |
-
"grad_norm": 0.04776179790496826,
|
| 1373 |
-
"learning_rate": 2.7905086971250393e-05,
|
| 1374 |
-
"loss": 0.028332751989364625,
|
| 1375 |
-
"step": 3880
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 1.2245404673346614,
|
| 1379 |
-
"grad_norm": 0.2371724396944046,
|
| 1380 |
-
"learning_rate": 2.771842995849098e-05,
|
| 1381 |
-
"loss": 0.027332887053489685,
|
| 1382 |
-
"step": 3900
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.2308213195677988,
|
| 1386 |
-
"grad_norm": 0.05050333961844444,
|
| 1387 |
-
"learning_rate": 2.753161946441546e-05,
|
| 1388 |
-
"loss": 0.02915753722190857,
|
| 1389 |
-
"step": 3920
|
| 1390 |
-
},
|
| 1391 |
-
{
|
| 1392 |
-
"epoch": 1.2371021718009363,
|
| 1393 |
-
"grad_norm": 0.08952230215072632,
|
| 1394 |
-
"learning_rate": 2.734466603625915e-05,
|
| 1395 |
-
"loss": 0.025665390491485595,
|
| 1396 |
-
"step": 3940
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 1.2433830240340735,
|
| 1400 |
-
"grad_norm": 0.20825526118278503,
|
| 1401 |
-
"learning_rate": 2.7157580229327385e-05,
|
| 1402 |
-
"loss": 0.2276303768157959,
|
| 1403 |
-
"step": 3960
|
| 1404 |
-
},
|
| 1405 |
-
{
|
| 1406 |
-
"epoch": 1.249663876267211,
|
| 1407 |
-
"grad_norm": 1.8631341457366943,
|
| 1408 |
-
"learning_rate": 2.6970372606399525e-05,
|
| 1409 |
-
"loss": 0.03601144254207611,
|
| 1410 |
-
"step": 3980
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"epoch": 1.2559447285003484,
|
| 1414 |
-
"grad_norm": 0.21078747510910034,
|
| 1415 |
-
"learning_rate": 2.6783053737132607e-05,
|
| 1416 |
-
"loss": 0.08526483774185181,
|
| 1417 |
-
"step": 4000
|
| 1418 |
-
},
|
| 1419 |
-
{
|
| 1420 |
-
"epoch": 1.2622255807334857,
|
| 1421 |
-
"grad_norm": 0.051980458199977875,
|
| 1422 |
-
"learning_rate": 2.6595634197464596e-05,
|
| 1423 |
-
"loss": 0.022251833975315095,
|
| 1424 |
-
"step": 4020
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.2685064329666231,
|
| 1428 |
-
"grad_norm": 0.3019862771034241,
|
| 1429 |
-
"learning_rate": 2.6408124569017257e-05,
|
| 1430 |
-
"loss": 0.05009118318557739,
|
| 1431 |
-
"step": 4040
|
| 1432 |
-
},
|
| 1433 |
-
{
|
| 1434 |
-
"epoch": 1.2747872851997606,
|
| 1435 |
-
"grad_norm": 0.6371468305587769,
|
| 1436 |
-
"learning_rate": 2.622053543849872e-05,
|
| 1437 |
-
"loss": 0.040181088447570804,
|
| 1438 |
-
"step": 4060
|
| 1439 |
-
},
|
| 1440 |
-
{
|
| 1441 |
-
"epoch": 1.2810681374328978,
|
| 1442 |
-
"grad_norm": 1.1055502891540527,
|
| 1443 |
-
"learning_rate": 2.6032877397105783e-05,
|
| 1444 |
-
"loss": 0.02929992079734802,
|
| 1445 |
-
"step": 4080
|
| 1446 |
-
},
|
| 1447 |
-
{
|
| 1448 |
-
"epoch": 1.2873489896660353,
|
| 1449 |
-
"grad_norm": 0.26957738399505615,
|
| 1450 |
-
"learning_rate": 2.58451610399259e-05,
|
| 1451 |
-
"loss": 0.020054836571216584,
|
| 1452 |
-
"step": 4100
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 1.2936298418991727,
|
| 1456 |
-
"grad_norm": 0.07279884070158005,
|
| 1457 |
-
"learning_rate": 2.5657396965339003e-05,
|
| 1458 |
-
"loss": 0.039886316657066344,
|
| 1459 |
-
"step": 4120
|
| 1460 |
-
},
|
| 1461 |
-
{
|
| 1462 |
-
"epoch": 1.29991069413231,
|
| 1463 |
-
"grad_norm": 0.09953224658966064,
|
| 1464 |
-
"learning_rate": 2.5469595774419158e-05,
|
| 1465 |
-
"loss": 0.03994681239128113,
|
| 1466 |
-
"step": 4140
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.3061915463654474,
|
| 1470 |
-
"grad_norm": 0.14353635907173157,
|
| 1471 |
-
"learning_rate": 2.5281768070335988e-05,
|
| 1472 |
-
"loss": 0.03711488544940948,
|
| 1473 |
-
"step": 4160
|
| 1474 |
-
},
|
| 1475 |
-
{
|
| 1476 |
-
"epoch": 1.3124723985985849,
|
| 1477 |
-
"grad_norm": 0.27422836422920227,
|
| 1478 |
-
"learning_rate": 2.5093924457756018e-05,
|
| 1479 |
-
"loss": 0.023467497527599336,
|
| 1480 |
-
"step": 4180
|
| 1481 |
-
},
|
| 1482 |
-
{
|
| 1483 |
-
"epoch": 1.3187532508317221,
|
| 1484 |
-
"grad_norm": 0.0728367418050766,
|
| 1485 |
-
"learning_rate": 2.490607554224399e-05,
|
| 1486 |
-
"loss": 0.018936164677143097,
|
| 1487 |
-
"step": 4200
|
| 1488 |
-
},
|
| 1489 |
-
{
|
| 1490 |
-
"epoch": 1.3250341030648596,
|
| 1491 |
-
"grad_norm": 0.24521099030971527,
|
| 1492 |
-
"learning_rate": 2.4718231929664025e-05,
|
| 1493 |
-
"loss": 0.03874744176864624,
|
| 1494 |
-
"step": 4220
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 1.331314955297997,
|
| 1498 |
-
"grad_norm": 0.13463203608989716,
|
| 1499 |
-
"learning_rate": 2.453040422558084e-05,
|
| 1500 |
-
"loss": 0.03414709270000458,
|
| 1501 |
-
"step": 4240
|
| 1502 |
-
},
|
| 1503 |
-
{
|
| 1504 |
-
"epoch": 1.3375958075311343,
|
| 1505 |
-
"grad_norm": 0.12021703273057938,
|
| 1506 |
-
"learning_rate": 2.4342603034661003e-05,
|
| 1507 |
-
"loss": 0.028204512596130372,
|
| 1508 |
-
"step": 4260
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.3438766597642717,
|
| 1512 |
-
"grad_norm": 0.09300895035266876,
|
| 1513 |
-
"learning_rate": 2.4154838960074117e-05,
|
| 1514 |
-
"loss": 0.03560083508491516,
|
| 1515 |
-
"step": 4280
|
| 1516 |
-
},
|
| 1517 |
-
{
|
| 1518 |
-
"epoch": 1.3501575119974092,
|
| 1519 |
-
"grad_norm": 0.07003216445446014,
|
| 1520 |
-
"learning_rate": 2.3967122602894223e-05,
|
| 1521 |
-
"loss": 0.04477728307247162,
|
| 1522 |
-
"step": 4300
|
| 1523 |
-
},
|
| 1524 |
-
{
|
| 1525 |
-
"epoch": 1.3564383642305464,
|
| 1526 |
-
"grad_norm": 0.15798458456993103,
|
| 1527 |
-
"learning_rate": 2.377946456150128e-05,
|
| 1528 |
-
"loss": 0.05230234265327453,
|
| 1529 |
-
"step": 4320
|
| 1530 |
-
},
|
| 1531 |
-
{
|
| 1532 |
-
"epoch": 1.3627192164636839,
|
| 1533 |
-
"grad_norm": 0.05332406610250473,
|
| 1534 |
-
"learning_rate": 2.3591875430982752e-05,
|
| 1535 |
-
"loss": 0.029861843585968016,
|
| 1536 |
-
"step": 4340
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 1.3690000686968213,
|
| 1540 |
-
"grad_norm": 0.05614598095417023,
|
| 1541 |
-
"learning_rate": 2.3404365802535413e-05,
|
| 1542 |
-
"loss": 0.03531986773014069,
|
| 1543 |
-
"step": 4360
|
| 1544 |
-
},
|
| 1545 |
-
{
|
| 1546 |
-
"epoch": 1.3752809209299586,
|
| 1547 |
-
"grad_norm": 0.33550721406936646,
|
| 1548 |
-
"learning_rate": 2.3216946262867396e-05,
|
| 1549 |
-
"loss": 0.031084492802619934,
|
| 1550 |
-
"step": 4380
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.381561773163096,
|
| 1554 |
-
"grad_norm": 0.3401692807674408,
|
| 1555 |
-
"learning_rate": 2.3029627393600484e-05,
|
| 1556 |
-
"loss": 0.02833726108074188,
|
| 1557 |
-
"step": 4400
|
| 1558 |
-
},
|
| 1559 |
-
{
|
| 1560 |
-
"epoch": 1.3878426253962335,
|
| 1561 |
-
"grad_norm": 0.15662887692451477,
|
| 1562 |
-
"learning_rate": 2.2842419770672624e-05,
|
| 1563 |
-
"loss": 0.03976942598819733,
|
| 1564 |
-
"step": 4420
|
| 1565 |
-
},
|
| 1566 |
-
{
|
| 1567 |
-
"epoch": 1.3941234776293707,
|
| 1568 |
-
"grad_norm": 0.0530722439289093,
|
| 1569 |
-
"learning_rate": 2.265533396374085e-05,
|
| 1570 |
-
"loss": 0.03581877648830414,
|
| 1571 |
-
"step": 4440
|
| 1572 |
-
},
|
| 1573 |
-
{
|
| 1574 |
-
"epoch": 1.4004043298625082,
|
| 1575 |
-
"grad_norm": 0.09495271742343903,
|
| 1576 |
-
"learning_rate": 2.246838053558455e-05,
|
| 1577 |
-
"loss": 0.02298358678817749,
|
| 1578 |
-
"step": 4460
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 1.4066851820956456,
|
| 1582 |
-
"grad_norm": 0.04610360413789749,
|
| 1583 |
-
"learning_rate": 2.2281570041509024e-05,
|
| 1584 |
-
"loss": 0.04205411374568939,
|
| 1585 |
-
"step": 4480
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"epoch": 1.4129660343287829,
|
| 1589 |
-
"grad_norm": 0.07970083504915237,
|
| 1590 |
-
"learning_rate": 2.209491302874961e-05,
|
| 1591 |
-
"loss": 0.028789132833480835,
|
| 1592 |
-
"step": 4500
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.4192468865619203,
|
| 1596 |
-
"grad_norm": 1.6510498523712158,
|
| 1597 |
-
"learning_rate": 2.1908420035876148e-05,
|
| 1598 |
-
"loss": 0.030281484127044678,
|
| 1599 |
-
"step": 4520
|
| 1600 |
-
},
|
| 1601 |
-
{
|
| 1602 |
-
"epoch": 1.4255277387950578,
|
| 1603 |
-
"grad_norm": 0.061110883951187134,
|
| 1604 |
-
"learning_rate": 2.1722101592197995e-05,
|
| 1605 |
-
"loss": 0.026464346051216125,
|
| 1606 |
-
"step": 4540
|
| 1607 |
-
},
|
| 1608 |
-
{
|
| 1609 |
-
"epoch": 1.431808591028195,
|
| 1610 |
-
"grad_norm": 0.10222575813531876,
|
| 1611 |
-
"learning_rate": 2.1535968217169535e-05,
|
| 1612 |
-
"loss": 0.024321816861629486,
|
| 1613 |
-
"step": 4560
|
| 1614 |
-
},
|
| 1615 |
-
{
|
| 1616 |
-
"epoch": 1.4380894432613325,
|
| 1617 |
-
"grad_norm": 0.13632528483867645,
|
| 1618 |
-
"learning_rate": 2.135003041979625e-05,
|
| 1619 |
-
"loss": 0.022314466536045074,
|
| 1620 |
-
"step": 4580
|
| 1621 |
-
},
|
| 1622 |
-
{
|
| 1623 |
-
"epoch": 1.44437029549447,
|
| 1624 |
-
"grad_norm": 0.3248070478439331,
|
| 1625 |
-
"learning_rate": 2.1164298698041432e-05,
|
| 1626 |
-
"loss": 0.05798411965370178,
|
| 1627 |
-
"step": 4600
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"epoch": 1.4506511477276072,
|
| 1631 |
-
"grad_norm": 0.12926240265369415,
|
| 1632 |
-
"learning_rate": 2.0978783538233405e-05,
|
| 1633 |
-
"loss": 0.02899252772331238,
|
| 1634 |
-
"step": 4620
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 1.4569319999607446,
|
| 1638 |
-
"grad_norm": 0.23640146851539612,
|
| 1639 |
-
"learning_rate": 2.0793495414473506e-05,
|
| 1640 |
-
"loss": 0.020399539172649382,
|
| 1641 |
-
"step": 4640
|
| 1642 |
-
},
|
| 1643 |
-
{
|
| 1644 |
-
"epoch": 1.463212852193882,
|
| 1645 |
-
"grad_norm": 0.07195625454187393,
|
| 1646 |
-
"learning_rate": 2.0608444788044744e-05,
|
| 1647 |
-
"loss": 0.02832261621952057,
|
| 1648 |
-
"step": 4660
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 1.4694937044270193,
|
| 1652 |
-
"grad_norm": 0.05917816609144211,
|
| 1653 |
-
"learning_rate": 2.0423642106821118e-05,
|
| 1654 |
-
"loss": 0.017405569553375244,
|
| 1655 |
-
"step": 4680
|
| 1656 |
-
},
|
| 1657 |
-
{
|
| 1658 |
-
"epoch": 1.4757745566601568,
|
| 1659 |
-
"grad_norm": 0.050064459443092346,
|
| 1660 |
-
"learning_rate": 2.0239097804677755e-05,
|
| 1661 |
-
"loss": 0.03790850043296814,
|
| 1662 |
-
"step": 4700
|
| 1663 |
-
},
|
| 1664 |
-
{
|
| 1665 |
-
"epoch": 1.4820554088932942,
|
| 1666 |
-
"grad_norm": 0.06861293315887451,
|
| 1667 |
-
"learning_rate": 2.005482230090181e-05,
|
| 1668 |
-
"loss": 0.02318151295185089,
|
| 1669 |
-
"step": 4720
|
| 1670 |
-
},
|
| 1671 |
-
{
|
| 1672 |
-
"epoch": 1.4883362611264315,
|
| 1673 |
-
"grad_norm": 0.5692355036735535,
|
| 1674 |
-
"learning_rate": 1.987082599960423e-05,
|
| 1675 |
-
"loss": 0.05805472135543823,
|
| 1676 |
-
"step": 4740
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.494617113359569,
|
| 1680 |
-
"grad_norm": 0.10575691610574722,
|
| 1681 |
-
"learning_rate": 1.9687119289132286e-05,
|
| 1682 |
-
"loss": 0.025386866927146912,
|
| 1683 |
-
"step": 4760
|
| 1684 |
-
},
|
| 1685 |
-
{
|
| 1686 |
-
"epoch": 1.5008979655927064,
|
| 1687 |
-
"grad_norm": 0.07561904937028885,
|
| 1688 |
-
"learning_rate": 1.9503712541483105e-05,
|
| 1689 |
-
"loss": 0.02556038498878479,
|
| 1690 |
-
"step": 4780
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 1.5071788178258436,
|
| 1694 |
-
"grad_norm": 0.06342057138681412,
|
| 1695 |
-
"learning_rate": 1.9320616111718055e-05,
|
| 1696 |
-
"loss": 0.026987478137016296,
|
| 1697 |
-
"step": 4800
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 1.513459670058981,
|
| 1701 |
-
"grad_norm": 0.22881099581718445,
|
| 1702 |
-
"learning_rate": 1.913784033737807e-05,
|
| 1703 |
-
"loss": 0.023304201662540436,
|
| 1704 |
-
"step": 4820
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 1.5197405222921185,
|
| 1708 |
-
"grad_norm": 0.34815022349357605,
|
| 1709 |
-
"learning_rate": 1.8955395537900045e-05,
|
| 1710 |
-
"loss": 0.023628510534763336,
|
| 1711 |
-
"step": 4840
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 1.5260213745252558,
|
| 1715 |
-
"grad_norm": 0.07004115730524063,
|
| 1716 |
-
"learning_rate": 1.877329201403417e-05,
|
| 1717 |
-
"loss": 0.01844399720430374,
|
| 1718 |
-
"step": 4860
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 1.5323022267583932,
|
| 1722 |
-
"grad_norm": 0.5382964015007019,
|
| 1723 |
-
"learning_rate": 1.8591540047262408e-05,
|
| 1724 |
-
"loss": 0.022429858148097993,
|
| 1725 |
-
"step": 4880
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 1.5385830789915307,
|
| 1729 |
-
"grad_norm": 0.10155873745679855,
|
| 1730 |
-
"learning_rate": 1.841014989921792e-05,
|
| 1731 |
-
"loss": 0.019565099477767946,
|
| 1732 |
-
"step": 4900
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 1.544863931224668,
|
| 1736 |
-
"grad_norm": 0.45096078515052795,
|
| 1737 |
-
"learning_rate": 1.8229131811105773e-05,
|
| 1738 |
-
"loss": 0.02386317253112793,
|
| 1739 |
-
"step": 4920
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 1.5511447834578054,
|
| 1743 |
-
"grad_norm": 1.6493111848831177,
|
| 1744 |
-
"learning_rate": 1.8048496003124692e-05,
|
| 1745 |
-
"loss": 0.022837825119495392,
|
| 1746 |
-
"step": 4940
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 1.5574256356909428,
|
| 1750 |
-
"grad_norm": 0.2704576849937439,
|
| 1751 |
-
"learning_rate": 1.786825267389005e-05,
|
| 1752 |
-
"loss": 0.02158147543668747,
|
| 1753 |
-
"step": 4960
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 1.56370648792408,
|
| 1757 |
-
"grad_norm": 0.14172030985355377,
|
| 1758 |
-
"learning_rate": 1.768841199985804e-05,
|
| 1759 |
-
"loss": 0.02343100756406784,
|
| 1760 |
-
"step": 4980
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 1.5699873401572177,
|
| 1764 |
-
"grad_norm": 0.3839413523674011,
|
| 1765 |
-
"learning_rate": 1.7508984134751115e-05,
|
| 1766 |
-
"loss": 0.0211731418967247,
|
| 1767 |
-
"step": 5000
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 1.576268192390355,
|
| 1771 |
-
"grad_norm": 0.17986074090003967,
|
| 1772 |
-
"learning_rate": 1.7329979208984757e-05,
|
| 1773 |
-
"loss": 0.050633370876312256,
|
| 1774 |
-
"step": 5020
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 1.5825490446234922,
|
| 1778 |
-
"grad_norm": 0.11192335188388824,
|
| 1779 |
-
"learning_rate": 1.715140732909545e-05,
|
| 1780 |
-
"loss": 0.019000351428985596,
|
| 1781 |
-
"step": 5040
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 1.58882989685663,
|
| 1785 |
-
"grad_norm": 0.06556563079357147,
|
| 1786 |
-
"learning_rate": 1.697327857717012e-05,
|
| 1787 |
-
"loss": 0.031407642364501956,
|
| 1788 |
-
"step": 5060
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 1.5951107490897671,
|
| 1792 |
-
"grad_norm": 0.5898942351341248,
|
| 1793 |
-
"learning_rate": 1.67956030102769e-05,
|
| 1794 |
-
"loss": 0.03214563131332397,
|
| 1795 |
-
"step": 5080
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 1.6013916013229044,
|
| 1799 |
-
"grad_norm": 0.11632583290338516,
|
| 1800 |
-
"learning_rate": 1.6618390659897297e-05,
|
| 1801 |
-
"loss": 0.12849061489105223,
|
| 1802 |
-
"step": 5100
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 1.607672453556042,
|
| 1806 |
-
"grad_norm": 0.26994606852531433,
|
| 1807 |
-
"learning_rate": 1.64416515313598e-05,
|
| 1808 |
-
"loss": 0.01800369322299957,
|
| 1809 |
-
"step": 5120
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 1.6139533057891793,
|
| 1813 |
-
"grad_norm": 0.3220551908016205,
|
| 1814 |
-
"learning_rate": 1.626539560327505e-05,
|
| 1815 |
-
"loss": 0.022022050619125367,
|
| 1816 |
-
"step": 5140
|
| 1817 |
-
},
|
| 1818 |
-
{
|
| 1819 |
-
"epoch": 1.6202341580223165,
|
| 1820 |
-
"grad_norm": 0.16517284512519836,
|
| 1821 |
-
"learning_rate": 1.6089632826972395e-05,
|
| 1822 |
-
"loss": 0.015372823178768157,
|
| 1823 |
-
"step": 5160
|
| 1824 |
-
},
|
| 1825 |
-
{
|
| 1826 |
-
"epoch": 1.6265150102554542,
|
| 1827 |
-
"grad_norm": 0.03851303830742836,
|
| 1828 |
-
"learning_rate": 1.591437312593804e-05,
|
| 1829 |
-
"loss": 0.0164533868432045,
|
| 1830 |
-
"step": 5180
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 1.6327958624885914,
|
| 1834 |
-
"grad_norm": 0.3491153419017792,
|
| 1835 |
-
"learning_rate": 1.5739626395254798e-05,
|
| 1836 |
-
"loss": 0.031122979521751405,
|
| 1837 |
-
"step": 5200
|
| 1838 |
-
},
|
| 1839 |
-
{
|
| 1840 |
-
"epoch": 1.6390767147217287,
|
| 1841 |
-
"grad_norm": 0.06409186869859695,
|
| 1842 |
-
"learning_rate": 1.556540250104343e-05,
|
| 1843 |
-
"loss": 0.0328720360994339,
|
| 1844 |
-
"step": 5220
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 1.6453575669548663,
|
| 1848 |
-
"grad_norm": 0.04858129099011421,
|
| 1849 |
-
"learning_rate": 1.539171127990557e-05,
|
| 1850 |
-
"loss": 0.05794671773910522,
|
| 1851 |
-
"step": 5240
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"epoch": 1.6516384191880036,
|
| 1855 |
-
"grad_norm": 0.262850821018219,
|
| 1856 |
-
"learning_rate": 1.5218562538368364e-05,
|
| 1857 |
-
"loss": 0.014299227297306061,
|
| 1858 |
-
"step": 5260
|
| 1859 |
-
},
|
| 1860 |
-
{
|
| 1861 |
-
"epoch": 1.6579192714211408,
|
| 1862 |
-
"grad_norm": 0.12737710773944855,
|
| 1863 |
-
"learning_rate": 1.504596605233084e-05,
|
| 1864 |
-
"loss": 0.02150038182735443,
|
| 1865 |
-
"step": 5280
|
| 1866 |
-
},
|
| 1867 |
-
{
|
| 1868 |
-
"epoch": 1.6642001236542785,
|
| 1869 |
-
"grad_norm": 0.06830194592475891,
|
| 1870 |
-
"learning_rate": 1.4873931566511901e-05,
|
| 1871 |
-
"loss": 0.01579444855451584,
|
| 1872 |
-
"step": 5300
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 1.6704809758874157,
|
| 1876 |
-
"grad_norm": 0.08396535366773605,
|
| 1877 |
-
"learning_rate": 1.4702468793900188e-05,
|
| 1878 |
-
"loss": 0.027909812331199647,
|
| 1879 |
-
"step": 5320
|
| 1880 |
-
},
|
| 1881 |
-
{
|
| 1882 |
-
"epoch": 1.676761828120553,
|
| 1883 |
-
"grad_norm": 0.26403504610061646,
|
| 1884 |
-
"learning_rate": 1.4531587415205667e-05,
|
| 1885 |
-
"loss": 0.02287571579217911,
|
| 1886 |
-
"step": 5340
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 1.6830426803536906,
|
| 1890 |
-
"grad_norm": 0.43412601947784424,
|
| 1891 |
-
"learning_rate": 1.436129707831308e-05,
|
| 1892 |
-
"loss": 0.017245300114154816,
|
| 1893 |
-
"step": 5360
|
| 1894 |
-
},
|
| 1895 |
-
{
|
| 1896 |
-
"epoch": 1.6893235325868279,
|
| 1897 |
-
"grad_norm": 0.18354500830173492,
|
| 1898 |
-
"learning_rate": 1.4191607397737206e-05,
|
| 1899 |
-
"loss": 0.031501269340515135,
|
| 1900 |
-
"step": 5380
|
| 1901 |
-
},
|
| 1902 |
-
{
|
| 1903 |
-
"epoch": 1.6956043848199651,
|
| 1904 |
-
"grad_norm": 0.04084882140159607,
|
| 1905 |
-
"learning_rate": 1.402252795408004e-05,
|
| 1906 |
-
"loss": 0.036852356791496274,
|
| 1907 |
-
"step": 5400
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 1.7018852370531028,
|
| 1911 |
-
"grad_norm": 0.06115623563528061,
|
| 1912 |
-
"learning_rate": 1.3854068293489909e-05,
|
| 1913 |
-
"loss": 0.028507155179977418,
|
| 1914 |
-
"step": 5420
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 1.70816608928624,
|
| 1918 |
-
"grad_norm": 0.32743072509765625,
|
| 1919 |
-
"learning_rate": 1.368623792712243e-05,
|
| 1920 |
-
"loss": 0.014020594954490661,
|
| 1921 |
-
"step": 5440
|
| 1922 |
-
},
|
| 1923 |
-
{
|
| 1924 |
-
"epoch": 1.7144469415193773,
|
| 1925 |
-
"grad_norm": 0.06222687289118767,
|
| 1926 |
-
"learning_rate": 1.3519046330603584e-05,
|
| 1927 |
-
"loss": 0.05104716420173645,
|
| 1928 |
-
"step": 5460
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 1.720727793752515,
|
| 1932 |
-
"grad_norm": 0.06253579258918762,
|
| 1933 |
-
"learning_rate": 1.3352502943494672e-05,
|
| 1934 |
-
"loss": 0.026476505398750304,
|
| 1935 |
-
"step": 5480
|
| 1936 |
-
},
|
| 1937 |
-
{
|
| 1938 |
-
"epoch": 1.7270086459856522,
|
| 1939 |
-
"grad_norm": 0.4662589430809021,
|
| 1940 |
-
"learning_rate": 1.318661716875942e-05,
|
| 1941 |
-
"loss": 0.01999303698539734,
|
| 1942 |
-
"step": 5500
|
| 1943 |
-
},
|
| 1944 |
-
{
|
| 1945 |
-
"epoch": 1.7332894982187894,
|
| 1946 |
-
"grad_norm": 0.1247578039765358,
|
| 1947 |
-
"learning_rate": 1.3021398372233035e-05,
|
| 1948 |
-
"loss": 0.02069215327501297,
|
| 1949 |
-
"step": 5520
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 1.739570350451927,
|
| 1953 |
-
"grad_norm": 0.3788617253303528,
|
| 1954 |
-
"learning_rate": 1.2856855882093432e-05,
|
| 1955 |
-
"loss": 0.018105532228946685,
|
| 1956 |
-
"step": 5540
|
| 1957 |
-
},
|
| 1958 |
-
{
|
| 1959 |
-
"epoch": 1.7458512026850643,
|
| 1960 |
-
"grad_norm": 1.8883055448532104,
|
| 1961 |
-
"learning_rate": 1.2692998988334612e-05,
|
| 1962 |
-
"loss": 0.04176854789257049,
|
| 1963 |
-
"step": 5560
|
| 1964 |
-
},
|
| 1965 |
-
{
|
| 1966 |
-
"epoch": 1.7521320549182016,
|
| 1967 |
-
"grad_norm": 0.21989472210407257,
|
| 1968 |
-
"learning_rate": 1.252983694224208e-05,
|
| 1969 |
-
"loss": 0.020176875591278075,
|
| 1970 |
-
"step": 5580
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 1.7584129071513392,
|
| 1974 |
-
"grad_norm": 0.16766223311424255,
|
| 1975 |
-
"learning_rate": 1.236737895587056e-05,
|
| 1976 |
-
"loss": 0.020288291573524474,
|
| 1977 |
-
"step": 5600
|
| 1978 |
-
},
|
| 1979 |
-
{
|
| 1980 |
-
"epoch": 1.7646937593844765,
|
| 1981 |
-
"grad_norm": 0.09795085340738297,
|
| 1982 |
-
"learning_rate": 1.2205634201523922e-05,
|
| 1983 |
-
"loss": 0.015002159774303437,
|
| 1984 |
-
"step": 5620
|
| 1985 |
-
},
|
| 1986 |
-
{
|
| 1987 |
-
"epoch": 1.7709746116176137,
|
| 1988 |
-
"grad_norm": 0.029697278514504433,
|
| 1989 |
-
"learning_rate": 1.2044611811237242e-05,
|
| 1990 |
-
"loss": 0.023318004608154298,
|
| 1991 |
-
"step": 5640
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 1.7772554638507514,
|
| 1995 |
-
"grad_norm": 0.3376557528972626,
|
| 1996 |
-
"learning_rate": 1.1884320876261268e-05,
|
| 1997 |
-
"loss": 0.0196805402636528,
|
| 1998 |
-
"step": 5660
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 1.7835363160838886,
|
| 2002 |
-
"grad_norm": 0.06531202793121338,
|
| 2003 |
-
"learning_rate": 1.17247704465491e-05,
|
| 2004 |
-
"loss": 0.020646215975284578,
|
| 2005 |
-
"step": 5680
|
| 2006 |
-
},
|
| 2007 |
-
{
|
| 2008 |
-
"epoch": 1.7898171683170259,
|
| 2009 |
-
"grad_norm": 0.36572521924972534,
|
| 2010 |
-
"learning_rate": 1.1565969530245277e-05,
|
| 2011 |
-
"loss": 0.025085961818695067,
|
| 2012 |
-
"step": 5700
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 1.7960980205501635,
|
| 2016 |
-
"grad_norm": 0.1814069002866745,
|
| 2017 |
-
"learning_rate": 1.140792709317713e-05,
|
| 2018 |
-
"loss": 0.030590158700942994,
|
| 2019 |
-
"step": 5720
|
| 2020 |
-
},
|
| 2021 |
-
{
|
| 2022 |
-
"epoch": 1.8023788727833008,
|
| 2023 |
-
"grad_norm": 0.24981825053691864,
|
| 2024 |
-
"learning_rate": 1.12506520583486e-05,
|
| 2025 |
-
"loss": 0.0215068444609642,
|
| 2026 |
-
"step": 5740
|
| 2027 |
-
},
|
| 2028 |
-
{
|
| 2029 |
-
"epoch": 1.808659725016438,
|
| 2030 |
-
"grad_norm": 0.35276567935943604,
|
| 2031 |
-
"learning_rate": 1.1094153305436453e-05,
|
| 2032 |
-
"loss": 0.01926688849925995,
|
| 2033 |
-
"step": 5760
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 1.8149405772495757,
|
| 2037 |
-
"grad_norm": 0.02407790906727314,
|
| 2038 |
-
"learning_rate": 1.0938439670288928e-05,
|
| 2039 |
-
"loss": 0.013058166205883025,
|
| 2040 |
-
"step": 5780
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 1.821221429482713,
|
| 2044 |
-
"grad_norm": 0.08507027477025986,
|
| 2045 |
-
"learning_rate": 1.0783519944426874e-05,
|
| 2046 |
-
"loss": 0.01877378821372986,
|
| 2047 |
-
"step": 5800
|
| 2048 |
-
},
|
| 2049 |
-
{
|
| 2050 |
-
"epoch": 1.8275022817158502,
|
| 2051 |
-
"grad_norm": 0.18674074113368988,
|
| 2052 |
-
"learning_rate": 1.0629402874547378e-05,
|
| 2053 |
-
"loss": 0.018952280282974243,
|
| 2054 |
-
"step": 5820
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 1.8337831339489878,
|
| 2058 |
-
"grad_norm": 0.057133324444293976,
|
| 2059 |
-
"learning_rate": 1.047609716202996e-05,
|
| 2060 |
-
"loss": 0.030013075470924376,
|
| 2061 |
-
"step": 5840
|
| 2062 |
-
},
|
| 2063 |
-
{
|
| 2064 |
-
"epoch": 1.840063986182125,
|
| 2065 |
-
"grad_norm": 0.10284405946731567,
|
| 2066 |
-
"learning_rate": 1.0323611462445246e-05,
|
| 2067 |
-
"loss": 0.020608581602573395,
|
| 2068 |
-
"step": 5860
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 1.8463448384152623,
|
| 2072 |
-
"grad_norm": 0.25261375308036804,
|
| 2073 |
-
"learning_rate": 1.0171954385066303e-05,
|
| 2074 |
-
"loss": 0.02777992784976959,
|
| 2075 |
-
"step": 5880
|
| 2076 |
-
},
|
| 2077 |
-
{
|
| 2078 |
-
"epoch": 1.8526256906484,
|
| 2079 |
-
"grad_norm": 0.38947853446006775,
|
| 2080 |
-
"learning_rate": 1.0021134492382606e-05,
|
| 2081 |
-
"loss": 0.02200171649456024,
|
| 2082 |
-
"step": 5900
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 1.8589065428815372,
|
| 2086 |
-
"grad_norm": 0.1173999160528183,
|
| 2087 |
-
"learning_rate": 9.87116029961652e-06,
|
| 2088 |
-
"loss": 0.019530104100704195,
|
| 2089 |
-
"step": 5920
|
| 2090 |
-
},
|
| 2091 |
-
{
|
| 2092 |
-
"epoch": 1.8651873951146745,
|
| 2093 |
-
"grad_norm": 2.3180384635925293,
|
| 2094 |
-
"learning_rate": 9.722040274242594e-06,
|
| 2095 |
-
"loss": 0.028142544627189636,
|
| 2096 |
-
"step": 5940
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 1.8714682473478121,
|
| 2100 |
-
"grad_norm": 0.4677383601665497,
|
| 2101 |
-
"learning_rate": 9.573782835509471e-06,
|
| 2102 |
-
"loss": 0.017200836539268495,
|
| 2103 |
-
"step": 5960
|
| 2104 |
-
},
|
| 2105 |
-
{
|
| 2106 |
-
"epoch": 1.8777490995809494,
|
| 2107 |
-
"grad_norm": 0.19456204771995544,
|
| 2108 |
-
"learning_rate": 9.426396353964565e-06,
|
| 2109 |
-
"loss": 0.015525393187999725,
|
| 2110 |
-
"step": 5980
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 1.8840299518140866,
|
| 2114 |
-
"grad_norm": 0.561476469039917,
|
| 2115 |
-
"learning_rate": 9.279889150981421e-06,
|
| 2116 |
-
"loss": 0.04386698603630066,
|
| 2117 |
-
"step": 6000
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 1.8903108040472243,
|
| 2121 |
-
"grad_norm": 0.0613991916179657,
|
| 2122 |
-
"learning_rate": 9.134269498289922e-06,
|
| 2123 |
-
"loss": 0.031869342923164366,
|
| 2124 |
-
"step": 6020
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 1.8965916562803615,
|
| 2128 |
-
"grad_norm": 0.047774072736501694,
|
| 2129 |
-
"learning_rate": 8.989545617509283e-06,
|
| 2130 |
-
"loss": 0.02780761420726776,
|
| 2131 |
-
"step": 6040
|
| 2132 |
-
},
|
| 2133 |
-
{
|
| 2134 |
-
"epoch": 1.9028725085134988,
|
| 2135 |
-
"grad_norm": 0.33418071269989014,
|
| 2136 |
-
"learning_rate": 8.845725679683822e-06,
|
| 2137 |
-
"loss": 0.04635312259197235,
|
| 2138 |
-
"step": 6060
|
| 2139 |
-
}
|
| 2140 |
-
],
|
| 2141 |
-
"logging_steps": 20,
|
| 2142 |
-
"max_steps": 8362,
|
| 2143 |
-
"num_input_tokens_seen": 0,
|
| 2144 |
-
"num_train_epochs": 3,
|
| 2145 |
-
"save_steps": 1000000000,
|
| 2146 |
-
"stateful_callbacks": {
|
| 2147 |
-
"TrainerControl": {
|
| 2148 |
-
"args": {
|
| 2149 |
-
"should_epoch_stop": false,
|
| 2150 |
-
"should_evaluate": false,
|
| 2151 |
-
"should_log": false,
|
| 2152 |
-
"should_save": true,
|
| 2153 |
-
"should_training_stop": false
|
| 2154 |
-
},
|
| 2155 |
-
"attributes": {}
|
| 2156 |
-
}
|
| 2157 |
-
},
|
| 2158 |
-
"total_flos": 1.920950799922944e+16,
|
| 2159 |
-
"train_batch_size": 1,
|
| 2160 |
-
"trial_name": null,
|
| 2161 |
-
"trial_params": null
|
| 2162 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
slots/7/checkpoint-6622/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/7/checkpoint-6622/trainer_state.json
DELETED
|
@@ -1,2358 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_global_step": null,
|
| 3 |
-
"best_metric": null,
|
| 4 |
-
"best_model_checkpoint": null,
|
| 5 |
-
"epoch": 2.079138738137531,
|
| 6 |
-
"eval_steps": 500,
|
| 7 |
-
"global_step": 6622,
|
| 8 |
-
"is_hyper_param_search": false,
|
| 9 |
-
"is_local_process_zero": true,
|
| 10 |
-
"is_world_process_zero": true,
|
| 11 |
-
"log_history": [
|
| 12 |
-
{
|
| 13 |
-
"epoch": 0.0003140426116568692,
|
| 14 |
-
"grad_norm": 11.236513137817383,
|
| 15 |
-
"learning_rate": 5e-05,
|
| 16 |
-
"loss": 14.929291725158691,
|
| 17 |
-
"step": 1
|
| 18 |
-
},
|
| 19 |
-
{
|
| 20 |
-
"epoch": 0.0062808522331373835,
|
| 21 |
-
"grad_norm": 8.230694770812988,
|
| 22 |
-
"learning_rate": 4.9999363065476e-05,
|
| 23 |
-
"loss": 13.227896439401727,
|
| 24 |
-
"step": 20
|
| 25 |
-
},
|
| 26 |
-
{
|
| 27 |
-
"epoch": 0.012561704466274767,
|
| 28 |
-
"grad_norm": 15.541277885437012,
|
| 29 |
-
"learning_rate": 4.999745229435885e-05,
|
| 30 |
-
"loss": 11.76095199584961,
|
| 31 |
-
"step": 40
|
| 32 |
-
},
|
| 33 |
-
{
|
| 34 |
-
"epoch": 0.018842556699412153,
|
| 35 |
-
"grad_norm": 19.88605308532715,
|
| 36 |
-
"learning_rate": 4.999406489735261e-05,
|
| 37 |
-
"loss": 10.630082702636718,
|
| 38 |
-
"step": 60
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.025123408932549534,
|
| 42 |
-
"grad_norm": 21.772388458251953,
|
| 43 |
-
"learning_rate": 4.998953978539926e-05,
|
| 44 |
-
"loss": 10.612107086181641,
|
| 45 |
-
"step": 80
|
| 46 |
-
},
|
| 47 |
-
{
|
| 48 |
-
"epoch": 0.03140426116568692,
|
| 49 |
-
"grad_norm": 24.13600730895996,
|
| 50 |
-
"learning_rate": 4.998374133739565e-05,
|
| 51 |
-
"loss": 8.763066864013672,
|
| 52 |
-
"step": 100
|
| 53 |
-
},
|
| 54 |
-
{
|
| 55 |
-
"epoch": 0.037685113398824306,
|
| 56 |
-
"grad_norm": 33.56028747558594,
|
| 57 |
-
"learning_rate": 4.997666984880032e-05,
|
| 58 |
-
"loss": 11.96340103149414,
|
| 59 |
-
"step": 120
|
| 60 |
-
},
|
| 61 |
-
{
|
| 62 |
-
"epoch": 0.04396596563196169,
|
| 63 |
-
"grad_norm": 51.29957962036133,
|
| 64 |
-
"learning_rate": 4.996785126662353e-05,
|
| 65 |
-
"loss": 7.662068939208984,
|
| 66 |
-
"step": 140
|
| 67 |
-
},
|
| 68 |
-
{
|
| 69 |
-
"epoch": 0.05024681786509907,
|
| 70 |
-
"grad_norm": 34.75215530395508,
|
| 71 |
-
"learning_rate": 4.9957623010947197e-05,
|
| 72 |
-
"loss": 7.163829040527344,
|
| 73 |
-
"step": 160
|
| 74 |
-
},
|
| 75 |
-
{
|
| 76 |
-
"epoch": 0.056527670098236456,
|
| 77 |
-
"grad_norm": 48.993743896484375,
|
| 78 |
-
"learning_rate": 4.994598565925399e-05,
|
| 79 |
-
"loss": 6.738508605957032,
|
| 80 |
-
"step": 180
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.06280852233137384,
|
| 84 |
-
"grad_norm": 37.503150939941406,
|
| 85 |
-
"learning_rate": 4.9932939868583476e-05,
|
| 86 |
-
"loss": 5.927839660644532,
|
| 87 |
-
"step": 200
|
| 88 |
-
},
|
| 89 |
-
{
|
| 90 |
-
"epoch": 0.06908937456451122,
|
| 91 |
-
"grad_norm": 34.306514739990234,
|
| 92 |
-
"learning_rate": 4.991848637549505e-05,
|
| 93 |
-
"loss": 5.548626327514649,
|
| 94 |
-
"step": 220
|
| 95 |
-
},
|
| 96 |
-
{
|
| 97 |
-
"epoch": 0.07537022679764861,
|
| 98 |
-
"grad_norm": 25.989212036132812,
|
| 99 |
-
"learning_rate": 4.990345241504926e-05,
|
| 100 |
-
"loss": 5.030820083618164,
|
| 101 |
-
"step": 240
|
| 102 |
-
},
|
| 103 |
-
{
|
| 104 |
-
"epoch": 0.08165107903078599,
|
| 105 |
-
"grad_norm": 23.606063842773438,
|
| 106 |
-
"learning_rate": 4.98862563217225e-05,
|
| 107 |
-
"loss": 4.923490524291992,
|
| 108 |
-
"step": 260
|
| 109 |
-
},
|
| 110 |
-
{
|
| 111 |
-
"epoch": 0.08793193126392337,
|
| 112 |
-
"grad_norm": 36.173194885253906,
|
| 113 |
-
"learning_rate": 4.9867655161709595e-05,
|
| 114 |
-
"loss": 4.673915100097656,
|
| 115 |
-
"step": 280
|
| 116 |
-
},
|
| 117 |
-
{
|
| 118 |
-
"epoch": 0.09421278349706076,
|
| 119 |
-
"grad_norm": 18.624378204345703,
|
| 120 |
-
"learning_rate": 4.984764998522357e-05,
|
| 121 |
-
"loss": 3.8839893341064453,
|
| 122 |
-
"step": 300
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.10049363573019814,
|
| 126 |
-
"grad_norm": 22.662349700927734,
|
| 127 |
-
"learning_rate": 4.9826241921747555e-05,
|
| 128 |
-
"loss": 3.063496971130371,
|
| 129 |
-
"step": 320
|
| 130 |
-
},
|
| 131 |
-
{
|
| 132 |
-
"epoch": 0.10677448796333552,
|
| 133 |
-
"grad_norm": 25.49854278564453,
|
| 134 |
-
"learning_rate": 4.9803432179971055e-05,
|
| 135 |
-
"loss": 3.0309188842773436,
|
| 136 |
-
"step": 340
|
| 137 |
-
},
|
| 138 |
-
{
|
| 139 |
-
"epoch": 0.11305534019647291,
|
| 140 |
-
"grad_norm": 15.640332221984863,
|
| 141 |
-
"learning_rate": 4.9779222047721693e-05,
|
| 142 |
-
"loss": 2.6161094665527345,
|
| 143 |
-
"step": 360
|
| 144 |
-
},
|
| 145 |
-
{
|
| 146 |
-
"epoch": 0.1193361924296103,
|
| 147 |
-
"grad_norm": 13.35746955871582,
|
| 148 |
-
"learning_rate": 4.9753612891892514e-05,
|
| 149 |
-
"loss": 2.3498945236206055,
|
| 150 |
-
"step": 380
|
| 151 |
-
},
|
| 152 |
-
{
|
| 153 |
-
"epoch": 0.12561704466274767,
|
| 154 |
-
"grad_norm": 10.242664337158203,
|
| 155 |
-
"learning_rate": 4.9726606158364765e-05,
|
| 156 |
-
"loss": 2.069006156921387,
|
| 157 |
-
"step": 400
|
| 158 |
-
},
|
| 159 |
-
{
|
| 160 |
-
"epoch": 0.13189789689588505,
|
| 161 |
-
"grad_norm": 8.599258422851562,
|
| 162 |
-
"learning_rate": 4.969965664306835e-05,
|
| 163 |
-
"loss": 1.7923540115356444,
|
| 164 |
-
"step": 420
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.13817874912902245,
|
| 168 |
-
"grad_norm": 5.654482364654541,
|
| 169 |
-
"learning_rate": 4.966992909049121e-05,
|
| 170 |
-
"loss": 1.7063348770141602,
|
| 171 |
-
"step": 440
|
| 172 |
-
},
|
| 173 |
-
{
|
| 174 |
-
"epoch": 0.14445960136215982,
|
| 175 |
-
"grad_norm": 9.39411449432373,
|
| 176 |
-
"learning_rate": 4.9638808684964765e-05,
|
| 177 |
-
"loss": 1.4251695632934571,
|
| 178 |
-
"step": 460
|
| 179 |
-
},
|
| 180 |
-
{
|
| 181 |
-
"epoch": 0.15074045359529722,
|
| 182 |
-
"grad_norm": 4.354184627532959,
|
| 183 |
-
"learning_rate": 4.960629718353295e-05,
|
| 184 |
-
"loss": 1.2388976097106934,
|
| 185 |
-
"step": 480
|
| 186 |
-
},
|
| 187 |
-
{
|
| 188 |
-
"epoch": 0.1570213058284346,
|
| 189 |
-
"grad_norm": 4.586623191833496,
|
| 190 |
-
"learning_rate": 4.9572396421780296e-05,
|
| 191 |
-
"loss": 1.2111310958862305,
|
| 192 |
-
"step": 500
|
| 193 |
-
},
|
| 194 |
-
{
|
| 195 |
-
"epoch": 0.16330215806157197,
|
| 196 |
-
"grad_norm": 3.979524850845337,
|
| 197 |
-
"learning_rate": 4.9537108313728365e-05,
|
| 198 |
-
"loss": 1.4075250625610352,
|
| 199 |
-
"step": 520
|
| 200 |
-
},
|
| 201 |
-
{
|
| 202 |
-
"epoch": 0.16958301029470937,
|
| 203 |
-
"grad_norm": 3.687380790710449,
|
| 204 |
-
"learning_rate": 4.9500434851727635e-05,
|
| 205 |
-
"loss": 0.9275534629821778,
|
| 206 |
-
"step": 540
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.17586386252784675,
|
| 210 |
-
"grad_norm": 2.872728109359741,
|
| 211 |
-
"learning_rate": 4.946431376374067e-05,
|
| 212 |
-
"loss": 1.418515110015869,
|
| 213 |
-
"step": 560
|
| 214 |
-
},
|
| 215 |
-
{
|
| 216 |
-
"epoch": 0.18214471476098412,
|
| 217 |
-
"grad_norm": 2.1888179779052734,
|
| 218 |
-
"learning_rate": 4.9424944888143395e-05,
|
| 219 |
-
"loss": 0.7708720684051513,
|
| 220 |
-
"step": 580
|
| 221 |
-
},
|
| 222 |
-
{
|
| 223 |
-
"epoch": 0.18842556699412152,
|
| 224 |
-
"grad_norm": 2.96533465385437,
|
| 225 |
-
"learning_rate": 4.9384196991293205e-05,
|
| 226 |
-
"loss": 0.6360139369964599,
|
| 227 |
-
"step": 600
|
| 228 |
-
},
|
| 229 |
-
{
|
| 230 |
-
"epoch": 0.1947064192272589,
|
| 231 |
-
"grad_norm": 3.389042615890503,
|
| 232 |
-
"learning_rate": 4.9342072373797774e-05,
|
| 233 |
-
"loss": 1.2306670188903808,
|
| 234 |
-
"step": 620
|
| 235 |
-
},
|
| 236 |
-
{
|
| 237 |
-
"epoch": 0.20098727146039627,
|
| 238 |
-
"grad_norm": 3.054168939590454,
|
| 239 |
-
"learning_rate": 4.9298573413993806e-05,
|
| 240 |
-
"loss": 0.684608793258667,
|
| 241 |
-
"step": 640
|
| 242 |
-
},
|
| 243 |
-
{
|
| 244 |
-
"epoch": 0.20726812369353367,
|
| 245 |
-
"grad_norm": 2.2155184745788574,
|
| 246 |
-
"learning_rate": 4.925370256781273e-05,
|
| 247 |
-
"loss": 0.6700978755950928,
|
| 248 |
-
"step": 660
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.21354897592667105,
|
| 252 |
-
"grad_norm": 1.223708987236023,
|
| 253 |
-
"learning_rate": 4.920746236864208e-05,
|
| 254 |
-
"loss": 0.707273292541504,
|
| 255 |
-
"step": 680
|
| 256 |
-
},
|
| 257 |
-
{
|
| 258 |
-
"epoch": 0.21982982815980842,
|
| 259 |
-
"grad_norm": 1.1626676321029663,
|
| 260 |
-
"learning_rate": 4.915985542718241e-05,
|
| 261 |
-
"loss": 0.6534492492675781,
|
| 262 |
-
"step": 700
|
| 263 |
-
},
|
| 264 |
-
{
|
| 265 |
-
"epoch": 0.22611068039294582,
|
| 266 |
-
"grad_norm": 1.1415362358093262,
|
| 267 |
-
"learning_rate": 4.9110884431299935e-05,
|
| 268 |
-
"loss": 0.45153131484985354,
|
| 269 |
-
"step": 720
|
| 270 |
-
},
|
| 271 |
-
{
|
| 272 |
-
"epoch": 0.2323915326260832,
|
| 273 |
-
"grad_norm": 1.0272735357284546,
|
| 274 |
-
"learning_rate": 4.906055214587476e-05,
|
| 275 |
-
"loss": 0.46799488067626954,
|
| 276 |
-
"step": 740
|
| 277 |
-
},
|
| 278 |
-
{
|
| 279 |
-
"epoch": 0.2386723848592206,
|
| 280 |
-
"grad_norm": 1.2896318435668945,
|
| 281 |
-
"learning_rate": 4.9008861412644756e-05,
|
| 282 |
-
"loss": 0.4087998390197754,
|
| 283 |
-
"step": 760
|
| 284 |
-
},
|
| 285 |
-
{
|
| 286 |
-
"epoch": 0.24495323709235797,
|
| 287 |
-
"grad_norm": 0.873556911945343,
|
| 288 |
-
"learning_rate": 4.895581515004517e-05,
|
| 289 |
-
"loss": 0.3405194044113159,
|
| 290 |
-
"step": 780
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.25123408932549535,
|
| 294 |
-
"grad_norm": 1.2751230001449585,
|
| 295 |
-
"learning_rate": 4.89014163530438e-05,
|
| 296 |
-
"loss": 0.401216459274292,
|
| 297 |
-
"step": 800
|
| 298 |
-
},
|
| 299 |
-
{
|
| 300 |
-
"epoch": 0.25751494155863275,
|
| 301 |
-
"grad_norm": 0.8582271337509155,
|
| 302 |
-
"learning_rate": 4.8845668092971934e-05,
|
| 303 |
-
"loss": 0.36631605625152586,
|
| 304 |
-
"step": 820
|
| 305 |
-
},
|
| 306 |
-
{
|
| 307 |
-
"epoch": 0.2637957937917701,
|
| 308 |
-
"grad_norm": 0.7967858910560608,
|
| 309 |
-
"learning_rate": 4.87885735173509e-05,
|
| 310 |
-
"loss": 0.33444316387176515,
|
| 311 |
-
"step": 840
|
| 312 |
-
},
|
| 313 |
-
{
|
| 314 |
-
"epoch": 0.2700766460249075,
|
| 315 |
-
"grad_norm": 0.7824229001998901,
|
| 316 |
-
"learning_rate": 4.873013584971442e-05,
|
| 317 |
-
"loss": 0.3244049549102783,
|
| 318 |
-
"step": 860
|
| 319 |
-
},
|
| 320 |
-
{
|
| 321 |
-
"epoch": 0.2763574982580449,
|
| 322 |
-
"grad_norm": 0.774782121181488,
|
| 323 |
-
"learning_rate": 4.867035838942657e-05,
|
| 324 |
-
"loss": 0.6262166023254394,
|
| 325 |
-
"step": 880
|
| 326 |
-
},
|
| 327 |
-
{
|
| 328 |
-
"epoch": 0.2826383504911823,
|
| 329 |
-
"grad_norm": 0.6737077832221985,
|
| 330 |
-
"learning_rate": 4.86092445114955e-05,
|
| 331 |
-
"loss": 0.28058154582977296,
|
| 332 |
-
"step": 900
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.28891920272431965,
|
| 336 |
-
"grad_norm": 1.0165168046951294,
|
| 337 |
-
"learning_rate": 4.8546797666382896e-05,
|
| 338 |
-
"loss": 0.3938837289810181,
|
| 339 |
-
"step": 920
|
| 340 |
-
},
|
| 341 |
-
{
|
| 342 |
-
"epoch": 0.29520005495745705,
|
| 343 |
-
"grad_norm": 1.093992829322815,
|
| 344 |
-
"learning_rate": 4.8483021379809154e-05,
|
| 345 |
-
"loss": 0.3142520904541016,
|
| 346 |
-
"step": 940
|
| 347 |
-
},
|
| 348 |
-
{
|
| 349 |
-
"epoch": 0.30148090719059445,
|
| 350 |
-
"grad_norm": 0.6402747631072998,
|
| 351 |
-
"learning_rate": 4.8417919252554354e-05,
|
| 352 |
-
"loss": 0.29062786102294924,
|
| 353 |
-
"step": 960
|
| 354 |
-
},
|
| 355 |
-
{
|
| 356 |
-
"epoch": 0.3077617594237318,
|
| 357 |
-
"grad_norm": 1.9970321655273438,
|
| 358 |
-
"learning_rate": 4.8351494960254916e-05,
|
| 359 |
-
"loss": 0.31338260173797605,
|
| 360 |
-
"step": 980
|
| 361 |
-
},
|
| 362 |
-
{
|
| 363 |
-
"epoch": 0.3140426116568692,
|
| 364 |
-
"grad_norm": 1.1727873086929321,
|
| 365 |
-
"learning_rate": 4.8283752253196094e-05,
|
| 366 |
-
"loss": 0.2333289623260498,
|
| 367 |
-
"step": 1000
|
| 368 |
-
},
|
| 369 |
-
{
|
| 370 |
-
"epoch": 0.3203234638900066,
|
| 371 |
-
"grad_norm": 0.7795476913452148,
|
| 372 |
-
"learning_rate": 4.8214694956100245e-05,
|
| 373 |
-
"loss": 0.2488691806793213,
|
| 374 |
-
"step": 1020
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.32660431612314395,
|
| 378 |
-
"grad_norm": 1.1675890684127808,
|
| 379 |
-
"learning_rate": 4.814432696791089e-05,
|
| 380 |
-
"loss": 0.22975556850433348,
|
| 381 |
-
"step": 1040
|
| 382 |
-
},
|
| 383 |
-
{
|
| 384 |
-
"epoch": 0.33288516835628135,
|
| 385 |
-
"grad_norm": 0.5809735059738159,
|
| 386 |
-
"learning_rate": 4.807265226157255e-05,
|
| 387 |
-
"loss": 0.2238070011138916,
|
| 388 |
-
"step": 1060
|
| 389 |
-
},
|
| 390 |
-
{
|
| 391 |
-
"epoch": 0.33916602058941875,
|
| 392 |
-
"grad_norm": 0.6172874569892883,
|
| 393 |
-
"learning_rate": 4.7999674883806465e-05,
|
| 394 |
-
"loss": 0.22853121757507325,
|
| 395 |
-
"step": 1080
|
| 396 |
-
},
|
| 397 |
-
{
|
| 398 |
-
"epoch": 0.3454468728225561,
|
| 399 |
-
"grad_norm": 0.9794221520423889,
|
| 400 |
-
"learning_rate": 4.792539895488213e-05,
|
| 401 |
-
"loss": 0.2538387060165405,
|
| 402 |
-
"step": 1100
|
| 403 |
-
},
|
| 404 |
-
{
|
| 405 |
-
"epoch": 0.3517277250556935,
|
| 406 |
-
"grad_norm": 0.6366945505142212,
|
| 407 |
-
"learning_rate": 4.784982866838462e-05,
|
| 408 |
-
"loss": 0.207783842086792,
|
| 409 |
-
"step": 1120
|
| 410 |
-
},
|
| 411 |
-
{
|
| 412 |
-
"epoch": 0.3580085772888309,
|
| 413 |
-
"grad_norm": 0.5059712529182434,
|
| 414 |
-
"learning_rate": 4.777296829097785e-05,
|
| 415 |
-
"loss": 0.2022031545639038,
|
| 416 |
-
"step": 1140
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.36428942952196824,
|
| 420 |
-
"grad_norm": 0.42327556014060974,
|
| 421 |
-
"learning_rate": 4.769482216216369e-05,
|
| 422 |
-
"loss": 0.19924311637878417,
|
| 423 |
-
"step": 1160
|
| 424 |
-
},
|
| 425 |
-
{
|
| 426 |
-
"epoch": 0.37057028175510565,
|
| 427 |
-
"grad_norm": 1.0261644124984741,
|
| 428 |
-
"learning_rate": 4.761539469403695e-05,
|
| 429 |
-
"loss": 0.1949863314628601,
|
| 430 |
-
"step": 1180
|
| 431 |
-
},
|
| 432 |
-
{
|
| 433 |
-
"epoch": 0.37685113398824305,
|
| 434 |
-
"grad_norm": 0.78790283203125,
|
| 435 |
-
"learning_rate": 4.7534690371036274e-05,
|
| 436 |
-
"loss": 0.1740255832672119,
|
| 437 |
-
"step": 1200
|
| 438 |
-
},
|
| 439 |
-
{
|
| 440 |
-
"epoch": 0.3831319862213804,
|
| 441 |
-
"grad_norm": 0.49199551343917847,
|
| 442 |
-
"learning_rate": 4.7452713749690934e-05,
|
| 443 |
-
"loss": 0.16614840030670167,
|
| 444 |
-
"step": 1220
|
| 445 |
-
},
|
| 446 |
-
{
|
| 447 |
-
"epoch": 0.3894128384545178,
|
| 448 |
-
"grad_norm": 0.5265403985977173,
|
| 449 |
-
"learning_rate": 4.73694694583636e-05,
|
| 450 |
-
"loss": 0.17591538429260253,
|
| 451 |
-
"step": 1240
|
| 452 |
-
},
|
| 453 |
-
{
|
| 454 |
-
"epoch": 0.3956936906876552,
|
| 455 |
-
"grad_norm": 0.5221320390701294,
|
| 456 |
-
"learning_rate": 4.728496219698902e-05,
|
| 457 |
-
"loss": 0.18058695793151855,
|
| 458 |
-
"step": 1260
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.40197454292079254,
|
| 462 |
-
"grad_norm": 1.2317312955856323,
|
| 463 |
-
"learning_rate": 4.719919673680863e-05,
|
| 464 |
-
"loss": 0.12510637044906617,
|
| 465 |
-
"step": 1280
|
| 466 |
-
},
|
| 467 |
-
{
|
| 468 |
-
"epoch": 0.40825539515392995,
|
| 469 |
-
"grad_norm": 0.21159659326076508,
|
| 470 |
-
"learning_rate": 4.711217792010122e-05,
|
| 471 |
-
"loss": 0.3407147884368896,
|
| 472 |
-
"step": 1300
|
| 473 |
-
},
|
| 474 |
-
{
|
| 475 |
-
"epoch": 0.41453624738706735,
|
| 476 |
-
"grad_norm": 1.4082627296447754,
|
| 477 |
-
"learning_rate": 4.70239106599095e-05,
|
| 478 |
-
"loss": 0.16848219633102418,
|
| 479 |
-
"step": 1320
|
| 480 |
-
},
|
| 481 |
-
{
|
| 482 |
-
"epoch": 0.4208170996202047,
|
| 483 |
-
"grad_norm": 0.35336634516716003,
|
| 484 |
-
"learning_rate": 4.6934399939762746e-05,
|
| 485 |
-
"loss": 0.13055137395858765,
|
| 486 |
-
"step": 1340
|
| 487 |
-
},
|
| 488 |
-
{
|
| 489 |
-
"epoch": 0.4270979518533421,
|
| 490 |
-
"grad_norm": 0.8723010420799255,
|
| 491 |
-
"learning_rate": 4.684365081339541e-05,
|
| 492 |
-
"loss": 0.1356166958808899,
|
| 493 |
-
"step": 1360
|
| 494 |
-
},
|
| 495 |
-
{
|
| 496 |
-
"epoch": 0.4333788040864795,
|
| 497 |
-
"grad_norm": 0.49977484345436096,
|
| 498 |
-
"learning_rate": 4.675166840446179e-05,
|
| 499 |
-
"loss": 0.17129049301147461,
|
| 500 |
-
"step": 1380
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.43965965631961684,
|
| 504 |
-
"grad_norm": 0.8447715640068054,
|
| 505 |
-
"learning_rate": 4.6658457906246754e-05,
|
| 506 |
-
"loss": 0.16764267683029174,
|
| 507 |
-
"step": 1400
|
| 508 |
-
},
|
| 509 |
-
{
|
| 510 |
-
"epoch": 0.44594050855275424,
|
| 511 |
-
"grad_norm": 0.491344153881073,
|
| 512 |
-
"learning_rate": 4.656402458137255e-05,
|
| 513 |
-
"loss": 0.2734020709991455,
|
| 514 |
-
"step": 1420
|
| 515 |
-
},
|
| 516 |
-
{
|
| 517 |
-
"epoch": 0.45222136078589165,
|
| 518 |
-
"grad_norm": 0.7019853591918945,
|
| 519 |
-
"learning_rate": 4.646837376150164e-05,
|
| 520 |
-
"loss": 0.15193712711334229,
|
| 521 |
-
"step": 1440
|
| 522 |
-
},
|
| 523 |
-
{
|
| 524 |
-
"epoch": 0.458502213019029,
|
| 525 |
-
"grad_norm": 0.5096240639686584,
|
| 526 |
-
"learning_rate": 4.637151084703571e-05,
|
| 527 |
-
"loss": 0.31453657150268555,
|
| 528 |
-
"step": 1460
|
| 529 |
-
},
|
| 530 |
-
{
|
| 531 |
-
"epoch": 0.4647830652521664,
|
| 532 |
-
"grad_norm": 0.457865446805954,
|
| 533 |
-
"learning_rate": 4.6273441306810724e-05,
|
| 534 |
-
"loss": 0.14785597324371338,
|
| 535 |
-
"step": 1480
|
| 536 |
-
},
|
| 537 |
-
{
|
| 538 |
-
"epoch": 0.4710639174853038,
|
| 539 |
-
"grad_norm": 0.5041202306747437,
|
| 540 |
-
"learning_rate": 4.617417067778825e-05,
|
| 541 |
-
"loss": 0.13907097578048705,
|
| 542 |
-
"step": 1500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.4773447697184412,
|
| 546 |
-
"grad_norm": 3.153047800064087,
|
| 547 |
-
"learning_rate": 4.6073704564742734e-05,
|
| 548 |
-
"loss": 0.1594400644302368,
|
| 549 |
-
"step": 1520
|
| 550 |
-
},
|
| 551 |
-
{
|
| 552 |
-
"epoch": 0.48362562195157854,
|
| 553 |
-
"grad_norm": 6.409017562866211,
|
| 554 |
-
"learning_rate": 4.5972048639945105e-05,
|
| 555 |
-
"loss": 0.1708819031715393,
|
| 556 |
-
"step": 1540
|
| 557 |
-
},
|
| 558 |
-
{
|
| 559 |
-
"epoch": 0.48990647418471595,
|
| 560 |
-
"grad_norm": 0.5795732736587524,
|
| 561 |
-
"learning_rate": 4.5869208642842536e-05,
|
| 562 |
-
"loss": 0.5376199245452881,
|
| 563 |
-
"step": 1560
|
| 564 |
-
},
|
| 565 |
-
{
|
| 566 |
-
"epoch": 0.49618732641785335,
|
| 567 |
-
"grad_norm": 1.402137279510498,
|
| 568 |
-
"learning_rate": 4.5765190379734384e-05,
|
| 569 |
-
"loss": 0.13104203939437867,
|
| 570 |
-
"step": 1580
|
| 571 |
-
},
|
| 572 |
-
{
|
| 573 |
-
"epoch": 0.5024681786509907,
|
| 574 |
-
"grad_norm": 1.01236093044281,
|
| 575 |
-
"learning_rate": 4.565999972344436e-05,
|
| 576 |
-
"loss": 0.14823287725448608,
|
| 577 |
-
"step": 1600
|
| 578 |
-
},
|
| 579 |
-
{
|
| 580 |
-
"epoch": 0.508749030884128,
|
| 581 |
-
"grad_norm": 0.47942015528678894,
|
| 582 |
-
"learning_rate": 4.5553642612988954e-05,
|
| 583 |
-
"loss": 0.13399872779846192,
|
| 584 |
-
"step": 1620
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.5150298831172655,
|
| 588 |
-
"grad_norm": 0.5736086964607239,
|
| 589 |
-
"learning_rate": 4.544612505324213e-05,
|
| 590 |
-
"loss": 0.09690673351287842,
|
| 591 |
-
"step": 1640
|
| 592 |
-
},
|
| 593 |
-
{
|
| 594 |
-
"epoch": 0.5213107353504028,
|
| 595 |
-
"grad_norm": 0.4799337387084961,
|
| 596 |
-
"learning_rate": 4.533745311459631e-05,
|
| 597 |
-
"loss": 0.12333678007125855,
|
| 598 |
-
"step": 1660
|
| 599 |
-
},
|
| 600 |
-
{
|
| 601 |
-
"epoch": 0.5275915875835402,
|
| 602 |
-
"grad_norm": 0.49539658427238464,
|
| 603 |
-
"learning_rate": 4.5227632932619604e-05,
|
| 604 |
-
"loss": 0.1237714409828186,
|
| 605 |
-
"step": 1680
|
| 606 |
-
},
|
| 607 |
-
{
|
| 608 |
-
"epoch": 0.5338724398166776,
|
| 609 |
-
"grad_norm": 3.5184292793273926,
|
| 610 |
-
"learning_rate": 4.5116670707709396e-05,
|
| 611 |
-
"loss": 0.12902543544769288,
|
| 612 |
-
"step": 1700
|
| 613 |
-
},
|
| 614 |
-
{
|
| 615 |
-
"epoch": 0.540153292049815,
|
| 616 |
-
"grad_norm": 0.543497622013092,
|
| 617 |
-
"learning_rate": 4.5004572704742345e-05,
|
| 618 |
-
"loss": 0.11819875240325928,
|
| 619 |
-
"step": 1720
|
| 620 |
-
},
|
| 621 |
-
{
|
| 622 |
-
"epoch": 0.5464341442829523,
|
| 623 |
-
"grad_norm": 0.47288012504577637,
|
| 624 |
-
"learning_rate": 4.489134525272057e-05,
|
| 625 |
-
"loss": 0.09024000763893128,
|
| 626 |
-
"step": 1740
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.5527149965160898,
|
| 630 |
-
"grad_norm": 0.5600817203521729,
|
| 631 |
-
"learning_rate": 4.477699474441441e-05,
|
| 632 |
-
"loss": 0.10022807121276855,
|
| 633 |
-
"step": 1760
|
| 634 |
-
},
|
| 635 |
-
{
|
| 636 |
-
"epoch": 0.5589958487492271,
|
| 637 |
-
"grad_norm": 0.38338902592658997,
|
| 638 |
-
"learning_rate": 4.466152763600142e-05,
|
| 639 |
-
"loss": 0.1357239603996277,
|
| 640 |
-
"step": 1780
|
| 641 |
-
},
|
| 642 |
-
{
|
| 643 |
-
"epoch": 0.5652767009823646,
|
| 644 |
-
"grad_norm": 0.12702441215515137,
|
| 645 |
-
"learning_rate": 4.454495044670189e-05,
|
| 646 |
-
"loss": 0.36903703212738037,
|
| 647 |
-
"step": 1800
|
| 648 |
-
},
|
| 649 |
-
{
|
| 650 |
-
"epoch": 0.571557553215502,
|
| 651 |
-
"grad_norm": 0.32971400022506714,
|
| 652 |
-
"learning_rate": 4.4427269758410806e-05,
|
| 653 |
-
"loss": 0.09078516364097595,
|
| 654 |
-
"step": 1820
|
| 655 |
-
},
|
| 656 |
-
{
|
| 657 |
-
"epoch": 0.5778384054486393,
|
| 658 |
-
"grad_norm": 0.5972228050231934,
|
| 659 |
-
"learning_rate": 4.4308492215326145e-05,
|
| 660 |
-
"loss": 0.13515870571136473,
|
| 661 |
-
"step": 1840
|
| 662 |
-
},
|
| 663 |
-
{
|
| 664 |
-
"epoch": 0.5841192576817767,
|
| 665 |
-
"grad_norm": 0.4297616183757782,
|
| 666 |
-
"learning_rate": 4.418862452357387e-05,
|
| 667 |
-
"loss": 0.16306908130645753,
|
| 668 |
-
"step": 1860
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.5904001099149141,
|
| 672 |
-
"grad_norm": 0.3069750666618347,
|
| 673 |
-
"learning_rate": 4.4067673450829184e-05,
|
| 674 |
-
"loss": 0.11015008687973023,
|
| 675 |
-
"step": 1880
|
| 676 |
-
},
|
| 677 |
-
{
|
| 678 |
-
"epoch": 0.5966809621480514,
|
| 679 |
-
"grad_norm": 0.2897976040840149,
|
| 680 |
-
"learning_rate": 4.394564582593452e-05,
|
| 681 |
-
"loss": 0.10074495077133179,
|
| 682 |
-
"step": 1900
|
| 683 |
-
},
|
| 684 |
-
{
|
| 685 |
-
"epoch": 0.6029618143811889,
|
| 686 |
-
"grad_norm": 0.2866128981113434,
|
| 687 |
-
"learning_rate": 4.382254853851396e-05,
|
| 688 |
-
"loss": 0.10123426914215088,
|
| 689 |
-
"step": 1920
|
| 690 |
-
},
|
| 691 |
-
{
|
| 692 |
-
"epoch": 0.6092426666143262,
|
| 693 |
-
"grad_norm": 0.611520528793335,
|
| 694 |
-
"learning_rate": 4.3698388538584214e-05,
|
| 695 |
-
"loss": 0.1677260398864746,
|
| 696 |
-
"step": 1940
|
| 697 |
-
},
|
| 698 |
-
{
|
| 699 |
-
"epoch": 0.6155235188474636,
|
| 700 |
-
"grad_norm": 0.6373476982116699,
|
| 701 |
-
"learning_rate": 4.357317283616228e-05,
|
| 702 |
-
"loss": 0.07637218236923218,
|
| 703 |
-
"step": 1960
|
| 704 |
-
},
|
| 705 |
-
{
|
| 706 |
-
"epoch": 0.621804371080601,
|
| 707 |
-
"grad_norm": 0.24775440990924835,
|
| 708 |
-
"learning_rate": 4.3446908500869624e-05,
|
| 709 |
-
"loss": 0.07576732635498047,
|
| 710 |
-
"step": 1980
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.6280852233137384,
|
| 714 |
-
"grad_norm": 0.36698204278945923,
|
| 715 |
-
"learning_rate": 4.3319602661533054e-05,
|
| 716 |
-
"loss": 0.08246396780014038,
|
| 717 |
-
"step": 2000
|
| 718 |
-
},
|
| 719 |
-
{
|
| 720 |
-
"epoch": 0.6343660755468757,
|
| 721 |
-
"grad_norm": 0.4456537365913391,
|
| 722 |
-
"learning_rate": 4.319126250578223e-05,
|
| 723 |
-
"loss": 0.08348861336708069,
|
| 724 |
-
"step": 2020
|
| 725 |
-
},
|
| 726 |
-
{
|
| 727 |
-
"epoch": 0.6406469277800132,
|
| 728 |
-
"grad_norm": 0.12586534023284912,
|
| 729 |
-
"learning_rate": 4.306189527964382e-05,
|
| 730 |
-
"loss": 0.11425427198410035,
|
| 731 |
-
"step": 2040
|
| 732 |
-
},
|
| 733 |
-
{
|
| 734 |
-
"epoch": 0.6469277800131505,
|
| 735 |
-
"grad_norm": 0.42404818534851074,
|
| 736 |
-
"learning_rate": 4.2931508287132435e-05,
|
| 737 |
-
"loss": 0.43116607666015627,
|
| 738 |
-
"step": 2060
|
| 739 |
-
},
|
| 740 |
-
{
|
| 741 |
-
"epoch": 0.6532086322462879,
|
| 742 |
-
"grad_norm": 0.322374165058136,
|
| 743 |
-
"learning_rate": 4.2800108889838244e-05,
|
| 744 |
-
"loss": 0.07275817394256592,
|
| 745 |
-
"step": 2080
|
| 746 |
-
},
|
| 747 |
-
{
|
| 748 |
-
"epoch": 0.6594894844794253,
|
| 749 |
-
"grad_norm": 0.24643594026565552,
|
| 750 |
-
"learning_rate": 4.26677045065113e-05,
|
| 751 |
-
"loss": 0.0800868034362793,
|
| 752 |
-
"step": 2100
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.6657703367125627,
|
| 756 |
-
"grad_norm": 0.32612672448158264,
|
| 757 |
-
"learning_rate": 4.253430261264272e-05,
|
| 758 |
-
"loss": 0.11347514390945435,
|
| 759 |
-
"step": 2120
|
| 760 |
-
},
|
| 761 |
-
{
|
| 762 |
-
"epoch": 0.6720511889457,
|
| 763 |
-
"grad_norm": 0.9363917708396912,
|
| 764 |
-
"learning_rate": 4.239991074004262e-05,
|
| 765 |
-
"loss": 0.14267417192459106,
|
| 766 |
-
"step": 2140
|
| 767 |
-
},
|
| 768 |
-
{
|
| 769 |
-
"epoch": 0.6783320411788375,
|
| 770 |
-
"grad_norm": 5.144097328186035,
|
| 771 |
-
"learning_rate": 4.226453647641484e-05,
|
| 772 |
-
"loss": 0.2039745807647705,
|
| 773 |
-
"step": 2160
|
| 774 |
-
},
|
| 775 |
-
{
|
| 776 |
-
"epoch": 0.6846128934119748,
|
| 777 |
-
"grad_norm": 0.275455117225647,
|
| 778 |
-
"learning_rate": 4.2128187464928604e-05,
|
| 779 |
-
"loss": 0.12193784713745118,
|
| 780 |
-
"step": 2180
|
| 781 |
-
},
|
| 782 |
-
{
|
| 783 |
-
"epoch": 0.6908937456451122,
|
| 784 |
-
"grad_norm": 0.23535878956317902,
|
| 785 |
-
"learning_rate": 4.1990871403786934e-05,
|
| 786 |
-
"loss": 0.08345263600349426,
|
| 787 |
-
"step": 2200
|
| 788 |
-
},
|
| 789 |
-
{
|
| 790 |
-
"epoch": 0.6971745978782496,
|
| 791 |
-
"grad_norm": 0.24043764173984528,
|
| 792 |
-
"learning_rate": 4.185259604579203e-05,
|
| 793 |
-
"loss": 0.06981859803199768,
|
| 794 |
-
"step": 2220
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.703455450111387,
|
| 798 |
-
"grad_norm": 0.6327846050262451,
|
| 799 |
-
"learning_rate": 4.171336919790759e-05,
|
| 800 |
-
"loss": 0.1702319025993347,
|
| 801 |
-
"step": 2240
|
| 802 |
-
},
|
| 803 |
-
{
|
| 804 |
-
"epoch": 0.7097363023445243,
|
| 805 |
-
"grad_norm": 0.059428345412015915,
|
| 806 |
-
"learning_rate": 4.1573198720817955e-05,
|
| 807 |
-
"loss": 0.07783352732658386,
|
| 808 |
-
"step": 2260
|
| 809 |
-
},
|
| 810 |
-
{
|
| 811 |
-
"epoch": 0.7160171545776618,
|
| 812 |
-
"grad_norm": 0.4627244770526886,
|
| 813 |
-
"learning_rate": 4.143209252848433e-05,
|
| 814 |
-
"loss": 0.07844600677490235,
|
| 815 |
-
"step": 2280
|
| 816 |
-
},
|
| 817 |
-
{
|
| 818 |
-
"epoch": 0.7222980068107991,
|
| 819 |
-
"grad_norm": 0.7847932577133179,
|
| 820 |
-
"learning_rate": 4.1290058587698054e-05,
|
| 821 |
-
"loss": 0.09436154365539551,
|
| 822 |
-
"step": 2300
|
| 823 |
-
},
|
| 824 |
-
{
|
| 825 |
-
"epoch": 0.7285788590439365,
|
| 826 |
-
"grad_norm": 0.6646613478660583,
|
| 827 |
-
"learning_rate": 4.114710491763063e-05,
|
| 828 |
-
"loss": 0.10847301483154297,
|
| 829 |
-
"step": 2320
|
| 830 |
-
},
|
| 831 |
-
{
|
| 832 |
-
"epoch": 0.734859711277074,
|
| 833 |
-
"grad_norm": 0.13904166221618652,
|
| 834 |
-
"learning_rate": 4.100323958938112e-05,
|
| 835 |
-
"loss": 0.10292520523071289,
|
| 836 |
-
"step": 2340
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.7411405635102113,
|
| 840 |
-
"grad_norm": 0.17684051394462585,
|
| 841 |
-
"learning_rate": 4.085847072552036e-05,
|
| 842 |
-
"loss": 0.10409547090530395,
|
| 843 |
-
"step": 2360
|
| 844 |
-
},
|
| 845 |
-
{
|
| 846 |
-
"epoch": 0.7474214157433486,
|
| 847 |
-
"grad_norm": 2.33097767829895,
|
| 848 |
-
"learning_rate": 4.0712806499632397e-05,
|
| 849 |
-
"loss": 0.07961801886558532,
|
| 850 |
-
"step": 2380
|
| 851 |
-
},
|
| 852 |
-
{
|
| 853 |
-
"epoch": 0.7537022679764861,
|
| 854 |
-
"grad_norm": 0.476825475692749,
|
| 855 |
-
"learning_rate": 4.0566255135853014e-05,
|
| 856 |
-
"loss": 0.07112305760383605,
|
| 857 |
-
"step": 2400
|
| 858 |
-
},
|
| 859 |
-
{
|
| 860 |
-
"epoch": 0.7599831202096234,
|
| 861 |
-
"grad_norm": 0.5014712810516357,
|
| 862 |
-
"learning_rate": 4.04188249084054e-05,
|
| 863 |
-
"loss": 0.1152498722076416,
|
| 864 |
-
"step": 2420
|
| 865 |
-
},
|
| 866 |
-
{
|
| 867 |
-
"epoch": 0.7662639724427608,
|
| 868 |
-
"grad_norm": 0.4747202694416046,
|
| 869 |
-
"learning_rate": 4.0270524141133e-05,
|
| 870 |
-
"loss": 0.12496150732040405,
|
| 871 |
-
"step": 2440
|
| 872 |
-
},
|
| 873 |
-
{
|
| 874 |
-
"epoch": 0.7725448246758982,
|
| 875 |
-
"grad_norm": 0.3705868721008301,
|
| 876 |
-
"learning_rate": 4.01213612070295e-05,
|
| 877 |
-
"loss": 0.09812939167022705,
|
| 878 |
-
"step": 2460
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 0.7788256769090356,
|
| 882 |
-
"grad_norm": 0.18966151773929596,
|
| 883 |
-
"learning_rate": 3.997134452776617e-05,
|
| 884 |
-
"loss": 0.09260130524635315,
|
| 885 |
-
"step": 2480
|
| 886 |
-
},
|
| 887 |
-
{
|
| 888 |
-
"epoch": 0.7851065291421729,
|
| 889 |
-
"grad_norm": 0.4178507328033447,
|
| 890 |
-
"learning_rate": 3.982048257321631e-05,
|
| 891 |
-
"loss": 0.08796253800392151,
|
| 892 |
-
"step": 2500
|
| 893 |
-
},
|
| 894 |
-
{
|
| 895 |
-
"epoch": 0.7913873813753104,
|
| 896 |
-
"grad_norm": 4.684873580932617,
|
| 897 |
-
"learning_rate": 3.966878386097711e-05,
|
| 898 |
-
"loss": 0.09675108194351197,
|
| 899 |
-
"step": 2520
|
| 900 |
-
},
|
| 901 |
-
{
|
| 902 |
-
"epoch": 0.7976682336084477,
|
| 903 |
-
"grad_norm": 0.7546767592430115,
|
| 904 |
-
"learning_rate": 3.95162569558887e-05,
|
| 905 |
-
"loss": 0.061058801412582395,
|
| 906 |
-
"step": 2540
|
| 907 |
-
},
|
| 908 |
-
{
|
| 909 |
-
"epoch": 0.8039490858415851,
|
| 910 |
-
"grad_norm": 1.2984436750411987,
|
| 911 |
-
"learning_rate": 3.936291046955061e-05,
|
| 912 |
-
"loss": 0.0732529878616333,
|
| 913 |
-
"step": 2560
|
| 914 |
-
},
|
| 915 |
-
{
|
| 916 |
-
"epoch": 0.8102299380747225,
|
| 917 |
-
"grad_norm": 0.08046652376651764,
|
| 918 |
-
"learning_rate": 3.920875305983553e-05,
|
| 919 |
-
"loss": 0.056511855125427245,
|
| 920 |
-
"step": 2580
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 0.8165107903078599,
|
| 924 |
-
"grad_norm": 0.24541246891021729,
|
| 925 |
-
"learning_rate": 3.905379343040054e-05,
|
| 926 |
-
"loss": 0.06492850780487061,
|
| 927 |
-
"step": 2600
|
| 928 |
-
},
|
| 929 |
-
{
|
| 930 |
-
"epoch": 0.8227916425409972,
|
| 931 |
-
"grad_norm": 0.34249547123908997,
|
| 932 |
-
"learning_rate": 3.889804033019567e-05,
|
| 933 |
-
"loss": 0.06372414827346802,
|
| 934 |
-
"step": 2620
|
| 935 |
-
},
|
| 936 |
-
{
|
| 937 |
-
"epoch": 0.8290724947741347,
|
| 938 |
-
"grad_norm": 0.6350385546684265,
|
| 939 |
-
"learning_rate": 3.8741502552969956e-05,
|
| 940 |
-
"loss": 0.07334259748458863,
|
| 941 |
-
"step": 2640
|
| 942 |
-
},
|
| 943 |
-
{
|
| 944 |
-
"epoch": 0.835353347007272,
|
| 945 |
-
"grad_norm": 0.5676931738853455,
|
| 946 |
-
"learning_rate": 3.8584188936774925e-05,
|
| 947 |
-
"loss": 0.051751017570495605,
|
| 948 |
-
"step": 2660
|
| 949 |
-
},
|
| 950 |
-
{
|
| 951 |
-
"epoch": 0.8416341992404094,
|
| 952 |
-
"grad_norm": 1.5374674797058105,
|
| 953 |
-
"learning_rate": 3.842610836346563e-05,
|
| 954 |
-
"loss": 0.13946666717529296,
|
| 955 |
-
"step": 2680
|
| 956 |
-
},
|
| 957 |
-
{
|
| 958 |
-
"epoch": 0.8479150514735468,
|
| 959 |
-
"grad_norm": 0.4549136459827423,
|
| 960 |
-
"learning_rate": 3.8267269758199175e-05,
|
| 961 |
-
"loss": 0.06678192019462585,
|
| 962 |
-
"step": 2700
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 0.8541959037066842,
|
| 966 |
-
"grad_norm": 0.4328550696372986,
|
| 967 |
-
"learning_rate": 3.8107682088930794e-05,
|
| 968 |
-
"loss": 0.0756498098373413,
|
| 969 |
-
"step": 2720
|
| 970 |
-
},
|
| 971 |
-
{
|
| 972 |
-
"epoch": 0.8604767559398215,
|
| 973 |
-
"grad_norm": 1.7214152812957764,
|
| 974 |
-
"learning_rate": 3.7947354365907536e-05,
|
| 975 |
-
"loss": 0.07417964935302734,
|
| 976 |
-
"step": 2740
|
| 977 |
-
},
|
| 978 |
-
{
|
| 979 |
-
"epoch": 0.866757608172959,
|
| 980 |
-
"grad_norm": 0.3970869481563568,
|
| 981 |
-
"learning_rate": 3.7786295641159544e-05,
|
| 982 |
-
"loss": 0.055640166997909545,
|
| 983 |
-
"step": 2760
|
| 984 |
-
},
|
| 985 |
-
{
|
| 986 |
-
"epoch": 0.8730384604060963,
|
| 987 |
-
"grad_norm": 0.16465847194194794,
|
| 988 |
-
"learning_rate": 3.7624515007989004e-05,
|
| 989 |
-
"loss": 0.057208091020584106,
|
| 990 |
-
"step": 2780
|
| 991 |
-
},
|
| 992 |
-
{
|
| 993 |
-
"epoch": 0.8793193126392337,
|
| 994 |
-
"grad_norm": 1.0308940410614014,
|
| 995 |
-
"learning_rate": 3.746202160045669e-05,
|
| 996 |
-
"loss": 0.1124604344367981,
|
| 997 |
-
"step": 2800
|
| 998 |
-
},
|
| 999 |
-
{
|
| 1000 |
-
"epoch": 0.8856001648723711,
|
| 1001 |
-
"grad_norm": 1.095229983329773,
|
| 1002 |
-
"learning_rate": 3.729882459286632e-05,
|
| 1003 |
-
"loss": 0.06749147176742554,
|
| 1004 |
-
"step": 2820
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 0.8918810171055085,
|
| 1008 |
-
"grad_norm": 0.7268246412277222,
|
| 1009 |
-
"learning_rate": 3.713493319924655e-05,
|
| 1010 |
-
"loss": 0.05816763639450073,
|
| 1011 |
-
"step": 2840
|
| 1012 |
-
},
|
| 1013 |
-
{
|
| 1014 |
-
"epoch": 0.8981618693386458,
|
| 1015 |
-
"grad_norm": 0.06477148085832596,
|
| 1016 |
-
"learning_rate": 3.697035667283072e-05,
|
| 1017 |
-
"loss": 0.07365052700042725,
|
| 1018 |
-
"step": 2860
|
| 1019 |
-
},
|
| 1020 |
-
{
|
| 1021 |
-
"epoch": 0.9044427215717833,
|
| 1022 |
-
"grad_norm": 0.14640307426452637,
|
| 1023 |
-
"learning_rate": 3.68051043055345e-05,
|
| 1024 |
-
"loss": 0.10839139223098755,
|
| 1025 |
-
"step": 2880
|
| 1026 |
-
},
|
| 1027 |
-
{
|
| 1028 |
-
"epoch": 0.9107235738049206,
|
| 1029 |
-
"grad_norm": 0.6694417595863342,
|
| 1030 |
-
"learning_rate": 3.66391854274312e-05,
|
| 1031 |
-
"loss": 0.0827883541584015,
|
| 1032 |
-
"step": 2900
|
| 1033 |
-
},
|
| 1034 |
-
{
|
| 1035 |
-
"epoch": 0.917004426038058,
|
| 1036 |
-
"grad_norm": 0.7503764629364014,
|
| 1037 |
-
"learning_rate": 3.647260940622504e-05,
|
| 1038 |
-
"loss": 0.05839287042617798,
|
| 1039 |
-
"step": 2920
|
| 1040 |
-
},
|
| 1041 |
-
{
|
| 1042 |
-
"epoch": 0.9232852782711954,
|
| 1043 |
-
"grad_norm": 0.6776005029678345,
|
| 1044 |
-
"learning_rate": 3.6305385646722224e-05,
|
| 1045 |
-
"loss": 0.08498733639717101,
|
| 1046 |
-
"step": 2940
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 0.9295661305043328,
|
| 1050 |
-
"grad_norm": 0.1537044644355774,
|
| 1051 |
-
"learning_rate": 3.613752359029998e-05,
|
| 1052 |
-
"loss": 0.1408205509185791,
|
| 1053 |
-
"step": 2960
|
| 1054 |
-
},
|
| 1055 |
-
{
|
| 1056 |
-
"epoch": 0.9358469827374702,
|
| 1057 |
-
"grad_norm": 0.18748170137405396,
|
| 1058 |
-
"learning_rate": 3.5969032714373495e-05,
|
| 1059 |
-
"loss": 0.07079569697380066,
|
| 1060 |
-
"step": 2980
|
| 1061 |
-
},
|
| 1062 |
-
{
|
| 1063 |
-
"epoch": 0.9421278349706076,
|
| 1064 |
-
"grad_norm": 1.4993492364883423,
|
| 1065 |
-
"learning_rate": 3.57999225318608e-05,
|
| 1066 |
-
"loss": 0.07452060580253601,
|
| 1067 |
-
"step": 3000
|
| 1068 |
-
},
|
| 1069 |
-
{
|
| 1070 |
-
"epoch": 0.9484086872037449,
|
| 1071 |
-
"grad_norm": 2.759636163711548,
|
| 1072 |
-
"learning_rate": 3.5638702921686935e-05,
|
| 1073 |
-
"loss": 0.30155444145202637,
|
| 1074 |
-
"step": 3020
|
| 1075 |
-
},
|
| 1076 |
-
{
|
| 1077 |
-
"epoch": 0.9546895394368824,
|
| 1078 |
-
"grad_norm": 0.35174673795700073,
|
| 1079 |
-
"learning_rate": 3.546841258479434e-05,
|
| 1080 |
-
"loss": 0.07043100595474243,
|
| 1081 |
-
"step": 3040
|
| 1082 |
-
},
|
| 1083 |
-
{
|
| 1084 |
-
"epoch": 0.9609703916700197,
|
| 1085 |
-
"grad_norm": 0.1061045303940773,
|
| 1086 |
-
"learning_rate": 3.529753120609982e-05,
|
| 1087 |
-
"loss": 0.06520496010780334,
|
| 1088 |
-
"step": 3060
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 0.9672512439031571,
|
| 1092 |
-
"grad_norm": 0.32999858260154724,
|
| 1093 |
-
"learning_rate": 3.51260684334881e-05,
|
| 1094 |
-
"loss": 0.11166098117828369,
|
| 1095 |
-
"step": 3080
|
| 1096 |
-
},
|
| 1097 |
-
{
|
| 1098 |
-
"epoch": 0.9735320961362945,
|
| 1099 |
-
"grad_norm": 0.527561366558075,
|
| 1100 |
-
"learning_rate": 3.4954033947669163e-05,
|
| 1101 |
-
"loss": 0.05884957313537598,
|
| 1102 |
-
"step": 3100
|
| 1103 |
-
},
|
| 1104 |
-
{
|
| 1105 |
-
"epoch": 0.9798129483694319,
|
| 1106 |
-
"grad_norm": 0.21894922852516174,
|
| 1107 |
-
"learning_rate": 3.478143746163164e-05,
|
| 1108 |
-
"loss": 0.04760840237140655,
|
| 1109 |
-
"step": 3120
|
| 1110 |
-
},
|
| 1111 |
-
{
|
| 1112 |
-
"epoch": 0.9860938006025692,
|
| 1113 |
-
"grad_norm": 0.39011839032173157,
|
| 1114 |
-
"learning_rate": 3.460828872009444e-05,
|
| 1115 |
-
"loss": 0.07473071217536927,
|
| 1116 |
-
"step": 3140
|
| 1117 |
-
},
|
| 1118 |
-
{
|
| 1119 |
-
"epoch": 0.9923746528357067,
|
| 1120 |
-
"grad_norm": 0.24014221131801605,
|
| 1121 |
-
"learning_rate": 3.443459749895658e-05,
|
| 1122 |
-
"loss": 0.05981050133705139,
|
| 1123 |
-
"step": 3160
|
| 1124 |
-
},
|
| 1125 |
-
{
|
| 1126 |
-
"epoch": 0.998655505068844,
|
| 1127 |
-
"grad_norm": 0.4369584918022156,
|
| 1128 |
-
"learning_rate": 3.4260373604745205e-05,
|
| 1129 |
-
"loss": 0.06494853496551514,
|
| 1130 |
-
"step": 3180
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.004710639174853,
|
| 1134 |
-
"grad_norm": 0.11435407400131226,
|
| 1135 |
-
"learning_rate": 3.4085626874061974e-05,
|
| 1136 |
-
"loss": 0.037298232316970825,
|
| 1137 |
-
"step": 3200
|
| 1138 |
-
},
|
| 1139 |
-
{
|
| 1140 |
-
"epoch": 1.0109914914079905,
|
| 1141 |
-
"grad_norm": 0.10405384004116058,
|
| 1142 |
-
"learning_rate": 3.391036717302761e-05,
|
| 1143 |
-
"loss": 0.03180496394634247,
|
| 1144 |
-
"step": 3220
|
| 1145 |
-
},
|
| 1146 |
-
{
|
| 1147 |
-
"epoch": 1.0172723436411277,
|
| 1148 |
-
"grad_norm": 0.2586648166179657,
|
| 1149 |
-
"learning_rate": 3.373460439672495e-05,
|
| 1150 |
-
"loss": 0.04955384433269501,
|
| 1151 |
-
"step": 3240
|
| 1152 |
-
},
|
| 1153 |
-
{
|
| 1154 |
-
"epoch": 1.0235531958742652,
|
| 1155 |
-
"grad_norm": 0.08358602225780487,
|
| 1156 |
-
"learning_rate": 3.35583484686402e-05,
|
| 1157 |
-
"loss": 0.05318580269813537,
|
| 1158 |
-
"step": 3260
|
| 1159 |
-
},
|
| 1160 |
-
{
|
| 1161 |
-
"epoch": 1.0298340481074026,
|
| 1162 |
-
"grad_norm": 0.07553794234991074,
|
| 1163 |
-
"learning_rate": 3.338160934010271e-05,
|
| 1164 |
-
"loss": 0.08674532771110535,
|
| 1165 |
-
"step": 3280
|
| 1166 |
-
},
|
| 1167 |
-
{
|
| 1168 |
-
"epoch": 1.0361149003405399,
|
| 1169 |
-
"grad_norm": 0.2271793782711029,
|
| 1170 |
-
"learning_rate": 3.32043969897231e-05,
|
| 1171 |
-
"loss": 0.04139872789382935,
|
| 1172 |
-
"step": 3300
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.0423957525736773,
|
| 1176 |
-
"grad_norm": 0.25413981080055237,
|
| 1177 |
-
"learning_rate": 3.302672142282988e-05,
|
| 1178 |
-
"loss": 0.035733970999717715,
|
| 1179 |
-
"step": 3320
|
| 1180 |
-
},
|
| 1181 |
-
{
|
| 1182 |
-
"epoch": 1.0486766048068148,
|
| 1183 |
-
"grad_norm": 0.09163467586040497,
|
| 1184 |
-
"learning_rate": 3.284859267090456e-05,
|
| 1185 |
-
"loss": 0.08379928469657898,
|
| 1186 |
-
"step": 3340
|
| 1187 |
-
},
|
| 1188 |
-
{
|
| 1189 |
-
"epoch": 1.054957457039952,
|
| 1190 |
-
"grad_norm": 0.07256153970956802,
|
| 1191 |
-
"learning_rate": 3.267002079101525e-05,
|
| 1192 |
-
"loss": 0.03688819110393524,
|
| 1193 |
-
"step": 3360
|
| 1194 |
-
},
|
| 1195 |
-
{
|
| 1196 |
-
"epoch": 1.0612383092730895,
|
| 1197 |
-
"grad_norm": 0.305923193693161,
|
| 1198 |
-
"learning_rate": 3.2491015865248884e-05,
|
| 1199 |
-
"loss": 0.049816560745239255,
|
| 1200 |
-
"step": 3380
|
| 1201 |
-
},
|
| 1202 |
-
{
|
| 1203 |
-
"epoch": 1.067519161506227,
|
| 1204 |
-
"grad_norm": 0.046878550201654434,
|
| 1205 |
-
"learning_rate": 3.231158800014197e-05,
|
| 1206 |
-
"loss": 0.03804933726787567,
|
| 1207 |
-
"step": 3400
|
| 1208 |
-
},
|
| 1209 |
-
{
|
| 1210 |
-
"epoch": 1.0738000137393642,
|
| 1211 |
-
"grad_norm": 0.6410548686981201,
|
| 1212 |
-
"learning_rate": 3.213174732610995e-05,
|
| 1213 |
-
"loss": 0.0333125501871109,
|
| 1214 |
-
"step": 3420
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.0800808659725016,
|
| 1218 |
-
"grad_norm": 0.06022074818611145,
|
| 1219 |
-
"learning_rate": 3.195150399687531e-05,
|
| 1220 |
-
"loss": 0.028510311245918275,
|
| 1221 |
-
"step": 3440
|
| 1222 |
-
},
|
| 1223 |
-
{
|
| 1224 |
-
"epoch": 1.086361718205639,
|
| 1225 |
-
"grad_norm": 0.41786953806877136,
|
| 1226 |
-
"learning_rate": 3.177086818889423e-05,
|
| 1227 |
-
"loss": 0.046647512912750246,
|
| 1228 |
-
"step": 3460
|
| 1229 |
-
},
|
| 1230 |
-
{
|
| 1231 |
-
"epoch": 1.0926425704387763,
|
| 1232 |
-
"grad_norm": 0.1293671578168869,
|
| 1233 |
-
"learning_rate": 3.1589850100782085e-05,
|
| 1234 |
-
"loss": 0.03643852770328522,
|
| 1235 |
-
"step": 3480
|
| 1236 |
-
},
|
| 1237 |
-
{
|
| 1238 |
-
"epoch": 1.0989234226719138,
|
| 1239 |
-
"grad_norm": 0.05348878353834152,
|
| 1240 |
-
"learning_rate": 3.1408459952737594e-05,
|
| 1241 |
-
"loss": 0.03161140382289886,
|
| 1242 |
-
"step": 3500
|
| 1243 |
-
},
|
| 1244 |
-
{
|
| 1245 |
-
"epoch": 1.1052042749050512,
|
| 1246 |
-
"grad_norm": 0.29638731479644775,
|
| 1247 |
-
"learning_rate": 3.122670798596583e-05,
|
| 1248 |
-
"loss": 0.04662284553050995,
|
| 1249 |
-
"step": 3520
|
| 1250 |
-
},
|
| 1251 |
-
{
|
| 1252 |
-
"epoch": 1.1114851271381885,
|
| 1253 |
-
"grad_norm": 0.11425400525331497,
|
| 1254 |
-
"learning_rate": 3.1044604462099965e-05,
|
| 1255 |
-
"loss": 0.04306844174861908,
|
| 1256 |
-
"step": 3540
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.117765979371326,
|
| 1260 |
-
"grad_norm": 13.269792556762695,
|
| 1261 |
-
"learning_rate": 3.0862159662621933e-05,
|
| 1262 |
-
"loss": 0.10026944875717163,
|
| 1263 |
-
"step": 3560
|
| 1264 |
-
},
|
| 1265 |
-
{
|
| 1266 |
-
"epoch": 1.1240468316044634,
|
| 1267 |
-
"grad_norm": 0.09814783930778503,
|
| 1268 |
-
"learning_rate": 3.067938388828195e-05,
|
| 1269 |
-
"loss": 0.03279039561748505,
|
| 1270 |
-
"step": 3580
|
| 1271 |
-
},
|
| 1272 |
-
{
|
| 1273 |
-
"epoch": 1.1303276838376006,
|
| 1274 |
-
"grad_norm": 2.8222439289093018,
|
| 1275 |
-
"learning_rate": 3.049628745851689e-05,
|
| 1276 |
-
"loss": 0.03593765795230865,
|
| 1277 |
-
"step": 3600
|
| 1278 |
-
},
|
| 1279 |
-
{
|
| 1280 |
-
"epoch": 1.136608536070738,
|
| 1281 |
-
"grad_norm": 0.14487864077091217,
|
| 1282 |
-
"learning_rate": 3.031288071086772e-05,
|
| 1283 |
-
"loss": 0.0726437509059906,
|
| 1284 |
-
"step": 3620
|
| 1285 |
-
},
|
| 1286 |
-
{
|
| 1287 |
-
"epoch": 1.1428893883038755,
|
| 1288 |
-
"grad_norm": 1.1210272312164307,
|
| 1289 |
-
"learning_rate": 3.0129174000395776e-05,
|
| 1290 |
-
"loss": 0.03795652091503143,
|
| 1291 |
-
"step": 3640
|
| 1292 |
-
},
|
| 1293 |
-
{
|
| 1294 |
-
"epoch": 1.1491702405370128,
|
| 1295 |
-
"grad_norm": 0.42077013850212097,
|
| 1296 |
-
"learning_rate": 2.99451776990982e-05,
|
| 1297 |
-
"loss": 0.04205833375453949,
|
| 1298 |
-
"step": 3660
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.1554510927701502,
|
| 1302 |
-
"grad_norm": 0.08231453597545624,
|
| 1303 |
-
"learning_rate": 2.976090219532226e-05,
|
| 1304 |
-
"loss": 0.04487053751945495,
|
| 1305 |
-
"step": 3680
|
| 1306 |
-
},
|
| 1307 |
-
{
|
| 1308 |
-
"epoch": 1.1617319450032877,
|
| 1309 |
-
"grad_norm": 0.1877751648426056,
|
| 1310 |
-
"learning_rate": 2.9576357893178885e-05,
|
| 1311 |
-
"loss": 0.034676852822303775,
|
| 1312 |
-
"step": 3700
|
| 1313 |
-
},
|
| 1314 |
-
{
|
| 1315 |
-
"epoch": 1.168012797236425,
|
| 1316 |
-
"grad_norm": 0.6968119144439697,
|
| 1317 |
-
"learning_rate": 2.939155521195526e-05,
|
| 1318 |
-
"loss": 0.03220001757144928,
|
| 1319 |
-
"step": 3720
|
| 1320 |
-
},
|
| 1321 |
-
{
|
| 1322 |
-
"epoch": 1.1742936494695624,
|
| 1323 |
-
"grad_norm": 0.06414724141359329,
|
| 1324 |
-
"learning_rate": 2.9206504585526493e-05,
|
| 1325 |
-
"loss": 0.041881483793258664,
|
| 1326 |
-
"step": 3740
|
| 1327 |
-
},
|
| 1328 |
-
{
|
| 1329 |
-
"epoch": 1.1805745017026998,
|
| 1330 |
-
"grad_norm": 0.10654035210609436,
|
| 1331 |
-
"learning_rate": 2.9021216461766604e-05,
|
| 1332 |
-
"loss": 0.03662925064563751,
|
| 1333 |
-
"step": 3760
|
| 1334 |
-
},
|
| 1335 |
-
{
|
| 1336 |
-
"epoch": 1.186855353935837,
|
| 1337 |
-
"grad_norm": 0.04183540865778923,
|
| 1338 |
-
"learning_rate": 2.8835701301958574e-05,
|
| 1339 |
-
"loss": 0.04763528108596802,
|
| 1340 |
-
"step": 3780
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.1931362061689745,
|
| 1344 |
-
"grad_norm": 2.18105411529541,
|
| 1345 |
-
"learning_rate": 2.864996958020375e-05,
|
| 1346 |
-
"loss": 0.06307411193847656,
|
| 1347 |
-
"step": 3800
|
| 1348 |
-
},
|
| 1349 |
-
{
|
| 1350 |
-
"epoch": 1.199417058402112,
|
| 1351 |
-
"grad_norm": 0.3592161536216736,
|
| 1352 |
-
"learning_rate": 2.8464031782830474e-05,
|
| 1353 |
-
"loss": 0.051524901390075685,
|
| 1354 |
-
"step": 3820
|
| 1355 |
-
},
|
| 1356 |
-
{
|
| 1357 |
-
"epoch": 1.2056979106352492,
|
| 1358 |
-
"grad_norm": 0.11931532621383667,
|
| 1359 |
-
"learning_rate": 2.827789840780201e-05,
|
| 1360 |
-
"loss": 0.035766962170600894,
|
| 1361 |
-
"step": 3840
|
| 1362 |
-
},
|
| 1363 |
-
{
|
| 1364 |
-
"epoch": 1.2119787628683867,
|
| 1365 |
-
"grad_norm": 0.06202741712331772,
|
| 1366 |
-
"learning_rate": 2.8091579964123854e-05,
|
| 1367 |
-
"loss": 0.04704097211360932,
|
| 1368 |
-
"step": 3860
|
| 1369 |
-
},
|
| 1370 |
-
{
|
| 1371 |
-
"epoch": 1.2182596151015241,
|
| 1372 |
-
"grad_norm": 0.04776179790496826,
|
| 1373 |
-
"learning_rate": 2.7905086971250393e-05,
|
| 1374 |
-
"loss": 0.028332751989364625,
|
| 1375 |
-
"step": 3880
|
| 1376 |
-
},
|
| 1377 |
-
{
|
| 1378 |
-
"epoch": 1.2245404673346614,
|
| 1379 |
-
"grad_norm": 0.2371724396944046,
|
| 1380 |
-
"learning_rate": 2.771842995849098e-05,
|
| 1381 |
-
"loss": 0.027332887053489685,
|
| 1382 |
-
"step": 3900
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.2308213195677988,
|
| 1386 |
-
"grad_norm": 0.05050333961844444,
|
| 1387 |
-
"learning_rate": 2.753161946441546e-05,
|
| 1388 |
-
"loss": 0.02915753722190857,
|
| 1389 |
-
"step": 3920
|
| 1390 |
-
},
|
| 1391 |
-
{
|
| 1392 |
-
"epoch": 1.2371021718009363,
|
| 1393 |
-
"grad_norm": 0.08952230215072632,
|
| 1394 |
-
"learning_rate": 2.734466603625915e-05,
|
| 1395 |
-
"loss": 0.025665390491485595,
|
| 1396 |
-
"step": 3940
|
| 1397 |
-
},
|
| 1398 |
-
{
|
| 1399 |
-
"epoch": 1.2433830240340735,
|
| 1400 |
-
"grad_norm": 0.20825526118278503,
|
| 1401 |
-
"learning_rate": 2.7157580229327385e-05,
|
| 1402 |
-
"loss": 0.2276303768157959,
|
| 1403 |
-
"step": 3960
|
| 1404 |
-
},
|
| 1405 |
-
{
|
| 1406 |
-
"epoch": 1.249663876267211,
|
| 1407 |
-
"grad_norm": 1.8631341457366943,
|
| 1408 |
-
"learning_rate": 2.6970372606399525e-05,
|
| 1409 |
-
"loss": 0.03601144254207611,
|
| 1410 |
-
"step": 3980
|
| 1411 |
-
},
|
| 1412 |
-
{
|
| 1413 |
-
"epoch": 1.2559447285003484,
|
| 1414 |
-
"grad_norm": 0.21078747510910034,
|
| 1415 |
-
"learning_rate": 2.6783053737132607e-05,
|
| 1416 |
-
"loss": 0.08526483774185181,
|
| 1417 |
-
"step": 4000
|
| 1418 |
-
},
|
| 1419 |
-
{
|
| 1420 |
-
"epoch": 1.2622255807334857,
|
| 1421 |
-
"grad_norm": 0.051980458199977875,
|
| 1422 |
-
"learning_rate": 2.6595634197464596e-05,
|
| 1423 |
-
"loss": 0.022251833975315095,
|
| 1424 |
-
"step": 4020
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.2685064329666231,
|
| 1428 |
-
"grad_norm": 0.3019862771034241,
|
| 1429 |
-
"learning_rate": 2.6408124569017257e-05,
|
| 1430 |
-
"loss": 0.05009118318557739,
|
| 1431 |
-
"step": 4040
|
| 1432 |
-
},
|
| 1433 |
-
{
|
| 1434 |
-
"epoch": 1.2747872851997606,
|
| 1435 |
-
"grad_norm": 0.6371468305587769,
|
| 1436 |
-
"learning_rate": 2.622053543849872e-05,
|
| 1437 |
-
"loss": 0.040181088447570804,
|
| 1438 |
-
"step": 4060
|
| 1439 |
-
},
|
| 1440 |
-
{
|
| 1441 |
-
"epoch": 1.2810681374328978,
|
| 1442 |
-
"grad_norm": 1.1055502891540527,
|
| 1443 |
-
"learning_rate": 2.6032877397105783e-05,
|
| 1444 |
-
"loss": 0.02929992079734802,
|
| 1445 |
-
"step": 4080
|
| 1446 |
-
},
|
| 1447 |
-
{
|
| 1448 |
-
"epoch": 1.2873489896660353,
|
| 1449 |
-
"grad_norm": 0.26957738399505615,
|
| 1450 |
-
"learning_rate": 2.58451610399259e-05,
|
| 1451 |
-
"loss": 0.020054836571216584,
|
| 1452 |
-
"step": 4100
|
| 1453 |
-
},
|
| 1454 |
-
{
|
| 1455 |
-
"epoch": 1.2936298418991727,
|
| 1456 |
-
"grad_norm": 0.07279884070158005,
|
| 1457 |
-
"learning_rate": 2.5657396965339003e-05,
|
| 1458 |
-
"loss": 0.039886316657066344,
|
| 1459 |
-
"step": 4120
|
| 1460 |
-
},
|
| 1461 |
-
{
|
| 1462 |
-
"epoch": 1.29991069413231,
|
| 1463 |
-
"grad_norm": 0.09953224658966064,
|
| 1464 |
-
"learning_rate": 2.5469595774419158e-05,
|
| 1465 |
-
"loss": 0.03994681239128113,
|
| 1466 |
-
"step": 4140
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.3061915463654474,
|
| 1470 |
-
"grad_norm": 0.14353635907173157,
|
| 1471 |
-
"learning_rate": 2.5281768070335988e-05,
|
| 1472 |
-
"loss": 0.03711488544940948,
|
| 1473 |
-
"step": 4160
|
| 1474 |
-
},
|
| 1475 |
-
{
|
| 1476 |
-
"epoch": 1.3124723985985849,
|
| 1477 |
-
"grad_norm": 0.27422836422920227,
|
| 1478 |
-
"learning_rate": 2.5093924457756018e-05,
|
| 1479 |
-
"loss": 0.023467497527599336,
|
| 1480 |
-
"step": 4180
|
| 1481 |
-
},
|
| 1482 |
-
{
|
| 1483 |
-
"epoch": 1.3187532508317221,
|
| 1484 |
-
"grad_norm": 0.0728367418050766,
|
| 1485 |
-
"learning_rate": 2.490607554224399e-05,
|
| 1486 |
-
"loss": 0.018936164677143097,
|
| 1487 |
-
"step": 4200
|
| 1488 |
-
},
|
| 1489 |
-
{
|
| 1490 |
-
"epoch": 1.3250341030648596,
|
| 1491 |
-
"grad_norm": 0.24521099030971527,
|
| 1492 |
-
"learning_rate": 2.4718231929664025e-05,
|
| 1493 |
-
"loss": 0.03874744176864624,
|
| 1494 |
-
"step": 4220
|
| 1495 |
-
},
|
| 1496 |
-
{
|
| 1497 |
-
"epoch": 1.331314955297997,
|
| 1498 |
-
"grad_norm": 0.13463203608989716,
|
| 1499 |
-
"learning_rate": 2.453040422558084e-05,
|
| 1500 |
-
"loss": 0.03414709270000458,
|
| 1501 |
-
"step": 4240
|
| 1502 |
-
},
|
| 1503 |
-
{
|
| 1504 |
-
"epoch": 1.3375958075311343,
|
| 1505 |
-
"grad_norm": 0.12021703273057938,
|
| 1506 |
-
"learning_rate": 2.4342603034661003e-05,
|
| 1507 |
-
"loss": 0.028204512596130372,
|
| 1508 |
-
"step": 4260
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.3438766597642717,
|
| 1512 |
-
"grad_norm": 0.09300895035266876,
|
| 1513 |
-
"learning_rate": 2.4154838960074117e-05,
|
| 1514 |
-
"loss": 0.03560083508491516,
|
| 1515 |
-
"step": 4280
|
| 1516 |
-
},
|
| 1517 |
-
{
|
| 1518 |
-
"epoch": 1.3501575119974092,
|
| 1519 |
-
"grad_norm": 0.07003216445446014,
|
| 1520 |
-
"learning_rate": 2.3967122602894223e-05,
|
| 1521 |
-
"loss": 0.04477728307247162,
|
| 1522 |
-
"step": 4300
|
| 1523 |
-
},
|
| 1524 |
-
{
|
| 1525 |
-
"epoch": 1.3564383642305464,
|
| 1526 |
-
"grad_norm": 0.15798458456993103,
|
| 1527 |
-
"learning_rate": 2.377946456150128e-05,
|
| 1528 |
-
"loss": 0.05230234265327453,
|
| 1529 |
-
"step": 4320
|
| 1530 |
-
},
|
| 1531 |
-
{
|
| 1532 |
-
"epoch": 1.3627192164636839,
|
| 1533 |
-
"grad_norm": 0.05332406610250473,
|
| 1534 |
-
"learning_rate": 2.3591875430982752e-05,
|
| 1535 |
-
"loss": 0.029861843585968016,
|
| 1536 |
-
"step": 4340
|
| 1537 |
-
},
|
| 1538 |
-
{
|
| 1539 |
-
"epoch": 1.3690000686968213,
|
| 1540 |
-
"grad_norm": 0.05614598095417023,
|
| 1541 |
-
"learning_rate": 2.3404365802535413e-05,
|
| 1542 |
-
"loss": 0.03531986773014069,
|
| 1543 |
-
"step": 4360
|
| 1544 |
-
},
|
| 1545 |
-
{
|
| 1546 |
-
"epoch": 1.3752809209299586,
|
| 1547 |
-
"grad_norm": 0.33550721406936646,
|
| 1548 |
-
"learning_rate": 2.3216946262867396e-05,
|
| 1549 |
-
"loss": 0.031084492802619934,
|
| 1550 |
-
"step": 4380
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.381561773163096,
|
| 1554 |
-
"grad_norm": 0.3401692807674408,
|
| 1555 |
-
"learning_rate": 2.3029627393600484e-05,
|
| 1556 |
-
"loss": 0.02833726108074188,
|
| 1557 |
-
"step": 4400
|
| 1558 |
-
},
|
| 1559 |
-
{
|
| 1560 |
-
"epoch": 1.3878426253962335,
|
| 1561 |
-
"grad_norm": 0.15662887692451477,
|
| 1562 |
-
"learning_rate": 2.2842419770672624e-05,
|
| 1563 |
-
"loss": 0.03976942598819733,
|
| 1564 |
-
"step": 4420
|
| 1565 |
-
},
|
| 1566 |
-
{
|
| 1567 |
-
"epoch": 1.3941234776293707,
|
| 1568 |
-
"grad_norm": 0.0530722439289093,
|
| 1569 |
-
"learning_rate": 2.265533396374085e-05,
|
| 1570 |
-
"loss": 0.03581877648830414,
|
| 1571 |
-
"step": 4440
|
| 1572 |
-
},
|
| 1573 |
-
{
|
| 1574 |
-
"epoch": 1.4004043298625082,
|
| 1575 |
-
"grad_norm": 0.09495271742343903,
|
| 1576 |
-
"learning_rate": 2.246838053558455e-05,
|
| 1577 |
-
"loss": 0.02298358678817749,
|
| 1578 |
-
"step": 4460
|
| 1579 |
-
},
|
| 1580 |
-
{
|
| 1581 |
-
"epoch": 1.4066851820956456,
|
| 1582 |
-
"grad_norm": 0.04610360413789749,
|
| 1583 |
-
"learning_rate": 2.2281570041509024e-05,
|
| 1584 |
-
"loss": 0.04205411374568939,
|
| 1585 |
-
"step": 4480
|
| 1586 |
-
},
|
| 1587 |
-
{
|
| 1588 |
-
"epoch": 1.4129660343287829,
|
| 1589 |
-
"grad_norm": 0.07970083504915237,
|
| 1590 |
-
"learning_rate": 2.209491302874961e-05,
|
| 1591 |
-
"loss": 0.028789132833480835,
|
| 1592 |
-
"step": 4500
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.4192468865619203,
|
| 1596 |
-
"grad_norm": 1.6510498523712158,
|
| 1597 |
-
"learning_rate": 2.1908420035876148e-05,
|
| 1598 |
-
"loss": 0.030281484127044678,
|
| 1599 |
-
"step": 4520
|
| 1600 |
-
},
|
| 1601 |
-
{
|
| 1602 |
-
"epoch": 1.4255277387950578,
|
| 1603 |
-
"grad_norm": 0.061110883951187134,
|
| 1604 |
-
"learning_rate": 2.1722101592197995e-05,
|
| 1605 |
-
"loss": 0.026464346051216125,
|
| 1606 |
-
"step": 4540
|
| 1607 |
-
},
|
| 1608 |
-
{
|
| 1609 |
-
"epoch": 1.431808591028195,
|
| 1610 |
-
"grad_norm": 0.10222575813531876,
|
| 1611 |
-
"learning_rate": 2.1535968217169535e-05,
|
| 1612 |
-
"loss": 0.024321816861629486,
|
| 1613 |
-
"step": 4560
|
| 1614 |
-
},
|
| 1615 |
-
{
|
| 1616 |
-
"epoch": 1.4380894432613325,
|
| 1617 |
-
"grad_norm": 0.13632528483867645,
|
| 1618 |
-
"learning_rate": 2.135003041979625e-05,
|
| 1619 |
-
"loss": 0.022314466536045074,
|
| 1620 |
-
"step": 4580
|
| 1621 |
-
},
|
| 1622 |
-
{
|
| 1623 |
-
"epoch": 1.44437029549447,
|
| 1624 |
-
"grad_norm": 0.3248070478439331,
|
| 1625 |
-
"learning_rate": 2.1164298698041432e-05,
|
| 1626 |
-
"loss": 0.05798411965370178,
|
| 1627 |
-
"step": 4600
|
| 1628 |
-
},
|
| 1629 |
-
{
|
| 1630 |
-
"epoch": 1.4506511477276072,
|
| 1631 |
-
"grad_norm": 0.12926240265369415,
|
| 1632 |
-
"learning_rate": 2.0978783538233405e-05,
|
| 1633 |
-
"loss": 0.02899252772331238,
|
| 1634 |
-
"step": 4620
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 1.4569319999607446,
|
| 1638 |
-
"grad_norm": 0.23640146851539612,
|
| 1639 |
-
"learning_rate": 2.0793495414473506e-05,
|
| 1640 |
-
"loss": 0.020399539172649382,
|
| 1641 |
-
"step": 4640
|
| 1642 |
-
},
|
| 1643 |
-
{
|
| 1644 |
-
"epoch": 1.463212852193882,
|
| 1645 |
-
"grad_norm": 0.07195625454187393,
|
| 1646 |
-
"learning_rate": 2.0608444788044744e-05,
|
| 1647 |
-
"loss": 0.02832261621952057,
|
| 1648 |
-
"step": 4660
|
| 1649 |
-
},
|
| 1650 |
-
{
|
| 1651 |
-
"epoch": 1.4694937044270193,
|
| 1652 |
-
"grad_norm": 0.05917816609144211,
|
| 1653 |
-
"learning_rate": 2.0423642106821118e-05,
|
| 1654 |
-
"loss": 0.017405569553375244,
|
| 1655 |
-
"step": 4680
|
| 1656 |
-
},
|
| 1657 |
-
{
|
| 1658 |
-
"epoch": 1.4757745566601568,
|
| 1659 |
-
"grad_norm": 0.050064459443092346,
|
| 1660 |
-
"learning_rate": 2.0239097804677755e-05,
|
| 1661 |
-
"loss": 0.03790850043296814,
|
| 1662 |
-
"step": 4700
|
| 1663 |
-
},
|
| 1664 |
-
{
|
| 1665 |
-
"epoch": 1.4820554088932942,
|
| 1666 |
-
"grad_norm": 0.06861293315887451,
|
| 1667 |
-
"learning_rate": 2.005482230090181e-05,
|
| 1668 |
-
"loss": 0.02318151295185089,
|
| 1669 |
-
"step": 4720
|
| 1670 |
-
},
|
| 1671 |
-
{
|
| 1672 |
-
"epoch": 1.4883362611264315,
|
| 1673 |
-
"grad_norm": 0.5692355036735535,
|
| 1674 |
-
"learning_rate": 1.987082599960423e-05,
|
| 1675 |
-
"loss": 0.05805472135543823,
|
| 1676 |
-
"step": 4740
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.494617113359569,
|
| 1680 |
-
"grad_norm": 0.10575691610574722,
|
| 1681 |
-
"learning_rate": 1.9687119289132286e-05,
|
| 1682 |
-
"loss": 0.025386866927146912,
|
| 1683 |
-
"step": 4760
|
| 1684 |
-
},
|
| 1685 |
-
{
|
| 1686 |
-
"epoch": 1.5008979655927064,
|
| 1687 |
-
"grad_norm": 0.07561904937028885,
|
| 1688 |
-
"learning_rate": 1.9503712541483105e-05,
|
| 1689 |
-
"loss": 0.02556038498878479,
|
| 1690 |
-
"step": 4780
|
| 1691 |
-
},
|
| 1692 |
-
{
|
| 1693 |
-
"epoch": 1.5071788178258436,
|
| 1694 |
-
"grad_norm": 0.06342057138681412,
|
| 1695 |
-
"learning_rate": 1.9320616111718055e-05,
|
| 1696 |
-
"loss": 0.026987478137016296,
|
| 1697 |
-
"step": 4800
|
| 1698 |
-
},
|
| 1699 |
-
{
|
| 1700 |
-
"epoch": 1.513459670058981,
|
| 1701 |
-
"grad_norm": 0.22881099581718445,
|
| 1702 |
-
"learning_rate": 1.913784033737807e-05,
|
| 1703 |
-
"loss": 0.023304201662540436,
|
| 1704 |
-
"step": 4820
|
| 1705 |
-
},
|
| 1706 |
-
{
|
| 1707 |
-
"epoch": 1.5197405222921185,
|
| 1708 |
-
"grad_norm": 0.34815022349357605,
|
| 1709 |
-
"learning_rate": 1.8955395537900045e-05,
|
| 1710 |
-
"loss": 0.023628510534763336,
|
| 1711 |
-
"step": 4840
|
| 1712 |
-
},
|
| 1713 |
-
{
|
| 1714 |
-
"epoch": 1.5260213745252558,
|
| 1715 |
-
"grad_norm": 0.07004115730524063,
|
| 1716 |
-
"learning_rate": 1.877329201403417e-05,
|
| 1717 |
-
"loss": 0.01844399720430374,
|
| 1718 |
-
"step": 4860
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 1.5323022267583932,
|
| 1722 |
-
"grad_norm": 0.5382964015007019,
|
| 1723 |
-
"learning_rate": 1.8591540047262408e-05,
|
| 1724 |
-
"loss": 0.022429858148097993,
|
| 1725 |
-
"step": 4880
|
| 1726 |
-
},
|
| 1727 |
-
{
|
| 1728 |
-
"epoch": 1.5385830789915307,
|
| 1729 |
-
"grad_norm": 0.10155873745679855,
|
| 1730 |
-
"learning_rate": 1.841014989921792e-05,
|
| 1731 |
-
"loss": 0.019565099477767946,
|
| 1732 |
-
"step": 4900
|
| 1733 |
-
},
|
| 1734 |
-
{
|
| 1735 |
-
"epoch": 1.544863931224668,
|
| 1736 |
-
"grad_norm": 0.45096078515052795,
|
| 1737 |
-
"learning_rate": 1.8229131811105773e-05,
|
| 1738 |
-
"loss": 0.02386317253112793,
|
| 1739 |
-
"step": 4920
|
| 1740 |
-
},
|
| 1741 |
-
{
|
| 1742 |
-
"epoch": 1.5511447834578054,
|
| 1743 |
-
"grad_norm": 1.6493111848831177,
|
| 1744 |
-
"learning_rate": 1.8048496003124692e-05,
|
| 1745 |
-
"loss": 0.022837825119495392,
|
| 1746 |
-
"step": 4940
|
| 1747 |
-
},
|
| 1748 |
-
{
|
| 1749 |
-
"epoch": 1.5574256356909428,
|
| 1750 |
-
"grad_norm": 0.2704576849937439,
|
| 1751 |
-
"learning_rate": 1.786825267389005e-05,
|
| 1752 |
-
"loss": 0.02158147543668747,
|
| 1753 |
-
"step": 4960
|
| 1754 |
-
},
|
| 1755 |
-
{
|
| 1756 |
-
"epoch": 1.56370648792408,
|
| 1757 |
-
"grad_norm": 0.14172030985355377,
|
| 1758 |
-
"learning_rate": 1.768841199985804e-05,
|
| 1759 |
-
"loss": 0.02343100756406784,
|
| 1760 |
-
"step": 4980
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 1.5699873401572177,
|
| 1764 |
-
"grad_norm": 0.3839413523674011,
|
| 1765 |
-
"learning_rate": 1.7508984134751115e-05,
|
| 1766 |
-
"loss": 0.0211731418967247,
|
| 1767 |
-
"step": 5000
|
| 1768 |
-
},
|
| 1769 |
-
{
|
| 1770 |
-
"epoch": 1.576268192390355,
|
| 1771 |
-
"grad_norm": 0.17986074090003967,
|
| 1772 |
-
"learning_rate": 1.7329979208984757e-05,
|
| 1773 |
-
"loss": 0.050633370876312256,
|
| 1774 |
-
"step": 5020
|
| 1775 |
-
},
|
| 1776 |
-
{
|
| 1777 |
-
"epoch": 1.5825490446234922,
|
| 1778 |
-
"grad_norm": 0.11192335188388824,
|
| 1779 |
-
"learning_rate": 1.715140732909545e-05,
|
| 1780 |
-
"loss": 0.019000351428985596,
|
| 1781 |
-
"step": 5040
|
| 1782 |
-
},
|
| 1783 |
-
{
|
| 1784 |
-
"epoch": 1.58882989685663,
|
| 1785 |
-
"grad_norm": 0.06556563079357147,
|
| 1786 |
-
"learning_rate": 1.697327857717012e-05,
|
| 1787 |
-
"loss": 0.031407642364501956,
|
| 1788 |
-
"step": 5060
|
| 1789 |
-
},
|
| 1790 |
-
{
|
| 1791 |
-
"epoch": 1.5951107490897671,
|
| 1792 |
-
"grad_norm": 0.5898942351341248,
|
| 1793 |
-
"learning_rate": 1.67956030102769e-05,
|
| 1794 |
-
"loss": 0.03214563131332397,
|
| 1795 |
-
"step": 5080
|
| 1796 |
-
},
|
| 1797 |
-
{
|
| 1798 |
-
"epoch": 1.6013916013229044,
|
| 1799 |
-
"grad_norm": 0.11632583290338516,
|
| 1800 |
-
"learning_rate": 1.6618390659897297e-05,
|
| 1801 |
-
"loss": 0.12849061489105223,
|
| 1802 |
-
"step": 5100
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 1.607672453556042,
|
| 1806 |
-
"grad_norm": 0.26994606852531433,
|
| 1807 |
-
"learning_rate": 1.64416515313598e-05,
|
| 1808 |
-
"loss": 0.01800369322299957,
|
| 1809 |
-
"step": 5120
|
| 1810 |
-
},
|
| 1811 |
-
{
|
| 1812 |
-
"epoch": 1.6139533057891793,
|
| 1813 |
-
"grad_norm": 0.3220551908016205,
|
| 1814 |
-
"learning_rate": 1.626539560327505e-05,
|
| 1815 |
-
"loss": 0.022022050619125367,
|
| 1816 |
-
"step": 5140
|
| 1817 |
-
},
|
| 1818 |
-
{
|
| 1819 |
-
"epoch": 1.6202341580223165,
|
| 1820 |
-
"grad_norm": 0.16517284512519836,
|
| 1821 |
-
"learning_rate": 1.6089632826972395e-05,
|
| 1822 |
-
"loss": 0.015372823178768157,
|
| 1823 |
-
"step": 5160
|
| 1824 |
-
},
|
| 1825 |
-
{
|
| 1826 |
-
"epoch": 1.6265150102554542,
|
| 1827 |
-
"grad_norm": 0.03851303830742836,
|
| 1828 |
-
"learning_rate": 1.591437312593804e-05,
|
| 1829 |
-
"loss": 0.0164533868432045,
|
| 1830 |
-
"step": 5180
|
| 1831 |
-
},
|
| 1832 |
-
{
|
| 1833 |
-
"epoch": 1.6327958624885914,
|
| 1834 |
-
"grad_norm": 0.3491153419017792,
|
| 1835 |
-
"learning_rate": 1.5739626395254798e-05,
|
| 1836 |
-
"loss": 0.031122979521751405,
|
| 1837 |
-
"step": 5200
|
| 1838 |
-
},
|
| 1839 |
-
{
|
| 1840 |
-
"epoch": 1.6390767147217287,
|
| 1841 |
-
"grad_norm": 0.06409186869859695,
|
| 1842 |
-
"learning_rate": 1.556540250104343e-05,
|
| 1843 |
-
"loss": 0.0328720360994339,
|
| 1844 |
-
"step": 5220
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 1.6453575669548663,
|
| 1848 |
-
"grad_norm": 0.04858129099011421,
|
| 1849 |
-
"learning_rate": 1.539171127990557e-05,
|
| 1850 |
-
"loss": 0.05794671773910522,
|
| 1851 |
-
"step": 5240
|
| 1852 |
-
},
|
| 1853 |
-
{
|
| 1854 |
-
"epoch": 1.6516384191880036,
|
| 1855 |
-
"grad_norm": 0.262850821018219,
|
| 1856 |
-
"learning_rate": 1.5218562538368364e-05,
|
| 1857 |
-
"loss": 0.014299227297306061,
|
| 1858 |
-
"step": 5260
|
| 1859 |
-
},
|
| 1860 |
-
{
|
| 1861 |
-
"epoch": 1.6579192714211408,
|
| 1862 |
-
"grad_norm": 0.12737710773944855,
|
| 1863 |
-
"learning_rate": 1.504596605233084e-05,
|
| 1864 |
-
"loss": 0.02150038182735443,
|
| 1865 |
-
"step": 5280
|
| 1866 |
-
},
|
| 1867 |
-
{
|
| 1868 |
-
"epoch": 1.6642001236542785,
|
| 1869 |
-
"grad_norm": 0.06830194592475891,
|
| 1870 |
-
"learning_rate": 1.4873931566511901e-05,
|
| 1871 |
-
"loss": 0.01579444855451584,
|
| 1872 |
-
"step": 5300
|
| 1873 |
-
},
|
| 1874 |
-
{
|
| 1875 |
-
"epoch": 1.6704809758874157,
|
| 1876 |
-
"grad_norm": 0.08396535366773605,
|
| 1877 |
-
"learning_rate": 1.4702468793900188e-05,
|
| 1878 |
-
"loss": 0.027909812331199647,
|
| 1879 |
-
"step": 5320
|
| 1880 |
-
},
|
| 1881 |
-
{
|
| 1882 |
-
"epoch": 1.676761828120553,
|
| 1883 |
-
"grad_norm": 0.26403504610061646,
|
| 1884 |
-
"learning_rate": 1.4531587415205667e-05,
|
| 1885 |
-
"loss": 0.02287571579217911,
|
| 1886 |
-
"step": 5340
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 1.6830426803536906,
|
| 1890 |
-
"grad_norm": 0.43412601947784424,
|
| 1891 |
-
"learning_rate": 1.436129707831308e-05,
|
| 1892 |
-
"loss": 0.017245300114154816,
|
| 1893 |
-
"step": 5360
|
| 1894 |
-
},
|
| 1895 |
-
{
|
| 1896 |
-
"epoch": 1.6893235325868279,
|
| 1897 |
-
"grad_norm": 0.18354500830173492,
|
| 1898 |
-
"learning_rate": 1.4191607397737206e-05,
|
| 1899 |
-
"loss": 0.031501269340515135,
|
| 1900 |
-
"step": 5380
|
| 1901 |
-
},
|
| 1902 |
-
{
|
| 1903 |
-
"epoch": 1.6956043848199651,
|
| 1904 |
-
"grad_norm": 0.04084882140159607,
|
| 1905 |
-
"learning_rate": 1.402252795408004e-05,
|
| 1906 |
-
"loss": 0.036852356791496274,
|
| 1907 |
-
"step": 5400
|
| 1908 |
-
},
|
| 1909 |
-
{
|
| 1910 |
-
"epoch": 1.7018852370531028,
|
| 1911 |
-
"grad_norm": 0.06115623563528061,
|
| 1912 |
-
"learning_rate": 1.3854068293489909e-05,
|
| 1913 |
-
"loss": 0.028507155179977418,
|
| 1914 |
-
"step": 5420
|
| 1915 |
-
},
|
| 1916 |
-
{
|
| 1917 |
-
"epoch": 1.70816608928624,
|
| 1918 |
-
"grad_norm": 0.32743072509765625,
|
| 1919 |
-
"learning_rate": 1.368623792712243e-05,
|
| 1920 |
-
"loss": 0.014020594954490661,
|
| 1921 |
-
"step": 5440
|
| 1922 |
-
},
|
| 1923 |
-
{
|
| 1924 |
-
"epoch": 1.7144469415193773,
|
| 1925 |
-
"grad_norm": 0.06222687289118767,
|
| 1926 |
-
"learning_rate": 1.3519046330603584e-05,
|
| 1927 |
-
"loss": 0.05104716420173645,
|
| 1928 |
-
"step": 5460
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 1.720727793752515,
|
| 1932 |
-
"grad_norm": 0.06253579258918762,
|
| 1933 |
-
"learning_rate": 1.3352502943494672e-05,
|
| 1934 |
-
"loss": 0.026476505398750304,
|
| 1935 |
-
"step": 5480
|
| 1936 |
-
},
|
| 1937 |
-
{
|
| 1938 |
-
"epoch": 1.7270086459856522,
|
| 1939 |
-
"grad_norm": 0.4662589430809021,
|
| 1940 |
-
"learning_rate": 1.318661716875942e-05,
|
| 1941 |
-
"loss": 0.01999303698539734,
|
| 1942 |
-
"step": 5500
|
| 1943 |
-
},
|
| 1944 |
-
{
|
| 1945 |
-
"epoch": 1.7332894982187894,
|
| 1946 |
-
"grad_norm": 0.1247578039765358,
|
| 1947 |
-
"learning_rate": 1.3021398372233035e-05,
|
| 1948 |
-
"loss": 0.02069215327501297,
|
| 1949 |
-
"step": 5520
|
| 1950 |
-
},
|
| 1951 |
-
{
|
| 1952 |
-
"epoch": 1.739570350451927,
|
| 1953 |
-
"grad_norm": 0.3788617253303528,
|
| 1954 |
-
"learning_rate": 1.2856855882093432e-05,
|
| 1955 |
-
"loss": 0.018105532228946685,
|
| 1956 |
-
"step": 5540
|
| 1957 |
-
},
|
| 1958 |
-
{
|
| 1959 |
-
"epoch": 1.7458512026850643,
|
| 1960 |
-
"grad_norm": 1.8883055448532104,
|
| 1961 |
-
"learning_rate": 1.2692998988334612e-05,
|
| 1962 |
-
"loss": 0.04176854789257049,
|
| 1963 |
-
"step": 5560
|
| 1964 |
-
},
|
| 1965 |
-
{
|
| 1966 |
-
"epoch": 1.7521320549182016,
|
| 1967 |
-
"grad_norm": 0.21989472210407257,
|
| 1968 |
-
"learning_rate": 1.252983694224208e-05,
|
| 1969 |
-
"loss": 0.020176875591278075,
|
| 1970 |
-
"step": 5580
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 1.7584129071513392,
|
| 1974 |
-
"grad_norm": 0.16766223311424255,
|
| 1975 |
-
"learning_rate": 1.236737895587056e-05,
|
| 1976 |
-
"loss": 0.020288291573524474,
|
| 1977 |
-
"step": 5600
|
| 1978 |
-
},
|
| 1979 |
-
{
|
| 1980 |
-
"epoch": 1.7646937593844765,
|
| 1981 |
-
"grad_norm": 0.09795085340738297,
|
| 1982 |
-
"learning_rate": 1.2205634201523922e-05,
|
| 1983 |
-
"loss": 0.015002159774303437,
|
| 1984 |
-
"step": 5620
|
| 1985 |
-
},
|
| 1986 |
-
{
|
| 1987 |
-
"epoch": 1.7709746116176137,
|
| 1988 |
-
"grad_norm": 0.029697278514504433,
|
| 1989 |
-
"learning_rate": 1.2044611811237242e-05,
|
| 1990 |
-
"loss": 0.023318004608154298,
|
| 1991 |
-
"step": 5640
|
| 1992 |
-
},
|
| 1993 |
-
{
|
| 1994 |
-
"epoch": 1.7772554638507514,
|
| 1995 |
-
"grad_norm": 0.3376557528972626,
|
| 1996 |
-
"learning_rate": 1.1884320876261268e-05,
|
| 1997 |
-
"loss": 0.0196805402636528,
|
| 1998 |
-
"step": 5660
|
| 1999 |
-
},
|
| 2000 |
-
{
|
| 2001 |
-
"epoch": 1.7835363160838886,
|
| 2002 |
-
"grad_norm": 0.06531202793121338,
|
| 2003 |
-
"learning_rate": 1.17247704465491e-05,
|
| 2004 |
-
"loss": 0.020646215975284578,
|
| 2005 |
-
"step": 5680
|
| 2006 |
-
},
|
| 2007 |
-
{
|
| 2008 |
-
"epoch": 1.7898171683170259,
|
| 2009 |
-
"grad_norm": 0.36572521924972534,
|
| 2010 |
-
"learning_rate": 1.1565969530245277e-05,
|
| 2011 |
-
"loss": 0.025085961818695067,
|
| 2012 |
-
"step": 5700
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 1.7960980205501635,
|
| 2016 |
-
"grad_norm": 0.1814069002866745,
|
| 2017 |
-
"learning_rate": 1.140792709317713e-05,
|
| 2018 |
-
"loss": 0.030590158700942994,
|
| 2019 |
-
"step": 5720
|
| 2020 |
-
},
|
| 2021 |
-
{
|
| 2022 |
-
"epoch": 1.8023788727833008,
|
| 2023 |
-
"grad_norm": 0.24981825053691864,
|
| 2024 |
-
"learning_rate": 1.12506520583486e-05,
|
| 2025 |
-
"loss": 0.0215068444609642,
|
| 2026 |
-
"step": 5740
|
| 2027 |
-
},
|
| 2028 |
-
{
|
| 2029 |
-
"epoch": 1.808659725016438,
|
| 2030 |
-
"grad_norm": 0.35276567935943604,
|
| 2031 |
-
"learning_rate": 1.1094153305436453e-05,
|
| 2032 |
-
"loss": 0.01926688849925995,
|
| 2033 |
-
"step": 5760
|
| 2034 |
-
},
|
| 2035 |
-
{
|
| 2036 |
-
"epoch": 1.8149405772495757,
|
| 2037 |
-
"grad_norm": 0.02407790906727314,
|
| 2038 |
-
"learning_rate": 1.0938439670288928e-05,
|
| 2039 |
-
"loss": 0.013058166205883025,
|
| 2040 |
-
"step": 5780
|
| 2041 |
-
},
|
| 2042 |
-
{
|
| 2043 |
-
"epoch": 1.821221429482713,
|
| 2044 |
-
"grad_norm": 0.08507027477025986,
|
| 2045 |
-
"learning_rate": 1.0783519944426874e-05,
|
| 2046 |
-
"loss": 0.01877378821372986,
|
| 2047 |
-
"step": 5800
|
| 2048 |
-
},
|
| 2049 |
-
{
|
| 2050 |
-
"epoch": 1.8275022817158502,
|
| 2051 |
-
"grad_norm": 0.18674074113368988,
|
| 2052 |
-
"learning_rate": 1.0629402874547378e-05,
|
| 2053 |
-
"loss": 0.018952280282974243,
|
| 2054 |
-
"step": 5820
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 1.8337831339489878,
|
| 2058 |
-
"grad_norm": 0.057133324444293976,
|
| 2059 |
-
"learning_rate": 1.047609716202996e-05,
|
| 2060 |
-
"loss": 0.030013075470924376,
|
| 2061 |
-
"step": 5840
|
| 2062 |
-
},
|
| 2063 |
-
{
|
| 2064 |
-
"epoch": 1.840063986182125,
|
| 2065 |
-
"grad_norm": 0.10284405946731567,
|
| 2066 |
-
"learning_rate": 1.0323611462445246e-05,
|
| 2067 |
-
"loss": 0.020608581602573395,
|
| 2068 |
-
"step": 5860
|
| 2069 |
-
},
|
| 2070 |
-
{
|
| 2071 |
-
"epoch": 1.8463448384152623,
|
| 2072 |
-
"grad_norm": 0.25261375308036804,
|
| 2073 |
-
"learning_rate": 1.0171954385066303e-05,
|
| 2074 |
-
"loss": 0.02777992784976959,
|
| 2075 |
-
"step": 5880
|
| 2076 |
-
},
|
| 2077 |
-
{
|
| 2078 |
-
"epoch": 1.8526256906484,
|
| 2079 |
-
"grad_norm": 0.38947853446006775,
|
| 2080 |
-
"learning_rate": 1.0021134492382606e-05,
|
| 2081 |
-
"loss": 0.02200171649456024,
|
| 2082 |
-
"step": 5900
|
| 2083 |
-
},
|
| 2084 |
-
{
|
| 2085 |
-
"epoch": 1.8589065428815372,
|
| 2086 |
-
"grad_norm": 0.1173999160528183,
|
| 2087 |
-
"learning_rate": 9.87116029961652e-06,
|
| 2088 |
-
"loss": 0.019530104100704195,
|
| 2089 |
-
"step": 5920
|
| 2090 |
-
},
|
| 2091 |
-
{
|
| 2092 |
-
"epoch": 1.8651873951146745,
|
| 2093 |
-
"grad_norm": 2.3180384635925293,
|
| 2094 |
-
"learning_rate": 9.722040274242594e-06,
|
| 2095 |
-
"loss": 0.028142544627189636,
|
| 2096 |
-
"step": 5940
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 1.8714682473478121,
|
| 2100 |
-
"grad_norm": 0.4677383601665497,
|
| 2101 |
-
"learning_rate": 9.573782835509471e-06,
|
| 2102 |
-
"loss": 0.017200836539268495,
|
| 2103 |
-
"step": 5960
|
| 2104 |
-
},
|
| 2105 |
-
{
|
| 2106 |
-
"epoch": 1.8777490995809494,
|
| 2107 |
-
"grad_norm": 0.19456204771995544,
|
| 2108 |
-
"learning_rate": 9.426396353964565e-06,
|
| 2109 |
-
"loss": 0.015525393187999725,
|
| 2110 |
-
"step": 5980
|
| 2111 |
-
},
|
| 2112 |
-
{
|
| 2113 |
-
"epoch": 1.8840299518140866,
|
| 2114 |
-
"grad_norm": 0.561476469039917,
|
| 2115 |
-
"learning_rate": 9.279889150981421e-06,
|
| 2116 |
-
"loss": 0.04386698603630066,
|
| 2117 |
-
"step": 6000
|
| 2118 |
-
},
|
| 2119 |
-
{
|
| 2120 |
-
"epoch": 1.8903108040472243,
|
| 2121 |
-
"grad_norm": 0.0613991916179657,
|
| 2122 |
-
"learning_rate": 9.134269498289922e-06,
|
| 2123 |
-
"loss": 0.031869342923164366,
|
| 2124 |
-
"step": 6020
|
| 2125 |
-
},
|
| 2126 |
-
{
|
| 2127 |
-
"epoch": 1.8965916562803615,
|
| 2128 |
-
"grad_norm": 0.047774072736501694,
|
| 2129 |
-
"learning_rate": 8.989545617509283e-06,
|
| 2130 |
-
"loss": 0.02780761420726776,
|
| 2131 |
-
"step": 6040
|
| 2132 |
-
},
|
| 2133 |
-
{
|
| 2134 |
-
"epoch": 1.9028725085134988,
|
| 2135 |
-
"grad_norm": 0.33418071269989014,
|
| 2136 |
-
"learning_rate": 8.845725679683822e-06,
|
| 2137 |
-
"loss": 0.04635312259197235,
|
| 2138 |
-
"step": 6060
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 1.9091533607466364,
|
| 2142 |
-
"grad_norm": 0.0721767470240593,
|
| 2143 |
-
"learning_rate": 8.702817804821664e-06,
|
| 2144 |
-
"loss": 0.01725383847951889,
|
| 2145 |
-
"step": 6080
|
| 2146 |
-
},
|
| 2147 |
-
{
|
| 2148 |
-
"epoch": 1.9154342129797737,
|
| 2149 |
-
"grad_norm": 0.04245443269610405,
|
| 2150 |
-
"learning_rate": 8.560830061436265e-06,
|
| 2151 |
-
"loss": 0.02144094407558441,
|
| 2152 |
-
"step": 6100
|
| 2153 |
-
},
|
| 2154 |
-
{
|
| 2155 |
-
"epoch": 1.921715065212911,
|
| 2156 |
-
"grad_norm": 0.1140967458486557,
|
| 2157 |
-
"learning_rate": 8.419770466090896e-06,
|
| 2158 |
-
"loss": 0.014450745284557342,
|
| 2159 |
-
"step": 6120
|
| 2160 |
-
},
|
| 2161 |
-
{
|
| 2162 |
-
"epoch": 1.9279959174460486,
|
| 2163 |
-
"grad_norm": 0.06213388964533806,
|
| 2164 |
-
"learning_rate": 8.27964698294599e-06,
|
| 2165 |
-
"loss": 0.023716664314270018,
|
| 2166 |
-
"step": 6140
|
| 2167 |
-
},
|
| 2168 |
-
{
|
| 2169 |
-
"epoch": 1.9342767696791858,
|
| 2170 |
-
"grad_norm": 0.2430906891822815,
|
| 2171 |
-
"learning_rate": 8.140467523309527e-06,
|
| 2172 |
-
"loss": 0.011777687072753906,
|
| 2173 |
-
"step": 6160
|
| 2174 |
-
},
|
| 2175 |
-
{
|
| 2176 |
-
"epoch": 1.9405576219123233,
|
| 2177 |
-
"grad_norm": 0.044944941997528076,
|
| 2178 |
-
"learning_rate": 8.002239945190345e-06,
|
| 2179 |
-
"loss": 0.01790996491909027,
|
| 2180 |
-
"step": 6180
|
| 2181 |
-
},
|
| 2182 |
-
{
|
| 2183 |
-
"epoch": 1.9468384741454607,
|
| 2184 |
-
"grad_norm": 0.032080262899398804,
|
| 2185 |
-
"learning_rate": 7.864972052854485e-06,
|
| 2186 |
-
"loss": 0.03710399568080902,
|
| 2187 |
-
"step": 6200
|
| 2188 |
-
},
|
| 2189 |
-
{
|
| 2190 |
-
"epoch": 1.953119326378598,
|
| 2191 |
-
"grad_norm": 0.4400108754634857,
|
| 2192 |
-
"learning_rate": 7.72867159638456e-06,
|
| 2193 |
-
"loss": 0.022929078340530394,
|
| 2194 |
-
"step": 6220
|
| 2195 |
-
},
|
| 2196 |
-
{
|
| 2197 |
-
"epoch": 1.9594001786117354,
|
| 2198 |
-
"grad_norm": 0.04446094483137131,
|
| 2199 |
-
"learning_rate": 7.593346271242216e-06,
|
| 2200 |
-
"loss": 0.01109272986650467,
|
| 2201 |
-
"step": 6240
|
| 2202 |
-
},
|
| 2203 |
-
{
|
| 2204 |
-
"epoch": 1.965681030844873,
|
| 2205 |
-
"grad_norm": 0.37326884269714355,
|
| 2206 |
-
"learning_rate": 7.459003717833607e-06,
|
| 2207 |
-
"loss": 0.04479179680347443,
|
| 2208 |
-
"step": 6260
|
| 2209 |
-
},
|
| 2210 |
-
{
|
| 2211 |
-
"epoch": 1.9719618830780101,
|
| 2212 |
-
"grad_norm": 0.03807030990719795,
|
| 2213 |
-
"learning_rate": 7.325651521078053e-06,
|
| 2214 |
-
"loss": 0.020693680644035338,
|
| 2215 |
-
"step": 6280
|
| 2216 |
-
},
|
| 2217 |
-
{
|
| 2218 |
-
"epoch": 1.9782427353111476,
|
| 2219 |
-
"grad_norm": 0.07757691293954849,
|
| 2220 |
-
"learning_rate": 7.193297209979777e-06,
|
| 2221 |
-
"loss": 0.015228751301765441,
|
| 2222 |
-
"step": 6300
|
| 2223 |
-
},
|
| 2224 |
-
{
|
| 2225 |
-
"epoch": 1.984523587544285,
|
| 2226 |
-
"grad_norm": 0.06824570149183273,
|
| 2227 |
-
"learning_rate": 7.061948257202858e-06,
|
| 2228 |
-
"loss": 0.01840972900390625,
|
| 2229 |
-
"step": 6320
|
| 2230 |
-
},
|
| 2231 |
-
{
|
| 2232 |
-
"epoch": 1.9908044397774223,
|
| 2233 |
-
"grad_norm": 0.1820923089981079,
|
| 2234 |
-
"learning_rate": 6.931612078649283e-06,
|
| 2235 |
-
"loss": 0.02284184694290161,
|
| 2236 |
-
"step": 6340
|
| 2237 |
-
},
|
| 2238 |
-
{
|
| 2239 |
-
"epoch": 1.9970852920105597,
|
| 2240 |
-
"grad_norm": 0.0430208295583725,
|
| 2241 |
-
"learning_rate": 6.8022960330402616e-06,
|
| 2242 |
-
"loss": 0.016644990444183348,
|
| 2243 |
-
"step": 6360
|
| 2244 |
-
},
|
| 2245 |
-
{
|
| 2246 |
-
"epoch": 2.003140426116569,
|
| 2247 |
-
"grad_norm": 0.050990670919418335,
|
| 2248 |
-
"learning_rate": 6.674007421500786e-06,
|
| 2249 |
-
"loss": 0.03776961266994476,
|
| 2250 |
-
"step": 6380
|
| 2251 |
-
},
|
| 2252 |
-
{
|
| 2253 |
-
"epoch": 2.009421278349706,
|
| 2254 |
-
"grad_norm": 0.13200590014457703,
|
| 2255 |
-
"learning_rate": 6.546753487147375e-06,
|
| 2256 |
-
"loss": 0.026686957478523253,
|
| 2257 |
-
"step": 6400
|
| 2258 |
-
},
|
| 2259 |
-
{
|
| 2260 |
-
"epoch": 2.0157021305828433,
|
| 2261 |
-
"grad_norm": 0.08269485086202621,
|
| 2262 |
-
"learning_rate": 6.42054141467914e-06,
|
| 2263 |
-
"loss": 0.01832299828529358,
|
| 2264 |
-
"step": 6420
|
| 2265 |
-
},
|
| 2266 |
-
{
|
| 2267 |
-
"epoch": 2.021982982815981,
|
| 2268 |
-
"grad_norm": 0.04629537835717201,
|
| 2269 |
-
"learning_rate": 6.2953783299721486e-06,
|
| 2270 |
-
"loss": 0.01113806888461113,
|
| 2271 |
-
"step": 6440
|
| 2272 |
-
},
|
| 2273 |
-
{
|
| 2274 |
-
"epoch": 2.0282638350491182,
|
| 2275 |
-
"grad_norm": 0.06518969684839249,
|
| 2276 |
-
"learning_rate": 6.171271299677111e-06,
|
| 2277 |
-
"loss": 0.014084245264530181,
|
| 2278 |
-
"step": 6460
|
| 2279 |
-
},
|
| 2280 |
-
{
|
| 2281 |
-
"epoch": 2.0345446872822555,
|
| 2282 |
-
"grad_norm": 0.0387808196246624,
|
| 2283 |
-
"learning_rate": 6.048227330820375e-06,
|
| 2284 |
-
"loss": 0.01689321845769882,
|
| 2285 |
-
"step": 6480
|
| 2286 |
-
},
|
| 2287 |
-
{
|
| 2288 |
-
"epoch": 2.040825539515393,
|
| 2289 |
-
"grad_norm": 0.04762839153409004,
|
| 2290 |
-
"learning_rate": 5.9262533704083135e-06,
|
| 2291 |
-
"loss": 0.009809207171201706,
|
| 2292 |
-
"step": 6500
|
| 2293 |
-
},
|
| 2294 |
-
{
|
| 2295 |
-
"epoch": 2.0471063917485304,
|
| 2296 |
-
"grad_norm": 0.051051948219537735,
|
| 2297 |
-
"learning_rate": 5.805356305035142e-06,
|
| 2298 |
-
"loss": 0.013716338574886322,
|
| 2299 |
-
"step": 6520
|
| 2300 |
-
},
|
| 2301 |
-
{
|
| 2302 |
-
"epoch": 2.0533872439816676,
|
| 2303 |
-
"grad_norm": 0.08170059323310852,
|
| 2304 |
-
"learning_rate": 5.685542960494042e-06,
|
| 2305 |
-
"loss": 0.014487373828887939,
|
| 2306 |
-
"step": 6540
|
| 2307 |
-
},
|
| 2308 |
-
{
|
| 2309 |
-
"epoch": 2.0596680962148053,
|
| 2310 |
-
"grad_norm": 0.09992720931768417,
|
| 2311 |
-
"learning_rate": 5.566820101391826e-06,
|
| 2312 |
-
"loss": 0.017018157243728637,
|
| 2313 |
-
"step": 6560
|
| 2314 |
-
},
|
| 2315 |
-
{
|
| 2316 |
-
"epoch": 2.0659489484479425,
|
| 2317 |
-
"grad_norm": 0.16989538073539734,
|
| 2318 |
-
"learning_rate": 5.449194430766991e-06,
|
| 2319 |
-
"loss": 0.021767497062683105,
|
| 2320 |
-
"step": 6580
|
| 2321 |
-
},
|
| 2322 |
-
{
|
| 2323 |
-
"epoch": 2.0722298006810798,
|
| 2324 |
-
"grad_norm": 0.05336920917034149,
|
| 2325 |
-
"learning_rate": 5.332672589711274e-06,
|
| 2326 |
-
"loss": 0.016090822219848634,
|
| 2327 |
-
"step": 6600
|
| 2328 |
-
},
|
| 2329 |
-
{
|
| 2330 |
-
"epoch": 2.0785106529142174,
|
| 2331 |
-
"grad_norm": 0.09633775055408478,
|
| 2332 |
-
"learning_rate": 5.217261156994699e-06,
|
| 2333 |
-
"loss": 0.016508692502975465,
|
| 2334 |
-
"step": 6620
|
| 2335 |
-
}
|
| 2336 |
-
],
|
| 2337 |
-
"logging_steps": 20,
|
| 2338 |
-
"max_steps": 8362,
|
| 2339 |
-
"num_input_tokens_seen": 0,
|
| 2340 |
-
"num_train_epochs": 3,
|
| 2341 |
-
"save_steps": 1000000000,
|
| 2342 |
-
"stateful_callbacks": {
|
| 2343 |
-
"TrainerControl": {
|
| 2344 |
-
"args": {
|
| 2345 |
-
"should_epoch_stop": false,
|
| 2346 |
-
"should_evaluate": false,
|
| 2347 |
-
"should_log": false,
|
| 2348 |
-
"should_save": true,
|
| 2349 |
-
"should_training_stop": false
|
| 2350 |
-
},
|
| 2351 |
-
"attributes": {}
|
| 2352 |
-
}
|
| 2353 |
-
},
|
| 2354 |
-
"total_flos": 2.092675047796224e+16,
|
| 2355 |
-
"train_batch_size": 1,
|
| 2356 |
-
"trial_name": null,
|
| 2357 |
-
"trial_params": null
|
| 2358 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|