pushing run wordflow_run_2026_07_20_06_58
Browse files- checkpoint_best.pt +3 -0
- config.json +31 -0
- tokenizer.json +0 -0
- training_logs.json +327 -0
checkpoint_best.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eb786600ef248df08bd75042a6eb1ead06b00d7442d2cb3548e87a05ffe44e14
|
| 3 |
+
size 20648869
|
config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"seed": 1212,
|
| 3 |
+
"use_accelerator": true,
|
| 4 |
+
"runs_dir": "runs/",
|
| 5 |
+
"data": {
|
| 6 |
+
"train_data_path": "data/processed/train.txt",
|
| 7 |
+
"val_data_path": "data/processed/val.txt",
|
| 8 |
+
"max_vocab_size": 25000,
|
| 9 |
+
"tokenizer_lowercase": false
|
| 10 |
+
},
|
| 11 |
+
"model": {
|
| 12 |
+
"embedding_dim": 400,
|
| 13 |
+
"hidden_dim": 400,
|
| 14 |
+
"num_layers": 1,
|
| 15 |
+
"tie_weights": true,
|
| 16 |
+
"emb_dropout_p": 0.2,
|
| 17 |
+
"rnn_dropout_p": 0.25,
|
| 18 |
+
"out_dropout_p": 0.2
|
| 19 |
+
},
|
| 20 |
+
"train": {
|
| 21 |
+
"seq_len": 256,
|
| 22 |
+
"batch_size": 128,
|
| 23 |
+
"n_epochs": 80,
|
| 24 |
+
"lr": 0.001,
|
| 25 |
+
"weight_decay": 0.01,
|
| 26 |
+
"eta_min": 5e-05,
|
| 27 |
+
"early_stopping_patience": 4,
|
| 28 |
+
"grad_clip_norm": 1.0,
|
| 29 |
+
"enable_mixed_precision": true
|
| 30 |
+
}
|
| 31 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
training_logs.json
ADDED
|
@@ -0,0 +1,327 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"train_loss": [
|
| 3 |
+
5.7151332003005,
|
| 4 |
+
4.189456285314357,
|
| 5 |
+
3.698368394628484,
|
| 6 |
+
3.393957954771975,
|
| 7 |
+
3.205798461082134,
|
| 8 |
+
3.0755185624386403,
|
| 9 |
+
2.9776548781293504,
|
| 10 |
+
2.896381350273782,
|
| 11 |
+
2.8301105575358614,
|
| 12 |
+
2.7728853986618365,
|
| 13 |
+
2.7234079787071717,
|
| 14 |
+
2.680263151513769,
|
| 15 |
+
2.6403521476907934,
|
| 16 |
+
2.605504444304933,
|
| 17 |
+
2.5738011750769108,
|
| 18 |
+
2.5445751484404218,
|
| 19 |
+
2.518137865878166,
|
| 20 |
+
2.494830164503544,
|
| 21 |
+
2.472218729080038,
|
| 22 |
+
2.450672078639903,
|
| 23 |
+
2.4311314542242823,
|
| 24 |
+
2.412825553975207,
|
| 25 |
+
2.395720403245155,
|
| 26 |
+
2.3794552772603135,
|
| 27 |
+
2.364531354701265,
|
| 28 |
+
2.3507255645508462,
|
| 29 |
+
2.3380369176255895,
|
| 30 |
+
2.3246318456974437,
|
| 31 |
+
2.3137892028118703,
|
| 32 |
+
2.30237237696952,
|
| 33 |
+
2.2920593632028456,
|
| 34 |
+
2.2828418549070966,
|
| 35 |
+
2.2732821855139225,
|
| 36 |
+
2.264455059741406,
|
| 37 |
+
2.256133515784081,
|
| 38 |
+
2.2488109735732382,
|
| 39 |
+
2.240724890790087,
|
| 40 |
+
2.233784703498191,
|
| 41 |
+
2.2271454511804785,
|
| 42 |
+
2.2200561584310328,
|
| 43 |
+
2.2150913958853864,
|
| 44 |
+
2.2088582997626447,
|
| 45 |
+
2.203660622556159,
|
| 46 |
+
2.198605068186496,
|
| 47 |
+
2.193435803372809,
|
| 48 |
+
2.1884521524956884,
|
| 49 |
+
2.1848263258629657,
|
| 50 |
+
2.179438033002488,
|
| 51 |
+
2.1760000396282115,
|
| 52 |
+
2.172977049300011,
|
| 53 |
+
2.1684880535653295,
|
| 54 |
+
2.165035402521174,
|
| 55 |
+
2.1621104757836522,
|
| 56 |
+
2.158827279476409,
|
| 57 |
+
2.1561670531617834,
|
| 58 |
+
2.1529580963418837,
|
| 59 |
+
2.1506490479124354,
|
| 60 |
+
2.148303678695192,
|
| 61 |
+
2.145878122207966,
|
| 62 |
+
2.1434922192959074,
|
| 63 |
+
2.1420343251938516,
|
| 64 |
+
2.139381418836878,
|
| 65 |
+
2.1379473691290998
|
| 66 |
+
],
|
| 67 |
+
"val_loss": [
|
| 68 |
+
4.589956140518188,
|
| 69 |
+
4.498831701278687,
|
| 70 |
+
3.450820231437683,
|
| 71 |
+
3.223701071739197,
|
| 72 |
+
3.079462504386902,
|
| 73 |
+
2.977644419670105,
|
| 74 |
+
2.8982948541641234,
|
| 75 |
+
2.8353756427764893,
|
| 76 |
+
2.782619667053223,
|
| 77 |
+
2.7403300523757936,
|
| 78 |
+
2.702825403213501,
|
| 79 |
+
2.673724317550659,
|
| 80 |
+
2.647006702423096,
|
| 81 |
+
2.623520851135254,
|
| 82 |
+
2.6043602228164673,
|
| 83 |
+
2.585438299179077,
|
| 84 |
+
2.569737935066223,
|
| 85 |
+
2.5546409845352174,
|
| 86 |
+
2.541029667854309,
|
| 87 |
+
2.52796356678009,
|
| 88 |
+
2.5161680221557616,
|
| 89 |
+
2.5063876867294312,
|
| 90 |
+
2.496633267402649,
|
| 91 |
+
2.4888595581054687,
|
| 92 |
+
2.4822137117385865,
|
| 93 |
+
2.4775348901748657,
|
| 94 |
+
2.469939041137695,
|
| 95 |
+
2.4642407655715943,
|
| 96 |
+
2.460089588165283,
|
| 97 |
+
2.455437707901001,
|
| 98 |
+
2.450346279144287,
|
| 99 |
+
2.4466224908828735,
|
| 100 |
+
2.443621039390564,
|
| 101 |
+
2.4403581619262695,
|
| 102 |
+
2.4369861125946044,
|
| 103 |
+
2.4344600677490233,
|
| 104 |
+
2.4310266017913817,
|
| 105 |
+
2.42836012840271,
|
| 106 |
+
2.4268747091293337,
|
| 107 |
+
2.4252954959869384,
|
| 108 |
+
2.4236485719680787,
|
| 109 |
+
2.4222830533981323,
|
| 110 |
+
2.421295237541199,
|
| 111 |
+
2.4200443267822265,
|
| 112 |
+
2.4181291103363036,
|
| 113 |
+
2.41718168258667,
|
| 114 |
+
2.415796399116516,
|
| 115 |
+
2.4140535593032837,
|
| 116 |
+
2.41275897026062,
|
| 117 |
+
2.4116687297821047,
|
| 118 |
+
2.4105141878128054,
|
| 119 |
+
2.40928852558136,
|
| 120 |
+
2.408313775062561,
|
| 121 |
+
2.4072006225585936,
|
| 122 |
+
2.406491279602051,
|
| 123 |
+
2.4066580057144167,
|
| 124 |
+
2.405950331687927,
|
| 125 |
+
2.4055263996124268,
|
| 126 |
+
2.404764199256897,
|
| 127 |
+
2.4043495416641236,
|
| 128 |
+
2.403930592536926,
|
| 129 |
+
2.403970170021057,
|
| 130 |
+
2.4034884929656983
|
| 131 |
+
],
|
| 132 |
+
"val_perplexity": [
|
| 133 |
+
98.49011034201129,
|
| 134 |
+
89.91202581038705,
|
| 135 |
+
31.526240520127526,
|
| 136 |
+
25.120922678041623,
|
| 137 |
+
21.74671049281597,
|
| 138 |
+
19.641494989744096,
|
| 139 |
+
18.14318220700166,
|
| 140 |
+
17.036798848958792,
|
| 141 |
+
16.161302774108204,
|
| 142 |
+
15.492097456191711,
|
| 143 |
+
14.921832414408936,
|
| 144 |
+
14.493848493032438,
|
| 145 |
+
14.111734741468034,
|
| 146 |
+
13.784170259773791,
|
| 147 |
+
13.522571102208923,
|
| 148 |
+
13.269103652854021,
|
| 149 |
+
13.06240079514605,
|
| 150 |
+
12.86667949129843,
|
| 151 |
+
12.692733544337681,
|
| 152 |
+
12.527967772157874,
|
| 153 |
+
12.381061691175436,
|
| 154 |
+
12.260560984108105,
|
| 155 |
+
12.14154772735496,
|
| 156 |
+
12.047528776369004,
|
| 157 |
+
11.967728215867426,
|
| 158 |
+
11.911864141950842,
|
| 159 |
+
11.821726190702183,
|
| 160 |
+
11.754554300927456,
|
| 161 |
+
11.705860199544748,
|
| 162 |
+
11.651532400684747,
|
| 163 |
+
11.592360216864181,
|
| 164 |
+
11.54927299560035,
|
| 165 |
+
11.514660382956134,
|
| 166 |
+
11.477150685135939,
|
| 167 |
+
11.43851434527095,
|
| 168 |
+
11.409656608381601,
|
| 169 |
+
11.370549116368428,
|
| 170 |
+
11.340270236589069,
|
| 171 |
+
11.323437685407361,
|
| 172 |
+
11.305569676208389,
|
| 173 |
+
11.286965585928396,
|
| 174 |
+
11.271563543103484,
|
| 175 |
+
11.26043481137694,
|
| 176 |
+
11.2463578186871,
|
| 177 |
+
11.224839222198575,
|
| 178 |
+
11.21420953426007,
|
| 179 |
+
11.198685430286304,
|
| 180 |
+
11.179184913537071,
|
| 181 |
+
11.164721827138647,
|
| 182 |
+
11.152556228390464,
|
| 183 |
+
11.139687564295668,
|
| 184 |
+
11.126042433846722,
|
| 185 |
+
11.115202602136222,
|
| 186 |
+
11.102836570444024,
|
| 187 |
+
11.094963644157655,
|
| 188 |
+
11.096813618528119,
|
| 189 |
+
11.088963469754303,
|
| 190 |
+
11.08426349876004,
|
| 191 |
+
11.075818288060649,
|
| 192 |
+
11.07122656797292,
|
| 193 |
+
11.066589258731419,
|
| 194 |
+
11.06702725515953,
|
| 195 |
+
11.061697805699493
|
| 196 |
+
],
|
| 197 |
+
"lr": [
|
| 198 |
+
0.001,
|
| 199 |
+
0.0009996337922143434,
|
| 200 |
+
0.0009985357335232357,
|
| 201 |
+
0.00099670751705359,
|
| 202 |
+
0.0009941519617826901,
|
| 203 |
+
0.0009908730081915342,
|
| 204 |
+
0.000986875712188896,
|
| 205 |
+
0.0009821662373154822,
|
| 206 |
+
0.0009767518452401978,
|
| 207 |
+
0.0009706408845631798,
|
| 208 |
+
0.000963842777942861,
|
| 209 |
+
0.0009563680075669134,
|
| 210 |
+
0.0009482280989894746,
|
| 211 |
+
0.0009394356033595785,
|
| 212 |
+
0.0009300040780681936,
|
| 213 |
+
0.0009199480658437089,
|
| 214 |
+
0.0009092830723281,
|
| 215 |
+
0.0008980255421683537,
|
| 216 |
+
0.0008861928336600147,
|
| 217 |
+
0.0008738031919819506,
|
| 218 |
+
0.00086087572106361,
|
| 219 |
+
0.0008474303541281473,
|
| 220 |
+
0.0008334878229568373,
|
| 221 |
+
0.000819069625922171,
|
| 222 |
+
0.0008041979948389247,
|
| 223 |
+
0.000788895860684311,
|
| 224 |
+
0.0007731868182400757,
|
| 225 |
+
0.0007570950897110536,
|
| 226 |
+
0.0007406454873762847,
|
| 227 |
+
0.0007238633753302783,
|
| 228 |
+
0.0007067746303734176,
|
| 229 |
+
0.0006894056021118091,
|
| 230 |
+
0.0006717830723281,
|
| 231 |
+
0.0006539342136859102,
|
| 232 |
+
0.0006358865478315551,
|
| 233 |
+
0.0006176679029576609,
|
| 234 |
+
0.0005993063708941097,
|
| 235 |
+
0.0005808302637924728,
|
| 236 |
+
0.0005622680704707263,
|
| 237 |
+
0.0005436484124855575,
|
| 238 |
+
0.0005249999999999999,
|
| 239 |
+
0.0005063515875144424,
|
| 240 |
+
0.00048773192952927363,
|
| 241 |
+
0.0004691697362075271,
|
| 242 |
+
0.0004506936291058904,
|
| 243 |
+
0.00043233209704233907,
|
| 244 |
+
0.00041411345216844496,
|
| 245 |
+
0.0003960657863140896,
|
| 246 |
+
0.0003782169276719,
|
| 247 |
+
0.0003605943978881909,
|
| 248 |
+
0.0003432253696265824,
|
| 249 |
+
0.0003261366246697218,
|
| 250 |
+
0.0003093545126237153,
|
| 251 |
+
0.00029290491028894635,
|
| 252 |
+
0.0002768131817599243,
|
| 253 |
+
0.00026110413931568893,
|
| 254 |
+
0.00024580200516107533,
|
| 255 |
+
0.000230930374077829,
|
| 256 |
+
0.00021651217704316285,
|
| 257 |
+
0.0002025696458718528,
|
| 258 |
+
0.00018912427893638998,
|
| 259 |
+
0.00017619680801804933,
|
| 260 |
+
0.00016380716633998532
|
| 261 |
+
],
|
| 262 |
+
"epoch_time": [
|
| 263 |
+
18.687887,
|
| 264 |
+
18.271976,
|
| 265 |
+
19.295498,
|
| 266 |
+
20.159063,
|
| 267 |
+
21.511919,
|
| 268 |
+
21.237398,
|
| 269 |
+
20.360741,
|
| 270 |
+
20.582748,
|
| 271 |
+
20.994744,
|
| 272 |
+
20.833688,
|
| 273 |
+
20.590812,
|
| 274 |
+
20.656061,
|
| 275 |
+
20.826784,
|
| 276 |
+
20.841734,
|
| 277 |
+
20.759313,
|
| 278 |
+
20.633515,
|
| 279 |
+
20.608371,
|
| 280 |
+
20.576107,
|
| 281 |
+
20.510795,
|
| 282 |
+
20.568413,
|
| 283 |
+
20.576285,
|
| 284 |
+
20.627223,
|
| 285 |
+
20.626052,
|
| 286 |
+
20.684826,
|
| 287 |
+
20.669935,
|
| 288 |
+
20.657214,
|
| 289 |
+
20.721946,
|
| 290 |
+
20.747953,
|
| 291 |
+
20.712902,
|
| 292 |
+
20.740117,
|
| 293 |
+
20.696147,
|
| 294 |
+
20.65011,
|
| 295 |
+
20.649358,
|
| 296 |
+
20.600996,
|
| 297 |
+
20.603366,
|
| 298 |
+
20.561034,
|
| 299 |
+
20.597736,
|
| 300 |
+
20.559257,
|
| 301 |
+
20.503821,
|
| 302 |
+
20.547216,
|
| 303 |
+
20.576612,
|
| 304 |
+
20.543663,
|
| 305 |
+
20.61156,
|
| 306 |
+
20.677359,
|
| 307 |
+
20.633286,
|
| 308 |
+
20.620682,
|
| 309 |
+
20.60477,
|
| 310 |
+
20.588634,
|
| 311 |
+
20.630487,
|
| 312 |
+
20.656216,
|
| 313 |
+
20.664338,
|
| 314 |
+
20.593479,
|
| 315 |
+
20.654712,
|
| 316 |
+
20.594688,
|
| 317 |
+
20.512631,
|
| 318 |
+
20.585156,
|
| 319 |
+
20.590198,
|
| 320 |
+
20.629543,
|
| 321 |
+
20.696788,
|
| 322 |
+
20.792584,
|
| 323 |
+
20.701093,
|
| 324 |
+
20.677725,
|
| 325 |
+
20.688729
|
| 326 |
+
]
|
| 327 |
+
}
|