Model save

all_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.9987959060806744,
     "total_flos": 4.152545054111269e+19,
-    "train_loss": 0.05819672941988573,
-    "train_runtime": 6053.7503,
-    "train_samples_per_second": 52.671,
-    "train_steps_per_second": 0.411
+    "train_loss": 0.0599212793472511,
+    "train_runtime": 7516.3059,
+    "train_samples_per_second": 42.422,
+    "train_steps_per_second": 0.331
 }

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 2.9987959060806744,
     "total_flos": 4.152545054111269e+19,
-    "train_loss": 0.05819672941988573,
-    "train_runtime": 6053.7503,
-    "train_samples_per_second": 52.671,
-    "train_steps_per_second": 0.411
+    "train_loss": 0.0599212793472511,
+    "train_runtime": 7516.3059,
+    "train_samples_per_second": 42.422,
+    "train_steps_per_second": 0.331
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.9948107465549906,
+  "best_metric": 0.994895873653879,
   "best_model_checkpoint": "convnext-base-224-finetuned-eurosat/checkpoint-2490",
   "epoch": 2.9987959060806744,
   "eval_steps": 500,
@@ -10,1782 +10,1782 @@
   "log_history": [
     {
       "epoch": 0.012040939193257074,
-      "grad_norm": 14.643075942993164,
+      "grad_norm": 13.194865226745605,
       "learning_rate": 2.0080321285140564e-06,
-      "loss": 1.4308,
+      "loss": 1.3766,
       "step": 10
     },
     {
       "epoch": 0.024081878386514148,
-      "grad_norm": 13.680133819580078,
+      "grad_norm": 10.99959945678711,
       "learning_rate": 4.016064257028113e-06,
-      "loss": 1.3994,
+      "loss": 1.3446,
       "step": 20
     },
     {
       "epoch": 0.036122817579771226,
-      "grad_norm": 19.79886817932129,
+      "grad_norm": 10.659394264221191,
       "learning_rate": 6.024096385542169e-06,
-      "loss": 1.338,
+      "loss": 1.285,
       "step": 30
     },
     {
       "epoch": 0.048163756773028296,
-      "grad_norm": 16.858301162719727,
+      "grad_norm": 14.758916854858398,
       "learning_rate": 8.032128514056226e-06,
-      "loss": 1.2521,
+      "loss": 1.2002,
       "step": 40
     },
     {
       "epoch": 0.060204695966285374,
-      "grad_norm": 13.908726692199707,
+      "grad_norm": 16.772724151611328,
       "learning_rate": 1.0040160642570281e-05,
-      "loss": 1.1221,
+      "loss": 1.0747,
       "step": 50
     },
     {
       "epoch": 0.07224563515954245,
-      "grad_norm": 21.977474212646484,
+      "grad_norm": 21.30389404296875,
       "learning_rate": 1.2048192771084338e-05,
-      "loss": 0.9512,
+      "loss": 0.9052,
       "step": 60
     },
     {
       "epoch": 0.08428657435279951,
-      "grad_norm": 17.320178985595703,
+      "grad_norm": 26.00191879272461,
       "learning_rate": 1.4056224899598394e-05,
-      "loss": 0.7717,
+      "loss": 0.7216,
       "step": 70
     },
     {
       "epoch": 0.09632751354605659,
-      "grad_norm": 12.808456420898438,
+      "grad_norm": 20.422157287597656,
       "learning_rate": 1.606425702811245e-05,
-      "loss": 0.597,
+      "loss": 0.5357,
       "step": 80
     },
     {
       "epoch": 0.10836845273931367,
-      "grad_norm": 12.416606903076172,
+      "grad_norm": 21.17734718322754,
       "learning_rate": 1.8072289156626505e-05,
-      "loss": 0.4373,
+      "loss": 0.414,
       "step": 90
     },
     {
       "epoch": 0.12040939193257075,
-      "grad_norm": 11.89809513092041,
+      "grad_norm": 24.643775939941406,
       "learning_rate": 2.0080321285140562e-05,
-      "loss": 0.347,
+      "loss": 0.3476,
       "step": 100
     },
     {
       "epoch": 0.13245033112582782,
-      "grad_norm": 20.761890411376953,
+      "grad_norm": 34.078800201416016,
       "learning_rate": 2.208835341365462e-05,
-      "loss": 0.2527,
+      "loss": 0.266,
       "step": 110
     },
     {
       "epoch": 0.1444912703190849,
-      "grad_norm": 26.88390350341797,
+      "grad_norm": 60.97035598754883,
       "learning_rate": 2.4096385542168677e-05,
-      "loss": 0.1933,
+      "loss": 0.2241,
       "step": 120
     },
     {
       "epoch": 0.15653220951234195,
-      "grad_norm": 21.959823608398438,
+      "grad_norm": 39.30607223510742,
       "learning_rate": 2.6104417670682734e-05,
-      "loss": 0.1726,
+      "loss": 0.1988,
       "step": 130
     },
     {
       "epoch": 0.16857314870559903,
-      "grad_norm": 34.98974609375,
+      "grad_norm": 34.965641021728516,
       "learning_rate": 2.8112449799196788e-05,
-      "loss": 0.1758,
+      "loss": 0.1989,
       "step": 140
     },
     {
       "epoch": 0.1806140878988561,
-      "grad_norm": 15.278114318847656,
+      "grad_norm": 26.677417755126953,
       "learning_rate": 3.012048192771085e-05,
-      "loss": 0.1277,
+      "loss": 0.1652,
       "step": 150
     },
     {
       "epoch": 0.19265502709211318,
-      "grad_norm": 68.78533172607422,
+      "grad_norm": 21.16180992126465,
       "learning_rate": 3.21285140562249e-05,
-      "loss": 0.1371,
+      "loss": 0.1566,
       "step": 160
     },
     {
       "epoch": 0.20469596628537026,
-      "grad_norm": 17.30523109436035,
+      "grad_norm": 35.19940948486328,
       "learning_rate": 3.413654618473896e-05,
-      "loss": 0.1391,
+      "loss": 0.1497,
       "step": 170
     },
     {
       "epoch": 0.21673690547862734,
-      "grad_norm": 18.58247184753418,
+      "grad_norm": 19.098440170288086,
       "learning_rate": 3.614457831325301e-05,
-      "loss": 0.1112,
+      "loss": 0.1356,
       "step": 180
     },
     {
       "epoch": 0.22877784467188442,
-      "grad_norm": 20.406381607055664,
+      "grad_norm": 42.77901077270508,
       "learning_rate": 3.815261044176707e-05,
-      "loss": 0.0882,
+      "loss": 0.1172,
       "step": 190
     },
     {
       "epoch": 0.2408187838651415,
-      "grad_norm": 27.328937530517578,
+      "grad_norm": 11.036260604858398,
       "learning_rate": 4.0160642570281125e-05,
-      "loss": 0.0827,
+      "loss": 0.0962,
       "step": 200
     },
     {
       "epoch": 0.25285972305839854,
-      "grad_norm": 7.420236587524414,
+      "grad_norm": 23.800642013549805,
       "learning_rate": 4.2168674698795186e-05,
-      "loss": 0.0782,
+      "loss": 0.0878,
       "step": 210
     },
     {
       "epoch": 0.26490066225165565,
-      "grad_norm": 30.249544143676758,
+      "grad_norm": 16.770301818847656,
       "learning_rate": 4.417670682730924e-05,
-      "loss": 0.0747,
+      "loss": 0.084,
       "step": 220
     },
     {
       "epoch": 0.2769416014449127,
-      "grad_norm": 6.613826274871826,
+      "grad_norm": 14.868804931640625,
       "learning_rate": 4.61847389558233e-05,
-      "loss": 0.0831,
+      "loss": 0.107,
       "step": 230
     },
     {
       "epoch": 0.2889825406381698,
-      "grad_norm": 17.4040584564209,
+      "grad_norm": 10.135931015014648,
       "learning_rate": 4.8192771084337354e-05,
-      "loss": 0.0731,
+      "loss": 0.0823,
       "step": 240
     },
     {
       "epoch": 0.30102347983142685,
-      "grad_norm": 3.861469268798828,
+      "grad_norm": 29.757469177246094,
       "learning_rate": 4.9977688531905406e-05,
-      "loss": 0.0561,
+      "loss": 0.07,
       "step": 250
     },
     {
       "epoch": 0.3130644190246839,
-      "grad_norm": 7.319100856781006,
+      "grad_norm": 8.043286323547363,
       "learning_rate": 4.97545738509594e-05,
-      "loss": 0.0505,
+      "loss": 0.0655,
       "step": 260
     },
     {
       "epoch": 0.325105358217941,
-      "grad_norm": 17.325563430786133,
+      "grad_norm": 11.818015098571777,
       "learning_rate": 4.953145917001339e-05,
-      "loss": 0.0519,
+      "loss": 0.0571,
       "step": 270
     },
     {
       "epoch": 0.33714629741119806,
-      "grad_norm": 18.715402603149414,
+      "grad_norm": 24.017484664916992,
       "learning_rate": 4.930834448906738e-05,
-      "loss": 0.0637,
+      "loss": 0.0673,
       "step": 280
     },
     {
       "epoch": 0.34918723660445516,
-      "grad_norm": 7.095179080963135,
+      "grad_norm": 9.49643325805664,
       "learning_rate": 4.908522980812137e-05,
-      "loss": 0.0624,
+      "loss": 0.0683,
       "step": 290
     },
     {
       "epoch": 0.3612281757977122,
-      "grad_norm": 7.166253089904785,
+      "grad_norm": 22.132099151611328,
       "learning_rate": 4.886211512717537e-05,
-      "loss": 0.0681,
+      "loss": 0.0735,
       "step": 300
     },
     {
       "epoch": 0.3732691149909693,
-      "grad_norm": 14.443611145019531,
+      "grad_norm": 17.463157653808594,
       "learning_rate": 4.8639000446229364e-05,
-      "loss": 0.0425,
+      "loss": 0.0448,
       "step": 310
     },
     {
       "epoch": 0.38531005418422637,
-      "grad_norm": 9.1868257522583,
+      "grad_norm": 11.628926277160645,
       "learning_rate": 4.8415885765283355e-05,
-      "loss": 0.0565,
+      "loss": 0.0638,
       "step": 320
     },
     {
       "epoch": 0.3973509933774834,
-      "grad_norm": 11.00070571899414,
+      "grad_norm": 17.63985252380371,
       "learning_rate": 4.8192771084337354e-05,
-      "loss": 0.0598,
+      "loss": 0.0587,
       "step": 330
     },
     {
       "epoch": 0.4093919325707405,
-      "grad_norm": 5.017125129699707,
+      "grad_norm": 12.73922061920166,
       "learning_rate": 4.7969656403391346e-05,
-      "loss": 0.0623,
+      "loss": 0.072,
       "step": 340
     },
     {
       "epoch": 0.4214328717639976,
-      "grad_norm": 10.400481224060059,
+      "grad_norm": 9.357921600341797,
       "learning_rate": 4.774654172244534e-05,
-      "loss": 0.0259,
+      "loss": 0.0305,
       "step": 350
     },
     {
       "epoch": 0.4334738109572547,
-      "grad_norm": 20.244483947753906,
+      "grad_norm": 10.350850105285645,
       "learning_rate": 4.7523427041499336e-05,
-      "loss": 0.051,
+      "loss": 0.0435,
       "step": 360
     },
     {
       "epoch": 0.44551475015051173,
-      "grad_norm": 19.375219345092773,
+      "grad_norm": 18.032106399536133,
       "learning_rate": 4.730031236055333e-05,
-      "loss": 0.0321,
+      "loss": 0.0353,
       "step": 370
     },
     {
       "epoch": 0.45755568934376883,
-      "grad_norm": 9.210841178894043,
+      "grad_norm": 18.358362197875977,
       "learning_rate": 4.707719767960732e-05,
-      "loss": 0.0312,
+      "loss": 0.0396,
       "step": 380
     },
     {
       "epoch": 0.4695966285370259,
-      "grad_norm": 11.146401405334473,
+      "grad_norm": 22.580930709838867,
       "learning_rate": 4.685408299866131e-05,
-      "loss": 0.0399,
+      "loss": 0.0569,
       "step": 390
     },
     {
       "epoch": 0.481637567730283,
-      "grad_norm": 7.5413994789123535,
+      "grad_norm": 12.24130916595459,
       "learning_rate": 4.663096831771531e-05,
-      "loss": 0.0472,
+      "loss": 0.0703,
       "step": 400
     },
     {
       "epoch": 0.49367850692354004,
-      "grad_norm": 6.88725471496582,
+      "grad_norm": 18.149843215942383,
       "learning_rate": 4.64078536367693e-05,
-      "loss": 0.046,
+      "loss": 0.052,
       "step": 410
     },
     {
       "epoch": 0.5057194461167971,
-      "grad_norm": 14.230225563049316,
+      "grad_norm": 7.217360496520996,
       "learning_rate": 4.61847389558233e-05,
       "loss": 0.0419,
       "step": 420
     },
     {
       "epoch": 0.5177603853100542,
-      "grad_norm": 8.377012252807617,
+      "grad_norm": 15.023098945617676,
       "learning_rate": 4.596162427487729e-05,
-      "loss": 0.0524,
+      "loss": 0.0603,
       "step": 430
     },
     {
       "epoch": 0.5298013245033113,
-      "grad_norm": 0.7863742709159851,
+      "grad_norm": 2.35644268989563,
       "learning_rate": 4.5738509593931284e-05,
-      "loss": 0.0229,
+      "loss": 0.0275,
       "step": 440
     },
     {
       "epoch": 0.5418422636965683,
-      "grad_norm": 4.985135555267334,
+      "grad_norm": 8.228934288024902,
       "learning_rate": 4.5515394912985275e-05,
-      "loss": 0.0464,
+      "loss": 0.0383,
       "step": 450
     },
     {
       "epoch": 0.5538832028898254,
-      "grad_norm": 10.999000549316406,
+      "grad_norm": 9.796806335449219,
       "learning_rate": 4.529228023203927e-05,
-      "loss": 0.029,
+      "loss": 0.0334,
       "step": 460
     },
     {
       "epoch": 0.5659241420830825,
-      "grad_norm": 16.4847412109375,
+      "grad_norm": 16.539167404174805,
       "learning_rate": 4.506916555109326e-05,
-      "loss": 0.0334,
+      "loss": 0.0323,
       "step": 470
     },
     {
       "epoch": 0.5779650812763396,
-      "grad_norm": 12.632153511047363,
+      "grad_norm": 19.142715454101562,
       "learning_rate": 4.484605087014726e-05,
-      "loss": 0.0354,
+      "loss": 0.0362,
       "step": 480
     },
     {
       "epoch": 0.5900060204695966,
-      "grad_norm": 13.491677284240723,
+      "grad_norm": 2.512476682662964,
       "learning_rate": 4.4622936189201256e-05,
-      "loss": 0.0226,
+      "loss": 0.0232,
       "step": 490
     },
     {
       "epoch": 0.6020469596628537,
-      "grad_norm": 3.246889352798462,
+      "grad_norm": 23.41269302368164,
       "learning_rate": 4.439982150825525e-05,
-      "loss": 0.0217,
+      "loss": 0.036,
       "step": 500
     },
     {
       "epoch": 0.6140878988561108,
-      "grad_norm": 0.35594967007637024,
+      "grad_norm": 7.777035713195801,
       "learning_rate": 4.417670682730924e-05,
-      "loss": 0.0405,
+      "loss": 0.0397,
       "step": 510
     },
     {
       "epoch": 0.6261288380493678,
-      "grad_norm": 4.427694797515869,
+      "grad_norm": 3.8365113735198975,
       "learning_rate": 4.395359214636323e-05,
-      "loss": 0.0207,
+      "loss": 0.0224,
       "step": 520
     },
     {
       "epoch": 0.6381697772426249,
-      "grad_norm": 10.66477108001709,
+      "grad_norm": 2.88848876953125,
       "learning_rate": 4.373047746541722e-05,
-      "loss": 0.0227,
+      "loss": 0.0105,
       "step": 530
     },
     {
       "epoch": 0.650210716435882,
-      "grad_norm": 15.648704528808594,
+      "grad_norm": 4.8202972412109375,
       "learning_rate": 4.350736278447122e-05,
-      "loss": 0.027,
+      "loss": 0.033,
       "step": 540
     },
     {
       "epoch": 0.6622516556291391,
-      "grad_norm": 0.560919463634491,
+      "grad_norm": 14.322851181030273,
       "learning_rate": 4.328424810352521e-05,
-      "loss": 0.0143,
+      "loss": 0.0323,
       "step": 550
     },
     {
       "epoch": 0.6742925948223961,
-      "grad_norm": 11.590463638305664,
+      "grad_norm": 7.548096179962158,
       "learning_rate": 4.306113342257921e-05,
-      "loss": 0.0272,
+      "loss": 0.0301,
       "step": 560
     },
     {
       "epoch": 0.6863335340156532,
-      "grad_norm": 3.3958561420440674,
+      "grad_norm": 14.765242576599121,
       "learning_rate": 4.2838018741633203e-05,
-      "loss": 0.0403,
+      "loss": 0.0446,
       "step": 570
     },
     {
       "epoch": 0.6983744732089103,
-      "grad_norm": 17.12464714050293,
+      "grad_norm": 3.6058449745178223,
       "learning_rate": 4.2614904060687195e-05,
-      "loss": 0.0277,
+      "loss": 0.0334,
       "step": 580
     },
     {
       "epoch": 0.7104154124021673,
-      "grad_norm": 5.0445876121521,
+      "grad_norm": 3.7084667682647705,
       "learning_rate": 4.239178937974119e-05,
-      "loss": 0.0172,
+      "loss": 0.0211,
       "step": 590
     },
     {
       "epoch": 0.7224563515954244,
-      "grad_norm": 12.882268905639648,
+      "grad_norm": 26.29864501953125,
       "learning_rate": 4.2168674698795186e-05,
-      "loss": 0.034,
+      "loss": 0.0444,
       "step": 600
     },
     {
       "epoch": 0.7344972907886815,
-      "grad_norm": 14.19309139251709,
+      "grad_norm": 16.656940460205078,
       "learning_rate": 4.194556001784918e-05,
-      "loss": 0.0339,
+      "loss": 0.03,
       "step": 610
     },
     {
       "epoch": 0.7465382299819386,
-      "grad_norm": 7.258231163024902,
+      "grad_norm": 14.499817848205566,
       "learning_rate": 4.172244533690317e-05,
-      "loss": 0.0193,
+      "loss": 0.0285,
       "step": 620
     },
     {
       "epoch": 0.7585791691751956,
-      "grad_norm": 2.1384546756744385,
+      "grad_norm": 18.41780662536621,
       "learning_rate": 4.149933065595716e-05,
-      "loss": 0.0179,
+      "loss": 0.0256,
       "step": 630
     },
     {
       "epoch": 0.7706201083684527,
-      "grad_norm": 2.1396892070770264,
+      "grad_norm": 9.299426078796387,
       "learning_rate": 4.127621597501116e-05,
-      "loss": 0.0259,
+      "loss": 0.0228,
       "step": 640
     },
     {
       "epoch": 0.7826610475617098,
-      "grad_norm": 2.213364839553833,
+      "grad_norm": 12.47787857055664,
       "learning_rate": 4.105310129406515e-05,
-      "loss": 0.0198,
+      "loss": 0.0255,
       "step": 650
     },
     {
       "epoch": 0.7947019867549668,
-      "grad_norm": 9.647805213928223,
+      "grad_norm": 4.619048595428467,
       "learning_rate": 4.082998661311915e-05,
-      "loss": 0.0429,
+      "loss": 0.0467,
       "step": 660
     },
     {
       "epoch": 0.8067429259482239,
-      "grad_norm": 13.320932388305664,
+      "grad_norm": 2.003941059112549,
       "learning_rate": 4.060687193217314e-05,
-      "loss": 0.022,
+      "loss": 0.015,
       "step": 670
     },
     {
       "epoch": 0.818783865141481,
-      "grad_norm": 6.148782253265381,
+      "grad_norm": 5.838876724243164,
       "learning_rate": 4.038375725122713e-05,
-      "loss": 0.0111,
+      "loss": 0.0192,
       "step": 680
     },
     {
       "epoch": 0.8308248043347382,
-      "grad_norm": 0.37077805399894714,
+      "grad_norm": 11.127952575683594,
       "learning_rate": 4.0160642570281125e-05,
-      "loss": 0.0233,
+      "loss": 0.0324,
       "step": 690
     },
     {
       "epoch": 0.8428657435279951,
-      "grad_norm": 9.790139198303223,
+      "grad_norm": 7.260125160217285,
       "learning_rate": 3.993752788933512e-05,
-      "loss": 0.0232,
+      "loss": 0.0282,
       "step": 700
     },
     {
       "epoch": 0.8549066827212523,
-      "grad_norm": 15.844186782836914,
+      "grad_norm": 8.641480445861816,
       "learning_rate": 3.9714413208389115e-05,
-      "loss": 0.0181,
+      "loss": 0.0465,
       "step": 710
     },
     {
       "epoch": 0.8669476219145094,
-      "grad_norm": 5.37238883972168,
+      "grad_norm": 20.085968017578125,
       "learning_rate": 3.949129852744311e-05,
-      "loss": 0.0195,
+      "loss": 0.0284,
       "step": 720
     },
     {
       "epoch": 0.8789885611077664,
-      "grad_norm": 19.885295867919922,
+      "grad_norm": 10.058391571044922,
       "learning_rate": 3.9268183846497105e-05,
-      "loss": 0.0282,
+      "loss": 0.0324,
       "step": 730
     },
     {
       "epoch": 0.8910295003010235,
-      "grad_norm": 14.080533981323242,
+      "grad_norm": 3.2667603492736816,
       "learning_rate": 3.90450691655511e-05,
-      "loss": 0.0101,
+      "loss": 0.0185,
       "step": 740
     },
     {
       "epoch": 0.9030704394942806,
-      "grad_norm": 1.8638286590576172,
+      "grad_norm": 14.557952880859375,
       "learning_rate": 3.882195448460509e-05,
-      "loss": 0.0135,
+      "loss": 0.0126,
       "step": 750
     },
     {
       "epoch": 0.9151113786875377,
-      "grad_norm": 0.49987268447875977,
+      "grad_norm": 8.564680099487305,
       "learning_rate": 3.859883980365908e-05,
-      "loss": 0.0281,
+      "loss": 0.0261,
       "step": 760
     },
     {
       "epoch": 0.9271523178807947,
-      "grad_norm": 7.230884552001953,
+      "grad_norm": 4.106870174407959,
       "learning_rate": 3.837572512271307e-05,
-      "loss": 0.0119,
+      "loss": 0.0207,
       "step": 770
     },
     {
       "epoch": 0.9391932570740518,
-      "grad_norm": 15.956979751586914,
+      "grad_norm": 7.618483543395996,
       "learning_rate": 3.815261044176707e-05,
-      "loss": 0.0307,
+      "loss": 0.0385,
       "step": 780
     },
     {
       "epoch": 0.9512341962673089,
-      "grad_norm": 6.89923095703125,
+      "grad_norm": 14.047365188598633,
       "learning_rate": 3.792949576082106e-05,
-      "loss": 0.0299,
+      "loss": 0.0331,
       "step": 790
     },
     {
       "epoch": 0.963275135460566,
-      "grad_norm": 3.6012470722198486,
+      "grad_norm": 0.3435573875904083,
       "learning_rate": 3.770638107987506e-05,
-      "loss": 0.0187,
+      "loss": 0.0169,
       "step": 800
     },
     {
       "epoch": 0.975316074653823,
-      "grad_norm": 0.42434361577033997,
+      "grad_norm": 0.7744494080543518,
       "learning_rate": 3.748326639892905e-05,
-      "loss": 0.0056,
+      "loss": 0.0154,
       "step": 810
     },
     {
       "epoch": 0.9873570138470801,
-      "grad_norm": 10.235840797424316,
+      "grad_norm": 5.615227699279785,
       "learning_rate": 3.7260151717983045e-05,
-      "loss": 0.022,
+      "loss": 0.0194,
       "step": 820
     },
     {
       "epoch": 0.9993979530403372,
-      "grad_norm": 14.793251037597656,
+      "grad_norm": 6.243287563323975,
       "learning_rate": 3.7037037037037037e-05,
-      "loss": 0.0219,
+      "loss": 0.0242,
       "step": 830
     },
     {
       "epoch": 0.9993979530403372,
-      "eval_f1": 0.9941767167416564,
-      "eval_loss": 0.012211144901812077,
-      "eval_runtime": 107.6031,
-      "eval_samples_per_second": 109.755,
-      "eval_steps_per_second": 3.439,
+      "eval_f1": 0.9902855517157121,
+      "eval_loss": 0.01683718152344227,
+      "eval_runtime": 158.7916,
+      "eval_samples_per_second": 74.374,
+      "eval_steps_per_second": 2.33,
       "step": 830
     },
     {
       "epoch": 1.0117399157134257,
-      "grad_norm": 0.2734871506690979,
+      "grad_norm": 5.433340072631836,
       "learning_rate": 3.6813922356091035e-05,
-      "loss": 0.0317,
+      "loss": 0.02,
       "step": 840
     },
     {
       "epoch": 1.0237808549066827,
-      "grad_norm": 1.6646366119384766,
+      "grad_norm": 0.711068332195282,
       "learning_rate": 3.659080767514503e-05,
-      "loss": 0.0058,
+      "loss": 0.0041,
       "step": 850
     },
     {
       "epoch": 1.03582179409994,
-      "grad_norm": 6.263219833374023,
+      "grad_norm": 0.14189627766609192,
       "learning_rate": 3.636769299419902e-05,
-      "loss": 0.0108,
+      "loss": 0.0139,
       "step": 860
     },
     {
       "epoch": 1.047862733293197,
-      "grad_norm": 13.928460121154785,
+      "grad_norm": 11.135564804077148,
       "learning_rate": 3.614457831325301e-05,
-      "loss": 0.0148,
+      "loss": 0.0175,
       "step": 870
     },
     {
       "epoch": 1.059903672486454,
-      "grad_norm": 0.08599582314491272,
+      "grad_norm": 0.0977085754275322,
       "learning_rate": 3.592146363230701e-05,
-      "loss": 0.0074,
+      "loss": 0.0075,
       "step": 880
     },
     {
       "epoch": 1.0719446116797111,
-      "grad_norm": 0.2059994339942932,
+      "grad_norm": 0.896812915802002,
       "learning_rate": 3.5698348951361e-05,
-      "loss": 0.0036,
+      "loss": 0.0083,
       "step": 890
     },
     {
       "epoch": 1.083985550872968,
-      "grad_norm": 5.329049587249756,
+      "grad_norm": 4.47445011138916,
       "learning_rate": 3.5475234270415e-05,
-      "loss": 0.0053,
+      "loss": 0.0058,
       "step": 900
     },
     {
       "epoch": 1.096026490066225,
-      "grad_norm": 1.9016759395599365,
+      "grad_norm": 6.7714457511901855,
       "learning_rate": 3.525211958946899e-05,
-      "loss": 0.0096,
+      "loss": 0.0126,
       "step": 910
     },
     {
       "epoch": 1.1080674292594823,
-      "grad_norm": 18.670433044433594,
+      "grad_norm": 23.34847068786621,
       "learning_rate": 3.502900490852298e-05,
-      "loss": 0.0125,
+      "loss": 0.0161,
       "step": 920
     },
     {
       "epoch": 1.1201083684527393,
-      "grad_norm": 0.5670663714408875,
+      "grad_norm": 0.047478608787059784,
       "learning_rate": 3.4805890227576974e-05,
-      "loss": 0.0228,
+      "loss": 0.0203,
       "step": 930
     },
     {
       "epoch": 1.1321493076459963,
-      "grad_norm": 0.1877487599849701,
+      "grad_norm": 0.1649002730846405,
       "learning_rate": 3.4582775546630966e-05,
-      "loss": 0.0143,
+      "loss": 0.0177,
       "step": 940
     },
     {
       "epoch": 1.1441902468392535,
-      "grad_norm": 0.13180898129940033,
+      "grad_norm": 15.473665237426758,
       "learning_rate": 3.4359660865684965e-05,
-      "loss": 0.0072,
+      "loss": 0.0108,
       "step": 950
     },
     {
       "epoch": 1.1562311860325105,
-      "grad_norm": 1.8746509552001953,
+      "grad_norm": 15.388993263244629,
       "learning_rate": 3.413654618473896e-05,
-      "loss": 0.0071,
+      "loss": 0.0111,
       "step": 960
     },
     {
       "epoch": 1.1682721252257675,
-      "grad_norm": 0.07824055105447769,
+      "grad_norm": 0.5350409746170044,
       "learning_rate": 3.3913431503792955e-05,
-      "loss": 0.0099,
+      "loss": 0.0104,
       "step": 970
     },
     {
       "epoch": 1.1803130644190247,
-      "grad_norm": 0.26168355345726013,
+      "grad_norm": 0.1209443062543869,
       "learning_rate": 3.369031682284695e-05,
-      "loss": 0.0193,
+      "loss": 0.0154,
       "step": 980
     },
     {
       "epoch": 1.1923540036122817,
-      "grad_norm": 27.489240646362305,
+      "grad_norm": 5.60445499420166,
       "learning_rate": 3.346720214190094e-05,
-      "loss": 0.0142,
+      "loss": 0.0132,
       "step": 990
     },
     {
       "epoch": 1.2043949428055387,
-      "grad_norm": 15.5933198928833,
+      "grad_norm": 2.625100612640381,
       "learning_rate": 3.324408746095493e-05,
-      "loss": 0.0132,
+      "loss": 0.0127,
       "step": 1000
     },
     {
       "epoch": 1.216435881998796,
-      "grad_norm": 0.03655429556965828,
+      "grad_norm": 3.1627309322357178,
       "learning_rate": 3.302097278000892e-05,
-      "loss": 0.0113,
+      "loss": 0.0043,
       "step": 1010
     },
     {
       "epoch": 1.228476821192053,
-      "grad_norm": 0.0989295095205307,
+      "grad_norm": 0.24617458879947662,
       "learning_rate": 3.279785809906292e-05,
-      "loss": 0.0093,
+      "loss": 0.0078,
       "step": 1020
     },
     {
       "epoch": 1.2405177603853101,
-      "grad_norm": 9.063276290893555,
+      "grad_norm": 0.4519745111465454,
       "learning_rate": 3.257474341811691e-05,
-      "loss": 0.0077,
+      "loss": 0.0051,
       "step": 1030
     },
     {
       "epoch": 1.2525586995785671,
-      "grad_norm": 0.42270582914352417,
+      "grad_norm": 0.5981425642967224,
       "learning_rate": 3.235162873717091e-05,
-      "loss": 0.006,
+      "loss": 0.0064,
       "step": 1040
     },
     {
       "epoch": 1.2645996387718241,
-      "grad_norm": 0.026505226269364357,
+      "grad_norm": 0.020264046266674995,
       "learning_rate": 3.21285140562249e-05,
-      "loss": 0.0131,
+      "loss": 0.0173,
       "step": 1050
     },
     {
       "epoch": 1.2766405779650813,
-      "grad_norm": 7.347019672393799,
+      "grad_norm": 21.119285583496094,
       "learning_rate": 3.1905399375278894e-05,
-      "loss": 0.0077,
+      "loss": 0.01,
       "step": 1060
     },
     {
       "epoch": 1.2886815171583383,
-      "grad_norm": 0.46574658155441284,
+      "grad_norm": 15.559020042419434,
       "learning_rate": 3.1682284694332886e-05,
-      "loss": 0.0032,
+      "loss": 0.0163,
       "step": 1070
     },
     {
       "epoch": 1.3007224563515956,
-      "grad_norm": 0.4452888071537018,
+      "grad_norm": 0.3815273642539978,
       "learning_rate": 3.1459170013386885e-05,
-      "loss": 0.0017,
+      "loss": 0.0064,
       "step": 1080
     },
     {
       "epoch": 1.3127633955448526,
-      "grad_norm": 0.03053126111626625,
+      "grad_norm": 24.21092414855957,
       "learning_rate": 3.1236055332440876e-05,
-      "loss": 0.0183,
+      "loss": 0.0135,
       "step": 1090
     },
     {
       "epoch": 1.3248043347381095,
-      "grad_norm": 0.01922285184264183,
+      "grad_norm": 0.1006922796368599,
       "learning_rate": 3.101294065149487e-05,
-      "loss": 0.0082,
+      "loss": 0.0314,
       "step": 1100
     },
     {
       "epoch": 1.3368452739313668,
-      "grad_norm": 8.291805267333984,
+      "grad_norm": 13.257355690002441,
       "learning_rate": 3.078982597054887e-05,
-      "loss": 0.0056,
+      "loss": 0.0077,
       "step": 1110
     },
     {
       "epoch": 1.3488862131246238,
-      "grad_norm": 0.8168500065803528,
+      "grad_norm": 0.17610830068588257,
       "learning_rate": 3.056671128960286e-05,
-      "loss": 0.0079,
+      "loss": 0.0169,
       "step": 1120
     },
     {
       "epoch": 1.3609271523178808,
-      "grad_norm": 0.8204330801963806,
+      "grad_norm": 10.659666061401367,
       "learning_rate": 3.034359660865685e-05,
-      "loss": 0.0292,
+      "loss": 0.0372,
       "step": 1130
     },
     {
       "epoch": 1.372968091511138,
-      "grad_norm": 0.0944126546382904,
+      "grad_norm": 13.006769180297852,
       "learning_rate": 3.012048192771085e-05,
-      "loss": 0.0137,
+      "loss": 0.0261,
       "step": 1140
     },
     {
       "epoch": 1.385009030704395,
-      "grad_norm": 22.353675842285156,
+      "grad_norm": 3.6800951957702637,
       "learning_rate": 2.989736724676484e-05,
-      "loss": 0.0103,
+      "loss": 0.0072,
       "step": 1150
     },
     {
       "epoch": 1.397049969897652,
-      "grad_norm": 0.9541614651679993,
+      "grad_norm": 0.10272738337516785,
       "learning_rate": 2.9674252565818832e-05,
-      "loss": 0.004,
+      "loss": 0.0115,
       "step": 1160
     },
     {
       "epoch": 1.4090909090909092,
-      "grad_norm": 0.01685040444135666,
+      "grad_norm": 0.10577847063541412,
       "learning_rate": 2.9451137884872827e-05,
-      "loss": 0.0098,
+      "loss": 0.0095,
       "step": 1170
     },
     {
       "epoch": 1.4211318482841662,
-      "grad_norm": 0.028883345425128937,
+      "grad_norm": 0.6034663915634155,
       "learning_rate": 2.922802320392682e-05,
-      "loss": 0.0188,
+      "loss": 0.0183,
       "step": 1180
     },
     {
       "epoch": 1.4331727874774232,
-      "grad_norm": 0.042280565947294235,
+      "grad_norm": 0.03119043819606304,
       "learning_rate": 2.900490852298081e-05,
-      "loss": 0.0225,
+      "loss": 0.0209,
       "step": 1190
     },
     {
       "epoch": 1.4452137266706804,
-      "grad_norm": 0.6758914589881897,
+      "grad_norm": 0.04898101091384888,
       "learning_rate": 2.878179384203481e-05,
-      "loss": 0.0088,
+      "loss": 0.0142,
       "step": 1200
     },
     {
       "epoch": 1.4572546658639374,
-      "grad_norm": 0.032702963799238205,
+      "grad_norm": 0.08179830759763718,
       "learning_rate": 2.85586791610888e-05,
-      "loss": 0.0127,
+      "loss": 0.0185,
       "step": 1210
     },
     {
       "epoch": 1.4692956050571944,
-      "grad_norm": 8.852455139160156,
+      "grad_norm": 0.37462204694747925,
       "learning_rate": 2.8335564480142796e-05,
-      "loss": 0.0205,
+      "loss": 0.0137,
       "step": 1220
     },
     {
       "epoch": 1.4813365442504516,
-      "grad_norm": 0.17334896326065063,
+      "grad_norm": 8.365055084228516,
       "learning_rate": 2.8112449799196788e-05,
-      "loss": 0.0009,
+      "loss": 0.0043,
       "step": 1230
     },
     {
       "epoch": 1.4933774834437086,
-      "grad_norm": 0.6626876592636108,
+      "grad_norm": 0.1767185479402542,
       "learning_rate": 2.788933511825078e-05,
-      "loss": 0.0147,
+      "loss": 0.0133,
       "step": 1240
     },
     {
       "epoch": 1.5054184226369656,
-      "grad_norm": 15.536624908447266,
+      "grad_norm": 10.049013137817383,
       "learning_rate": 2.7666220437304775e-05,
-      "loss": 0.0147,
+      "loss": 0.0204,
       "step": 1250
     },
     {
       "epoch": 1.5174593618302228,
-      "grad_norm": 0.8571311235427856,
+      "grad_norm": 9.319756507873535,
       "learning_rate": 2.7443105756358774e-05,
-      "loss": 0.0052,
+      "loss": 0.0099,
       "step": 1260
     },
     {
       "epoch": 1.5295003010234798,
-      "grad_norm": 20.779983520507812,
+      "grad_norm": 0.5939036011695862,
       "learning_rate": 2.7219991075412765e-05,
-      "loss": 0.0077,
+      "loss": 0.0192,
       "step": 1270
     },
     {
       "epoch": 1.5415412402167368,
-      "grad_norm": 0.022782055661082268,
+      "grad_norm": 0.03510546311736107,
       "learning_rate": 2.6996876394466757e-05,
-      "loss": 0.0028,
+      "loss": 0.0065,
       "step": 1280
     },
     {
       "epoch": 1.553582179409994,
-      "grad_norm": 0.04790806025266647,
+      "grad_norm": 0.03960377350449562,
       "learning_rate": 2.6773761713520752e-05,
-      "loss": 0.0025,
+      "loss": 0.0085,
       "step": 1290
     },
     {
       "epoch": 1.5656231186032512,
-      "grad_norm": 0.48444584012031555,
+      "grad_norm": 11.613543510437012,
       "learning_rate": 2.6550647032574744e-05,
-      "loss": 0.0211,
+      "loss": 0.0146,
       "step": 1300
     },
     {
       "epoch": 1.577664057796508,
-      "grad_norm": 0.025251317769289017,
+      "grad_norm": 2.0522215366363525,
       "learning_rate": 2.6327532351628736e-05,
-      "loss": 0.0064,
+      "loss": 0.0014,
       "step": 1310
     },
     {
       "epoch": 1.5897049969897652,
-      "grad_norm": 29.587995529174805,
+      "grad_norm": 32.89421844482422,
       "learning_rate": 2.6104417670682734e-05,
-      "loss": 0.0016,
+      "loss": 0.01,
       "step": 1320
     },
     {
       "epoch": 1.6017459361830224,
-      "grad_norm": 0.678594708442688,
+      "grad_norm": 0.1284407079219818,
       "learning_rate": 2.5881302989736726e-05,
-      "loss": 0.0092,
+      "loss": 0.0118,
       "step": 1330
     },
     {
       "epoch": 1.6137868753762792,
-      "grad_norm": 7.527316093444824,
+      "grad_norm": 7.301222324371338,
       "learning_rate": 2.565818830879072e-05,
-      "loss": 0.002,
+      "loss": 0.0146,
       "step": 1340
     },
     {
       "epoch": 1.6258278145695364,
-      "grad_norm": 15.365883827209473,
+      "grad_norm": 3.3705849647521973,
       "learning_rate": 2.5435073627844713e-05,
-      "loss": 0.01,
+      "loss": 0.0113,
       "step": 1350
     },
     {
       "epoch": 1.6378687537627936,
-      "grad_norm": 11.4562349319458,
+      "grad_norm": 4.357827663421631,
       "learning_rate": 2.5211958946898705e-05,
-      "loss": 0.0078,
+      "loss": 0.0056,
       "step": 1360
     },
     {
       "epoch": 1.6499096929560506,
-      "grad_norm": 1.4225062131881714,
+      "grad_norm": 5.27785587310791,
       "learning_rate": 2.4988844265952703e-05,
-      "loss": 0.0116,
+      "loss": 0.0153,
       "step": 1370
     },
     {
       "epoch": 1.6619506321493076,
-      "grad_norm": 18.761125564575195,
+      "grad_norm": 6.584476470947266,
       "learning_rate": 2.4765729585006695e-05,
-      "loss": 0.0122,
+      "loss": 0.0234,
       "step": 1380
     },
     {
       "epoch": 1.6739915713425648,
-      "grad_norm": 0.042121145874261856,
+      "grad_norm": 0.07857084274291992,
       "learning_rate": 2.4542614904060687e-05,
-      "loss": 0.0036,
+      "loss": 0.0097,
       "step": 1390
     },
     {
       "epoch": 1.6860325105358218,
-      "grad_norm": 0.7333714365959167,
+      "grad_norm": 8.162418365478516,
       "learning_rate": 2.4319500223114682e-05,
-      "loss": 0.0052,
+      "loss": 0.0134,
       "step": 1400
     },
     {
       "epoch": 1.6980734497290788,
-      "grad_norm": 0.4470759928226471,
+      "grad_norm": 0.4992442727088928,
       "learning_rate": 2.4096385542168677e-05,
-      "loss": 0.0118,
+      "loss": 0.0144,
       "step": 1410
     },
     {
       "epoch": 1.710114388922336,
-      "grad_norm": 0.04939042404294014,
+      "grad_norm": 0.058108534663915634,
       "learning_rate": 2.387327086122267e-05,
-      "loss": 0.0067,
+      "loss": 0.0105,
       "step": 1420
     },
     {
       "epoch": 1.722155328115593,
-      "grad_norm": 5.569286823272705,
+      "grad_norm": 3.4073092937469482,
       "learning_rate": 2.3650156180276664e-05,
-      "loss": 0.0349,
+      "loss": 0.0327,
       "step": 1430
     },
     {
       "epoch": 1.73419626730885,
-      "grad_norm": 12.6468505859375,
+      "grad_norm": 6.666907787322998,
       "learning_rate": 2.3427041499330656e-05,
-      "loss": 0.0095,
+      "loss": 0.0148,
       "step": 1440
     },
     {
       "epoch": 1.7462372065021072,
-      "grad_norm": 0.034750111401081085,
+      "grad_norm": 0.18498630821704865,
       "learning_rate": 2.320392681838465e-05,
-      "loss": 0.0097,
+      "loss": 0.0262,
       "step": 1450
     },
     {
       "epoch": 1.7582781456953642,
-      "grad_norm": 0.970057487487793,
+      "grad_norm": 11.826184272766113,
       "learning_rate": 2.2980812137438646e-05,
-      "loss": 0.0042,
+      "loss": 0.0023,
       "step": 1460
     },
     {
       "epoch": 1.7703190848886212,
-      "grad_norm": 1.737916350364685,
+      "grad_norm": 30.59642791748047,
       "learning_rate": 2.2757697456492638e-05,
-      "loss": 0.0119,
+      "loss": 0.0209,
       "step": 1470
     },
     {
       "epoch": 1.7823600240818784,
-      "grad_norm": 0.40598204731941223,
+      "grad_norm": 0.48171374201774597,
       "learning_rate": 2.253458277554663e-05,
-      "loss": 0.007,
+      "loss": 0.0141,
       "step": 1480
     },
     {
       "epoch": 1.7944009632751354,
-      "grad_norm": 32.843143463134766,
+      "grad_norm": 0.16140851378440857,
       "learning_rate": 2.2311468094600628e-05,
-      "loss": 0.004,
+      "loss": 0.0038,
       "step": 1490
     },
     {
       "epoch": 1.8064419024683924,
-      "grad_norm": 0.09000721573829651,
+      "grad_norm": 11.184020042419434,
       "learning_rate": 2.208835341365462e-05,
-      "loss": 0.0045,
+      "loss": 0.0191,
       "step": 1500
     },
     {
       "epoch": 1.8184828416616496,
-      "grad_norm": 0.016685089096426964,
+      "grad_norm": 0.04441598430275917,
       "learning_rate": 2.186523873270861e-05,
-      "loss": 0.0015,
+      "loss": 0.0056,
       "step": 1510
     },
     {
       "epoch": 1.8305237808549066,
-      "grad_norm": 0.013028348796069622,
+      "grad_norm": 0.02468044124543667,
       "learning_rate": 2.1642124051762607e-05,
-      "loss": 0.0075,
+      "loss": 0.0019,
       "step": 1520
     },
     {
       "epoch": 1.8425647200481636,
-      "grad_norm": 4.790024757385254,
+      "grad_norm": 0.9031987190246582,
       "learning_rate": 2.1419009370816602e-05,
-      "loss": 0.0101,
+      "loss": 0.0057,
       "step": 1530
     },
     {
       "epoch": 1.8546056592414208,
-      "grad_norm": 9.446985244750977,
+      "grad_norm": 24.169157028198242,
       "learning_rate": 2.1195894689870593e-05,
-      "loss": 0.0166,
+      "loss": 0.023,
       "step": 1540
     },
     {
       "epoch": 1.866646598434678,
-      "grad_norm": 15.988990783691406,
+      "grad_norm": 5.31037712097168,
       "learning_rate": 2.097278000892459e-05,
-      "loss": 0.0029,
+      "loss": 0.005,
       "step": 1550
     },
     {
       "epoch": 1.8786875376279348,
-      "grad_norm": 0.04741289094090462,
+      "grad_norm": 28.011241912841797,
       "learning_rate": 2.074966532797858e-05,
-      "loss": 0.0041,
+      "loss": 0.0032,
       "step": 1560
     },
     {
       "epoch": 1.890728476821192,
-      "grad_norm": 1.0360485315322876,
+      "grad_norm": 20.326263427734375,
       "learning_rate": 2.0526550647032576e-05,
-      "loss": 0.0131,
+      "loss": 0.0154,
       "step": 1570
     },
     {
       "epoch": 1.9027694160144493,
-      "grad_norm": 12.175363540649414,
+      "grad_norm": 16.641881942749023,
       "learning_rate": 2.030343596608657e-05,
-      "loss": 0.009,
+      "loss": 0.0165,
       "step": 1580
     },
     {
       "epoch": 1.914810355207706,
-      "grad_norm": 1.8163396120071411,
+      "grad_norm": 20.551401138305664,
       "learning_rate": 2.0080321285140562e-05,
-      "loss": 0.0082,
+      "loss": 0.0038,
       "step": 1590
     },
     {
       "epoch": 1.9268512944009633,
-      "grad_norm": 1.1811190843582153,
+      "grad_norm": 3.062608480453491,
       "learning_rate": 1.9857206604194558e-05,
-      "loss": 0.0128,
+      "loss": 0.005,
       "step": 1600
     },
     {
       "epoch": 1.9388922335942205,
-      "grad_norm": 1.4999727010726929,
+      "grad_norm": 13.328275680541992,
       "learning_rate": 1.9634091923248553e-05,
-      "loss": 0.0108,
+      "loss": 0.0199,
       "step": 1610
     },
     {
       "epoch": 1.9509331727874775,
-      "grad_norm": 0.02650211751461029,
+      "grad_norm": 1.0551005601882935,
       "learning_rate": 1.9410977242302544e-05,
-      "loss": 0.0035,
+      "loss": 0.011,
       "step": 1620
     },
     {
       "epoch": 1.9629741119807345,
-      "grad_norm": 1.4226313829421997,
+      "grad_norm": 0.9096565842628479,
       "learning_rate": 1.9187862561356536e-05,
-      "loss": 0.0191,
+      "loss": 0.0227,
       "step": 1630
     },
     {
       "epoch": 1.9750150511739917,
-      "grad_norm": 0.030506979674100876,
+      "grad_norm": 0.07952883094549179,
       "learning_rate": 1.896474788041053e-05,
       "loss": 0.0009,
       "step": 1640
     },
     {
       "epoch": 1.9870559903672487,
-      "grad_norm": 0.035177238285541534,
+      "grad_norm": 0.0597514845430851,
       "learning_rate": 1.8741633199464527e-05,
-      "loss": 0.0141,
+      "loss": 0.0013,
       "step": 1650
     },
     {
       "epoch": 1.9990969295605057,
-      "grad_norm": 3.573192834854126,
+      "grad_norm": 4.350442886352539,
       "learning_rate": 1.8518518518518518e-05,
-      "loss": 0.0145,
+      "loss": 0.0127,
       "step": 1660
     },
     {
       "epoch": 1.9990969295605057,
-      "eval_f1": 0.9946425546192459,
-      "eval_loss": 0.008867617696523666,
-      "eval_runtime": 108.4658,
-      "eval_samples_per_second": 108.882,
-      "eval_steps_per_second": 3.411,
+      "eval_f1": 0.9941047948626572,
+      "eval_loss": 0.009086688049137592,
+      "eval_runtime": 158.1432,
+      "eval_samples_per_second": 74.679,
+      "eval_steps_per_second": 2.34,
       "step": 1660
     },
     {
       "epoch": 2.011438892233594,
-      "grad_norm": 0.10145142674446106,
+      "grad_norm": 0.3171527683734894,
       "learning_rate": 1.8295403837572513e-05,
-      "loss": 0.0072,
+      "loss": 0.007,
       "step": 1670
     },
     {
       "epoch": 2.0234798314268514,
-      "grad_norm": 0.0519409105181694,
+      "grad_norm": 0.17591948807239532,
       "learning_rate": 1.8072289156626505e-05,
-      "loss": 0.0087,
+      "loss": 0.0034,
       "step": 1680
     },
     {
       "epoch": 2.035520770620108,
-      "grad_norm": 0.0766972079873085,
+      "grad_norm": 0.019866205751895905,
       "learning_rate": 1.78491744756805e-05,
       "loss": 0.0006,
       "step": 1690
     },
     {
       "epoch": 2.0475617098133654,
-      "grad_norm": 0.764567494392395,
+      "grad_norm": 0.042357683181762695,
       "learning_rate": 1.7626059794734495e-05,
-      "loss": 0.0055,
+      "loss": 0.0018,
       "step": 1700
     },
     {
       "epoch": 2.0596026490066226,
-      "grad_norm": 3.932276964187622,
+      "grad_norm": 6.617299556732178,
       "learning_rate": 1.7402945113788487e-05,
-      "loss": 0.0119,
+      "loss": 0.0144,
       "step": 1710
     },
     {
       "epoch": 2.07164358819988,
-      "grad_norm": 1.2542475461959839,
+      "grad_norm": 28.643131256103516,
       "learning_rate": 1.7179830432842482e-05,
-      "loss": 0.0041,
+      "loss": 0.0073,
       "step": 1720
     },
     {
       "epoch": 2.0836845273931366,
-      "grad_norm": 0.028969695791602135,
+      "grad_norm": 1.9134842157363892,
       "learning_rate": 1.6956715751896478e-05,
-      "loss": 0.0083,
+      "loss": 0.016,
       "step": 1730
     },
     {
       "epoch": 2.095725466586394,
-      "grad_norm": 7.430561542510986,
+      "grad_norm": 0.40688356757164,
       "learning_rate": 1.673360107095047e-05,
-      "loss": 0.0149,
+      "loss": 0.0092,
       "step": 1740
     },
     {
       "epoch": 2.107766405779651,
-      "grad_norm": 0.057673774659633636,
+      "grad_norm": 1.3027302026748657,
       "learning_rate": 1.651048639000446e-05,
-      "loss": 0.0014,
+      "loss": 0.0066,
       "step": 1750
     },
     {
       "epoch": 2.119807344972908,
-      "grad_norm": 0.02054547891020775,
+      "grad_norm": 0.02679573744535446,
       "learning_rate": 1.6287371709058456e-05,
-      "loss": 0.0012,
+      "loss": 0.0004,
       "step": 1760
     },
     {
       "epoch": 2.131848284166165,
-      "grad_norm": 0.030434923246502876,
+      "grad_norm": 0.02949480153620243,
       "learning_rate": 1.606425702811245e-05,
-      "loss": 0.0113,
+      "loss": 0.0133,
       "step": 1770
     },
     {
       "epoch": 2.1438892233594222,
-      "grad_norm": 0.016015587374567986,
+      "grad_norm": 0.09444648772478104,
       "learning_rate": 1.5841142347166443e-05,
-      "loss": 0.0003,
+      "loss": 0.001,
       "step": 1780
     },
     {
       "epoch": 2.155930162552679,
-      "grad_norm": 0.6460728049278259,
+      "grad_norm": 11.101505279541016,
       "learning_rate": 1.5618027666220438e-05,
-      "loss": 0.0175,
+      "loss": 0.0247,
       "step": 1790
     },
     {
       "epoch": 2.167971101745936,
-      "grad_norm": 0.05660252273082733,
+      "grad_norm": 0.03616014122962952,
       "learning_rate": 1.5394912985274433e-05,
       "loss": 0.0026,
       "step": 1800
     },
     {
       "epoch": 2.1800120409391934,
-      "grad_norm": 0.7767325043678284,
+      "grad_norm": 0.058971330523490906,
       "learning_rate": 1.5171798304328425e-05,
-      "loss": 0.0004,
+      "loss": 0.0015,
       "step": 1810
     },
     {
       "epoch": 2.19205298013245,
-      "grad_norm": 0.01565532386302948,
+      "grad_norm": 0.017834767699241638,
       "learning_rate": 1.494868362338242e-05,
-      "loss": 0.0009,
+      "loss": 0.0013,
       "step": 1820
     },
     {
       "epoch": 2.2040939193257074,
-      "grad_norm": 0.07455216348171234,
+      "grad_norm": 1.4042654037475586,
       "learning_rate": 1.4725568942436414e-05,
-      "loss": 0.0003,
+      "loss": 0.0016,
       "step": 1830
     },
     {
       "epoch": 2.2161348585189646,
-      "grad_norm": 0.01474799681454897,
+      "grad_norm": 20.515151977539062,
       "learning_rate": 1.4502454261490405e-05,
-      "loss": 0.0102,
+      "loss": 0.017,
       "step": 1840
     },
     {
       "epoch": 2.2281757977122214,
-      "grad_norm": 2.596896171569824,
+      "grad_norm": 3.5099308490753174,
       "learning_rate": 1.42793395805444e-05,
-      "loss": 0.0072,
+      "loss": 0.0067,
       "step": 1850
     },
     {
       "epoch": 2.2402167369054786,
-      "grad_norm": 0.07300672680139542,
+      "grad_norm": 0.5224878787994385,
       "learning_rate": 1.4056224899598394e-05,
-      "loss": 0.0057,
+      "loss": 0.0023,
       "step": 1860
     },
     {
       "epoch": 2.252257676098736,
-      "grad_norm": 0.02680467627942562,
+      "grad_norm": 0.05775818973779678,
       "learning_rate": 1.3833110218652387e-05,
-      "loss": 0.0003,
+      "loss": 0.0016,
       "step": 1870
     },
     {
       "epoch": 2.2642986152919926,
-      "grad_norm": 0.489324152469635,
+      "grad_norm": 0.032823625952005386,
       "learning_rate": 1.3609995537706383e-05,
-      "loss": 0.0007,
+      "loss": 0.0029,
       "step": 1880
     },
     {
       "epoch": 2.27633955448525,
-      "grad_norm": 0.019087281078100204,
+      "grad_norm": 0.02662728726863861,
       "learning_rate": 1.3386880856760376e-05,
-      "loss": 0.0005,
+      "loss": 0.0014,
       "step": 1890
     },
     {
       "epoch": 2.288380493678507,
-      "grad_norm": 0.20010869204998016,
+      "grad_norm": 7.03336763381958,
       "learning_rate": 1.3163766175814368e-05,
-      "loss": 0.0051,
+      "loss": 0.0112,
       "step": 1900
     },
     {
       "epoch": 2.300421432871764,
-      "grad_norm": 1.1036587953567505,
+      "grad_norm": 6.6170973777771,
       "learning_rate": 1.2940651494868363e-05,
-      "loss": 0.0073,
+      "loss": 0.0075,
       "step": 1910
     },
     {
       "epoch": 2.312462372065021,
-      "grad_norm": 0.03441654518246651,
+      "grad_norm": 0.025626573711633682,
       "learning_rate": 1.2717536813922356e-05,
-      "loss": 0.0016,
+      "loss": 0.0014,
       "step": 1920
     },
     {
       "epoch": 2.3245033112582782,
-      "grad_norm": 0.29569727182388306,
+      "grad_norm": 21.287378311157227,
       "learning_rate": 1.2494422132976352e-05,
-      "loss": 0.0004,
+      "loss": 0.0047,
       "step": 1930
     },
     {
       "epoch": 2.336544250451535,
-      "grad_norm": 0.014553075656294823,
+      "grad_norm": 0.013641959056258202,
       "learning_rate": 1.2271307452030343e-05,
       "loss": 0.0005,
       "step": 1940
     },
     {
       "epoch": 2.3485851896447922,
-      "grad_norm": 0.012233784422278404,
+      "grad_norm": 0.01311859954148531,
       "learning_rate": 1.2048192771084338e-05,
-      "loss": 0.0042,
+      "loss": 0.0006,
       "step": 1950
     },
     {
       "epoch": 2.3606261288380495,
-      "grad_norm": 0.04088492691516876,
+      "grad_norm": 0.05053797364234924,
       "learning_rate": 1.1825078090138332e-05,
-      "loss": 0.0105,
+      "loss": 0.0129,
       "step": 1960
     },
     {
       "epoch": 2.3726670680313067,
-      "grad_norm": 0.08669020235538483,
+      "grad_norm": 0.41378432512283325,
       "learning_rate": 1.1601963409192325e-05,
-      "loss": 0.0073,
+      "loss": 0.0076,
       "step": 1970
     },
     {
       "epoch": 2.3847080072245634,
-      "grad_norm": 0.04787085950374603,
+      "grad_norm": 0.014416687190532684,
       "learning_rate": 1.1378848728246319e-05,
-      "loss": 0.0022,
+      "loss": 0.0019,
       "step": 1980
     },
     {
       "epoch": 2.3967489464178207,
-      "grad_norm": 0.018111156299710274,
+      "grad_norm": 0.07235641777515411,
       "learning_rate": 1.1155734047300314e-05,
       "loss": 0.0004,
       "step": 1990
     },
     {
       "epoch": 2.4087898856110774,
-      "grad_norm": 0.10384394973516464,
+      "grad_norm": 0.10537643730640411,
       "learning_rate": 1.0932619366354306e-05,
-      "loss": 0.0107,
+      "loss": 0.0088,
       "step": 2000
     },
     {
       "epoch": 2.4208308248043346,
-      "grad_norm": 0.020993240177631378,
+      "grad_norm": 0.014000581577420235,
       "learning_rate": 1.0709504685408301e-05,
-      "loss": 0.0009,
+      "loss": 0.0008,
       "step": 2010
     },
     {
       "epoch": 2.432871763997592,
-      "grad_norm": 6.286465167999268,
+      "grad_norm": 16.16079330444336,
       "learning_rate": 1.0486390004462294e-05,
-      "loss": 0.0055,
+      "loss": 0.0026,
       "step": 2020
     },
     {
       "epoch": 2.444912703190849,
-      "grad_norm": 0.16340582072734833,
+      "grad_norm": 1.1684482097625732,
       "learning_rate": 1.0263275323516288e-05,
-      "loss": 0.0059,
+      "loss": 0.0041,
       "step": 2030
     },
     {
       "epoch": 2.456953642384106,
-      "grad_norm": 0.015615738928318024,
+      "grad_norm": 0.016048986464738846,
       "learning_rate": 1.0040160642570281e-05,
-      "loss": 0.0071,
+      "loss": 0.0075,
       "step": 2040
     },
     {
       "epoch": 2.468994581577363,
-      "grad_norm": 0.015223161317408085,
+      "grad_norm": 0.017376257106661797,
       "learning_rate": 9.817045961624276e-06,
-      "loss": 0.001,
+      "loss": 0.0046,
       "step": 2050
     },
     {
       "epoch": 2.4810355207706203,
-      "grad_norm": 0.04220983758568764,
+      "grad_norm": 0.016579024493694305,
       "learning_rate": 9.593931280678268e-06,
-      "loss": 0.0003,
+      "loss": 0.0012,
       "step": 2060
     },
     {
       "epoch": 2.493076459963877,
-      "grad_norm": 0.013744501397013664,
+      "grad_norm": 0.01798885501921177,
       "learning_rate": 9.370816599732263e-06,
-      "loss": 0.0005,
+      "loss": 0.0021,
       "step": 2070
     },
     {
       "epoch": 2.5051173991571343,
-      "grad_norm": 0.021128827705979347,
+      "grad_norm": 0.03361746296286583,
       "learning_rate": 9.147701918786257e-06,
-      "loss": 0.0046,
+      "loss": 0.0003,
       "step": 2080
     },
     {
       "epoch": 2.5171583383503915,
-      "grad_norm": 25.261341094970703,
+      "grad_norm": 0.03460362181067467,
       "learning_rate": 8.92458723784025e-06,
-      "loss": 0.0015,
+      "loss": 0.0021,
       "step": 2090
     },
     {
       "epoch": 2.5291992775436483,
-      "grad_norm": 27.442256927490234,
+      "grad_norm": 0.17655785381793976,
       "learning_rate": 8.701472556894244e-06,
-      "loss": 0.0052,
+      "loss": 0.0088,
       "step": 2100
     },
     {
       "epoch": 2.5412402167369055,
-      "grad_norm": 0.03090606816112995,
+      "grad_norm": 0.013977100141346455,
       "learning_rate": 8.478357875948239e-06,
-      "loss": 0.0003,
+      "loss": 0.0008,
       "step": 2110
     },
     {
       "epoch": 2.5532811559301627,
-      "grad_norm": 0.028400469571352005,
+      "grad_norm": 0.012227312661707401,
       "learning_rate": 8.25524319500223e-06,
-      "loss": 0.0009,
+      "loss": 0.0045,
       "step": 2120
     },
     {
       "epoch": 2.5653220951234195,
-      "grad_norm": 0.01129979733377695,
+      "grad_norm": 0.01880710944533348,
       "learning_rate": 8.032128514056226e-06,
-      "loss": 0.0026,
+      "loss": 0.0055,
       "step": 2130
     },
     {
       "epoch": 2.5773630343166767,
-      "grad_norm": 0.011010928079485893,
+      "grad_norm": 0.017849355936050415,
       "learning_rate": 7.809013833110219e-06,
-      "loss": 0.005,
+      "loss": 0.0014,
       "step": 2140
     },
     {
       "epoch": 2.589403973509934,
-      "grad_norm": 0.016233067959547043,
+      "grad_norm": 0.030449502170085907,
       "learning_rate": 7.5858991521642126e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2150
     },
     {
       "epoch": 2.601444912703191,
-      "grad_norm": 0.011614818125963211,
+      "grad_norm": 0.016365479677915573,
       "learning_rate": 7.362784471218207e-06,
-      "loss": 0.0066,
+      "loss": 0.0069,
       "step": 2160
     },
     {
       "epoch": 2.613485851896448,
-      "grad_norm": 0.009341931901872158,
+      "grad_norm": 0.017494693398475647,
       "learning_rate": 7.1396697902722e-06,
-      "loss": 0.0014,
+      "loss": 0.0003,
       "step": 2170
     },
     {
       "epoch": 2.625526791089705,
-      "grad_norm": 0.009875635616481304,
+      "grad_norm": 0.01181891281157732,
       "learning_rate": 6.916555109326194e-06,
-      "loss": 0.0017,
+      "loss": 0.001,
       "step": 2180
     },
     {
       "epoch": 2.637567730282962,
-      "grad_norm": 4.177189350128174,
+      "grad_norm": 2.1435914039611816,
       "learning_rate": 6.693440428380188e-06,
       "loss": 0.0124,
       "step": 2190
     },
     {
       "epoch": 2.649608669476219,
-      "grad_norm": 0.49204662442207336,
+      "grad_norm": 0.03837867081165314,
       "learning_rate": 6.4703257474341815e-06,
-      "loss": 0.0009,
+      "loss": 0.0003,
       "step": 2200
     },
     {
       "epoch": 2.6616496086694763,
-      "grad_norm": 0.022006221115589142,
+      "grad_norm": 0.026179147884249687,
       "learning_rate": 6.247211066488176e-06,
-      "loss": 0.0084,
+      "loss": 0.0029,
       "step": 2210
     },
     {
       "epoch": 2.6736905478627335,
-      "grad_norm": 0.0962655320763588,
+      "grad_norm": 0.07299380749464035,
       "learning_rate": 6.024096385542169e-06,
-      "loss": 0.0073,
+      "loss": 0.0131,
       "step": 2220
     },
     {
       "epoch": 2.6857314870559903,
-      "grad_norm": 0.7483286261558533,
+      "grad_norm": 0.05025468394160271,
       "learning_rate": 5.800981704596163e-06,
-      "loss": 0.0003,
+      "loss": 0.0006,
       "step": 2230
     },
     {
       "epoch": 2.6977724262492475,
-      "grad_norm": 0.08093533664941788,
+      "grad_norm": 0.19755429029464722,
       "learning_rate": 5.577867023650157e-06,
-      "loss": 0.0063,
+      "loss": 0.0164,
       "step": 2240
     },
     {
       "epoch": 2.7098133654425043,
-      "grad_norm": 0.017655853182077408,
+      "grad_norm": 0.1809852570295334,
       "learning_rate": 5.3547523427041504e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2250
     },
     {
       "epoch": 2.7218543046357615,
-      "grad_norm": 0.011604108847677708,
+      "grad_norm": 0.1278763711452484,
       "learning_rate": 5.131637661758144e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2260
     },
     {
       "epoch": 2.7338952438290187,
-      "grad_norm": 0.09758894145488739,
+      "grad_norm": 0.491524338722229,
       "learning_rate": 4.908522980812138e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2270
     },
     {
       "epoch": 2.745936183022276,
-      "grad_norm": 22.756349563598633,
+      "grad_norm": 13.114213943481445,
       "learning_rate": 4.685408299866132e-06,
-      "loss": 0.0024,
+      "loss": 0.0011,
       "step": 2280
     },
     {
       "epoch": 2.7579771222155327,
-      "grad_norm": 1.9275836944580078,
+      "grad_norm": 1.5844202041625977,
       "learning_rate": 4.462293618920125e-06,
-      "loss": 0.0045,
+      "loss": 0.0068,
       "step": 2290
     },
     {
       "epoch": 2.77001806140879,
-      "grad_norm": 40.43008041381836,
+      "grad_norm": 0.4727950394153595,
       "learning_rate": 4.239178937974119e-06,
-      "loss": 0.0035,
+      "loss": 0.0012,
       "step": 2300
     },
     {
       "epoch": 2.7820590006020467,
-      "grad_norm": 0.009210176765918732,
+      "grad_norm": 0.00994125846773386,
       "learning_rate": 4.016064257028113e-06,
-      "loss": 0.0008,
+      "loss": 0.0003,
       "step": 2310
     },
     {
       "epoch": 2.794099939795304,
-      "grad_norm": 0.008633752353489399,
+      "grad_norm": 0.06683458387851715,
       "learning_rate": 3.7929495760821063e-06,
       "loss": 0.0002,
       "step": 2320
     },
     {
       "epoch": 2.806140878988561,
-      "grad_norm": 0.008137097582221031,
+      "grad_norm": 0.40406838059425354,
       "learning_rate": 3.5698348951361e-06,
-      "loss": 0.0002,
+      "loss": 0.0048,
       "step": 2330
     },
     {
       "epoch": 2.8181818181818183,
-      "grad_norm": 0.011585449799895287,
+      "grad_norm": 2.3229804039001465,
       "learning_rate": 3.346720214190094e-06,
-      "loss": 0.003,
+      "loss": 0.0054,
       "step": 2340
     },
     {
       "epoch": 2.830222757375075,
-      "grad_norm": 0.00909222662448883,
+      "grad_norm": 0.09825894981622696,
       "learning_rate": 3.123605533244088e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2350
     },
     {
       "epoch": 2.8422636965683323,
-      "grad_norm": 1.3680801391601562,
+      "grad_norm": 2.2067606449127197,
       "learning_rate": 2.9004908522980813e-06,
       "loss": 0.0003,
       "step": 2360
     },
     {
       "epoch": 2.8543046357615895,
-      "grad_norm": 0.09263092279434204,
+      "grad_norm": 0.10803830623626709,
       "learning_rate": 2.6773761713520752e-06,
-      "loss": 0.002,
+      "loss": 0.0014,
       "step": 2370
     },
     {
       "epoch": 2.8663455749548463,
-      "grad_norm": 0.012016590684652328,
+      "grad_norm": 0.15049925446510315,
       "learning_rate": 2.454261490406069e-06,
-      "loss": 0.0002,
+      "loss": 0.0044,
       "step": 2380
     },
     {
       "epoch": 2.8783865141481035,
-      "grad_norm": 0.02138102613389492,
+      "grad_norm": 0.012403941713273525,
       "learning_rate": 2.2311468094600625e-06,
-      "loss": 0.0002,
+      "loss": 0.006,
       "step": 2390
     },
     {
       "epoch": 2.8904274533413608,
-      "grad_norm": 0.00862041860818863,
+      "grad_norm": 0.02164722979068756,
       "learning_rate": 2.0080321285140564e-06,
       "loss": 0.0002,
       "step": 2400
     },
     {
       "epoch": 2.902468392534618,
-      "grad_norm": 0.008481325581669807,
+      "grad_norm": 0.008666186593472958,
       "learning_rate": 1.78491744756805e-06,
       "loss": 0.0002,
       "step": 2410
     },
     {
       "epoch": 2.9145093317278747,
-      "grad_norm": 0.01207977719604969,
+      "grad_norm": 0.07253342866897583,
       "learning_rate": 1.561802766622044e-06,
-      "loss": 0.0002,
+      "loss": 0.0003,
       "step": 2420
     },
     {
       "epoch": 2.926550270921132,
-      "grad_norm": 0.03627391904592514,
+      "grad_norm": 39.6471061706543,
       "learning_rate": 1.3386880856760376e-06,
-      "loss": 0.0002,
+      "loss": 0.0028,
       "step": 2430
     },
     {
       "epoch": 2.9385912101143887,
-      "grad_norm": 0.01422710157930851,
+      "grad_norm": 1.279341220855713,
       "learning_rate": 1.1155734047300313e-06,
-      "loss": 0.0002,
+      "loss": 0.0035,
       "step": 2440
     },
     {
       "epoch": 2.950632149307646,
-      "grad_norm": 0.007475409656763077,
+      "grad_norm": 0.010345235466957092,
       "learning_rate": 8.92458723784025e-07,
-      "loss": 0.0004,
+      "loss": 0.0002,
       "step": 2450
     },
     {
       "epoch": 2.962673088500903,
-      "grad_norm": 0.13591405749320984,
+      "grad_norm": 0.8630630970001221,
       "learning_rate": 6.693440428380188e-07,
-      "loss": 0.0002,
+      "loss": 0.0047,
       "step": 2460
     },
     {
       "epoch": 2.9747140276941604,
-      "grad_norm": 0.13127803802490234,
+      "grad_norm": 0.5396146178245544,
       "learning_rate": 4.462293618920125e-07,
-      "loss": 0.002,
+      "loss": 0.0058,
       "step": 2470
     },
     {
       "epoch": 2.986754966887417,
-      "grad_norm": 0.010220134630799294,
+      "grad_norm": 0.009292367845773697,
       "learning_rate": 2.2311468094600626e-07,
-      "loss": 0.0002,
+      "loss": 0.0059,
       "step": 2480
     },
     {
       "epoch": 2.9987959060806744,
-      "grad_norm": 0.008627759292721748,
+      "grad_norm": 0.017796576023101807,
       "learning_rate": 0.0,
-      "loss": 0.0002,
+      "loss": 0.0075,
       "step": 2490
     },
     {
       "epoch": 2.9987959060806744,
-      "eval_f1": 0.9948107465549906,
-      "eval_loss": 0.008508403785526752,
-      "eval_runtime": 109.2642,
-      "eval_samples_per_second": 108.087,
-      "eval_steps_per_second": 3.386,
+      "eval_f1": 0.994895873653879,
+      "eval_loss": 0.008104476146399975,
+      "eval_runtime": 158.5918,
+      "eval_samples_per_second": 74.468,
+      "eval_steps_per_second": 2.333,
       "step": 2490
     },
     {
       "epoch": 2.9987959060806744,
       "step": 2490,
       "total_flos": 4.152545054111269e+19,
-      "train_loss": 0.05819672941988573,
-      "train_runtime": 6053.7503,
-      "train_samples_per_second": 52.671,
-      "train_steps_per_second": 0.411
+      "train_loss": 0.0599212793472511,
+      "train_runtime": 7516.3059,
+      "train_samples_per_second": 42.422,
+      "train_steps_per_second": 0.331
     }
   ],
   "logging_steps": 10,