diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,2686 +1,2686 @@
-{
-  "best_metric": 0.9920562868815251,
-  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat\\checkpoint-3856",
-  "epoch": 9.988331388564761,
-  "eval_steps": 500,
-  "global_step": 4280,
-  "is_hyper_param_search": false,
-  "is_local_process_zero": true,
-  "is_world_process_zero": true,
-  "log_history": [
-    {
-      "epoch": 0.02,
-      "learning_rate": 1.1682242990654206e-06,
-      "loss": 1.8617,
-      "step": 10
-    },
-    {
-      "epoch": 0.05,
-      "learning_rate": 2.3364485981308413e-06,
-      "loss": 1.818,
-      "step": 20
-    },
-    {
-      "epoch": 0.07,
-      "learning_rate": 3.5046728971962617e-06,
-      "loss": 1.7345,
-      "step": 30
-    },
-    {
-      "epoch": 0.09,
-      "learning_rate": 4.6728971962616825e-06,
-      "loss": 1.5938,
-      "step": 40
-    },
-    {
-      "epoch": 0.12,
-      "learning_rate": 5.841121495327103e-06,
-      "loss": 1.3981,
-      "step": 50
-    },
-    {
-      "epoch": 0.14,
-      "learning_rate": 7.009345794392523e-06,
-      "loss": 1.2735,
-      "step": 60
-    },
-    {
-      "epoch": 0.16,
-      "learning_rate": 8.177570093457943e-06,
-      "loss": 1.0981,
-      "step": 70
-    },
-    {
-      "epoch": 0.19,
-      "learning_rate": 9.345794392523365e-06,
-      "loss": 0.942,
-      "step": 80
-    },
-    {
-      "epoch": 0.21,
-      "learning_rate": 1.0514018691588785e-05,
-      "loss": 0.7897,
-      "step": 90
-    },
-    {
-      "epoch": 0.23,
-      "learning_rate": 1.1682242990654207e-05,
-      "loss": 0.7545,
-      "step": 100
-    },
-    {
-      "epoch": 0.26,
-      "learning_rate": 1.2850467289719625e-05,
-      "loss": 0.6081,
-      "step": 110
-    },
-    {
-      "epoch": 0.28,
-      "learning_rate": 1.4018691588785047e-05,
-      "loss": 0.5312,
-      "step": 120
-    },
-    {
-      "epoch": 0.3,
-      "learning_rate": 1.5186915887850467e-05,
-      "loss": 0.5601,
-      "step": 130
-    },
-    {
-      "epoch": 0.33,
-      "learning_rate": 1.6355140186915887e-05,
-      "loss": 0.4363,
-      "step": 140
-    },
-    {
-      "epoch": 0.35,
-      "learning_rate": 1.752336448598131e-05,
-      "loss": 0.3406,
-      "step": 150
-    },
-    {
-      "epoch": 0.37,
-      "learning_rate": 1.869158878504673e-05,
-      "loss": 0.3539,
-      "step": 160
-    },
-    {
-      "epoch": 0.4,
-      "learning_rate": 1.985981308411215e-05,
-      "loss": 0.2825,
-      "step": 170
-    },
-    {
-      "epoch": 0.42,
-      "learning_rate": 2.102803738317757e-05,
-      "loss": 0.3021,
-      "step": 180
-    },
-    {
-      "epoch": 0.44,
-      "learning_rate": 2.2196261682242992e-05,
-      "loss": 0.2469,
-      "step": 190
-    },
-    {
-      "epoch": 0.47,
-      "learning_rate": 2.3364485981308414e-05,
-      "loss": 0.3184,
-      "step": 200
-    },
-    {
-      "epoch": 0.49,
-      "learning_rate": 2.4532710280373832e-05,
-      "loss": 0.265,
-      "step": 210
-    },
-    {
-      "epoch": 0.51,
-      "learning_rate": 2.570093457943925e-05,
-      "loss": 0.2356,
-      "step": 220
-    },
-    {
-      "epoch": 0.54,
-      "learning_rate": 2.6869158878504675e-05,
-      "loss": 0.2352,
-      "step": 230
-    },
-    {
-      "epoch": 0.56,
-      "learning_rate": 2.8037383177570094e-05,
-      "loss": 0.19,
-      "step": 240
-    },
-    {
-      "epoch": 0.58,
-      "learning_rate": 2.9205607476635515e-05,
-      "loss": 0.1716,
-      "step": 250
-    },
-    {
-      "epoch": 0.61,
-      "learning_rate": 3.0373831775700934e-05,
-      "loss": 0.1793,
-      "step": 260
-    },
-    {
-      "epoch": 0.63,
-      "learning_rate": 3.1542056074766355e-05,
-      "loss": 0.1959,
-      "step": 270
-    },
-    {
-      "epoch": 0.65,
-      "learning_rate": 3.2710280373831774e-05,
-      "loss": 0.1464,
-      "step": 280
-    },
-    {
-      "epoch": 0.68,
-      "learning_rate": 3.38785046728972e-05,
-      "loss": 0.2572,
-      "step": 290
-    },
-    {
-      "epoch": 0.7,
-      "learning_rate": 3.504672897196262e-05,
-      "loss": 0.3013,
-      "step": 300
-    },
-    {
-      "epoch": 0.72,
-      "learning_rate": 3.621495327102804e-05,
-      "loss": 0.1792,
-      "step": 310
-    },
-    {
-      "epoch": 0.75,
-      "learning_rate": 3.738317757009346e-05,
-      "loss": 0.17,
-      "step": 320
-    },
-    {
-      "epoch": 0.77,
-      "learning_rate": 3.855140186915888e-05,
-      "loss": 0.1853,
-      "step": 330
-    },
-    {
-      "epoch": 0.79,
-      "learning_rate": 3.97196261682243e-05,
-      "loss": 0.1536,
-      "step": 340
-    },
-    {
-      "epoch": 0.82,
-      "learning_rate": 4.088785046728972e-05,
-      "loss": 0.1631,
-      "step": 350
-    },
-    {
-      "epoch": 0.84,
-      "learning_rate": 4.205607476635514e-05,
-      "loss": 0.2002,
-      "step": 360
-    },
-    {
-      "epoch": 0.86,
-      "learning_rate": 4.3224299065420565e-05,
-      "loss": 0.2862,
-      "step": 370
-    },
-    {
-      "epoch": 0.89,
-      "learning_rate": 4.4392523364485984e-05,
-      "loss": 0.2736,
-      "step": 380
-    },
-    {
-      "epoch": 0.91,
-      "learning_rate": 4.556074766355141e-05,
-      "loss": 0.0934,
-      "step": 390
-    },
-    {
-      "epoch": 0.93,
-      "learning_rate": 4.672897196261683e-05,
-      "loss": 0.1664,
-      "step": 400
-    },
-    {
-      "epoch": 0.96,
-      "learning_rate": 4.7897196261682245e-05,
-      "loss": 0.1944,
-      "step": 410
-    },
-    {
-      "epoch": 0.98,
-      "learning_rate": 4.9065420560747664e-05,
-      "loss": 0.2347,
-      "step": 420
-    },
-    {
-      "epoch": 1.0,
-      "eval_accuracy": 0.9720835224693599,
-      "eval_loss": 0.08239880204200745,
-      "eval_runtime": 287.4814,
-      "eval_samples_per_second": 15.326,
-      "eval_steps_per_second": 2.557,
-      "step": 428
-    },
-    {
-      "epoch": 1.0,
-      "learning_rate": 4.9974039460020774e-05,
-      "loss": 0.1616,
-      "step": 430
-    },
-    {
-      "epoch": 1.03,
-      "learning_rate": 4.9844236760124614e-05,
-      "loss": 0.15,
-      "step": 440
-    },
-    {
-      "epoch": 1.05,
-      "learning_rate": 4.971443406022845e-05,
-      "loss": 0.1774,
-      "step": 450
-    },
-    {
-      "epoch": 1.07,
-      "learning_rate": 4.958463136033229e-05,
-      "loss": 0.1845,
-      "step": 460
-    },
-    {
-      "epoch": 1.1,
-      "learning_rate": 4.9454828660436145e-05,
-      "loss": 0.1343,
-      "step": 470
-    },
-    {
-      "epoch": 1.12,
-      "learning_rate": 4.9325025960539985e-05,
-      "loss": 0.122,
-      "step": 480
-    },
-    {
-      "epoch": 1.14,
-      "learning_rate": 4.9195223260643824e-05,
-      "loss": 0.1828,
-      "step": 490
-    },
-    {
-      "epoch": 1.17,
-      "learning_rate": 4.9065420560747664e-05,
-      "loss": 0.1556,
-      "step": 500
-    },
-    {
-      "epoch": 1.19,
-      "learning_rate": 4.893561786085151e-05,
-      "loss": 0.2151,
-      "step": 510
-    },
-    {
-      "epoch": 1.21,
-      "learning_rate": 4.880581516095535e-05,
-      "loss": 0.1906,
-      "step": 520
-    },
-    {
-      "epoch": 1.24,
-      "learning_rate": 4.8676012461059195e-05,
-      "loss": 0.114,
-      "step": 530
-    },
-    {
-      "epoch": 1.26,
-      "learning_rate": 4.8546209761163035e-05,
-      "loss": 0.1115,
-      "step": 540
-    },
-    {
-      "epoch": 1.28,
-      "learning_rate": 4.8416407061266874e-05,
-      "loss": 0.0965,
-      "step": 550
-    },
-    {
-      "epoch": 1.31,
-      "learning_rate": 4.828660436137072e-05,
-      "loss": 0.114,
-      "step": 560
-    },
-    {
-      "epoch": 1.33,
-      "learning_rate": 4.815680166147456e-05,
-      "loss": 0.1171,
-      "step": 570
-    },
-    {
-      "epoch": 1.35,
-      "learning_rate": 4.80269989615784e-05,
-      "loss": 0.1471,
-      "step": 580
-    },
-    {
-      "epoch": 1.38,
-      "learning_rate": 4.7897196261682245e-05,
-      "loss": 0.2203,
-      "step": 590
-    },
-    {
-      "epoch": 1.4,
-      "learning_rate": 4.776739356178609e-05,
-      "loss": 0.1091,
-      "step": 600
-    },
-    {
-      "epoch": 1.42,
-      "learning_rate": 4.763759086188993e-05,
-      "loss": 0.2073,
-      "step": 610
-    },
-    {
-      "epoch": 1.45,
-      "learning_rate": 4.750778816199377e-05,
-      "loss": 0.1692,
-      "step": 620
-    },
-    {
-      "epoch": 1.47,
-      "learning_rate": 4.737798546209761e-05,
-      "loss": 0.0989,
-      "step": 630
-    },
-    {
-      "epoch": 1.49,
-      "learning_rate": 4.7248182762201456e-05,
-      "loss": 0.2112,
-      "step": 640
-    },
-    {
-      "epoch": 1.52,
-      "learning_rate": 4.71183800623053e-05,
-      "loss": 0.1584,
-      "step": 650
-    },
-    {
-      "epoch": 1.54,
-      "learning_rate": 4.698857736240914e-05,
-      "loss": 0.1364,
-      "step": 660
-    },
-    {
-      "epoch": 1.56,
-      "learning_rate": 4.685877466251298e-05,
-      "loss": 0.1532,
-      "step": 670
-    },
-    {
-      "epoch": 1.59,
-      "learning_rate": 4.672897196261683e-05,
-      "loss": 0.1373,
-      "step": 680
-    },
-    {
-      "epoch": 1.61,
-      "learning_rate": 4.6599169262720666e-05,
-      "loss": 0.2829,
-      "step": 690
-    },
-    {
-      "epoch": 1.63,
-      "learning_rate": 4.6469366562824506e-05,
-      "loss": 0.1439,
-      "step": 700
-    },
-    {
-      "epoch": 1.66,
-      "learning_rate": 4.633956386292835e-05,
-      "loss": 0.1535,
-      "step": 710
-    },
-    {
-      "epoch": 1.68,
-      "learning_rate": 4.620976116303219e-05,
-      "loss": 0.1178,
-      "step": 720
-    },
-    {
-      "epoch": 1.7,
-      "learning_rate": 4.607995846313604e-05,
-      "loss": 0.1518,
-      "step": 730
-    },
-    {
-      "epoch": 1.73,
-      "learning_rate": 4.595015576323988e-05,
-      "loss": 0.2072,
-      "step": 740
-    },
-    {
-      "epoch": 1.75,
-      "learning_rate": 4.5820353063343716e-05,
-      "loss": 0.1121,
-      "step": 750
-    },
-    {
-      "epoch": 1.77,
-      "learning_rate": 4.569055036344756e-05,
-      "loss": 0.0619,
-      "step": 760
-    },
-    {
-      "epoch": 1.8,
-      "learning_rate": 4.556074766355141e-05,
-      "loss": 0.132,
-      "step": 770
-    },
-    {
-      "epoch": 1.82,
-      "learning_rate": 4.543094496365525e-05,
-      "loss": 0.1038,
-      "step": 780
-    },
-    {
-      "epoch": 1.84,
-      "learning_rate": 4.530114226375909e-05,
-      "loss": 0.2358,
-      "step": 790
-    },
-    {
-      "epoch": 1.87,
-      "learning_rate": 4.517133956386293e-05,
-      "loss": 0.1398,
-      "step": 800
-    },
-    {
-      "epoch": 1.89,
-      "learning_rate": 4.504153686396677e-05,
-      "loss": 0.1683,
-      "step": 810
-    },
-    {
-      "epoch": 1.91,
-      "learning_rate": 4.491173416407061e-05,
-      "loss": 0.1513,
-      "step": 820
-    },
-    {
-      "epoch": 1.94,
-      "learning_rate": 4.478193146417446e-05,
-      "loss": 0.088,
-      "step": 830
-    },
-    {
-      "epoch": 1.96,
-      "learning_rate": 4.46521287642783e-05,
-      "loss": 0.0728,
-      "step": 840
-    },
-    {
-      "epoch": 1.98,
-      "learning_rate": 4.4522326064382144e-05,
-      "loss": 0.217,
-      "step": 850
-    },
-    {
-      "epoch": 2.0,
-      "eval_accuracy": 0.9768497503404449,
-      "eval_loss": 0.09057731926441193,
-      "eval_runtime": 312.4423,
-      "eval_samples_per_second": 14.102,
-      "eval_steps_per_second": 2.352,
-      "step": 857
-    },
-    {
-      "epoch": 2.01,
-      "learning_rate": 4.4392523364485984e-05,
-      "loss": 0.1882,
-      "step": 860
-    },
-    {
-      "epoch": 2.03,
-      "learning_rate": 4.426272066458982e-05,
-      "loss": 0.134,
-      "step": 870
-    },
-    {
-      "epoch": 2.05,
-      "learning_rate": 4.413291796469366e-05,
-      "loss": 0.1215,
-      "step": 880
-    },
-    {
-      "epoch": 2.08,
-      "learning_rate": 4.400311526479751e-05,
-      "loss": 0.1035,
-      "step": 890
-    },
-    {
-      "epoch": 2.1,
-      "learning_rate": 4.3873312564901355e-05,
-      "loss": 0.167,
-      "step": 900
-    },
-    {
-      "epoch": 2.12,
-      "learning_rate": 4.3743509865005194e-05,
-      "loss": 0.1287,
-      "step": 910
-    },
-    {
-      "epoch": 2.15,
-      "learning_rate": 4.3613707165109034e-05,
-      "loss": 0.114,
-      "step": 920
-    },
-    {
-      "epoch": 2.17,
-      "learning_rate": 4.348390446521288e-05,
-      "loss": 0.1652,
-      "step": 930
-    },
-    {
-      "epoch": 2.19,
-      "learning_rate": 4.335410176531672e-05,
-      "loss": 0.0745,
-      "step": 940
-    },
-    {
-      "epoch": 2.22,
-      "learning_rate": 4.3224299065420565e-05,
-      "loss": 0.122,
-      "step": 950
-    },
-    {
-      "epoch": 2.24,
-      "learning_rate": 4.3094496365524405e-05,
-      "loss": 0.0581,
-      "step": 960
-    },
-    {
-      "epoch": 2.26,
-      "learning_rate": 4.2964693665628244e-05,
-      "loss": 0.0924,
-      "step": 970
-    },
-    {
-      "epoch": 2.29,
-      "learning_rate": 4.283489096573209e-05,
-      "loss": 0.1636,
-      "step": 980
-    },
-    {
-      "epoch": 2.31,
-      "learning_rate": 4.270508826583593e-05,
-      "loss": 0.0971,
-      "step": 990
-    },
-    {
-      "epoch": 2.33,
-      "learning_rate": 4.257528556593977e-05,
-      "loss": 0.1453,
-      "step": 1000
-    },
-    {
-      "epoch": 2.36,
-      "learning_rate": 4.2445482866043615e-05,
-      "loss": 0.0725,
-      "step": 1010
-    },
-    {
-      "epoch": 2.38,
-      "learning_rate": 4.231568016614746e-05,
-      "loss": 0.1206,
-      "step": 1020
-    },
-    {
-      "epoch": 2.4,
-      "learning_rate": 4.21858774662513e-05,
-      "loss": 0.1614,
-      "step": 1030
-    },
-    {
-      "epoch": 2.43,
-      "learning_rate": 4.205607476635514e-05,
-      "loss": 0.117,
-      "step": 1040
-    },
-    {
-      "epoch": 2.45,
-      "learning_rate": 4.192627206645898e-05,
-      "loss": 0.1531,
-      "step": 1050
-    },
-    {
-      "epoch": 2.47,
-      "learning_rate": 4.1796469366562826e-05,
-      "loss": 0.1848,
-      "step": 1060
-    },
-    {
-      "epoch": 2.5,
-      "learning_rate": 4.166666666666667e-05,
-      "loss": 0.0604,
-      "step": 1070
-    },
-    {
-      "epoch": 2.52,
-      "learning_rate": 4.153686396677051e-05,
-      "loss": 0.0921,
-      "step": 1080
-    },
-    {
-      "epoch": 2.54,
-      "learning_rate": 4.140706126687435e-05,
-      "loss": 0.1631,
-      "step": 1090
-    },
-    {
-      "epoch": 2.57,
-      "learning_rate": 4.12772585669782e-05,
-      "loss": 0.1608,
-      "step": 1100
-    },
-    {
-      "epoch": 2.59,
-      "learning_rate": 4.1147455867082036e-05,
-      "loss": 0.1363,
-      "step": 1110
-    },
-    {
-      "epoch": 2.61,
-      "learning_rate": 4.101765316718588e-05,
-      "loss": 0.0879,
-      "step": 1120
-    },
-    {
-      "epoch": 2.64,
-      "learning_rate": 4.088785046728972e-05,
-      "loss": 0.1471,
-      "step": 1130
-    },
-    {
-      "epoch": 2.66,
-      "learning_rate": 4.075804776739356e-05,
-      "loss": 0.0747,
-      "step": 1140
-    },
-    {
-      "epoch": 2.68,
-      "learning_rate": 4.062824506749741e-05,
-      "loss": 0.1141,
-      "step": 1150
-    },
-    {
-      "epoch": 2.71,
-      "learning_rate": 4.049844236760125e-05,
-      "loss": 0.1144,
-      "step": 1160
-    },
-    {
-      "epoch": 2.73,
-      "learning_rate": 4.0368639667705086e-05,
-      "loss": 0.1359,
-      "step": 1170
-    },
-    {
-      "epoch": 2.75,
-      "learning_rate": 4.023883696780893e-05,
-      "loss": 0.1119,
-      "step": 1180
-    },
-    {
-      "epoch": 2.78,
-      "learning_rate": 4.010903426791278e-05,
-      "loss": 0.1034,
-      "step": 1190
-    },
-    {
-      "epoch": 2.8,
-      "learning_rate": 3.997923156801662e-05,
-      "loss": 0.1054,
-      "step": 1200
-    },
-    {
-      "epoch": 2.82,
-      "learning_rate": 3.984942886812046e-05,
-      "loss": 0.099,
-      "step": 1210
-    },
-    {
-      "epoch": 2.85,
-      "learning_rate": 3.97196261682243e-05,
-      "loss": 0.0826,
-      "step": 1220
-    },
-    {
-      "epoch": 2.87,
-      "learning_rate": 3.958982346832814e-05,
-      "loss": 0.182,
-      "step": 1230
-    },
-    {
-      "epoch": 2.89,
-      "learning_rate": 3.946002076843199e-05,
-      "loss": 0.0568,
-      "step": 1240
-    },
-    {
-      "epoch": 2.92,
-      "learning_rate": 3.933021806853583e-05,
-      "loss": 0.1613,
-      "step": 1250
-    },
-    {
-      "epoch": 2.94,
-      "learning_rate": 3.920041536863967e-05,
-      "loss": 0.0888,
-      "step": 1260
-    },
-    {
-      "epoch": 2.96,
-      "learning_rate": 3.9070612668743514e-05,
-      "loss": 0.0523,
-      "step": 1270
-    },
-    {
-      "epoch": 2.99,
-      "learning_rate": 3.8940809968847354e-05,
-      "loss": 0.1023,
-      "step": 1280
-    },
-    {
-      "epoch": 3.0,
-      "eval_accuracy": 0.986609169314571,
-      "eval_loss": 0.05608086660504341,
-      "eval_runtime": 272.5412,
-      "eval_samples_per_second": 16.166,
-      "eval_steps_per_second": 2.697,
-      "step": 1285
-    },
-    {
-      "epoch": 3.01,
-      "learning_rate": 3.881100726895119e-05,
-      "loss": 0.1112,
-      "step": 1290
-    },
-    {
-      "epoch": 3.03,
-      "learning_rate": 3.868120456905504e-05,
-      "loss": 0.0433,
-      "step": 1300
-    },
-    {
-      "epoch": 3.06,
-      "learning_rate": 3.855140186915888e-05,
-      "loss": 0.1266,
-      "step": 1310
-    },
-    {
-      "epoch": 3.08,
-      "learning_rate": 3.8421599169262725e-05,
-      "loss": 0.1462,
-      "step": 1320
-    },
-    {
-      "epoch": 3.1,
-      "learning_rate": 3.8291796469366564e-05,
-      "loss": 0.1397,
-      "step": 1330
-    },
-    {
-      "epoch": 3.13,
-      "learning_rate": 3.8161993769470404e-05,
-      "loss": 0.0813,
-      "step": 1340
-    },
-    {
-      "epoch": 3.15,
-      "learning_rate": 3.803219106957425e-05,
-      "loss": 0.0967,
-      "step": 1350
-    },
-    {
-      "epoch": 3.17,
-      "learning_rate": 3.7902388369678096e-05,
-      "loss": 0.0631,
-      "step": 1360
-    },
-    {
-      "epoch": 3.2,
-      "learning_rate": 3.7772585669781935e-05,
-      "loss": 0.1784,
-      "step": 1370
-    },
-    {
-      "epoch": 3.22,
-      "learning_rate": 3.7642782969885775e-05,
-      "loss": 0.0413,
-      "step": 1380
-    },
-    {
-      "epoch": 3.24,
-      "learning_rate": 3.7512980269989614e-05,
-      "loss": 0.0924,
-      "step": 1390
-    },
-    {
-      "epoch": 3.27,
-      "learning_rate": 3.738317757009346e-05,
-      "loss": 0.1043,
-      "step": 1400
-    },
-    {
-      "epoch": 3.29,
-      "learning_rate": 3.72533748701973e-05,
-      "loss": 0.0909,
-      "step": 1410
-    },
-    {
-      "epoch": 3.31,
-      "learning_rate": 3.7123572170301146e-05,
-      "loss": 0.1211,
-      "step": 1420
-    },
-    {
-      "epoch": 3.34,
-      "learning_rate": 3.6993769470404985e-05,
-      "loss": 0.0909,
-      "step": 1430
-    },
-    {
-      "epoch": 3.36,
-      "learning_rate": 3.686396677050883e-05,
-      "loss": 0.053,
-      "step": 1440
-    },
-    {
-      "epoch": 3.38,
-      "learning_rate": 3.673416407061267e-05,
-      "loss": 0.1416,
-      "step": 1450
-    },
-    {
-      "epoch": 3.41,
-      "learning_rate": 3.660436137071651e-05,
-      "loss": 0.0664,
-      "step": 1460
-    },
-    {
-      "epoch": 3.43,
-      "learning_rate": 3.647455867082035e-05,
-      "loss": 0.0572,
-      "step": 1470
-    },
-    {
-      "epoch": 3.45,
-      "learning_rate": 3.6344755970924196e-05,
-      "loss": 0.2081,
-      "step": 1480
-    },
-    {
-      "epoch": 3.48,
-      "learning_rate": 3.621495327102804e-05,
-      "loss": 0.0666,
-      "step": 1490
-    },
-    {
-      "epoch": 3.5,
-      "learning_rate": 3.608515057113188e-05,
-      "loss": 0.1286,
-      "step": 1500
-    },
-    {
-      "epoch": 3.52,
-      "learning_rate": 3.595534787123572e-05,
-      "loss": 0.1309,
-      "step": 1510
-    },
-    {
-      "epoch": 3.55,
-      "learning_rate": 3.582554517133957e-05,
-      "loss": 0.0854,
-      "step": 1520
-    },
-    {
-      "epoch": 3.57,
-      "learning_rate": 3.5695742471443406e-05,
-      "loss": 0.129,
-      "step": 1530
-    },
-    {
-      "epoch": 3.59,
-      "learning_rate": 3.556593977154725e-05,
-      "loss": 0.1091,
-      "step": 1540
-    },
-    {
-      "epoch": 3.62,
-      "learning_rate": 3.543613707165109e-05,
-      "loss": 0.1034,
-      "step": 1550
-    },
-    {
-      "epoch": 3.64,
-      "learning_rate": 3.530633437175493e-05,
-      "loss": 0.1008,
-      "step": 1560
-    },
-    {
-      "epoch": 3.66,
-      "learning_rate": 3.517653167185878e-05,
-      "loss": 0.1413,
-      "step": 1570
-    },
-    {
-      "epoch": 3.69,
-      "learning_rate": 3.504672897196262e-05,
-      "loss": 0.1446,
-      "step": 1580
-    },
-    {
-      "epoch": 3.71,
-      "learning_rate": 3.4916926272066456e-05,
-      "loss": 0.0846,
-      "step": 1590
-    },
-    {
-      "epoch": 3.73,
-      "learning_rate": 3.47871235721703e-05,
-      "loss": 0.0485,
-      "step": 1600
-    },
-    {
-      "epoch": 3.76,
-      "learning_rate": 3.465732087227415e-05,
-      "loss": 0.069,
-      "step": 1610
-    },
-    {
-      "epoch": 3.78,
-      "learning_rate": 3.452751817237799e-05,
-      "loss": 0.0664,
-      "step": 1620
-    },
-    {
-      "epoch": 3.8,
-      "learning_rate": 3.439771547248183e-05,
-      "loss": 0.0909,
-      "step": 1630
-    },
-    {
-      "epoch": 3.83,
-      "learning_rate": 3.426791277258567e-05,
-      "loss": 0.1046,
-      "step": 1640
-    },
-    {
-      "epoch": 3.85,
-      "learning_rate": 3.413811007268951e-05,
-      "loss": 0.0977,
-      "step": 1650
-    },
-    {
-      "epoch": 3.87,
-      "learning_rate": 3.400830737279336e-05,
-      "loss": 0.1135,
-      "step": 1660
-    },
-    {
-      "epoch": 3.9,
-      "learning_rate": 3.38785046728972e-05,
-      "loss": 0.0386,
-      "step": 1670
-    },
-    {
-      "epoch": 3.92,
-      "learning_rate": 3.374870197300104e-05,
-      "loss": 0.1839,
-      "step": 1680
-    },
-    {
-      "epoch": 3.94,
-      "learning_rate": 3.3618899273104884e-05,
-      "loss": 0.0722,
-      "step": 1690
-    },
-    {
-      "epoch": 3.97,
-      "learning_rate": 3.3489096573208724e-05,
-      "loss": 0.136,
-      "step": 1700
-    },
-    {
-      "epoch": 3.99,
-      "learning_rate": 3.335929387331256e-05,
-      "loss": 0.1007,
-      "step": 1710
-    },
-    {
-      "epoch": 4.0,
-      "eval_accuracy": 0.9902405810258738,
-      "eval_loss": 0.042905017733573914,
-      "eval_runtime": 280.0162,
-      "eval_samples_per_second": 15.735,
-      "eval_steps_per_second": 2.625,
-      "step": 1714
-    },
-    {
-      "epoch": 4.01,
-      "learning_rate": 3.322949117341641e-05,
-      "loss": 0.1086,
-      "step": 1720
-    },
-    {
-      "epoch": 4.04,
-      "learning_rate": 3.309968847352025e-05,
-      "loss": 0.1085,
-      "step": 1730
-    },
-    {
-      "epoch": 4.06,
-      "learning_rate": 3.2969885773624095e-05,
-      "loss": 0.1395,
-      "step": 1740
-    },
-    {
-      "epoch": 4.08,
-      "learning_rate": 3.2840083073727934e-05,
-      "loss": 0.1635,
-      "step": 1750
-    },
-    {
-      "epoch": 4.11,
-      "learning_rate": 3.2710280373831774e-05,
-      "loss": 0.0952,
-      "step": 1760
-    },
-    {
-      "epoch": 4.13,
-      "learning_rate": 3.258047767393562e-05,
-      "loss": 0.0691,
-      "step": 1770
-    },
-    {
-      "epoch": 4.15,
-      "learning_rate": 3.2450674974039466e-05,
-      "loss": 0.0535,
-      "step": 1780
-    },
-    {
-      "epoch": 4.18,
-      "learning_rate": 3.2320872274143305e-05,
-      "loss": 0.0514,
-      "step": 1790
-    },
-    {
-      "epoch": 4.2,
-      "learning_rate": 3.2191069574247145e-05,
-      "loss": 0.0875,
-      "step": 1800
-    },
-    {
-      "epoch": 4.22,
-      "learning_rate": 3.2061266874350984e-05,
-      "loss": 0.0517,
-      "step": 1810
-    },
-    {
-      "epoch": 4.25,
-      "learning_rate": 3.193146417445483e-05,
-      "loss": 0.1014,
-      "step": 1820
-    },
-    {
-      "epoch": 4.27,
-      "learning_rate": 3.1801661474558676e-05,
-      "loss": 0.0374,
-      "step": 1830
-    },
-    {
-      "epoch": 4.29,
-      "learning_rate": 3.1671858774662516e-05,
-      "loss": 0.1028,
-      "step": 1840
-    },
-    {
-      "epoch": 4.32,
-      "learning_rate": 3.1542056074766355e-05,
-      "loss": 0.1316,
-      "step": 1850
-    },
-    {
-      "epoch": 4.34,
-      "learning_rate": 3.14122533748702e-05,
-      "loss": 0.1412,
-      "step": 1860
-    },
-    {
-      "epoch": 4.36,
-      "learning_rate": 3.128245067497404e-05,
-      "loss": 0.0216,
-      "step": 1870
-    },
-    {
-      "epoch": 4.39,
-      "learning_rate": 3.115264797507788e-05,
-      "loss": 0.0441,
-      "step": 1880
-    },
-    {
-      "epoch": 4.41,
-      "learning_rate": 3.1022845275181726e-05,
-      "loss": 0.0754,
-      "step": 1890
-    },
-    {
-      "epoch": 4.43,
-      "learning_rate": 3.0893042575285566e-05,
-      "loss": 0.1619,
-      "step": 1900
-    },
-    {
-      "epoch": 4.46,
-      "learning_rate": 3.076323987538941e-05,
-      "loss": 0.03,
-      "step": 1910
-    },
-    {
-      "epoch": 4.48,
-      "learning_rate": 3.063343717549325e-05,
-      "loss": 0.0197,
-      "step": 1920
-    },
-    {
-      "epoch": 4.5,
-      "learning_rate": 3.0503634475597094e-05,
-      "loss": 0.0638,
-      "step": 1930
-    },
-    {
-      "epoch": 4.53,
-      "learning_rate": 3.0373831775700934e-05,
-      "loss": 0.0764,
-      "step": 1940
-    },
-    {
-      "epoch": 4.55,
-      "learning_rate": 3.024402907580478e-05,
-      "loss": 0.0439,
-      "step": 1950
-    },
-    {
-      "epoch": 4.57,
-      "learning_rate": 3.0114226375908622e-05,
-      "loss": 0.0468,
-      "step": 1960
-    },
-    {
-      "epoch": 4.6,
-      "learning_rate": 2.9984423676012462e-05,
-      "loss": 0.1002,
-      "step": 1970
-    },
-    {
-      "epoch": 4.62,
-      "learning_rate": 2.9854620976116305e-05,
-      "loss": 0.0707,
-      "step": 1980
-    },
-    {
-      "epoch": 4.64,
-      "learning_rate": 2.9724818276220144e-05,
-      "loss": 0.1629,
-      "step": 1990
-    },
-    {
-      "epoch": 4.67,
-      "learning_rate": 2.9595015576323987e-05,
-      "loss": 0.1135,
-      "step": 2000
-    },
-    {
-      "epoch": 4.69,
-      "learning_rate": 2.9465212876427833e-05,
-      "loss": 0.1017,
-      "step": 2010
-    },
-    {
-      "epoch": 4.71,
-      "learning_rate": 2.9335410176531676e-05,
-      "loss": 0.0981,
-      "step": 2020
-    },
-    {
-      "epoch": 4.74,
-      "learning_rate": 2.9205607476635515e-05,
-      "loss": 0.0931,
-      "step": 2030
-    },
-    {
-      "epoch": 4.76,
-      "learning_rate": 2.9075804776739358e-05,
-      "loss": 0.1317,
-      "step": 2040
-    },
-    {
-      "epoch": 4.78,
-      "learning_rate": 2.8946002076843197e-05,
-      "loss": 0.0295,
-      "step": 2050
-    },
-    {
-      "epoch": 4.81,
-      "learning_rate": 2.881619937694704e-05,
-      "loss": 0.0658,
-      "step": 2060
-    },
-    {
-      "epoch": 4.83,
-      "learning_rate": 2.8686396677050886e-05,
-      "loss": 0.1634,
-      "step": 2070
-    },
-    {
-      "epoch": 4.85,
-      "learning_rate": 2.855659397715473e-05,
-      "loss": 0.0865,
-      "step": 2080
-    },
-    {
-      "epoch": 4.88,
-      "learning_rate": 2.842679127725857e-05,
-      "loss": 0.0807,
-      "step": 2090
-    },
-    {
-      "epoch": 4.9,
-      "learning_rate": 2.829698857736241e-05,
-      "loss": 0.0434,
-      "step": 2100
-    },
-    {
-      "epoch": 4.92,
-      "learning_rate": 2.816718587746625e-05,
-      "loss": 0.0414,
-      "step": 2110
-    },
-    {
-      "epoch": 4.95,
-      "learning_rate": 2.8037383177570094e-05,
-      "loss": 0.0973,
-      "step": 2120
-    },
-    {
-      "epoch": 4.97,
-      "learning_rate": 2.790758047767394e-05,
-      "loss": 0.0677,
-      "step": 2130
-    },
-    {
-      "epoch": 4.99,
-      "learning_rate": 2.777777777777778e-05,
-      "loss": 0.0687,
-      "step": 2140
-    },
-    {
-      "epoch": 5.0,
-      "eval_accuracy": 0.9877439854743532,
-      "eval_loss": 0.04716098681092262,
-      "eval_runtime": 277.6857,
-      "eval_samples_per_second": 15.867,
-      "eval_steps_per_second": 2.647,
-      "step": 2142
-    },
-    {
-      "epoch": 5.02,
-      "learning_rate": 2.7647975077881622e-05,
-      "loss": 0.1053,
-      "step": 2150
-    },
-    {
-      "epoch": 5.04,
-      "learning_rate": 2.751817237798546e-05,
-      "loss": 0.0516,
-      "step": 2160
-    },
-    {
-      "epoch": 5.06,
-      "learning_rate": 2.7388369678089304e-05,
-      "loss": 0.0555,
-      "step": 2170
-    },
-    {
-      "epoch": 5.09,
-      "learning_rate": 2.7258566978193147e-05,
-      "loss": 0.1344,
-      "step": 2180
-    },
-    {
-      "epoch": 5.11,
-      "learning_rate": 2.7128764278296993e-05,
-      "loss": 0.1401,
-      "step": 2190
-    },
-    {
-      "epoch": 5.13,
-      "learning_rate": 2.6998961578400832e-05,
-      "loss": 0.0238,
-      "step": 2200
-    },
-    {
-      "epoch": 5.16,
-      "learning_rate": 2.6869158878504675e-05,
-      "loss": 0.0966,
-      "step": 2210
-    },
-    {
-      "epoch": 5.18,
-      "learning_rate": 2.6739356178608515e-05,
-      "loss": 0.078,
-      "step": 2220
-    },
-    {
-      "epoch": 5.2,
-      "learning_rate": 2.6609553478712357e-05,
-      "loss": 0.129,
-      "step": 2230
-    },
-    {
-      "epoch": 5.23,
-      "learning_rate": 2.6479750778816197e-05,
-      "loss": 0.019,
-      "step": 2240
-    },
-    {
-      "epoch": 5.25,
-      "learning_rate": 2.6349948078920046e-05,
-      "loss": 0.0505,
-      "step": 2250
-    },
-    {
-      "epoch": 5.27,
-      "learning_rate": 2.6220145379023886e-05,
-      "loss": 0.0413,
-      "step": 2260
-    },
-    {
-      "epoch": 5.3,
-      "learning_rate": 2.609034267912773e-05,
-      "loss": 0.0639,
-      "step": 2270
-    },
-    {
-      "epoch": 5.32,
-      "learning_rate": 2.5960539979231568e-05,
-      "loss": 0.1113,
-      "step": 2280
-    },
-    {
-      "epoch": 5.34,
-      "learning_rate": 2.583073727933541e-05,
-      "loss": 0.0955,
-      "step": 2290
-    },
-    {
-      "epoch": 5.37,
-      "learning_rate": 2.570093457943925e-05,
-      "loss": 0.0259,
-      "step": 2300
-    },
-    {
-      "epoch": 5.39,
-      "learning_rate": 2.5571131879543096e-05,
-      "loss": 0.0876,
-      "step": 2310
-    },
-    {
-      "epoch": 5.41,
-      "learning_rate": 2.544132917964694e-05,
-      "loss": 0.1074,
-      "step": 2320
-    },
-    {
-      "epoch": 5.44,
-      "learning_rate": 2.531152647975078e-05,
-      "loss": 0.0382,
-      "step": 2330
-    },
-    {
-      "epoch": 5.46,
-      "learning_rate": 2.518172377985462e-05,
-      "loss": 0.1101,
-      "step": 2340
-    },
-    {
-      "epoch": 5.48,
-      "learning_rate": 2.5051921079958464e-05,
-      "loss": 0.1023,
-      "step": 2350
-    },
-    {
-      "epoch": 5.51,
-      "learning_rate": 2.4922118380062307e-05,
-      "loss": 0.0437,
-      "step": 2360
-    },
-    {
-      "epoch": 5.53,
-      "learning_rate": 2.4792315680166146e-05,
-      "loss": 0.1524,
-      "step": 2370
-    },
-    {
-      "epoch": 5.55,
-      "learning_rate": 2.4662512980269992e-05,
-      "loss": 0.0495,
-      "step": 2380
-    },
-    {
-      "epoch": 5.58,
-      "learning_rate": 2.4532710280373832e-05,
-      "loss": 0.1185,
-      "step": 2390
-    },
-    {
-      "epoch": 5.6,
-      "learning_rate": 2.4402907580477675e-05,
-      "loss": 0.0849,
-      "step": 2400
-    },
-    {
-      "epoch": 5.62,
-      "learning_rate": 2.4273104880581517e-05,
-      "loss": 0.0454,
-      "step": 2410
-    },
-    {
-      "epoch": 5.65,
-      "learning_rate": 2.414330218068536e-05,
-      "loss": 0.072,
-      "step": 2420
-    },
-    {
-      "epoch": 5.67,
-      "learning_rate": 2.40134994807892e-05,
-      "loss": 0.0634,
-      "step": 2430
-    },
-    {
-      "epoch": 5.69,
-      "learning_rate": 2.3883696780893046e-05,
-      "loss": 0.1499,
-      "step": 2440
-    },
-    {
-      "epoch": 5.72,
-      "learning_rate": 2.3753894080996885e-05,
-      "loss": 0.0858,
-      "step": 2450
-    },
-    {
-      "epoch": 5.74,
-      "learning_rate": 2.3624091381100728e-05,
-      "loss": 0.1055,
-      "step": 2460
-    },
-    {
-      "epoch": 5.76,
-      "learning_rate": 2.349428868120457e-05,
-      "loss": 0.0382,
-      "step": 2470
-    },
-    {
-      "epoch": 5.79,
-      "learning_rate": 2.3364485981308414e-05,
-      "loss": 0.0385,
-      "step": 2480
-    },
-    {
-      "epoch": 5.81,
-      "learning_rate": 2.3234683281412253e-05,
-      "loss": 0.0402,
-      "step": 2490
-    },
-    {
-      "epoch": 5.83,
-      "learning_rate": 2.3104880581516096e-05,
-      "loss": 0.0712,
-      "step": 2500
-    },
-    {
-      "epoch": 5.86,
-      "learning_rate": 2.297507788161994e-05,
-      "loss": 0.0524,
-      "step": 2510
-    },
-    {
-      "epoch": 5.88,
-      "learning_rate": 2.284527518172378e-05,
-      "loss": 0.0494,
-      "step": 2520
-    },
-    {
-      "epoch": 5.9,
-      "learning_rate": 2.2715472481827624e-05,
-      "loss": 0.0917,
-      "step": 2530
-    },
-    {
-      "epoch": 5.93,
-      "learning_rate": 2.2585669781931463e-05,
-      "loss": 0.0737,
-      "step": 2540
-    },
-    {
-      "epoch": 5.95,
-      "learning_rate": 2.2455867082035306e-05,
-      "loss": 0.0682,
-      "step": 2550
-    },
-    {
-      "epoch": 5.97,
-      "learning_rate": 2.232606438213915e-05,
-      "loss": 0.1312,
-      "step": 2560
-    },
-    {
-      "epoch": 6.0,
-      "learning_rate": 2.2196261682242992e-05,
-      "loss": 0.09,
-      "step": 2570
-    },
-    {
-      "epoch": 6.0,
-      "eval_accuracy": 0.9884248751702224,
-      "eval_loss": 0.04844553396105766,
-      "eval_runtime": 281.2078,
-      "eval_samples_per_second": 15.668,
-      "eval_steps_per_second": 2.614,
-      "step": 2571
-    },
-    {
-      "epoch": 6.02,
-      "learning_rate": 2.206645898234683e-05,
-      "loss": 0.0954,
-      "step": 2580
-    },
-    {
-      "epoch": 6.04,
-      "learning_rate": 2.1936656282450677e-05,
-      "loss": 0.0657,
-      "step": 2590
-    },
-    {
-      "epoch": 6.07,
-      "learning_rate": 2.1806853582554517e-05,
-      "loss": 0.0274,
-      "step": 2600
-    },
-    {
-      "epoch": 6.09,
-      "learning_rate": 2.167705088265836e-05,
-      "loss": 0.0353,
-      "step": 2610
-    },
-    {
-      "epoch": 6.11,
-      "learning_rate": 2.1547248182762202e-05,
-      "loss": 0.0646,
-      "step": 2620
-    },
-    {
-      "epoch": 6.14,
-      "learning_rate": 2.1417445482866045e-05,
-      "loss": 0.094,
-      "step": 2630
-    },
-    {
-      "epoch": 6.16,
-      "learning_rate": 2.1287642782969885e-05,
-      "loss": 0.1035,
-      "step": 2640
-    },
-    {
-      "epoch": 6.18,
-      "learning_rate": 2.115784008307373e-05,
-      "loss": 0.0774,
-      "step": 2650
-    },
-    {
-      "epoch": 6.21,
-      "learning_rate": 2.102803738317757e-05,
-      "loss": 0.0823,
-      "step": 2660
-    },
-    {
-      "epoch": 6.23,
-      "learning_rate": 2.0898234683281413e-05,
-      "loss": 0.0855,
-      "step": 2670
-    },
-    {
-      "epoch": 6.25,
-      "learning_rate": 2.0768431983385256e-05,
-      "loss": 0.089,
-      "step": 2680
-    },
-    {
-      "epoch": 6.28,
-      "learning_rate": 2.06386292834891e-05,
-      "loss": 0.0706,
-      "step": 2690
-    },
-    {
-      "epoch": 6.3,
-      "learning_rate": 2.050882658359294e-05,
-      "loss": 0.0474,
-      "step": 2700
-    },
-    {
-      "epoch": 6.32,
-      "learning_rate": 2.037902388369678e-05,
-      "loss": 0.0867,
-      "step": 2710
-    },
-    {
-      "epoch": 6.35,
-      "learning_rate": 2.0249221183800623e-05,
-      "loss": 0.1014,
-      "step": 2720
-    },
-    {
-      "epoch": 6.37,
-      "learning_rate": 2.0119418483904466e-05,
-      "loss": 0.0826,
-      "step": 2730
-    },
-    {
-      "epoch": 6.39,
-      "learning_rate": 1.998961578400831e-05,
-      "loss": 0.0848,
-      "step": 2740
-    },
-    {
-      "epoch": 6.42,
-      "learning_rate": 1.985981308411215e-05,
-      "loss": 0.037,
-      "step": 2750
-    },
-    {
-      "epoch": 6.44,
-      "learning_rate": 1.9730010384215995e-05,
-      "loss": 0.0557,
-      "step": 2760
-    },
-    {
-      "epoch": 6.46,
-      "learning_rate": 1.9600207684319834e-05,
-      "loss": 0.0257,
-      "step": 2770
-    },
-    {
-      "epoch": 6.49,
-      "learning_rate": 1.9470404984423677e-05,
-      "loss": 0.0323,
-      "step": 2780
-    },
-    {
-      "epoch": 6.51,
-      "learning_rate": 1.934060228452752e-05,
-      "loss": 0.1008,
-      "step": 2790
-    },
-    {
-      "epoch": 6.53,
-      "learning_rate": 1.9210799584631362e-05,
-      "loss": 0.0456,
-      "step": 2800
-    },
-    {
-      "epoch": 6.56,
-      "learning_rate": 1.9080996884735202e-05,
-      "loss": 0.0613,
-      "step": 2810
-    },
-    {
-      "epoch": 6.58,
-      "learning_rate": 1.8951194184839048e-05,
-      "loss": 0.0573,
-      "step": 2820
-    },
-    {
-      "epoch": 6.6,
-      "learning_rate": 1.8821391484942887e-05,
-      "loss": 0.0205,
-      "step": 2830
-    },
-    {
-      "epoch": 6.63,
-      "learning_rate": 1.869158878504673e-05,
-      "loss": 0.1002,
-      "step": 2840
-    },
-    {
-      "epoch": 6.65,
-      "learning_rate": 1.8561786085150573e-05,
-      "loss": 0.0575,
-      "step": 2850
-    },
-    {
-      "epoch": 6.67,
-      "learning_rate": 1.8431983385254416e-05,
-      "loss": 0.0201,
-      "step": 2860
-    },
-    {
-      "epoch": 6.7,
-      "learning_rate": 1.8302180685358255e-05,
-      "loss": 0.1134,
-      "step": 2870
-    },
-    {
-      "epoch": 6.72,
-      "learning_rate": 1.8172377985462098e-05,
-      "loss": 0.0187,
-      "step": 2880
-    },
-    {
-      "epoch": 6.74,
-      "learning_rate": 1.804257528556594e-05,
-      "loss": 0.0503,
-      "step": 2890
-    },
-    {
-      "epoch": 6.77,
-      "learning_rate": 1.7912772585669783e-05,
-      "loss": 0.0975,
-      "step": 2900
-    },
-    {
-      "epoch": 6.79,
-      "learning_rate": 1.7782969885773626e-05,
-      "loss": 0.0436,
-      "step": 2910
-    },
-    {
-      "epoch": 6.81,
-      "learning_rate": 1.7653167185877466e-05,
-      "loss": 0.1462,
-      "step": 2920
-    },
-    {
-      "epoch": 6.84,
-      "learning_rate": 1.752336448598131e-05,
-      "loss": 0.0965,
-      "step": 2930
-    },
-    {
-      "epoch": 6.86,
-      "learning_rate": 1.739356178608515e-05,
-      "loss": 0.049,
-      "step": 2940
-    },
-    {
-      "epoch": 6.88,
-      "learning_rate": 1.7263759086188994e-05,
-      "loss": 0.0501,
-      "step": 2950
-    },
-    {
-      "epoch": 6.91,
-      "learning_rate": 1.7133956386292833e-05,
-      "loss": 0.0318,
-      "step": 2960
-    },
-    {
-      "epoch": 6.93,
-      "learning_rate": 1.700415368639668e-05,
-      "loss": 0.097,
-      "step": 2970
-    },
-    {
-      "epoch": 6.95,
-      "learning_rate": 1.687435098650052e-05,
-      "loss": 0.0539,
-      "step": 2980
-    },
-    {
-      "epoch": 6.98,
-      "learning_rate": 1.6744548286604362e-05,
-      "loss": 0.1087,
-      "step": 2990
-    },
-    {
-      "epoch": 7.0,
-      "eval_accuracy": 0.9897866545619609,
-      "eval_loss": 0.040911152958869934,
-      "eval_runtime": 301.3693,
-      "eval_samples_per_second": 14.62,
-      "eval_steps_per_second": 2.439,
-      "step": 2999
-    },
-    {
-      "epoch": 7.0,
-      "learning_rate": 1.6614745586708205e-05,
-      "loss": 0.0103,
-      "step": 3000
-    },
-    {
-      "epoch": 7.02,
-      "learning_rate": 1.6484942886812047e-05,
-      "loss": 0.043,
-      "step": 3010
-    },
-    {
-      "epoch": 7.05,
-      "learning_rate": 1.6355140186915887e-05,
-      "loss": 0.0529,
-      "step": 3020
-    },
-    {
-      "epoch": 7.07,
-      "learning_rate": 1.6225337487019733e-05,
-      "loss": 0.0245,
-      "step": 3030
-    },
-    {
-      "epoch": 7.09,
-      "learning_rate": 1.6095534787123572e-05,
-      "loss": 0.0222,
-      "step": 3040
-    },
-    {
-      "epoch": 7.12,
-      "learning_rate": 1.5965732087227415e-05,
-      "loss": 0.0123,
-      "step": 3050
-    },
-    {
-      "epoch": 7.14,
-      "learning_rate": 1.5835929387331258e-05,
-      "loss": 0.057,
-      "step": 3060
-    },
-    {
-      "epoch": 7.16,
-      "learning_rate": 1.57061266874351e-05,
-      "loss": 0.0724,
-      "step": 3070
-    },
-    {
-      "epoch": 7.19,
-      "learning_rate": 1.557632398753894e-05,
-      "loss": 0.062,
-      "step": 3080
-    },
-    {
-      "epoch": 7.21,
-      "learning_rate": 1.5446521287642783e-05,
-      "loss": 0.0768,
-      "step": 3090
-    },
-    {
-      "epoch": 7.23,
-      "learning_rate": 1.5316718587746626e-05,
-      "loss": 0.0415,
-      "step": 3100
-    },
-    {
-      "epoch": 7.26,
-      "learning_rate": 1.5186915887850467e-05,
-      "loss": 0.0894,
-      "step": 3110
-    },
-    {
-      "epoch": 7.28,
-      "learning_rate": 1.5057113187954311e-05,
-      "loss": 0.0789,
-      "step": 3120
-    },
-    {
-      "epoch": 7.3,
-      "learning_rate": 1.4927310488058152e-05,
-      "loss": 0.0344,
-      "step": 3130
-    },
-    {
-      "epoch": 7.33,
-      "learning_rate": 1.4797507788161993e-05,
-      "loss": 0.0326,
-      "step": 3140
-    },
-    {
-      "epoch": 7.35,
-      "learning_rate": 1.4667705088265838e-05,
-      "loss": 0.0724,
-      "step": 3150
-    },
-    {
-      "epoch": 7.37,
-      "learning_rate": 1.4537902388369679e-05,
-      "loss": 0.0513,
-      "step": 3160
-    },
-    {
-      "epoch": 7.4,
-      "learning_rate": 1.440809968847352e-05,
-      "loss": 0.0463,
-      "step": 3170
-    },
-    {
-      "epoch": 7.42,
-      "learning_rate": 1.4278296988577365e-05,
-      "loss": 0.0419,
-      "step": 3180
-    },
-    {
-      "epoch": 7.44,
-      "learning_rate": 1.4148494288681206e-05,
-      "loss": 0.1547,
-      "step": 3190
-    },
-    {
-      "epoch": 7.47,
-      "learning_rate": 1.4018691588785047e-05,
-      "loss": 0.0882,
-      "step": 3200
-    },
-    {
-      "epoch": 7.49,
-      "learning_rate": 1.388888888888889e-05,
-      "loss": 0.0385,
-      "step": 3210
-    },
-    {
-      "epoch": 7.51,
-      "learning_rate": 1.375908618899273e-05,
-      "loss": 0.0333,
-      "step": 3220
-    },
-    {
-      "epoch": 7.54,
-      "learning_rate": 1.3629283489096573e-05,
-      "loss": 0.0839,
-      "step": 3230
-    },
-    {
-      "epoch": 7.56,
-      "learning_rate": 1.3499480789200416e-05,
-      "loss": 0.0623,
-      "step": 3240
-    },
-    {
-      "epoch": 7.58,
-      "learning_rate": 1.3369678089304257e-05,
-      "loss": 0.094,
-      "step": 3250
-    },
-    {
-      "epoch": 7.61,
-      "learning_rate": 1.3239875389408098e-05,
-      "loss": 0.0429,
-      "step": 3260
-    },
-    {
-      "epoch": 7.63,
-      "learning_rate": 1.3110072689511943e-05,
-      "loss": 0.0015,
-      "step": 3270
-    },
-    {
-      "epoch": 7.65,
-      "learning_rate": 1.2980269989615784e-05,
-      "loss": 0.0122,
-      "step": 3280
-    },
-    {
-      "epoch": 7.68,
-      "learning_rate": 1.2850467289719625e-05,
-      "loss": 0.0476,
-      "step": 3290
-    },
-    {
-      "epoch": 7.7,
-      "learning_rate": 1.272066458982347e-05,
-      "loss": 0.0371,
-      "step": 3300
-    },
-    {
-      "epoch": 7.72,
-      "learning_rate": 1.259086188992731e-05,
-      "loss": 0.0905,
-      "step": 3310
-    },
-    {
-      "epoch": 7.75,
-      "learning_rate": 1.2461059190031153e-05,
-      "loss": 0.028,
-      "step": 3320
-    },
-    {
-      "epoch": 7.77,
-      "learning_rate": 1.2331256490134996e-05,
-      "loss": 0.0375,
-      "step": 3330
-    },
-    {
-      "epoch": 7.79,
-      "learning_rate": 1.2201453790238837e-05,
-      "loss": 0.0711,
-      "step": 3340
-    },
-    {
-      "epoch": 7.82,
-      "learning_rate": 1.207165109034268e-05,
-      "loss": 0.0787,
-      "step": 3350
-    },
-    {
-      "epoch": 7.84,
-      "learning_rate": 1.1941848390446523e-05,
-      "loss": 0.0192,
-      "step": 3360
-    },
-    {
-      "epoch": 7.86,
-      "learning_rate": 1.1812045690550364e-05,
-      "loss": 0.055,
-      "step": 3370
-    },
-    {
-      "epoch": 7.89,
-      "learning_rate": 1.1682242990654207e-05,
-      "loss": 0.1195,
-      "step": 3380
-    },
-    {
-      "epoch": 7.91,
-      "learning_rate": 1.1552440290758048e-05,
-      "loss": 0.0388,
-      "step": 3390
-    },
-    {
-      "epoch": 7.93,
-      "learning_rate": 1.142263759086189e-05,
-      "loss": 0.021,
-      "step": 3400
-    },
-    {
-      "epoch": 7.96,
-      "learning_rate": 1.1292834890965732e-05,
-      "loss": 0.0992,
-      "step": 3410
-    },
-    {
-      "epoch": 7.98,
-      "learning_rate": 1.1163032191069575e-05,
-      "loss": 0.0477,
-      "step": 3420
-    },
-    {
-      "epoch": 8.0,
-      "eval_accuracy": 0.9909214707217431,
-      "eval_loss": 0.04681561514735222,
-      "eval_runtime": 281.3706,
-      "eval_samples_per_second": 15.659,
-      "eval_steps_per_second": 2.612,
-      "step": 3428
-    },
-    {
-      "epoch": 8.0,
-      "learning_rate": 1.1033229491173416e-05,
-      "loss": 0.0154,
-      "step": 3430
-    },
-    {
-      "epoch": 8.03,
-      "learning_rate": 1.0903426791277258e-05,
-      "loss": 0.0787,
-      "step": 3440
-    },
-    {
-      "epoch": 8.05,
-      "learning_rate": 1.0773624091381101e-05,
-      "loss": 0.0948,
-      "step": 3450
-    },
-    {
-      "epoch": 8.07,
-      "learning_rate": 1.0643821391484942e-05,
-      "loss": 0.0322,
-      "step": 3460
-    },
-    {
-      "epoch": 8.1,
-      "learning_rate": 1.0514018691588785e-05,
-      "loss": 0.0836,
-      "step": 3470
-    },
-    {
-      "epoch": 8.12,
-      "learning_rate": 1.0384215991692628e-05,
-      "loss": 0.0754,
-      "step": 3480
-    },
-    {
-      "epoch": 8.14,
-      "learning_rate": 1.025441329179647e-05,
-      "loss": 0.1152,
-      "step": 3490
-    },
-    {
-      "epoch": 8.17,
-      "learning_rate": 1.0124610591900312e-05,
-      "loss": 0.0096,
-      "step": 3500
-    },
-    {
-      "epoch": 8.19,
-      "learning_rate": 9.994807892004155e-06,
-      "loss": 0.0449,
-      "step": 3510
-    },
-    {
-      "epoch": 8.21,
-      "learning_rate": 9.865005192107997e-06,
-      "loss": 0.0418,
-      "step": 3520
-    },
-    {
-      "epoch": 8.24,
-      "learning_rate": 9.735202492211838e-06,
-      "loss": 0.0434,
-      "step": 3530
-    },
-    {
-      "epoch": 8.26,
-      "learning_rate": 9.605399792315681e-06,
-      "loss": 0.0466,
-      "step": 3540
-    },
-    {
-      "epoch": 8.28,
-      "learning_rate": 9.475597092419524e-06,
-      "loss": 0.115,
-      "step": 3550
-    },
-    {
-      "epoch": 8.31,
-      "learning_rate": 9.345794392523365e-06,
-      "loss": 0.1147,
-      "step": 3560
-    },
-    {
-      "epoch": 8.33,
-      "learning_rate": 9.215991692627208e-06,
-      "loss": 0.0387,
-      "step": 3570
-    },
-    {
-      "epoch": 8.35,
-      "learning_rate": 9.086188992731049e-06,
-      "loss": 0.0574,
-      "step": 3580
-    },
-    {
-      "epoch": 8.38,
-      "learning_rate": 8.956386292834892e-06,
-      "loss": 0.0519,
-      "step": 3590
-    },
-    {
-      "epoch": 8.4,
-      "learning_rate": 8.826583592938733e-06,
-      "loss": 0.078,
-      "step": 3600
-    },
-    {
-      "epoch": 8.42,
-      "learning_rate": 8.696780893042576e-06,
-      "loss": 0.0301,
-      "step": 3610
-    },
-    {
-      "epoch": 8.45,
-      "learning_rate": 8.566978193146417e-06,
-      "loss": 0.0545,
-      "step": 3620
-    },
-    {
-      "epoch": 8.47,
-      "learning_rate": 8.43717549325026e-06,
-      "loss": 0.0423,
-      "step": 3630
-    },
-    {
-      "epoch": 8.49,
-      "learning_rate": 8.307372793354102e-06,
-      "loss": 0.0456,
-      "step": 3640
-    },
-    {
-      "epoch": 8.52,
-      "learning_rate": 8.177570093457943e-06,
-      "loss": 0.1198,
-      "step": 3650
-    },
-    {
-      "epoch": 8.54,
-      "learning_rate": 8.047767393561786e-06,
-      "loss": 0.0277,
-      "step": 3660
-    },
-    {
-      "epoch": 8.56,
-      "learning_rate": 7.917964693665629e-06,
-      "loss": 0.1402,
-      "step": 3670
-    },
-    {
-      "epoch": 8.59,
-      "learning_rate": 7.78816199376947e-06,
-      "loss": 0.0258,
-      "step": 3680
-    },
-    {
-      "epoch": 8.61,
-      "learning_rate": 7.658359293873313e-06,
-      "loss": 0.0477,
-      "step": 3690
-    },
-    {
-      "epoch": 8.63,
-      "learning_rate": 7.528556593977156e-06,
-      "loss": 0.0255,
-      "step": 3700
-    },
-    {
-      "epoch": 8.66,
-      "learning_rate": 7.398753894080997e-06,
-      "loss": 0.0699,
-      "step": 3710
-    },
-    {
-      "epoch": 8.68,
-      "learning_rate": 7.2689511941848395e-06,
-      "loss": 0.0368,
-      "step": 3720
-    },
-    {
-      "epoch": 8.7,
-      "learning_rate": 7.139148494288682e-06,
-      "loss": 0.0344,
-      "step": 3730
-    },
-    {
-      "epoch": 8.73,
-      "learning_rate": 7.009345794392523e-06,
-      "loss": 0.1283,
-      "step": 3740
-    },
-    {
-      "epoch": 8.75,
-      "learning_rate": 6.879543094496365e-06,
-      "loss": 0.0407,
-      "step": 3750
-    },
-    {
-      "epoch": 8.77,
-      "learning_rate": 6.749740394600208e-06,
-      "loss": 0.0833,
-      "step": 3760
-    },
-    {
-      "epoch": 8.8,
-      "learning_rate": 6.619937694704049e-06,
-      "loss": 0.0392,
-      "step": 3770
-    },
-    {
-      "epoch": 8.82,
-      "learning_rate": 6.490134994807892e-06,
-      "loss": 0.0896,
-      "step": 3780
-    },
-    {
-      "epoch": 8.84,
-      "learning_rate": 6.360332294911735e-06,
-      "loss": 0.0401,
-      "step": 3790
-    },
-    {
-      "epoch": 8.87,
-      "learning_rate": 6.230529595015577e-06,
-      "loss": 0.0219,
-      "step": 3800
-    },
-    {
-      "epoch": 8.89,
-      "learning_rate": 6.100726895119419e-06,
-      "loss": 0.0416,
-      "step": 3810
-    },
-    {
-      "epoch": 8.91,
-      "learning_rate": 5.9709241952232614e-06,
-      "loss": 0.0987,
-      "step": 3820
-    },
-    {
-      "epoch": 8.94,
-      "learning_rate": 5.841121495327103e-06,
-      "loss": 0.0462,
-      "step": 3830
-    },
-    {
-      "epoch": 8.96,
-      "learning_rate": 5.711318795430945e-06,
-      "loss": 0.0524,
-      "step": 3840
-    },
-    {
-      "epoch": 8.98,
-      "learning_rate": 5.581516095534787e-06,
-      "loss": 0.0677,
-      "step": 3850
-    },
-    {
-      "epoch": 9.0,
-      "eval_accuracy": 0.9920562868815251,
-      "eval_loss": 0.040971919894218445,
-      "eval_runtime": 266.271,
-      "eval_samples_per_second": 16.547,
-      "eval_steps_per_second": 2.76,
-      "step": 3856
-    },
-    {
-      "epoch": 9.01,
-      "learning_rate": 5.451713395638629e-06,
-      "loss": 0.0256,
-      "step": 3860
-    },
-    {
-      "epoch": 9.03,
-      "learning_rate": 5.321910695742471e-06,
-      "loss": 0.0595,
-      "step": 3870
-    },
-    {
-      "epoch": 9.05,
-      "learning_rate": 5.192107995846314e-06,
-      "loss": 0.077,
-      "step": 3880
-    },
-    {
-      "epoch": 9.08,
-      "learning_rate": 5.062305295950156e-06,
-      "loss": 0.109,
-      "step": 3890
-    },
-    {
-      "epoch": 9.1,
-      "learning_rate": 4.932502596053999e-06,
-      "loss": 0.0975,
-      "step": 3900
-    },
-    {
-      "epoch": 9.12,
-      "learning_rate": 4.802699896157841e-06,
-      "loss": 0.0157,
-      "step": 3910
-    },
-    {
-      "epoch": 9.15,
-      "learning_rate": 4.6728971962616825e-06,
-      "loss": 0.0457,
-      "step": 3920
-    },
-    {
-      "epoch": 9.17,
-      "learning_rate": 4.5430944963655245e-06,
-      "loss": 0.0122,
-      "step": 3930
-    },
-    {
-      "epoch": 9.19,
-      "learning_rate": 4.413291796469366e-06,
-      "loss": 0.0826,
-      "step": 3940
-    },
-    {
-      "epoch": 9.22,
-      "learning_rate": 4.283489096573208e-06,
-      "loss": 0.0302,
-      "step": 3950
-    },
-    {
-      "epoch": 9.24,
-      "learning_rate": 4.153686396677051e-06,
-      "loss": 0.0578,
-      "step": 3960
-    },
-    {
-      "epoch": 9.26,
-      "learning_rate": 4.023883696780893e-06,
-      "loss": 0.0359,
-      "step": 3970
-    },
-    {
-      "epoch": 9.29,
-      "learning_rate": 3.894080996884735e-06,
-      "loss": 0.1037,
-      "step": 3980
-    },
-    {
-      "epoch": 9.31,
-      "learning_rate": 3.764278296988578e-06,
-      "loss": 0.1095,
-      "step": 3990
-    },
-    {
-      "epoch": 9.33,
-      "learning_rate": 3.6344755970924198e-06,
-      "loss": 0.0605,
-      "step": 4000
-    },
-    {
-      "epoch": 9.36,
-      "learning_rate": 3.5046728971962617e-06,
-      "loss": 0.0118,
-      "step": 4010
-    },
-    {
-      "epoch": 9.38,
-      "learning_rate": 3.374870197300104e-06,
-      "loss": 0.0246,
-      "step": 4020
-    },
-    {
-      "epoch": 9.4,
-      "learning_rate": 3.245067497403946e-06,
-      "loss": 0.0354,
-      "step": 4030
-    },
-    {
-      "epoch": 9.43,
-      "learning_rate": 3.1152647975077884e-06,
-      "loss": 0.0229,
-      "step": 4040
-    },
-    {
-      "epoch": 9.45,
-      "learning_rate": 2.9854620976116307e-06,
-      "loss": 0.0696,
-      "step": 4050
-    },
-    {
-      "epoch": 9.47,
-      "learning_rate": 2.8556593977154727e-06,
-      "loss": 0.0075,
-      "step": 4060
-    },
-    {
-      "epoch": 9.5,
-      "learning_rate": 2.7258566978193146e-06,
-      "loss": 0.0672,
-      "step": 4070
-    },
-    {
-      "epoch": 9.52,
-      "learning_rate": 2.596053997923157e-06,
-      "loss": 0.0709,
-      "step": 4080
-    },
-    {
-      "epoch": 9.54,
-      "learning_rate": 2.4662512980269993e-06,
-      "loss": 0.0191,
-      "step": 4090
-    },
-    {
-      "epoch": 9.57,
-      "learning_rate": 2.3364485981308413e-06,
-      "loss": 0.0473,
-      "step": 4100
-    },
-    {
-      "epoch": 9.59,
-      "learning_rate": 2.206645898234683e-06,
-      "loss": 0.036,
-      "step": 4110
-    },
-    {
-      "epoch": 9.61,
-      "learning_rate": 2.0768431983385256e-06,
-      "loss": 0.0877,
-      "step": 4120
-    },
-    {
-      "epoch": 9.64,
-      "learning_rate": 1.9470404984423675e-06,
-      "loss": 0.0537,
-      "step": 4130
-    },
-    {
-      "epoch": 9.66,
-      "learning_rate": 1.8172377985462099e-06,
-      "loss": 0.014,
-      "step": 4140
-    },
-    {
-      "epoch": 9.68,
-      "learning_rate": 1.687435098650052e-06,
-      "loss": 0.0342,
-      "step": 4150
-    },
-    {
-      "epoch": 9.71,
-      "learning_rate": 1.5576323987538942e-06,
-      "loss": 0.001,
-      "step": 4160
-    },
-    {
-      "epoch": 9.73,
-      "learning_rate": 1.4278296988577363e-06,
-      "loss": 0.0394,
-      "step": 4170
-    },
-    {
-      "epoch": 9.75,
-      "learning_rate": 1.2980269989615785e-06,
-      "loss": 0.0157,
-      "step": 4180
-    },
-    {
-      "epoch": 9.78,
-      "learning_rate": 1.1682242990654206e-06,
-      "loss": 0.0451,
-      "step": 4190
-    },
-    {
-      "epoch": 9.8,
-      "learning_rate": 1.0384215991692628e-06,
-      "loss": 0.0807,
-      "step": 4200
-    },
-    {
-      "epoch": 9.82,
-      "learning_rate": 9.086188992731049e-07,
-      "loss": 0.1119,
-      "step": 4210
-    },
-    {
-      "epoch": 9.85,
-      "learning_rate": 7.788161993769471e-07,
-      "loss": 0.029,
-      "step": 4220
-    },
-    {
-      "epoch": 9.87,
-      "learning_rate": 6.490134994807892e-07,
-      "loss": 0.0812,
-      "step": 4230
-    },
-    {
-      "epoch": 9.89,
-      "learning_rate": 5.192107995846314e-07,
-      "loss": 0.1648,
-      "step": 4240
-    },
-    {
-      "epoch": 9.92,
-      "learning_rate": 3.8940809968847354e-07,
-      "loss": 0.054,
-      "step": 4250
-    },
-    {
-      "epoch": 9.94,
-      "learning_rate": 2.596053997923157e-07,
-      "loss": 0.0953,
-      "step": 4260
-    },
-    {
-      "epoch": 9.96,
-      "learning_rate": 1.2980269989615785e-07,
-      "loss": 0.0724,
-      "step": 4270
-    },
-    {
-      "epoch": 9.99,
-      "learning_rate": 0.0,
-      "loss": 0.0315,
-      "step": 4280
-    },
-    {
-      "epoch": 9.99,
-      "eval_accuracy": 0.9906945074897866,
-      "eval_loss": 0.04262673109769821,
-      "eval_runtime": 279.4653,
-      "eval_samples_per_second": 15.766,
-      "eval_steps_per_second": 2.63,
-      "step": 4280
-    },
-    {
-      "epoch": 9.99,
-      "step": 4280,
-      "total_flos": 2.552597644040995e+18,
-      "train_loss": 0.12827744820725312,
-      "train_runtime": 15501.4351,
-      "train_samples_per_second": 6.632,
-      "train_steps_per_second": 0.276
-    }
-  ],
-  "logging_steps": 10,
-  "max_steps": 4280,
-  "num_train_epochs": 10,
-  "save_steps": 500,
-  "total_flos": 2.552597644040995e+18,
-  "trial_name": null,
-  "trial_params": null
-}
+{
+  "best_metric": 0.9918293236495688,
+  "best_model_checkpoint": "swin-tiny-patch4-window7-224-finetuned-eurosat\\checkpoint-3428",
+  "epoch": 9.988331388564761,
+  "eval_steps": 500,
+  "global_step": 4280,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1682242990654206e-06,
+      "loss": 0.1233,
+      "step": 10
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.3364485981308413e-06,
+      "loss": 0.1409,
+      "step": 20
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 0.1085,
+      "step": 30
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 0.1851,
+      "step": 40
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.841121495327103e-06,
+      "loss": 0.0843,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 0.1936,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 0.2033,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 0.1218,
+      "step": 80
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.0514018691588785e-05,
+      "loss": 0.2019,
+      "step": 90
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.1682242990654207e-05,
+      "loss": 0.1738,
+      "step": 100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.2850467289719625e-05,
+      "loss": 0.0796,
+      "step": 110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 0.072,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.5186915887850467e-05,
+      "loss": 0.167,
+      "step": 130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 0.1465,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.752336448598131e-05,
+      "loss": 0.0922,
+      "step": 150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.869158878504673e-05,
+      "loss": 0.1146,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.985981308411215e-05,
+      "loss": 0.099,
+      "step": 170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.102803738317757e-05,
+      "loss": 0.0943,
+      "step": 180
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.2196261682242992e-05,
+      "loss": 0.1045,
+      "step": 190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.3364485981308414e-05,
+      "loss": 0.1733,
+      "step": 200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.4532710280373832e-05,
+      "loss": 0.0712,
+      "step": 210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.570093457943925e-05,
+      "loss": 0.0344,
+      "step": 220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.6869158878504675e-05,
+      "loss": 0.0413,
+      "step": 230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 0.0555,
+      "step": 240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9205607476635515e-05,
+      "loss": 0.0897,
+      "step": 250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.0373831775700934e-05,
+      "loss": 0.0867,
+      "step": 260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1542056074766355e-05,
+      "loss": 0.0228,
+      "step": 270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 0.0299,
+      "step": 280
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.38785046728972e-05,
+      "loss": 0.0955,
+      "step": 290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.504672897196262e-05,
+      "loss": 0.1222,
+      "step": 300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.621495327102804e-05,
+      "loss": 0.0528,
+      "step": 310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 0.1051,
+      "step": 320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.855140186915888e-05,
+      "loss": 0.0238,
+      "step": 330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.97196261682243e-05,
+      "loss": 0.027,
+      "step": 340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.088785046728972e-05,
+      "loss": 0.1789,
+      "step": 350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 0.1676,
+      "step": 360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3224299065420565e-05,
+      "loss": 0.0539,
+      "step": 370
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.4392523364485984e-05,
+      "loss": 0.0618,
+      "step": 380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556074766355141e-05,
+      "loss": 0.0228,
+      "step": 390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 0.1006,
+      "step": 400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7897196261682245e-05,
+      "loss": 0.0511,
+      "step": 410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.9065420560747664e-05,
+      "loss": 0.0765,
+      "step": 420
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.9773036768043577,
+      "eval_loss": 0.11013483256101608,
+      "eval_runtime": 392.9035,
+      "eval_samples_per_second": 11.214,
+      "eval_steps_per_second": 1.871,
+      "step": 428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.9974039460020774e-05,
+      "loss": 0.0892,
+      "step": 430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 4.9844236760124614e-05,
+      "loss": 0.0984,
+      "step": 440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.971443406022845e-05,
+      "loss": 0.1396,
+      "step": 450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.958463136033229e-05,
+      "loss": 0.1533,
+      "step": 460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.9454828660436145e-05,
+      "loss": 0.1496,
+      "step": 470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.9325025960539985e-05,
+      "loss": 0.2945,
+      "step": 480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.9195223260643824e-05,
+      "loss": 0.1244,
+      "step": 490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.9065420560747664e-05,
+      "loss": 0.2377,
+      "step": 500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.893561786085151e-05,
+      "loss": 0.2172,
+      "step": 510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.880581516095535e-05,
+      "loss": 0.1937,
+      "step": 520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.8676012461059195e-05,
+      "loss": 0.1183,
+      "step": 530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 4.8546209761163035e-05,
+      "loss": 0.1339,
+      "step": 540
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.8416407061266874e-05,
+      "loss": 0.1859,
+      "step": 550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.828660436137072e-05,
+      "loss": 0.0697,
+      "step": 560
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.815680166147456e-05,
+      "loss": 0.1691,
+      "step": 570
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.80269989615784e-05,
+      "loss": 0.3344,
+      "step": 580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.7897196261682245e-05,
+      "loss": 0.169,
+      "step": 590
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.776739356178609e-05,
+      "loss": 0.1024,
+      "step": 600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.763759086188993e-05,
+      "loss": 0.2665,
+      "step": 610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.750778816199377e-05,
+      "loss": 0.1175,
+      "step": 620
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.737798546209761e-05,
+      "loss": 0.1271,
+      "step": 630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 4.7248182762201456e-05,
+      "loss": 0.1421,
+      "step": 640
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.71183800623053e-05,
+      "loss": 0.165,
+      "step": 650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.698857736240914e-05,
+      "loss": 0.2327,
+      "step": 660
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.685877466251298e-05,
+      "loss": 0.1064,
+      "step": 670
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.672897196261683e-05,
+      "loss": 0.0849,
+      "step": 680
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.6599169262720666e-05,
+      "loss": 0.1302,
+      "step": 690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.6469366562824506e-05,
+      "loss": 0.091,
+      "step": 700
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.633956386292835e-05,
+      "loss": 0.0988,
+      "step": 710
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.620976116303219e-05,
+      "loss": 0.1034,
+      "step": 720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.607995846313604e-05,
+      "loss": 0.2778,
+      "step": 730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.595015576323988e-05,
+      "loss": 0.1861,
+      "step": 740
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.5820353063343716e-05,
+      "loss": 0.1253,
+      "step": 750
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.569055036344756e-05,
+      "loss": 0.1729,
+      "step": 760
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.556074766355141e-05,
+      "loss": 0.2417,
+      "step": 770
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.543094496365525e-05,
+      "loss": 0.0998,
+      "step": 780
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.530114226375909e-05,
+      "loss": 0.1067,
+      "step": 790
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.517133956386293e-05,
+      "loss": 0.0742,
+      "step": 800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.504153686396677e-05,
+      "loss": 0.1658,
+      "step": 810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.491173416407061e-05,
+      "loss": 0.0984,
+      "step": 820
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.478193146417446e-05,
+      "loss": 0.1219,
+      "step": 830
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.46521287642783e-05,
+      "loss": 0.0771,
+      "step": 840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.4522326064382144e-05,
+      "loss": 0.1014,
+      "step": 850
+    },
+    {
+      "epoch": 2.0,
+      "eval_accuracy": 0.9825238311393554,
+      "eval_loss": 0.06920906156301498,
+      "eval_runtime": 261.1543,
+      "eval_samples_per_second": 16.871,
+      "eval_steps_per_second": 2.814,
+      "step": 857
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 4.4392523364485984e-05,
+      "loss": 0.1961,
+      "step": 860
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.426272066458982e-05,
+      "loss": 0.0966,
+      "step": 870
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.413291796469366e-05,
+      "loss": 0.1267,
+      "step": 880
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.400311526479751e-05,
+      "loss": 0.0499,
+      "step": 890
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3873312564901355e-05,
+      "loss": 0.0841,
+      "step": 900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.3743509865005194e-05,
+      "loss": 0.1192,
+      "step": 910
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 4.3613707165109034e-05,
+      "loss": 0.0639,
+      "step": 920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 4.348390446521288e-05,
+      "loss": 0.1042,
+      "step": 930
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 4.335410176531672e-05,
+      "loss": 0.0612,
+      "step": 940
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 4.3224299065420565e-05,
+      "loss": 0.1319,
+      "step": 950
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 4.3094496365524405e-05,
+      "loss": 0.077,
+      "step": 960
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 4.2964693665628244e-05,
+      "loss": 0.2134,
+      "step": 970
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 4.283489096573209e-05,
+      "loss": 0.1452,
+      "step": 980
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 4.270508826583593e-05,
+      "loss": 0.1919,
+      "step": 990
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 4.257528556593977e-05,
+      "loss": 0.1121,
+      "step": 1000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 4.2445482866043615e-05,
+      "loss": 0.1547,
+      "step": 1010
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 4.231568016614746e-05,
+      "loss": 0.1347,
+      "step": 1020
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 4.21858774662513e-05,
+      "loss": 0.1598,
+      "step": 1030
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 4.205607476635514e-05,
+      "loss": 0.1122,
+      "step": 1040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 4.192627206645898e-05,
+      "loss": 0.1312,
+      "step": 1050
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 4.1796469366562826e-05,
+      "loss": 0.0814,
+      "step": 1060
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.0596,
+      "step": 1070
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 4.153686396677051e-05,
+      "loss": 0.0764,
+      "step": 1080
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 4.140706126687435e-05,
+      "loss": 0.1605,
+      "step": 1090
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 4.12772585669782e-05,
+      "loss": 0.1239,
+      "step": 1100
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.1147455867082036e-05,
+      "loss": 0.0592,
+      "step": 1110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.101765316718588e-05,
+      "loss": 0.2212,
+      "step": 1120
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.088785046728972e-05,
+      "loss": 0.0901,
+      "step": 1130
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.075804776739356e-05,
+      "loss": 0.1151,
+      "step": 1140
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.062824506749741e-05,
+      "loss": 0.076,
+      "step": 1150
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.049844236760125e-05,
+      "loss": 0.1206,
+      "step": 1160
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0368639667705086e-05,
+      "loss": 0.1452,
+      "step": 1170
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.023883696780893e-05,
+      "loss": 0.1499,
+      "step": 1180
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.010903426791278e-05,
+      "loss": 0.0686,
+      "step": 1190
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.997923156801662e-05,
+      "loss": 0.1197,
+      "step": 1200
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.984942886812046e-05,
+      "loss": 0.1705,
+      "step": 1210
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.97196261682243e-05,
+      "loss": 0.1328,
+      "step": 1220
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.958982346832814e-05,
+      "loss": 0.0919,
+      "step": 1230
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.946002076843199e-05,
+      "loss": 0.0554,
+      "step": 1240
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.933021806853583e-05,
+      "loss": 0.0425,
+      "step": 1250
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.920041536863967e-05,
+      "loss": 0.0751,
+      "step": 1260
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.9070612668743514e-05,
+      "loss": 0.1737,
+      "step": 1270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.8940809968847354e-05,
+      "loss": 0.0425,
+      "step": 1280
+    },
+    {
+      "epoch": 3.0,
+      "eval_accuracy": 0.9813890149795733,
+      "eval_loss": 0.07660907506942749,
+      "eval_runtime": 282.9189,
+      "eval_samples_per_second": 15.573,
+      "eval_steps_per_second": 2.598,
+      "step": 1285
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.881100726895119e-05,
+      "loss": 0.2164,
+      "step": 1290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.868120456905504e-05,
+      "loss": 0.1103,
+      "step": 1300
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.855140186915888e-05,
+      "loss": 0.1125,
+      "step": 1310
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.8421599169262725e-05,
+      "loss": 0.1068,
+      "step": 1320
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.8291796469366564e-05,
+      "loss": 0.1056,
+      "step": 1330
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.8161993769470404e-05,
+      "loss": 0.1497,
+      "step": 1340
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.803219106957425e-05,
+      "loss": 0.0829,
+      "step": 1350
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.7902388369678096e-05,
+      "loss": 0.1006,
+      "step": 1360
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.7772585669781935e-05,
+      "loss": 0.1013,
+      "step": 1370
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.7642782969885775e-05,
+      "loss": 0.0448,
+      "step": 1380
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.7512980269989614e-05,
+      "loss": 0.1184,
+      "step": 1390
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.738317757009346e-05,
+      "loss": 0.0498,
+      "step": 1400
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.72533748701973e-05,
+      "loss": 0.0694,
+      "step": 1410
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.7123572170301146e-05,
+      "loss": 0.1008,
+      "step": 1420
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.6993769470404985e-05,
+      "loss": 0.1204,
+      "step": 1430
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.686396677050883e-05,
+      "loss": 0.0562,
+      "step": 1440
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.673416407061267e-05,
+      "loss": 0.0419,
+      "step": 1450
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.660436137071651e-05,
+      "loss": 0.1405,
+      "step": 1460
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.647455867082035e-05,
+      "loss": 0.1039,
+      "step": 1470
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.6344755970924196e-05,
+      "loss": 0.1325,
+      "step": 1480
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.621495327102804e-05,
+      "loss": 0.0748,
+      "step": 1490
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.608515057113188e-05,
+      "loss": 0.0853,
+      "step": 1500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.595534787123572e-05,
+      "loss": 0.1364,
+      "step": 1510
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.582554517133957e-05,
+      "loss": 0.1138,
+      "step": 1520
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.5695742471443406e-05,
+      "loss": 0.0747,
+      "step": 1530
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.556593977154725e-05,
+      "loss": 0.1053,
+      "step": 1540
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.543613707165109e-05,
+      "loss": 0.0531,
+      "step": 1550
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.530633437175493e-05,
+      "loss": 0.0956,
+      "step": 1560
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.517653167185878e-05,
+      "loss": 0.166,
+      "step": 1570
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.504672897196262e-05,
+      "loss": 0.1193,
+      "step": 1580
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.4916926272066456e-05,
+      "loss": 0.1299,
+      "step": 1590
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.47871235721703e-05,
+      "loss": 0.0678,
+      "step": 1600
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.465732087227415e-05,
+      "loss": 0.1396,
+      "step": 1610
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.452751817237799e-05,
+      "loss": 0.035,
+      "step": 1620
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.439771547248183e-05,
+      "loss": 0.0584,
+      "step": 1630
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.426791277258567e-05,
+      "loss": 0.1252,
+      "step": 1640
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.413811007268951e-05,
+      "loss": 0.0645,
+      "step": 1650
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.400830737279336e-05,
+      "loss": 0.0999,
+      "step": 1660
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.38785046728972e-05,
+      "loss": 0.0915,
+      "step": 1670
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.374870197300104e-05,
+      "loss": 0.0922,
+      "step": 1680
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.3618899273104884e-05,
+      "loss": 0.122,
+      "step": 1690
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.3489096573208724e-05,
+      "loss": 0.1602,
+      "step": 1700
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.335929387331256e-05,
+      "loss": 0.1229,
+      "step": 1710
+    },
+    {
+      "epoch": 4.0,
+      "eval_accuracy": 0.9872900590104403,
+      "eval_loss": 0.0515071265399456,
+      "eval_runtime": 314.2248,
+      "eval_samples_per_second": 14.022,
+      "eval_steps_per_second": 2.339,
+      "step": 1714
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.322949117341641e-05,
+      "loss": 0.0587,
+      "step": 1720
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.309968847352025e-05,
+      "loss": 0.0739,
+      "step": 1730
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.2969885773624095e-05,
+      "loss": 0.0994,
+      "step": 1740
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.2840083073727934e-05,
+      "loss": 0.0877,
+      "step": 1750
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.2710280373831774e-05,
+      "loss": 0.0859,
+      "step": 1760
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.258047767393562e-05,
+      "loss": 0.0813,
+      "step": 1770
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.2450674974039466e-05,
+      "loss": 0.1231,
+      "step": 1780
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2320872274143305e-05,
+      "loss": 0.0774,
+      "step": 1790
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2191069574247145e-05,
+      "loss": 0.1181,
+      "step": 1800
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.2061266874350984e-05,
+      "loss": 0.0231,
+      "step": 1810
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.193146417445483e-05,
+      "loss": 0.0641,
+      "step": 1820
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 3.1801661474558676e-05,
+      "loss": 0.1327,
+      "step": 1830
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 3.1671858774662516e-05,
+      "loss": 0.0296,
+      "step": 1840
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 3.1542056074766355e-05,
+      "loss": 0.1249,
+      "step": 1850
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 3.14122533748702e-05,
+      "loss": 0.1184,
+      "step": 1860
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 3.128245067497404e-05,
+      "loss": 0.0827,
+      "step": 1870
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.115264797507788e-05,
+      "loss": 0.0275,
+      "step": 1880
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.1022845275181726e-05,
+      "loss": 0.149,
+      "step": 1890
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.0893042575285566e-05,
+      "loss": 0.1072,
+      "step": 1900
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.076323987538941e-05,
+      "loss": 0.0576,
+      "step": 1910
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 3.063343717549325e-05,
+      "loss": 0.1569,
+      "step": 1920
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 3.0503634475597094e-05,
+      "loss": 0.0939,
+      "step": 1930
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 3.0373831775700934e-05,
+      "loss": 0.1736,
+      "step": 1940
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 3.024402907580478e-05,
+      "loss": 0.0525,
+      "step": 1950
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.0114226375908622e-05,
+      "loss": 0.0718,
+      "step": 1960
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.9984423676012462e-05,
+      "loss": 0.0781,
+      "step": 1970
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9854620976116305e-05,
+      "loss": 0.0699,
+      "step": 1980
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.9724818276220144e-05,
+      "loss": 0.1131,
+      "step": 1990
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.9595015576323987e-05,
+      "loss": 0.0737,
+      "step": 2000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.9465212876427833e-05,
+      "loss": 0.1204,
+      "step": 2010
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.9335410176531676e-05,
+      "loss": 0.0636,
+      "step": 2020
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.9205607476635515e-05,
+      "loss": 0.058,
+      "step": 2030
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.9075804776739358e-05,
+      "loss": 0.0497,
+      "step": 2040
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.8946002076843197e-05,
+      "loss": 0.1449,
+      "step": 2050
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.881619937694704e-05,
+      "loss": 0.0601,
+      "step": 2060
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.8686396677050886e-05,
+      "loss": 0.1041,
+      "step": 2070
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.855659397715473e-05,
+      "loss": 0.1027,
+      "step": 2080
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.842679127725857e-05,
+      "loss": 0.0834,
+      "step": 2090
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.829698857736241e-05,
+      "loss": 0.0744,
+      "step": 2100
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 2.816718587746625e-05,
+      "loss": 0.0192,
+      "step": 2110
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.8037383177570094e-05,
+      "loss": 0.0566,
+      "step": 2120
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.790758047767394e-05,
+      "loss": 0.1073,
+      "step": 2130
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.074,
+      "step": 2140
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.9891057648660917,
+      "eval_loss": 0.04968353733420372,
+      "eval_runtime": 408.1815,
+      "eval_samples_per_second": 10.794,
+      "eval_steps_per_second": 1.801,
+      "step": 2142
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.7647975077881622e-05,
+      "loss": 0.1103,
+      "step": 2150
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.751817237798546e-05,
+      "loss": 0.0873,
+      "step": 2160
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.7388369678089304e-05,
+      "loss": 0.0941,
+      "step": 2170
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.7258566978193147e-05,
+      "loss": 0.0281,
+      "step": 2180
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.7128764278296993e-05,
+      "loss": 0.0588,
+      "step": 2190
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.6998961578400832e-05,
+      "loss": 0.071,
+      "step": 2200
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.6869158878504675e-05,
+      "loss": 0.0778,
+      "step": 2210
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.6739356178608515e-05,
+      "loss": 0.0789,
+      "step": 2220
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.6609553478712357e-05,
+      "loss": 0.0311,
+      "step": 2230
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.6479750778816197e-05,
+      "loss": 0.113,
+      "step": 2240
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.6349948078920046e-05,
+      "loss": 0.0819,
+      "step": 2250
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.6220145379023886e-05,
+      "loss": 0.0734,
+      "step": 2260
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.609034267912773e-05,
+      "loss": 0.0557,
+      "step": 2270
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.5960539979231568e-05,
+      "loss": 0.0982,
+      "step": 2280
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.583073727933541e-05,
+      "loss": 0.0671,
+      "step": 2290
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.570093457943925e-05,
+      "loss": 0.1331,
+      "step": 2300
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.5571131879543096e-05,
+      "loss": 0.0783,
+      "step": 2310
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.544132917964694e-05,
+      "loss": 0.0404,
+      "step": 2320
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.531152647975078e-05,
+      "loss": 0.0957,
+      "step": 2330
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.518172377985462e-05,
+      "loss": 0.0711,
+      "step": 2340
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.5051921079958464e-05,
+      "loss": 0.0977,
+      "step": 2350
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.4922118380062307e-05,
+      "loss": 0.045,
+      "step": 2360
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.4792315680166146e-05,
+      "loss": 0.0734,
+      "step": 2370
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.4662512980269992e-05,
+      "loss": 0.0506,
+      "step": 2380
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.4532710280373832e-05,
+      "loss": 0.0977,
+      "step": 2390
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.4402907580477675e-05,
+      "loss": 0.05,
+      "step": 2400
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.4273104880581517e-05,
+      "loss": 0.1653,
+      "step": 2410
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.414330218068536e-05,
+      "loss": 0.0688,
+      "step": 2420
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.40134994807892e-05,
+      "loss": 0.0867,
+      "step": 2430
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.3883696780893046e-05,
+      "loss": 0.0591,
+      "step": 2440
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.3753894080996885e-05,
+      "loss": 0.1166,
+      "step": 2450
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.3624091381100728e-05,
+      "loss": 0.0306,
+      "step": 2460
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.349428868120457e-05,
+      "loss": 0.1629,
+      "step": 2470
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.3364485981308414e-05,
+      "loss": 0.064,
+      "step": 2480
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.3234683281412253e-05,
+      "loss": 0.0581,
+      "step": 2490
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.3104880581516096e-05,
+      "loss": 0.102,
+      "step": 2500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.297507788161994e-05,
+      "loss": 0.1491,
+      "step": 2510
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.284527518172378e-05,
+      "loss": 0.0416,
+      "step": 2520
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.2715472481827624e-05,
+      "loss": 0.0276,
+      "step": 2530
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.2585669781931463e-05,
+      "loss": 0.0627,
+      "step": 2540
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.2455867082035306e-05,
+      "loss": 0.0869,
+      "step": 2550
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.232606438213915e-05,
+      "loss": 0.0408,
+      "step": 2560
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 2.2196261682242992e-05,
+      "loss": 0.0133,
+      "step": 2570
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.988197911938266,
+      "eval_loss": 0.053730662912130356,
+      "eval_runtime": 389.9644,
+      "eval_samples_per_second": 11.298,
+      "eval_steps_per_second": 1.885,
+      "step": 2571
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 2.206645898234683e-05,
+      "loss": 0.2029,
+      "step": 2580
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 2.1936656282450677e-05,
+      "loss": 0.0382,
+      "step": 2590
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 2.1806853582554517e-05,
+      "loss": 0.0434,
+      "step": 2600
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 2.167705088265836e-05,
+      "loss": 0.0436,
+      "step": 2610
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 2.1547248182762202e-05,
+      "loss": 0.0284,
+      "step": 2620
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 2.1417445482866045e-05,
+      "loss": 0.0556,
+      "step": 2630
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 2.1287642782969885e-05,
+      "loss": 0.0995,
+      "step": 2640
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 2.115784008307373e-05,
+      "loss": 0.0908,
+      "step": 2650
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 2.102803738317757e-05,
+      "loss": 0.0693,
+      "step": 2660
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 2.0898234683281413e-05,
+      "loss": 0.0138,
+      "step": 2670
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 2.0768431983385256e-05,
+      "loss": 0.0134,
+      "step": 2680
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 2.06386292834891e-05,
+      "loss": 0.0152,
+      "step": 2690
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 2.050882658359294e-05,
+      "loss": 0.11,
+      "step": 2700
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 2.037902388369678e-05,
+      "loss": 0.0621,
+      "step": 2710
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 2.0249221183800623e-05,
+      "loss": 0.1248,
+      "step": 2720
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 2.0119418483904466e-05,
+      "loss": 0.0746,
+      "step": 2730
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.998961578400831e-05,
+      "loss": 0.0401,
+      "step": 2740
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.985981308411215e-05,
+      "loss": 0.0364,
+      "step": 2750
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.9730010384215995e-05,
+      "loss": 0.0772,
+      "step": 2760
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.9600207684319834e-05,
+      "loss": 0.0328,
+      "step": 2770
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.9470404984423677e-05,
+      "loss": 0.0694,
+      "step": 2780
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.934060228452752e-05,
+      "loss": 0.048,
+      "step": 2790
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.9210799584631362e-05,
+      "loss": 0.0638,
+      "step": 2800
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.9080996884735202e-05,
+      "loss": 0.0099,
+      "step": 2810
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.8951194184839048e-05,
+      "loss": 0.0255,
+      "step": 2820
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.8821391484942887e-05,
+      "loss": 0.1108,
+      "step": 2830
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.869158878504673e-05,
+      "loss": 0.1095,
+      "step": 2840
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.8561786085150573e-05,
+      "loss": 0.1304,
+      "step": 2850
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.8431983385254416e-05,
+      "loss": 0.0378,
+      "step": 2860
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.8302180685358255e-05,
+      "loss": 0.04,
+      "step": 2870
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.8172377985462098e-05,
+      "loss": 0.0503,
+      "step": 2880
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.804257528556594e-05,
+      "loss": 0.103,
+      "step": 2890
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.7912772585669783e-05,
+      "loss": 0.1043,
+      "step": 2900
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.7782969885773626e-05,
+      "loss": 0.0712,
+      "step": 2910
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.7653167185877466e-05,
+      "loss": 0.0595,
+      "step": 2920
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.752336448598131e-05,
+      "loss": 0.0302,
+      "step": 2930
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.739356178608515e-05,
+      "loss": 0.0642,
+      "step": 2940
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.7263759086188994e-05,
+      "loss": 0.0957,
+      "step": 2950
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.7133956386292833e-05,
+      "loss": 0.0988,
+      "step": 2960
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.700415368639668e-05,
+      "loss": 0.0764,
+      "step": 2970
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.687435098650052e-05,
+      "loss": 0.097,
+      "step": 2980
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 1.6744548286604362e-05,
+      "loss": 0.0753,
+      "step": 2990
+    },
+    {
+      "epoch": 7.0,
+      "eval_accuracy": 0.9911484339536994,
+      "eval_loss": 0.049025967717170715,
+      "eval_runtime": 402.4957,
+      "eval_samples_per_second": 10.947,
+      "eval_steps_per_second": 1.826,
+      "step": 2999
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.6614745586708205e-05,
+      "loss": 0.0981,
+      "step": 3000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.6484942886812047e-05,
+      "loss": 0.0299,
+      "step": 3010
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.6355140186915887e-05,
+      "loss": 0.031,
+      "step": 3020
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.6225337487019733e-05,
+      "loss": 0.0933,
+      "step": 3030
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.6095534787123572e-05,
+      "loss": 0.0405,
+      "step": 3040
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.5965732087227415e-05,
+      "loss": 0.0784,
+      "step": 3050
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.5835929387331258e-05,
+      "loss": 0.0771,
+      "step": 3060
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.57061266874351e-05,
+      "loss": 0.03,
+      "step": 3070
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.557632398753894e-05,
+      "loss": 0.1243,
+      "step": 3080
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 1.5446521287642783e-05,
+      "loss": 0.0118,
+      "step": 3090
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.5316718587746626e-05,
+      "loss": 0.0662,
+      "step": 3100
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.5186915887850467e-05,
+      "loss": 0.0657,
+      "step": 3110
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.5057113187954311e-05,
+      "loss": 0.1301,
+      "step": 3120
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.4927310488058152e-05,
+      "loss": 0.0506,
+      "step": 3130
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.4797507788161993e-05,
+      "loss": 0.0587,
+      "step": 3140
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.4667705088265838e-05,
+      "loss": 0.078,
+      "step": 3150
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.4537902388369679e-05,
+      "loss": 0.0459,
+      "step": 3160
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.440809968847352e-05,
+      "loss": 0.045,
+      "step": 3170
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.4278296988577365e-05,
+      "loss": 0.0641,
+      "step": 3180
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 1.4148494288681206e-05,
+      "loss": 0.068,
+      "step": 3190
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.4018691588785047e-05,
+      "loss": 0.1033,
+      "step": 3200
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.0339,
+      "step": 3210
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.375908618899273e-05,
+      "loss": 0.0675,
+      "step": 3220
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.3629283489096573e-05,
+      "loss": 0.0228,
+      "step": 3230
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.3499480789200416e-05,
+      "loss": 0.0888,
+      "step": 3240
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.3369678089304257e-05,
+      "loss": 0.0919,
+      "step": 3250
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.3239875389408098e-05,
+      "loss": 0.0292,
+      "step": 3260
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.3110072689511943e-05,
+      "loss": 0.0752,
+      "step": 3270
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.2980269989615784e-05,
+      "loss": 0.0644,
+      "step": 3280
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.2850467289719625e-05,
+      "loss": 0.1123,
+      "step": 3290
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.272066458982347e-05,
+      "loss": 0.0674,
+      "step": 3300
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.259086188992731e-05,
+      "loss": 0.0344,
+      "step": 3310
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.2461059190031153e-05,
+      "loss": 0.0164,
+      "step": 3320
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.2331256490134996e-05,
+      "loss": 0.1099,
+      "step": 3330
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.2201453790238837e-05,
+      "loss": 0.0204,
+      "step": 3340
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.207165109034268e-05,
+      "loss": 0.0398,
+      "step": 3350
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.1941848390446523e-05,
+      "loss": 0.0686,
+      "step": 3360
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.1812045690550364e-05,
+      "loss": 0.0572,
+      "step": 3370
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 1.1682242990654207e-05,
+      "loss": 0.0552,
+      "step": 3380
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.1552440290758048e-05,
+      "loss": 0.0434,
+      "step": 3390
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.142263759086189e-05,
+      "loss": 0.0799,
+      "step": 3400
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.1292834890965732e-05,
+      "loss": 0.0582,
+      "step": 3410
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.1163032191069575e-05,
+      "loss": 0.0263,
+      "step": 3420
+    },
+    {
+      "epoch": 8.0,
+      "eval_accuracy": 0.9918293236495688,
+      "eval_loss": 0.0519937239587307,
+      "eval_runtime": 392.283,
+      "eval_samples_per_second": 11.232,
+      "eval_steps_per_second": 1.874,
+      "step": 3428
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.1033229491173416e-05,
+      "loss": 0.0371,
+      "step": 3430
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.0903426791277258e-05,
+      "loss": 0.1521,
+      "step": 3440
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.0773624091381101e-05,
+      "loss": 0.0398,
+      "step": 3450
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.0643821391484942e-05,
+      "loss": 0.0326,
+      "step": 3460
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.0514018691588785e-05,
+      "loss": 0.03,
+      "step": 3470
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.0384215991692628e-05,
+      "loss": 0.0934,
+      "step": 3480
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.025441329179647e-05,
+      "loss": 0.0456,
+      "step": 3490
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.0124610591900312e-05,
+      "loss": 0.1047,
+      "step": 3500
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.994807892004155e-06,
+      "loss": 0.058,
+      "step": 3510
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 9.865005192107997e-06,
+      "loss": 0.0345,
+      "step": 3520
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 9.735202492211838e-06,
+      "loss": 0.0547,
+      "step": 3530
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 9.605399792315681e-06,
+      "loss": 0.1301,
+      "step": 3540
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 9.475597092419524e-06,
+      "loss": 0.0654,
+      "step": 3550
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 9.345794392523365e-06,
+      "loss": 0.0524,
+      "step": 3560
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 9.215991692627208e-06,
+      "loss": 0.0579,
+      "step": 3570
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 9.086188992731049e-06,
+      "loss": 0.0635,
+      "step": 3580
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.956386292834892e-06,
+      "loss": 0.0678,
+      "step": 3590
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.826583592938733e-06,
+      "loss": 0.031,
+      "step": 3600
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 8.696780893042576e-06,
+      "loss": 0.0493,
+      "step": 3610
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 8.566978193146417e-06,
+      "loss": 0.0744,
+      "step": 3620
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 8.43717549325026e-06,
+      "loss": 0.0538,
+      "step": 3630
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 8.307372793354102e-06,
+      "loss": 0.0246,
+      "step": 3640
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 8.177570093457943e-06,
+      "loss": 0.0271,
+      "step": 3650
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 8.047767393561786e-06,
+      "loss": 0.0354,
+      "step": 3660
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.917964693665629e-06,
+      "loss": 0.0219,
+      "step": 3670
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.78816199376947e-06,
+      "loss": 0.0562,
+      "step": 3680
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 7.658359293873313e-06,
+      "loss": 0.0772,
+      "step": 3690
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 7.528556593977156e-06,
+      "loss": 0.1255,
+      "step": 3700
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 7.398753894080997e-06,
+      "loss": 0.0412,
+      "step": 3710
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 7.2689511941848395e-06,
+      "loss": 0.038,
+      "step": 3720
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 7.139148494288682e-06,
+      "loss": 0.0596,
+      "step": 3730
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 7.009345794392523e-06,
+      "loss": 0.0663,
+      "step": 3740
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.879543094496365e-06,
+      "loss": 0.0741,
+      "step": 3750
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.749740394600208e-06,
+      "loss": 0.0647,
+      "step": 3760
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 6.619937694704049e-06,
+      "loss": 0.0405,
+      "step": 3770
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 6.490134994807892e-06,
+      "loss": 0.0905,
+      "step": 3780
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.360332294911735e-06,
+      "loss": 0.0511,
+      "step": 3790
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.230529595015577e-06,
+      "loss": 0.0226,
+      "step": 3800
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.100726895119419e-06,
+      "loss": 0.0089,
+      "step": 3810
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.9709241952232614e-06,
+      "loss": 0.0592,
+      "step": 3820
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.841121495327103e-06,
+      "loss": 0.0733,
+      "step": 3830
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.711318795430945e-06,
+      "loss": 0.0604,
+      "step": 3840
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.581516095534787e-06,
+      "loss": 0.0423,
+      "step": 3850
+    },
+    {
+      "epoch": 9.0,
+      "eval_accuracy": 0.9913753971856559,
+      "eval_loss": 0.051343705505132675,
+      "eval_runtime": 313.2302,
+      "eval_samples_per_second": 14.066,
+      "eval_steps_per_second": 2.347,
+      "step": 3856
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.451713395638629e-06,
+      "loss": 0.0815,
+      "step": 3860
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 5.321910695742471e-06,
+      "loss": 0.0789,
+      "step": 3870
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 5.192107995846314e-06,
+      "loss": 0.1404,
+      "step": 3880
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 5.062305295950156e-06,
+      "loss": 0.0572,
+      "step": 3890
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.932502596053999e-06,
+      "loss": 0.0346,
+      "step": 3900
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.802699896157841e-06,
+      "loss": 0.0093,
+      "step": 3910
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.6728971962616825e-06,
+      "loss": 0.0447,
+      "step": 3920
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.5430944963655245e-06,
+      "loss": 0.0183,
+      "step": 3930
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.413291796469366e-06,
+      "loss": 0.0511,
+      "step": 3940
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 4.283489096573208e-06,
+      "loss": 0.0682,
+      "step": 3950
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 4.153686396677051e-06,
+      "loss": 0.0748,
+      "step": 3960
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 4.023883696780893e-06,
+      "loss": 0.0349,
+      "step": 3970
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.894080996884735e-06,
+      "loss": 0.0826,
+      "step": 3980
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.764278296988578e-06,
+      "loss": 0.0436,
+      "step": 3990
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.6344755970924198e-06,
+      "loss": 0.0563,
+      "step": 4000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.5046728971962617e-06,
+      "loss": 0.0369,
+      "step": 4010
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.374870197300104e-06,
+      "loss": 0.027,
+      "step": 4020
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 3.245067497403946e-06,
+      "loss": 0.098,
+      "step": 4030
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 3.1152647975077884e-06,
+      "loss": 0.0427,
+      "step": 4040
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.9854620976116307e-06,
+      "loss": 0.042,
+      "step": 4050
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.8556593977154727e-06,
+      "loss": 0.0318,
+      "step": 4060
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 2.7258566978193146e-06,
+      "loss": 0.1009,
+      "step": 4070
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.596053997923157e-06,
+      "loss": 0.047,
+      "step": 4080
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.4662512980269993e-06,
+      "loss": 0.0384,
+      "step": 4090
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.3364485981308413e-06,
+      "loss": 0.0518,
+      "step": 4100
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.206645898234683e-06,
+      "loss": 0.0184,
+      "step": 4110
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 2.0768431983385256e-06,
+      "loss": 0.0824,
+      "step": 4120
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.9470404984423675e-06,
+      "loss": 0.0695,
+      "step": 4130
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.8172377985462099e-06,
+      "loss": 0.011,
+      "step": 4140
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.687435098650052e-06,
+      "loss": 0.0303,
+      "step": 4150
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.5576323987538942e-06,
+      "loss": 0.0551,
+      "step": 4160
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 1.4278296988577363e-06,
+      "loss": 0.0799,
+      "step": 4170
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.2980269989615785e-06,
+      "loss": 0.066,
+      "step": 4180
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1682242990654206e-06,
+      "loss": 0.0906,
+      "step": 4190
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0384215991692628e-06,
+      "loss": 0.0583,
+      "step": 4200
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.086188992731049e-07,
+      "loss": 0.026,
+      "step": 4210
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.788161993769471e-07,
+      "loss": 0.0498,
+      "step": 4220
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.490134994807892e-07,
+      "loss": 0.0197,
+      "step": 4230
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.192107995846314e-07,
+      "loss": 0.0549,
+      "step": 4240
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.8940809968847354e-07,
+      "loss": 0.0782,
+      "step": 4250
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.596053997923157e-07,
+      "loss": 0.0889,
+      "step": 4260
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 1.2980269989615785e-07,
+      "loss": 0.0209,
+      "step": 4270
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 0.0,
+      "loss": 0.0266,
+      "step": 4280
+    },
+    {
+      "epoch": 9.99,
+      "eval_accuracy": 0.9916023604176123,
+      "eval_loss": 0.04850451648235321,
+      "eval_runtime": 280.7261,
+      "eval_samples_per_second": 15.695,
+      "eval_steps_per_second": 2.618,
+      "step": 4280
+    },
+    {
+      "epoch": 9.99,
+      "step": 4280,
+      "total_flos": 2.552597644040995e+18,
+      "train_loss": 0.08735134438306928,
+      "train_runtime": 17846.498,
+      "train_samples_per_second": 5.76,
+      "train_steps_per_second": 0.24
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 4280,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 2.552597644040995e+18,
+  "trial_name": null,
+  "trial_params": null
+}