diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7043 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 38.61003861003861,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.03861003861003861,
+      "grad_norm": 13.287198066711426,
+      "learning_rate": 1.8e-06,
+      "loss": 1.38,
+      "step": 10
+    },
+    {
+      "epoch": 0.07722007722007722,
+      "grad_norm": 6.655487537384033,
+      "learning_rate": 3.8e-06,
+      "loss": 0.9573,
+      "step": 20
+    },
+    {
+      "epoch": 0.11583011583011583,
+      "grad_norm": 4.606926441192627,
+      "learning_rate": 5.8e-06,
+      "loss": 0.6618,
+      "step": 30
+    },
+    {
+      "epoch": 0.15444015444015444,
+      "grad_norm": 3.794658660888672,
+      "learning_rate": 7.8e-06,
+      "loss": 0.4893,
+      "step": 40
+    },
+    {
+      "epoch": 0.19305019305019305,
+      "grad_norm": 3.480128049850464,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 0.3991,
+      "step": 50
+    },
+    {
+      "epoch": 0.23166023166023167,
+      "grad_norm": 2.990638494491577,
+      "learning_rate": 1.18e-05,
+      "loss": 0.3731,
+      "step": 60
+    },
+    {
+      "epoch": 0.2702702702702703,
+      "grad_norm": 3.4375221729278564,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.3468,
+      "step": 70
+    },
+    {
+      "epoch": 0.3088803088803089,
+      "grad_norm": 3.048978328704834,
+      "learning_rate": 1.58e-05,
+      "loss": 0.299,
+      "step": 80
+    },
+    {
+      "epoch": 0.3474903474903475,
+      "grad_norm": 2.6503806114196777,
+      "learning_rate": 1.78e-05,
+      "loss": 0.29,
+      "step": 90
+    },
+    {
+      "epoch": 0.3861003861003861,
+      "grad_norm": 2.6138811111450195,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.2716,
+      "step": 100
+    },
+    {
+      "epoch": 0.4247104247104247,
+      "grad_norm": 2.6181092262268066,
+      "learning_rate": 2.18e-05,
+      "loss": 0.2841,
+      "step": 110
+    },
+    {
+      "epoch": 0.46332046332046334,
+      "grad_norm": 2.2942309379577637,
+      "learning_rate": 2.38e-05,
+      "loss": 0.2886,
+      "step": 120
+    },
+    {
+      "epoch": 0.5019305019305019,
+      "grad_norm": 2.857469081878662,
+      "learning_rate": 2.58e-05,
+      "loss": 0.2244,
+      "step": 130
+    },
+    {
+      "epoch": 0.5405405405405406,
+      "grad_norm": 2.050870418548584,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.249,
+      "step": 140
+    },
+    {
+      "epoch": 0.5791505791505791,
+      "grad_norm": 2.1127007007598877,
+      "learning_rate": 2.98e-05,
+      "loss": 0.242,
+      "step": 150
+    },
+    {
+      "epoch": 0.6177606177606177,
+      "grad_norm": 2.074040651321411,
+      "learning_rate": 3.18e-05,
+      "loss": 0.1953,
+      "step": 160
+    },
+    {
+      "epoch": 0.6563706563706564,
+      "grad_norm": 1.7310792207717896,
+      "learning_rate": 3.38e-05,
+      "loss": 0.2109,
+      "step": 170
+    },
+    {
+      "epoch": 0.694980694980695,
+      "grad_norm": 1.7166472673416138,
+      "learning_rate": 3.58e-05,
+      "loss": 0.1997,
+      "step": 180
+    },
+    {
+      "epoch": 0.7335907335907336,
+      "grad_norm": 2.083514451980591,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.1976,
+      "step": 190
+    },
+    {
+      "epoch": 0.7722007722007722,
+      "grad_norm": 2.140868663787842,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.1892,
+      "step": 200
+    },
+    {
+      "epoch": 0.8108108108108109,
+      "grad_norm": 1.3370658159255981,
+      "learning_rate": 4.18e-05,
+      "loss": 0.1806,
+      "step": 210
+    },
+    {
+      "epoch": 0.8494208494208494,
+      "grad_norm": 1.2409409284591675,
+      "learning_rate": 4.38e-05,
+      "loss": 0.2032,
+      "step": 220
+    },
+    {
+      "epoch": 0.888030888030888,
+      "grad_norm": 1.60843026638031,
+      "learning_rate": 4.58e-05,
+      "loss": 0.2125,
+      "step": 230
+    },
+    {
+      "epoch": 0.9266409266409267,
+      "grad_norm": 1.6353750228881836,
+      "learning_rate": 4.78e-05,
+      "loss": 0.1991,
+      "step": 240
+    },
+    {
+      "epoch": 0.9652509652509652,
+      "grad_norm": 1.9422987699508667,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.1939,
+      "step": 250
+    },
+    {
+      "epoch": 1.0038610038610039,
+      "grad_norm": 2.1851253509521484,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.1796,
+      "step": 260
+    },
+    {
+      "epoch": 1.0424710424710424,
+      "grad_norm": 1.8957761526107788,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 0.1836,
+      "step": 270
+    },
+    {
+      "epoch": 1.0810810810810811,
+      "grad_norm": 1.7820072174072266,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.1887,
+      "step": 280
+    },
+    {
+      "epoch": 1.1196911196911197,
+      "grad_norm": 1.6855354309082031,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.2198,
+      "step": 290
+    },
+    {
+      "epoch": 1.1583011583011582,
+      "grad_norm": 1.629774570465088,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.2079,
+      "step": 300
+    },
+    {
+      "epoch": 1.196911196911197,
+      "grad_norm": 1.340558648109436,
+      "learning_rate": 6.18e-05,
+      "loss": 0.2119,
+      "step": 310
+    },
+    {
+      "epoch": 1.2355212355212355,
+      "grad_norm": 0.9606438875198364,
+      "learning_rate": 6.38e-05,
+      "loss": 0.1559,
+      "step": 320
+    },
+    {
+      "epoch": 1.2741312741312742,
+      "grad_norm": 1.6456347703933716,
+      "learning_rate": 6.58e-05,
+      "loss": 0.2021,
+      "step": 330
+    },
+    {
+      "epoch": 1.3127413127413128,
+      "grad_norm": 1.6377339363098145,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.1979,
+      "step": 340
+    },
+    {
+      "epoch": 1.3513513513513513,
+      "grad_norm": 1.6483289003372192,
+      "learning_rate": 6.98e-05,
+      "loss": 0.1935,
+      "step": 350
+    },
+    {
+      "epoch": 1.3899613899613898,
+      "grad_norm": 1.1614036560058594,
+      "learning_rate": 7.18e-05,
+      "loss": 0.1608,
+      "step": 360
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 1.685562014579773,
+      "learning_rate": 7.38e-05,
+      "loss": 0.1895,
+      "step": 370
+    },
+    {
+      "epoch": 1.4671814671814671,
+      "grad_norm": 1.646178960800171,
+      "learning_rate": 7.58e-05,
+      "loss": 0.195,
+      "step": 380
+    },
+    {
+      "epoch": 1.505791505791506,
+      "grad_norm": 1.4592519998550415,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.2059,
+      "step": 390
+    },
+    {
+      "epoch": 1.5444015444015444,
+      "grad_norm": 0.9366406798362732,
+      "learning_rate": 7.98e-05,
+      "loss": 0.2017,
+      "step": 400
+    },
+    {
+      "epoch": 1.583011583011583,
+      "grad_norm": 0.8088350296020508,
+      "learning_rate": 8.18e-05,
+      "loss": 0.1956,
+      "step": 410
+    },
+    {
+      "epoch": 1.6216216216216215,
+      "grad_norm": 1.1527583599090576,
+      "learning_rate": 8.38e-05,
+      "loss": 0.1754,
+      "step": 420
+    },
+    {
+      "epoch": 1.6602316602316602,
+      "grad_norm": 1.142561435699463,
+      "learning_rate": 8.58e-05,
+      "loss": 0.2244,
+      "step": 430
+    },
+    {
+      "epoch": 1.698841698841699,
+      "grad_norm": 1.2559890747070312,
+      "learning_rate": 8.78e-05,
+      "loss": 0.1964,
+      "step": 440
+    },
+    {
+      "epoch": 1.7374517374517375,
+      "grad_norm": 1.1182104349136353,
+      "learning_rate": 8.98e-05,
+      "loss": 0.1857,
+      "step": 450
+    },
+    {
+      "epoch": 1.776061776061776,
+      "grad_norm": 1.0533323287963867,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.1765,
+      "step": 460
+    },
+    {
+      "epoch": 1.8146718146718146,
+      "grad_norm": 0.9969793558120728,
+      "learning_rate": 9.38e-05,
+      "loss": 0.1737,
+      "step": 470
+    },
+    {
+      "epoch": 1.8532818532818531,
+      "grad_norm": 1.2537906169891357,
+      "learning_rate": 9.58e-05,
+      "loss": 0.1935,
+      "step": 480
+    },
+    {
+      "epoch": 1.8918918918918919,
+      "grad_norm": 0.8507728576660156,
+      "learning_rate": 9.78e-05,
+      "loss": 0.1926,
+      "step": 490
+    },
+    {
+      "epoch": 1.9305019305019306,
+      "grad_norm": 1.64980947971344,
+      "learning_rate": 9.98e-05,
+      "loss": 0.1639,
+      "step": 500
+    },
+    {
+      "epoch": 1.9691119691119692,
+      "grad_norm": 1.5939923524856567,
+      "learning_rate": 9.9999778549206e-05,
+      "loss": 0.1925,
+      "step": 510
+    },
+    {
+      "epoch": 2.0077220077220077,
+      "grad_norm": 1.505319595336914,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.1872,
+      "step": 520
+    },
+    {
+      "epoch": 2.0463320463320462,
+      "grad_norm": 1.1174558401107788,
+      "learning_rate": 9.999770075521164e-05,
+      "loss": 0.2034,
+      "step": 530
+    },
+    {
+      "epoch": 2.0849420849420848,
+      "grad_norm": 0.7865064144134521,
+      "learning_rate": 9.99958417007713e-05,
+      "loss": 0.1539,
+      "step": 540
+    },
+    {
+      "epoch": 2.1235521235521237,
+      "grad_norm": 1.0896923542022705,
+      "learning_rate": 9.999343589981615e-05,
+      "loss": 0.2019,
+      "step": 550
+    },
+    {
+      "epoch": 2.1621621621621623,
+      "grad_norm": 0.8688647747039795,
+      "learning_rate": 9.999048337865568e-05,
+      "loss": 0.2035,
+      "step": 560
+    },
+    {
+      "epoch": 2.200772200772201,
+      "grad_norm": 1.3430683612823486,
+      "learning_rate": 9.998698416957815e-05,
+      "loss": 0.1787,
+      "step": 570
+    },
+    {
+      "epoch": 2.2393822393822393,
+      "grad_norm": 1.0575079917907715,
+      "learning_rate": 9.998293831085037e-05,
+      "loss": 0.1488,
+      "step": 580
+    },
+    {
+      "epoch": 2.277992277992278,
+      "grad_norm": 1.3032593727111816,
+      "learning_rate": 9.997834584671719e-05,
+      "loss": 0.1664,
+      "step": 590
+    },
+    {
+      "epoch": 2.3166023166023164,
+      "grad_norm": 1.1514618396759033,
+      "learning_rate": 9.997320682740107e-05,
+      "loss": 0.1885,
+      "step": 600
+    },
+    {
+      "epoch": 2.3552123552123554,
+      "grad_norm": 0.8944717049598694,
+      "learning_rate": 9.996752130910149e-05,
+      "loss": 0.1736,
+      "step": 610
+    },
+    {
+      "epoch": 2.393822393822394,
+      "grad_norm": 1.0690749883651733,
+      "learning_rate": 9.99612893539944e-05,
+      "loss": 0.1568,
+      "step": 620
+    },
+    {
+      "epoch": 2.4324324324324325,
+      "grad_norm": 0.8698034286499023,
+      "learning_rate": 9.995451103023144e-05,
+      "loss": 0.1628,
+      "step": 630
+    },
+    {
+      "epoch": 2.471042471042471,
+      "grad_norm": 0.7732095718383789,
+      "learning_rate": 9.994718641193928e-05,
+      "loss": 0.1761,
+      "step": 640
+    },
+    {
+      "epoch": 2.5096525096525095,
+      "grad_norm": 0.8875847458839417,
+      "learning_rate": 9.993931557921874e-05,
+      "loss": 0.1659,
+      "step": 650
+    },
+    {
+      "epoch": 2.5482625482625485,
+      "grad_norm": 0.719978392124176,
+      "learning_rate": 9.993089861814402e-05,
+      "loss": 0.1846,
+      "step": 660
+    },
+    {
+      "epoch": 2.586872586872587,
+      "grad_norm": 1.2403405904769897,
+      "learning_rate": 9.992193562076166e-05,
+      "loss": 0.1613,
+      "step": 670
+    },
+    {
+      "epoch": 2.6254826254826256,
+      "grad_norm": 1.445119857788086,
+      "learning_rate": 9.991242668508954e-05,
+      "loss": 0.1797,
+      "step": 680
+    },
+    {
+      "epoch": 2.664092664092664,
+      "grad_norm": 0.6605463624000549,
+      "learning_rate": 9.990237191511587e-05,
+      "loss": 0.1607,
+      "step": 690
+    },
+    {
+      "epoch": 2.7027027027027026,
+      "grad_norm": 0.8701773285865784,
+      "learning_rate": 9.989177142079802e-05,
+      "loss": 0.17,
+      "step": 700
+    },
+    {
+      "epoch": 2.741312741312741,
+      "grad_norm": 0.7130756974220276,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.1433,
+      "step": 710
+    },
+    {
+      "epoch": 2.7799227799227797,
+      "grad_norm": 0.9633455872535706,
+      "learning_rate": 9.986893372879762e-05,
+      "loss": 0.1565,
+      "step": 720
+    },
+    {
+      "epoch": 2.8185328185328187,
+      "grad_norm": 1.4280503988265991,
+      "learning_rate": 9.985669678086443e-05,
+      "loss": 0.1426,
+      "step": 730
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 1.0674346685409546,
+      "learning_rate": 9.984391460808298e-05,
+      "loss": 0.1429,
+      "step": 740
+    },
+    {
+      "epoch": 2.8957528957528957,
+      "grad_norm": 0.8652824759483337,
+      "learning_rate": 9.983058735023709e-05,
+      "loss": 0.1796,
+      "step": 750
+    },
+    {
+      "epoch": 2.9343629343629343,
+      "grad_norm": 1.0494927167892456,
+      "learning_rate": 9.98167151530715e-05,
+      "loss": 0.1449,
+      "step": 760
+    },
+    {
+      "epoch": 2.972972972972973,
+      "grad_norm": 1.1388959884643555,
+      "learning_rate": 9.980229816829034e-05,
+      "loss": 0.1535,
+      "step": 770
+    },
+    {
+      "epoch": 3.011583011583012,
+      "grad_norm": 1.15719473361969,
+      "learning_rate": 9.978733655355544e-05,
+      "loss": 0.1746,
+      "step": 780
+    },
+    {
+      "epoch": 3.0501930501930503,
+      "grad_norm": 0.8897600769996643,
+      "learning_rate": 9.977183047248464e-05,
+      "loss": 0.1499,
+      "step": 790
+    },
+    {
+      "epoch": 3.088803088803089,
+      "grad_norm": 0.865238606929779,
+      "learning_rate": 9.975578009464992e-05,
+      "loss": 0.1429,
+      "step": 800
+    },
+    {
+      "epoch": 3.1274131274131274,
+      "grad_norm": 0.8140860795974731,
+      "learning_rate": 9.97391855955757e-05,
+      "loss": 0.1563,
+      "step": 810
+    },
+    {
+      "epoch": 3.166023166023166,
+      "grad_norm": 0.7682332396507263,
+      "learning_rate": 9.972204715673669e-05,
+      "loss": 0.1493,
+      "step": 820
+    },
+    {
+      "epoch": 3.2046332046332044,
+      "grad_norm": 0.9081019759178162,
+      "learning_rate": 9.970436496555617e-05,
+      "loss": 0.1887,
+      "step": 830
+    },
+    {
+      "epoch": 3.2432432432432434,
+      "grad_norm": 1.0015044212341309,
+      "learning_rate": 9.968613921540373e-05,
+      "loss": 0.1574,
+      "step": 840
+    },
+    {
+      "epoch": 3.281853281853282,
+      "grad_norm": 1.160292625427246,
+      "learning_rate": 9.966737010559326e-05,
+      "loss": 0.1714,
+      "step": 850
+    },
+    {
+      "epoch": 3.3204633204633205,
+      "grad_norm": 0.9939375519752502,
+      "learning_rate": 9.964805784138072e-05,
+      "loss": 0.1456,
+      "step": 860
+    },
+    {
+      "epoch": 3.359073359073359,
+      "grad_norm": 0.7976135015487671,
+      "learning_rate": 9.962820263396195e-05,
+      "loss": 0.1891,
+      "step": 870
+    },
+    {
+      "epoch": 3.3976833976833976,
+      "grad_norm": 1.28683602809906,
+      "learning_rate": 9.960780470047033e-05,
+      "loss": 0.1496,
+      "step": 880
+    },
+    {
+      "epoch": 3.436293436293436,
+      "grad_norm": 0.7454524636268616,
+      "learning_rate": 9.958686426397437e-05,
+      "loss": 0.1568,
+      "step": 890
+    },
+    {
+      "epoch": 3.474903474903475,
+      "grad_norm": 1.0883091688156128,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.1612,
+      "step": 900
+    },
+    {
+      "epoch": 3.5135135135135136,
+      "grad_norm": 1.1127383708953857,
+      "learning_rate": 9.95433568039047e-05,
+      "loss": 0.1541,
+      "step": 910
+    },
+    {
+      "epoch": 3.552123552123552,
+      "grad_norm": 0.8523925542831421,
+      "learning_rate": 9.952079025612162e-05,
+      "loss": 0.157,
+      "step": 920
+    },
+    {
+      "epoch": 3.5907335907335907,
+      "grad_norm": 0.7145108580589294,
+      "learning_rate": 9.949768215691022e-05,
+      "loss": 0.1639,
+      "step": 930
+    },
+    {
+      "epoch": 3.629343629343629,
+      "grad_norm": 0.9005628824234009,
+      "learning_rate": 9.9474032758977e-05,
+      "loss": 0.16,
+      "step": 940
+    },
+    {
+      "epoch": 3.667953667953668,
+      "grad_norm": 1.1173086166381836,
+      "learning_rate": 9.944984232094794e-05,
+      "loss": 0.1246,
+      "step": 950
+    },
+    {
+      "epoch": 3.7065637065637067,
+      "grad_norm": 1.0558993816375732,
+      "learning_rate": 9.942511110736584e-05,
+      "loss": 0.1411,
+      "step": 960
+    },
+    {
+      "epoch": 3.7451737451737452,
+      "grad_norm": 1.1991870403289795,
+      "learning_rate": 9.939983938868726e-05,
+      "loss": 0.1791,
+      "step": 970
+    },
+    {
+      "epoch": 3.7837837837837838,
+      "grad_norm": 0.6233035922050476,
+      "learning_rate": 9.93740274412797e-05,
+      "loss": 0.1197,
+      "step": 980
+    },
+    {
+      "epoch": 3.8223938223938223,
+      "grad_norm": 0.9723224639892578,
+      "learning_rate": 9.934767554741846e-05,
+      "loss": 0.1134,
+      "step": 990
+    },
+    {
+      "epoch": 3.861003861003861,
+      "grad_norm": 0.6947616934776306,
+      "learning_rate": 9.932078399528361e-05,
+      "loss": 0.1405,
+      "step": 1000
+    },
+    {
+      "epoch": 3.8996138996138994,
+      "grad_norm": 0.5903553366661072,
+      "learning_rate": 9.929335307895689e-05,
+      "loss": 0.1422,
+      "step": 1010
+    },
+    {
+      "epoch": 3.9382239382239383,
+      "grad_norm": 0.5389763116836548,
+      "learning_rate": 9.926538309841839e-05,
+      "loss": 0.1426,
+      "step": 1020
+    },
+    {
+      "epoch": 3.976833976833977,
+      "grad_norm": 0.8586219549179077,
+      "learning_rate": 9.923687435954334e-05,
+      "loss": 0.1351,
+      "step": 1030
+    },
+    {
+      "epoch": 4.015444015444015,
+      "grad_norm": 0.6858639121055603,
+      "learning_rate": 9.920782717409873e-05,
+      "loss": 0.1168,
+      "step": 1040
+    },
+    {
+      "epoch": 4.054054054054054,
+      "grad_norm": 0.9002066254615784,
+      "learning_rate": 9.917824185973994e-05,
+      "loss": 0.1275,
+      "step": 1050
+    },
+    {
+      "epoch": 4.0926640926640925,
+      "grad_norm": 0.6978407502174377,
+      "learning_rate": 9.914811874000723e-05,
+      "loss": 0.1505,
+      "step": 1060
+    },
+    {
+      "epoch": 4.1312741312741315,
+      "grad_norm": 1.1149928569793701,
+      "learning_rate": 9.911745814432218e-05,
+      "loss": 0.163,
+      "step": 1070
+    },
+    {
+      "epoch": 4.1698841698841695,
+      "grad_norm": 0.642062246799469,
+      "learning_rate": 9.90862604079842e-05,
+      "loss": 0.1669,
+      "step": 1080
+    },
+    {
+      "epoch": 4.2084942084942085,
+      "grad_norm": 0.5878561735153198,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.1483,
+      "step": 1090
+    },
+    {
+      "epoch": 4.2471042471042475,
+      "grad_norm": 0.707226037979126,
+      "learning_rate": 9.90222548839135e-05,
+      "loss": 0.1175,
+      "step": 1100
+    },
+    {
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.6299799084663391,
+      "learning_rate": 9.898944779613495e-05,
+      "loss": 0.1541,
+      "step": 1110
+    },
+    {
+      "epoch": 4.324324324324325,
+      "grad_norm": 0.594017505645752,
+      "learning_rate": 9.89561049676041e-05,
+      "loss": 0.1224,
+      "step": 1120
+    },
+    {
+      "epoch": 4.362934362934363,
+      "grad_norm": 0.9055441617965698,
+      "learning_rate": 9.89222267629528e-05,
+      "loss": 0.1474,
+      "step": 1130
+    },
+    {
+      "epoch": 4.401544401544402,
+      "grad_norm": 1.1802901029586792,
+      "learning_rate": 9.888781355266763e-05,
+      "loss": 0.1697,
+      "step": 1140
+    },
+    {
+      "epoch": 4.440154440154441,
+      "grad_norm": 0.6890931725502014,
+      "learning_rate": 9.885286571308598e-05,
+      "loss": 0.148,
+      "step": 1150
+    },
+    {
+      "epoch": 4.478764478764479,
+      "grad_norm": 0.5606589317321777,
+      "learning_rate": 9.881738362639182e-05,
+      "loss": 0.1344,
+      "step": 1160
+    },
+    {
+      "epoch": 4.517374517374518,
+      "grad_norm": 0.6417577266693115,
+      "learning_rate": 9.878136768061154e-05,
+      "loss": 0.1375,
+      "step": 1170
+    },
+    {
+      "epoch": 4.555984555984556,
+      "grad_norm": 0.8422744870185852,
+      "learning_rate": 9.874481826960979e-05,
+      "loss": 0.1222,
+      "step": 1180
+    },
+    {
+      "epoch": 4.594594594594595,
+      "grad_norm": 0.7227635979652405,
+      "learning_rate": 9.870773579308503e-05,
+      "loss": 0.1407,
+      "step": 1190
+    },
+    {
+      "epoch": 4.633204633204633,
+      "grad_norm": 0.9805036187171936,
+      "learning_rate": 9.867012065656533e-05,
+      "loss": 0.1473,
+      "step": 1200
+    },
+    {
+      "epoch": 4.671814671814672,
+      "grad_norm": 0.6718726754188538,
+      "learning_rate": 9.863197327140376e-05,
+      "loss": 0.1456,
+      "step": 1210
+    },
+    {
+      "epoch": 4.710424710424711,
+      "grad_norm": 0.9643238186836243,
+      "learning_rate": 9.859329405477403e-05,
+      "loss": 0.1352,
+      "step": 1220
+    },
+    {
+      "epoch": 4.749034749034749,
+      "grad_norm": 0.9292412400245667,
+      "learning_rate": 9.855408342966585e-05,
+      "loss": 0.1409,
+      "step": 1230
+    },
+    {
+      "epoch": 4.787644787644788,
+      "grad_norm": 0.7480415105819702,
+      "learning_rate": 9.851434182488033e-05,
+      "loss": 0.1378,
+      "step": 1240
+    },
+    {
+      "epoch": 4.826254826254826,
+      "grad_norm": 0.938568115234375,
+      "learning_rate": 9.84740696750253e-05,
+      "loss": 0.1175,
+      "step": 1250
+    },
+    {
+      "epoch": 4.864864864864865,
+      "grad_norm": 0.6812225580215454,
+      "learning_rate": 9.843326742051055e-05,
+      "loss": 0.1351,
+      "step": 1260
+    },
+    {
+      "epoch": 4.903474903474903,
+      "grad_norm": 0.5791270732879639,
+      "learning_rate": 9.839193550754297e-05,
+      "loss": 0.1182,
+      "step": 1270
+    },
+    {
+      "epoch": 4.942084942084942,
+      "grad_norm": 0.7875615954399109,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.137,
+      "step": 1280
+    },
+    {
+      "epoch": 4.980694980694981,
+      "grad_norm": 0.49673154950141907,
+      "learning_rate": 9.830768452003341e-05,
+      "loss": 0.1387,
+      "step": 1290
+    },
+    {
+      "epoch": 5.019305019305019,
+      "grad_norm": 0.701742947101593,
+      "learning_rate": 9.826476636684671e-05,
+      "loss": 0.1337,
+      "step": 1300
+    },
+    {
+      "epoch": 5.057915057915058,
+      "grad_norm": 0.5022175312042236,
+      "learning_rate": 9.822132039790773e-05,
+      "loss": 0.1316,
+      "step": 1310
+    },
+    {
+      "epoch": 5.096525096525096,
+      "grad_norm": 0.7052822113037109,
+      "learning_rate": 9.817734708833461e-05,
+      "loss": 0.1272,
+      "step": 1320
+    },
+    {
+      "epoch": 5.135135135135135,
+      "grad_norm": 0.9381476640701294,
+      "learning_rate": 9.813284691901243e-05,
+      "loss": 0.1201,
+      "step": 1330
+    },
+    {
+      "epoch": 5.173745173745174,
+      "grad_norm": 0.7570599317550659,
+      "learning_rate": 9.808782037658792e-05,
+      "loss": 0.1149,
+      "step": 1340
+    },
+    {
+      "epoch": 5.212355212355212,
+      "grad_norm": 0.7424160242080688,
+      "learning_rate": 9.804226795346411e-05,
+      "loss": 0.1311,
+      "step": 1350
+    },
+    {
+      "epoch": 5.250965250965251,
+      "grad_norm": 0.7185351252555847,
+      "learning_rate": 9.799619014779503e-05,
+      "loss": 0.1431,
+      "step": 1360
+    },
+    {
+      "epoch": 5.289575289575289,
+      "grad_norm": 0.8532226085662842,
+      "learning_rate": 9.794958746348013e-05,
+      "loss": 0.1392,
+      "step": 1370
+    },
+    {
+      "epoch": 5.328185328185328,
+      "grad_norm": 0.6534046530723572,
+      "learning_rate": 9.790246041015896e-05,
+      "loss": 0.123,
+      "step": 1380
+    },
+    {
+      "epoch": 5.366795366795367,
+      "grad_norm": 0.5626727938652039,
+      "learning_rate": 9.785480950320538e-05,
+      "loss": 0.1367,
+      "step": 1390
+    },
+    {
+      "epoch": 5.405405405405405,
+      "grad_norm": 0.8905888795852661,
+      "learning_rate": 9.78066352637221e-05,
+      "loss": 0.122,
+      "step": 1400
+    },
+    {
+      "epoch": 5.444015444015444,
+      "grad_norm": 0.818387508392334,
+      "learning_rate": 9.775793821853488e-05,
+      "loss": 0.1366,
+      "step": 1410
+    },
+    {
+      "epoch": 5.482625482625482,
+      "grad_norm": 0.9816891551017761,
+      "learning_rate": 9.77087189001868e-05,
+      "loss": 0.1332,
+      "step": 1420
+    },
+    {
+      "epoch": 5.521235521235521,
+      "grad_norm": 0.46126094460487366,
+      "learning_rate": 9.765897784693243e-05,
+      "loss": 0.1001,
+      "step": 1430
+    },
+    {
+      "epoch": 5.559845559845559,
+      "grad_norm": 0.6799705028533936,
+      "learning_rate": 9.760871560273197e-05,
+      "loss": 0.1407,
+      "step": 1440
+    },
+    {
+      "epoch": 5.598455598455598,
+      "grad_norm": 0.7840109467506409,
+      "learning_rate": 9.755793271724526e-05,
+      "loss": 0.1494,
+      "step": 1450
+    },
+    {
+      "epoch": 5.637065637065637,
+      "grad_norm": 0.7920464277267456,
+      "learning_rate": 9.750662974582584e-05,
+      "loss": 0.1332,
+      "step": 1460
+    },
+    {
+      "epoch": 5.675675675675675,
+      "grad_norm": 0.5265321135520935,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.1317,
+      "step": 1470
+    },
+    {
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.679107666015625,
+      "learning_rate": 9.740246579503447e-05,
+      "loss": 0.1062,
+      "step": 1480
+    },
+    {
+      "epoch": 5.752895752895753,
+      "grad_norm": 0.6273481845855713,
+      "learning_rate": 9.734960595478284e-05,
+      "loss": 0.1289,
+      "step": 1490
+    },
+    {
+      "epoch": 5.7915057915057915,
+      "grad_norm": 0.691649317741394,
+      "learning_rate": 9.729622830682657e-05,
+      "loss": 0.1186,
+      "step": 1500
+    },
+    {
+      "epoch": 5.8301158301158305,
+      "grad_norm": 0.8244628310203552,
+      "learning_rate": 9.724233343489504e-05,
+      "loss": 0.1225,
+      "step": 1510
+    },
+    {
+      "epoch": 5.8687258687258685,
+      "grad_norm": 1.0755887031555176,
+      "learning_rate": 9.718792192837396e-05,
+      "loss": 0.1317,
+      "step": 1520
+    },
+    {
+      "epoch": 5.9073359073359075,
+      "grad_norm": 0.8774343132972717,
+      "learning_rate": 9.713299438229886e-05,
+      "loss": 0.1312,
+      "step": 1530
+    },
+    {
+      "epoch": 5.945945945945946,
+      "grad_norm": 0.606974184513092,
+      "learning_rate": 9.707755139734855e-05,
+      "loss": 0.1145,
+      "step": 1540
+    },
+    {
+      "epoch": 5.984555984555985,
+      "grad_norm": 0.5077939033508301,
+      "learning_rate": 9.702159357983866e-05,
+      "loss": 0.1241,
+      "step": 1550
+    },
+    {
+      "epoch": 6.023166023166024,
+      "grad_norm": 0.5576596856117249,
+      "learning_rate": 9.696512154171492e-05,
+      "loss": 0.1247,
+      "step": 1560
+    },
+    {
+      "epoch": 6.061776061776062,
+      "grad_norm": 1.0629464387893677,
+      "learning_rate": 9.690813590054645e-05,
+      "loss": 0.1075,
+      "step": 1570
+    },
+    {
+      "epoch": 6.100386100386101,
+      "grad_norm": 0.5680471062660217,
+      "learning_rate": 9.685063727951914e-05,
+      "loss": 0.1225,
+      "step": 1580
+    },
+    {
+      "epoch": 6.138996138996139,
+      "grad_norm": 0.6717116832733154,
+      "learning_rate": 9.679262630742865e-05,
+      "loss": 0.129,
+      "step": 1590
+    },
+    {
+      "epoch": 6.177606177606178,
+      "grad_norm": 0.7363089919090271,
+      "learning_rate": 9.673410361867373e-05,
+      "loss": 0.1436,
+      "step": 1600
+    },
+    {
+      "epoch": 6.216216216216216,
+      "grad_norm": 0.670937716960907,
+      "learning_rate": 9.667506985324909e-05,
+      "loss": 0.1029,
+      "step": 1610
+    },
+    {
+      "epoch": 6.254826254826255,
+      "grad_norm": 0.6786972284317017,
+      "learning_rate": 9.661552565673855e-05,
+      "loss": 0.1215,
+      "step": 1620
+    },
+    {
+      "epoch": 6.293436293436294,
+      "grad_norm": 0.5797168612480164,
+      "learning_rate": 9.655547168030789e-05,
+      "loss": 0.1499,
+      "step": 1630
+    },
+    {
+      "epoch": 6.332046332046332,
+      "grad_norm": 1.157994031906128,
+      "learning_rate": 9.649490858069777e-05,
+      "loss": 0.1376,
+      "step": 1640
+    },
+    {
+      "epoch": 6.370656370656371,
+      "grad_norm": 0.7316443920135498,
+      "learning_rate": 9.643383702021658e-05,
+      "loss": 0.1226,
+      "step": 1650
+    },
+    {
+      "epoch": 6.409266409266409,
+      "grad_norm": 0.5234395265579224,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.1194,
+      "step": 1660
+    },
+    {
+      "epoch": 6.447876447876448,
+      "grad_norm": 0.852341890335083,
+      "learning_rate": 9.631017119366922e-05,
+      "loss": 0.1053,
+      "step": 1670
+    },
+    {
+      "epoch": 6.486486486486487,
+      "grad_norm": 0.8800656795501709,
+      "learning_rate": 9.624757827999273e-05,
+      "loss": 0.114,
+      "step": 1680
+    },
+    {
+      "epoch": 6.525096525096525,
+      "grad_norm": 0.8578097820281982,
+      "learning_rate": 9.618447961020971e-05,
+      "loss": 0.1141,
+      "step": 1690
+    },
+    {
+      "epoch": 6.563706563706564,
+      "grad_norm": 0.8600365519523621,
+      "learning_rate": 9.612087587435707e-05,
+      "loss": 0.1128,
+      "step": 1700
+    },
+    {
+      "epoch": 6.602316602316602,
+      "grad_norm": 0.6659354567527771,
+      "learning_rate": 9.605676776799508e-05,
+      "loss": 0.1185,
+      "step": 1710
+    },
+    {
+      "epoch": 6.640926640926641,
+      "grad_norm": 0.7638758420944214,
+      "learning_rate": 9.599215599219973e-05,
+      "loss": 0.1429,
+      "step": 1720
+    },
+    {
+      "epoch": 6.67953667953668,
+      "grad_norm": 0.6346604824066162,
+      "learning_rate": 9.592704125355505e-05,
+      "loss": 0.1079,
+      "step": 1730
+    },
+    {
+      "epoch": 6.718146718146718,
+      "grad_norm": 0.7118030190467834,
+      "learning_rate": 9.586142426414538e-05,
+      "loss": 0.1174,
+      "step": 1740
+    },
+    {
+      "epoch": 6.756756756756757,
+      "grad_norm": 0.6951404213905334,
+      "learning_rate": 9.57953057415476e-05,
+      "loss": 0.1141,
+      "step": 1750
+    },
+    {
+      "epoch": 6.795366795366795,
+      "grad_norm": 0.8285521864891052,
+      "learning_rate": 9.572868640882328e-05,
+      "loss": 0.1168,
+      "step": 1760
+    },
+    {
+      "epoch": 6.833976833976834,
+      "grad_norm": 0.7002620697021484,
+      "learning_rate": 9.56615669945108e-05,
+      "loss": 0.109,
+      "step": 1770
+    },
+    {
+      "epoch": 6.872586872586872,
+      "grad_norm": 0.5306683778762817,
+      "learning_rate": 9.55939482326173e-05,
+      "loss": 0.1324,
+      "step": 1780
+    },
+    {
+      "epoch": 6.911196911196911,
+      "grad_norm": 0.7958775162696838,
+      "learning_rate": 9.552583086261069e-05,
+      "loss": 0.1098,
+      "step": 1790
+    },
+    {
+      "epoch": 6.94980694980695,
+      "grad_norm": 0.5308017730712891,
+      "learning_rate": 9.545721562941168e-05,
+      "loss": 0.1135,
+      "step": 1800
+    },
+    {
+      "epoch": 6.988416988416988,
+      "grad_norm": 0.5937675833702087,
+      "learning_rate": 9.538810328338543e-05,
+      "loss": 0.1073,
+      "step": 1810
+    },
+    {
+      "epoch": 7.027027027027027,
+      "grad_norm": 0.72931307554245,
+      "learning_rate": 9.531849458033349e-05,
+      "loss": 0.1119,
+      "step": 1820
+    },
+    {
+      "epoch": 7.065637065637065,
+      "grad_norm": 0.6811597347259521,
+      "learning_rate": 9.524839028148547e-05,
+      "loss": 0.1127,
+      "step": 1830
+    },
+    {
+      "epoch": 7.104247104247104,
+      "grad_norm": 0.6665561199188232,
+      "learning_rate": 9.517779115349077e-05,
+      "loss": 0.1136,
+      "step": 1840
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.6536176800727844,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.1464,
+      "step": 1850
+    },
+    {
+      "epoch": 7.181467181467181,
+      "grad_norm": 0.8365440964698792,
+      "learning_rate": 9.503511150370727e-05,
+      "loss": 0.1455,
+      "step": 1860
+    },
+    {
+      "epoch": 7.22007722007722,
+      "grad_norm": 0.6462043523788452,
+      "learning_rate": 9.496303254224024e-05,
+      "loss": 0.0999,
+      "step": 1870
+    },
+    {
+      "epoch": 7.258687258687258,
+      "grad_norm": 0.558756411075592,
+      "learning_rate": 9.489046187225306e-05,
+      "loss": 0.1062,
+      "step": 1880
+    },
+    {
+      "epoch": 7.297297297297297,
+      "grad_norm": 0.8646144866943359,
+      "learning_rate": 9.481740028736692e-05,
+      "loss": 0.114,
+      "step": 1890
+    },
+    {
+      "epoch": 7.335907335907336,
+      "grad_norm": 0.6492713689804077,
+      "learning_rate": 9.474384858657164e-05,
+      "loss": 0.1157,
+      "step": 1900
+    },
+    {
+      "epoch": 7.374517374517374,
+      "grad_norm": 0.8752506375312805,
+      "learning_rate": 9.466980757421679e-05,
+      "loss": 0.1351,
+      "step": 1910
+    },
+    {
+      "epoch": 7.413127413127413,
+      "grad_norm": 0.6606900095939636,
+      "learning_rate": 9.459527806000305e-05,
+      "loss": 0.1032,
+      "step": 1920
+    },
+    {
+      "epoch": 7.4517374517374515,
+      "grad_norm": 0.6861739754676819,
+      "learning_rate": 9.452026085897325e-05,
+      "loss": 0.113,
+      "step": 1930
+    },
+    {
+      "epoch": 7.4903474903474905,
+      "grad_norm": 0.6919797658920288,
+      "learning_rate": 9.444475679150348e-05,
+      "loss": 0.1155,
+      "step": 1940
+    },
+    {
+      "epoch": 7.528957528957529,
+      "grad_norm": 0.6717447638511658,
+      "learning_rate": 9.436876668329411e-05,
+      "loss": 0.1057,
+      "step": 1950
+    },
+    {
+      "epoch": 7.5675675675675675,
+      "grad_norm": 0.6557009220123291,
+      "learning_rate": 9.429229136536079e-05,
+      "loss": 0.1316,
+      "step": 1960
+    },
+    {
+      "epoch": 7.6061776061776065,
+      "grad_norm": 0.46149829030036926,
+      "learning_rate": 9.421533167402534e-05,
+      "loss": 0.0889,
+      "step": 1970
+    },
+    {
+      "epoch": 7.644787644787645,
+      "grad_norm": 0.6347934603691101,
+      "learning_rate": 9.413788845090666e-05,
+      "loss": 0.1269,
+      "step": 1980
+    },
+    {
+      "epoch": 7.683397683397684,
+      "grad_norm": 0.688076913356781,
+      "learning_rate": 9.405996254291136e-05,
+      "loss": 0.0905,
+      "step": 1990
+    },
+    {
+      "epoch": 7.722007722007722,
+      "grad_norm": 0.8914374113082886,
+      "learning_rate": 9.398155480222474e-05,
+      "loss": 0.1108,
+      "step": 2000
+    },
+    {
+      "epoch": 7.760617760617761,
+      "grad_norm": 0.7588872313499451,
+      "learning_rate": 9.390266608630128e-05,
+      "loss": 0.1348,
+      "step": 2010
+    },
+    {
+      "epoch": 7.799227799227799,
+      "grad_norm": 0.7361499071121216,
+      "learning_rate": 9.38232972578553e-05,
+      "loss": 0.1198,
+      "step": 2020
+    },
+    {
+      "epoch": 7.837837837837838,
+      "grad_norm": 0.6843787431716919,
+      "learning_rate": 9.374344918485164e-05,
+      "loss": 0.0972,
+      "step": 2030
+    },
+    {
+      "epoch": 7.876447876447877,
+      "grad_norm": 0.6163288950920105,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.1171,
+      "step": 2040
+    },
+    {
+      "epoch": 7.915057915057915,
+      "grad_norm": 0.8062847256660461,
+      "learning_rate": 9.358231880322554e-05,
+      "loss": 0.1143,
+      "step": 2050
+    },
+    {
+      "epoch": 7.953667953667954,
+      "grad_norm": 0.7317461371421814,
+      "learning_rate": 9.350103825669916e-05,
+      "loss": 0.1101,
+      "step": 2060
+    },
+    {
+      "epoch": 7.992277992277993,
+      "grad_norm": 0.6277090311050415,
+      "learning_rate": 9.341928198978787e-05,
+      "loss": 0.0967,
+      "step": 2070
+    },
+    {
+      "epoch": 8.03088803088803,
+      "grad_norm": 0.5492644309997559,
+      "learning_rate": 9.333705089656512e-05,
+      "loss": 0.0988,
+      "step": 2080
+    },
+    {
+      "epoch": 8.069498069498069,
+      "grad_norm": 0.9302526116371155,
+      "learning_rate": 9.325434587629698e-05,
+      "loss": 0.1069,
+      "step": 2090
+    },
+    {
+      "epoch": 8.108108108108109,
+      "grad_norm": 0.6697160005569458,
+      "learning_rate": 9.31711678334323e-05,
+      "loss": 0.1049,
+      "step": 2100
+    },
+    {
+      "epoch": 8.146718146718147,
+      "grad_norm": 0.844696581363678,
+      "learning_rate": 9.308751767759282e-05,
+      "loss": 0.1387,
+      "step": 2110
+    },
+    {
+      "epoch": 8.185328185328185,
+      "grad_norm": 0.5563222169876099,
+      "learning_rate": 9.300339632356325e-05,
+      "loss": 0.1177,
+      "step": 2120
+    },
+    {
+      "epoch": 8.223938223938223,
+      "grad_norm": 0.585287868976593,
+      "learning_rate": 9.291880469128124e-05,
+      "loss": 0.0899,
+      "step": 2130
+    },
+    {
+      "epoch": 8.262548262548263,
+      "grad_norm": 0.6320898532867432,
+      "learning_rate": 9.283374370582732e-05,
+      "loss": 0.0855,
+      "step": 2140
+    },
+    {
+      "epoch": 8.301158301158301,
+      "grad_norm": 0.8141903281211853,
+      "learning_rate": 9.274821429741482e-05,
+      "loss": 0.1256,
+      "step": 2150
+    },
+    {
+      "epoch": 8.339768339768339,
+      "grad_norm": 0.5062960982322693,
+      "learning_rate": 9.266221740137961e-05,
+      "loss": 0.1093,
+      "step": 2160
+    },
+    {
+      "epoch": 8.378378378378379,
+      "grad_norm": 0.42139625549316406,
+      "learning_rate": 9.257575395817001e-05,
+      "loss": 0.102,
+      "step": 2170
+    },
+    {
+      "epoch": 8.416988416988417,
+      "grad_norm": 0.515316367149353,
+      "learning_rate": 9.248882491333637e-05,
+      "loss": 0.112,
+      "step": 2180
+    },
+    {
+      "epoch": 8.455598455598455,
+      "grad_norm": 0.7278834581375122,
+      "learning_rate": 9.240143121752076e-05,
+      "loss": 0.1132,
+      "step": 2190
+    },
+    {
+      "epoch": 8.494208494208495,
+      "grad_norm": 0.6631847023963928,
+      "learning_rate": 9.23135738264467e-05,
+      "loss": 0.0936,
+      "step": 2200
+    },
+    {
+      "epoch": 8.532818532818533,
+      "grad_norm": 0.5388717651367188,
+      "learning_rate": 9.222525370090849e-05,
+      "loss": 0.1109,
+      "step": 2210
+    },
+    {
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.826079785823822,
+      "learning_rate": 9.213647180676088e-05,
+      "loss": 0.1007,
+      "step": 2220
+    },
+    {
+      "epoch": 8.61003861003861,
+      "grad_norm": 0.7166194915771484,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.1059,
+      "step": 2230
+    },
+    {
+      "epoch": 8.64864864864865,
+      "grad_norm": 0.5368038415908813,
+      "learning_rate": 9.1957526601295e-05,
+      "loss": 0.1064,
+      "step": 2240
+    },
+    {
+      "epoch": 8.687258687258687,
+      "grad_norm": 0.7046194076538086,
+      "learning_rate": 9.186736524689281e-05,
+      "loss": 0.1185,
+      "step": 2250
+    },
+    {
+      "epoch": 8.725868725868725,
+      "grad_norm": 0.7599208950996399,
+      "learning_rate": 9.177674603769204e-05,
+      "loss": 0.0974,
+      "step": 2260
+    },
+    {
+      "epoch": 8.764478764478765,
+      "grad_norm": 0.7834563255310059,
+      "learning_rate": 9.168566996468983e-05,
+      "loss": 0.0964,
+      "step": 2270
+    },
+    {
+      "epoch": 8.803088803088803,
+      "grad_norm": 0.6925577521324158,
+      "learning_rate": 9.159413802387951e-05,
+      "loss": 0.1009,
+      "step": 2280
+    },
+    {
+      "epoch": 8.841698841698841,
+      "grad_norm": 0.5668750405311584,
+      "learning_rate": 9.150215121623974e-05,
+      "loss": 0.129,
+      "step": 2290
+    },
+    {
+      "epoch": 8.880308880308881,
+      "grad_norm": 0.469650536775589,
+      "learning_rate": 9.140971054772349e-05,
+      "loss": 0.0954,
+      "step": 2300
+    },
+    {
+      "epoch": 8.91891891891892,
+      "grad_norm": 0.5168886780738831,
+      "learning_rate": 9.131681702924713e-05,
+      "loss": 0.11,
+      "step": 2310
+    },
+    {
+      "epoch": 8.957528957528957,
+      "grad_norm": 0.5102772116661072,
+      "learning_rate": 9.122347167667926e-05,
+      "loss": 0.0921,
+      "step": 2320
+    },
+    {
+      "epoch": 8.996138996138995,
+      "grad_norm": 0.6147180795669556,
+      "learning_rate": 9.112967551082973e-05,
+      "loss": 0.1241,
+      "step": 2330
+    },
+    {
+      "epoch": 9.034749034749035,
+      "grad_norm": 0.39069607853889465,
+      "learning_rate": 9.103542955743835e-05,
+      "loss": 0.0937,
+      "step": 2340
+    },
+    {
+      "epoch": 9.073359073359073,
+      "grad_norm": 0.4520420730113983,
+      "learning_rate": 9.094073484716381e-05,
+      "loss": 0.0984,
+      "step": 2350
+    },
+    {
+      "epoch": 9.111969111969112,
+      "grad_norm": 0.5111847519874573,
+      "learning_rate": 9.084559241557226e-05,
+      "loss": 0.112,
+      "step": 2360
+    },
+    {
+      "epoch": 9.150579150579151,
+      "grad_norm": 0.7047441601753235,
+      "learning_rate": 9.075000330312608e-05,
+      "loss": 0.1154,
+      "step": 2370
+    },
+    {
+      "epoch": 9.18918918918919,
+      "grad_norm": 0.4314562976360321,
+      "learning_rate": 9.065396855517253e-05,
+      "loss": 0.1205,
+      "step": 2380
+    },
+    {
+      "epoch": 9.227799227799228,
+      "grad_norm": 0.5829694271087646,
+      "learning_rate": 9.055748922193219e-05,
+      "loss": 0.1033,
+      "step": 2390
+    },
+    {
+      "epoch": 9.266409266409266,
+      "grad_norm": 0.5105317234992981,
+      "learning_rate": 9.046056635848761e-05,
+      "loss": 0.1015,
+      "step": 2400
+    },
+    {
+      "epoch": 9.305019305019306,
+      "grad_norm": 0.7209296822547913,
+      "learning_rate": 9.036320102477169e-05,
+      "loss": 0.1052,
+      "step": 2410
+    },
+    {
+      "epoch": 9.343629343629344,
+      "grad_norm": 0.5549200773239136,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.1383,
+      "step": 2420
+    },
+    {
+      "epoch": 9.382239382239382,
+      "grad_norm": 0.7225512266159058,
+      "learning_rate": 9.016714721043971e-05,
+      "loss": 0.1168,
+      "step": 2430
+    },
+    {
+      "epoch": 9.420849420849422,
+      "grad_norm": 0.5784018635749817,
+      "learning_rate": 9.006846087383675e-05,
+      "loss": 0.1168,
+      "step": 2440
+    },
+    {
+      "epoch": 9.45945945945946,
+      "grad_norm": 0.7874284982681274,
+      "learning_rate": 8.996933635496523e-05,
+      "loss": 0.1087,
+      "step": 2450
+    },
+    {
+      "epoch": 9.498069498069498,
+      "grad_norm": 0.4949483871459961,
+      "learning_rate": 8.986977473783498e-05,
+      "loss": 0.0976,
+      "step": 2460
+    },
+    {
+      "epoch": 9.536679536679536,
+      "grad_norm": 0.8492587208747864,
+      "learning_rate": 8.97697771112359e-05,
+      "loss": 0.095,
+      "step": 2470
+    },
+    {
+      "epoch": 9.575289575289576,
+      "grad_norm": 0.5878174901008606,
+      "learning_rate": 8.966934456872602e-05,
+      "loss": 0.1085,
+      "step": 2480
+    },
+    {
+      "epoch": 9.613899613899614,
+      "grad_norm": 0.6225532293319702,
+      "learning_rate": 8.95684782086195e-05,
+      "loss": 0.0889,
+      "step": 2490
+    },
+    {
+      "epoch": 9.652509652509652,
+      "grad_norm": 0.5066150426864624,
+      "learning_rate": 8.946717913397476e-05,
+      "loss": 0.087,
+      "step": 2500
+    },
+    {
+      "epoch": 9.691119691119692,
+      "grad_norm": 0.48042768239974976,
+      "learning_rate": 8.93654484525822e-05,
+      "loss": 0.0767,
+      "step": 2510
+    },
+    {
+      "epoch": 9.72972972972973,
+      "grad_norm": 0.6562822461128235,
+      "learning_rate": 8.926328727695226e-05,
+      "loss": 0.0901,
+      "step": 2520
+    },
+    {
+      "epoch": 9.768339768339768,
+      "grad_norm": 0.723908543586731,
+      "learning_rate": 8.916069672430319e-05,
+      "loss": 0.0899,
+      "step": 2530
+    },
+    {
+      "epoch": 9.806949806949808,
+      "grad_norm": 0.8529309630393982,
+      "learning_rate": 8.905767791654884e-05,
+      "loss": 0.1066,
+      "step": 2540
+    },
+    {
+      "epoch": 9.845559845559846,
+      "grad_norm": 0.483698308467865,
+      "learning_rate": 8.895423198028638e-05,
+      "loss": 0.1059,
+      "step": 2550
+    },
+    {
+      "epoch": 9.884169884169884,
+      "grad_norm": 0.4942833185195923,
+      "learning_rate": 8.885036004678402e-05,
+      "loss": 0.1062,
+      "step": 2560
+    },
+    {
+      "epoch": 9.922779922779922,
+      "grad_norm": 0.28115925192832947,
+      "learning_rate": 8.874606325196857e-05,
+      "loss": 0.0947,
+      "step": 2570
+    },
+    {
+      "epoch": 9.961389961389962,
+      "grad_norm": 0.41657555103302,
+      "learning_rate": 8.864134273641304e-05,
+      "loss": 0.1056,
+      "step": 2580
+    },
+    {
+      "epoch": 10.0,
+      "grad_norm": 0.9054367542266846,
+      "learning_rate": 8.853619964532427e-05,
+      "loss": 0.0991,
+      "step": 2590
+    },
+    {
+      "epoch": 10.038610038610038,
+      "grad_norm": 0.4605831503868103,
+      "learning_rate": 8.843063512853019e-05,
+      "loss": 0.1027,
+      "step": 2600
+    },
+    {
+      "epoch": 10.077220077220078,
+      "grad_norm": 0.5121803283691406,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.0974,
+      "step": 2610
+    },
+    {
+      "epoch": 10.115830115830116,
+      "grad_norm": 0.5194014310836792,
+      "learning_rate": 8.821824644016882e-05,
+      "loss": 0.0852,
+      "step": 2620
+    },
+    {
+      "epoch": 10.154440154440154,
+      "grad_norm": 0.49289292097091675,
+      "learning_rate": 8.811142459125019e-05,
+      "loss": 0.1083,
+      "step": 2630
+    },
+    {
+      "epoch": 10.193050193050192,
+      "grad_norm": 0.47594016790390015,
+      "learning_rate": 8.800418596189822e-05,
+      "loss": 0.1142,
+      "step": 2640
+    },
+    {
+      "epoch": 10.231660231660232,
+      "grad_norm": 0.5656114816665649,
+      "learning_rate": 8.789653172485737e-05,
+      "loss": 0.1038,
+      "step": 2650
+    },
+    {
+      "epoch": 10.27027027027027,
+      "grad_norm": 0.575564980506897,
+      "learning_rate": 8.778846305741715e-05,
+      "loss": 0.1076,
+      "step": 2660
+    },
+    {
+      "epoch": 10.308880308880308,
+      "grad_norm": 0.46538421511650085,
+      "learning_rate": 8.767998114139918e-05,
+      "loss": 0.1136,
+      "step": 2670
+    },
+    {
+      "epoch": 10.347490347490348,
+      "grad_norm": 0.5079081654548645,
+      "learning_rate": 8.757108716314429e-05,
+      "loss": 0.1235,
+      "step": 2680
+    },
+    {
+      "epoch": 10.386100386100386,
+      "grad_norm": 0.5175929069519043,
+      "learning_rate": 8.746178231349962e-05,
+      "loss": 0.0942,
+      "step": 2690
+    },
+    {
+      "epoch": 10.424710424710424,
+      "grad_norm": 0.5166839957237244,
+      "learning_rate": 8.735206778780549e-05,
+      "loss": 0.0936,
+      "step": 2700
+    },
+    {
+      "epoch": 10.463320463320464,
+      "grad_norm": 0.6094571948051453,
+      "learning_rate": 8.724194478588234e-05,
+      "loss": 0.1108,
+      "step": 2710
+    },
+    {
+      "epoch": 10.501930501930502,
+      "grad_norm": 0.40622076392173767,
+      "learning_rate": 8.713141451201772e-05,
+      "loss": 0.1263,
+      "step": 2720
+    },
+    {
+      "epoch": 10.54054054054054,
+      "grad_norm": 0.5963900685310364,
+      "learning_rate": 8.702047817495295e-05,
+      "loss": 0.102,
+      "step": 2730
+    },
+    {
+      "epoch": 10.579150579150578,
+      "grad_norm": 0.8011244535446167,
+      "learning_rate": 8.69091369878701e-05,
+      "loss": 0.0936,
+      "step": 2740
+    },
+    {
+      "epoch": 10.617760617760618,
+      "grad_norm": 0.4866567552089691,
+      "learning_rate": 8.679739216837849e-05,
+      "loss": 0.0888,
+      "step": 2750
+    },
+    {
+      "epoch": 10.656370656370656,
+      "grad_norm": 0.5055763125419617,
+      "learning_rate": 8.66852449385016e-05,
+      "loss": 0.0917,
+      "step": 2760
+    },
+    {
+      "epoch": 10.694980694980694,
+      "grad_norm": 0.539553701877594,
+      "learning_rate": 8.657269652466356e-05,
+      "loss": 0.0902,
+      "step": 2770
+    },
+    {
+      "epoch": 10.733590733590734,
+      "grad_norm": 0.5983613729476929,
+      "learning_rate": 8.645974815767577e-05,
+      "loss": 0.1082,
+      "step": 2780
+    },
+    {
+      "epoch": 10.772200772200772,
+      "grad_norm": 0.6280248761177063,
+      "learning_rate": 8.634640107272351e-05,
+      "loss": 0.1028,
+      "step": 2790
+    },
+    {
+      "epoch": 10.81081081081081,
+      "grad_norm": 0.6110442280769348,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.0965,
+      "step": 2800
+    },
+    {
+      "epoch": 10.849420849420849,
+      "grad_norm": 0.7813714742660522,
+      "learning_rate": 8.611851571145456e-05,
+      "loss": 0.1389,
+      "step": 2810
+    },
+    {
+      "epoch": 10.888030888030888,
+      "grad_norm": 0.6177837252616882,
+      "learning_rate": 8.600397992725566e-05,
+      "loss": 0.1098,
+      "step": 2820
+    },
+    {
+      "epoch": 10.926640926640927,
+      "grad_norm": 0.5030504465103149,
+      "learning_rate": 8.588905040930061e-05,
+      "loss": 0.0877,
+      "step": 2830
+    },
+    {
+      "epoch": 10.965250965250965,
+      "grad_norm": 0.7109388709068298,
+      "learning_rate": 8.577372841444022e-05,
+      "loss": 0.1064,
+      "step": 2840
+    },
+    {
+      "epoch": 11.003861003861005,
+      "grad_norm": 0.5707525014877319,
+      "learning_rate": 8.565801520381736e-05,
+      "loss": 0.115,
+      "step": 2850
+    },
+    {
+      "epoch": 11.042471042471043,
+      "grad_norm": 0.5837938785552979,
+      "learning_rate": 8.554191204285313e-05,
+      "loss": 0.0986,
+      "step": 2860
+    },
+    {
+      "epoch": 11.08108108108108,
+      "grad_norm": 0.556896984577179,
+      "learning_rate": 8.542542020123315e-05,
+      "loss": 0.0921,
+      "step": 2870
+    },
+    {
+      "epoch": 11.11969111969112,
+      "grad_norm": 0.47775787115097046,
+      "learning_rate": 8.530854095289347e-05,
+      "loss": 0.0973,
+      "step": 2880
+    },
+    {
+      "epoch": 11.158301158301159,
+      "grad_norm": 0.47152382135391235,
+      "learning_rate": 8.519127557600688e-05,
+      "loss": 0.1008,
+      "step": 2890
+    },
+    {
+      "epoch": 11.196911196911197,
+      "grad_norm": 0.40532755851745605,
+      "learning_rate": 8.507362535296871e-05,
+      "loss": 0.1108,
+      "step": 2900
+    },
+    {
+      "epoch": 11.235521235521235,
+      "grad_norm": 0.5269514918327332,
+      "learning_rate": 8.495559157038299e-05,
+      "loss": 0.0884,
+      "step": 2910
+    },
+    {
+      "epoch": 11.274131274131275,
+      "grad_norm": 0.48876118659973145,
+      "learning_rate": 8.483717551904823e-05,
+      "loss": 0.109,
+      "step": 2920
+    },
+    {
+      "epoch": 11.312741312741313,
+      "grad_norm": 0.507595956325531,
+      "learning_rate": 8.47183784939434e-05,
+      "loss": 0.0859,
+      "step": 2930
+    },
+    {
+      "epoch": 11.35135135135135,
+      "grad_norm": 0.5442178845405579,
+      "learning_rate": 8.459920179421374e-05,
+      "loss": 0.0884,
+      "step": 2940
+    },
+    {
+      "epoch": 11.38996138996139,
+      "grad_norm": 0.44095930457115173,
+      "learning_rate": 8.447964672315656e-05,
+      "loss": 0.0829,
+      "step": 2950
+    },
+    {
+      "epoch": 11.428571428571429,
+      "grad_norm": 0.7041499018669128,
+      "learning_rate": 8.435971458820692e-05,
+      "loss": 0.0959,
+      "step": 2960
+    },
+    {
+      "epoch": 11.467181467181467,
+      "grad_norm": 0.7257633209228516,
+      "learning_rate": 8.423940670092345e-05,
+      "loss": 0.0875,
+      "step": 2970
+    },
+    {
+      "epoch": 11.505791505791505,
+      "grad_norm": 0.5796602368354797,
+      "learning_rate": 8.411872437697394e-05,
+      "loss": 0.0986,
+      "step": 2980
+    },
+    {
+      "epoch": 11.544401544401545,
+      "grad_norm": 0.6739242672920227,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.1102,
+      "step": 2990
+    },
+    {
+      "epoch": 11.583011583011583,
+      "grad_norm": 0.6059641242027283,
+      "learning_rate": 8.38762417022074e-05,
+      "loss": 0.1094,
+      "step": 3000
+    },
+    {
+      "epoch": 11.621621621621621,
+      "grad_norm": 0.9224318265914917,
+      "learning_rate": 8.375444400314204e-05,
+      "loss": 0.0836,
+      "step": 3010
+    },
+    {
+      "epoch": 11.660231660231661,
+      "grad_norm": 0.5493955016136169,
+      "learning_rate": 8.3632277170885e-05,
+      "loss": 0.0835,
+      "step": 3020
+    },
+    {
+      "epoch": 11.698841698841699,
+      "grad_norm": 0.5447977781295776,
+      "learning_rate": 8.350974254143318e-05,
+      "loss": 0.0906,
+      "step": 3030
+    },
+    {
+      "epoch": 11.737451737451737,
+      "grad_norm": 0.6024081707000732,
+      "learning_rate": 8.338684145480566e-05,
+      "loss": 0.0977,
+      "step": 3040
+    },
+    {
+      "epoch": 11.776061776061777,
+      "grad_norm": 0.7188856601715088,
+      "learning_rate": 8.326357525502904e-05,
+      "loss": 0.0939,
+      "step": 3050
+    },
+    {
+      "epoch": 11.814671814671815,
+      "grad_norm": 0.4234011769294739,
+      "learning_rate": 8.313994529012273e-05,
+      "loss": 0.1001,
+      "step": 3060
+    },
+    {
+      "epoch": 11.853281853281853,
+      "grad_norm": 0.47279033064842224,
+      "learning_rate": 8.301595291208422e-05,
+      "loss": 0.082,
+      "step": 3070
+    },
+    {
+      "epoch": 11.891891891891891,
+      "grad_norm": 0.6048724055290222,
+      "learning_rate": 8.289159947687427e-05,
+      "loss": 0.1029,
+      "step": 3080
+    },
+    {
+      "epoch": 11.930501930501931,
+      "grad_norm": 0.6834560036659241,
+      "learning_rate": 8.276688634440216e-05,
+      "loss": 0.1116,
+      "step": 3090
+    },
+    {
+      "epoch": 11.96911196911197,
+      "grad_norm": 0.730119526386261,
+      "learning_rate": 8.26418148785107e-05,
+      "loss": 0.0909,
+      "step": 3100
+    },
+    {
+      "epoch": 12.007722007722007,
+      "grad_norm": 0.5458978414535522,
+      "learning_rate": 8.251638644696141e-05,
+      "loss": 0.0786,
+      "step": 3110
+    },
+    {
+      "epoch": 12.046332046332047,
+      "grad_norm": 0.7734132409095764,
+      "learning_rate": 8.23906024214195e-05,
+      "loss": 0.0772,
+      "step": 3120
+    },
+    {
+      "epoch": 12.084942084942085,
+      "grad_norm": 0.6241917610168457,
+      "learning_rate": 8.226446417743897e-05,
+      "loss": 0.0725,
+      "step": 3130
+    },
+    {
+      "epoch": 12.123552123552123,
+      "grad_norm": 0.7433600425720215,
+      "learning_rate": 8.213797309444742e-05,
+      "loss": 0.0903,
+      "step": 3140
+    },
+    {
+      "epoch": 12.162162162162161,
+      "grad_norm": 0.511486291885376,
+      "learning_rate": 8.201113055573105e-05,
+      "loss": 0.0969,
+      "step": 3150
+    },
+    {
+      "epoch": 12.200772200772201,
+      "grad_norm": 0.5927137732505798,
+      "learning_rate": 8.188393794841958e-05,
+      "loss": 0.1011,
+      "step": 3160
+    },
+    {
+      "epoch": 12.23938223938224,
+      "grad_norm": 0.3921918272972107,
+      "learning_rate": 8.175639666347094e-05,
+      "loss": 0.0793,
+      "step": 3170
+    },
+    {
+      "epoch": 12.277992277992277,
+      "grad_norm": 0.719854474067688,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.1059,
+      "step": 3180
+    },
+    {
+      "epoch": 12.316602316602317,
+      "grad_norm": 0.5687322616577148,
+      "learning_rate": 8.150027364354431e-05,
+      "loss": 0.1051,
+      "step": 3190
+    },
+    {
+      "epoch": 12.355212355212355,
+      "grad_norm": 0.6002088189125061,
+      "learning_rate": 8.137169470948662e-05,
+      "loss": 0.0915,
+      "step": 3200
+    },
+    {
+      "epoch": 12.393822393822393,
+      "grad_norm": 0.5544345378875732,
+      "learning_rate": 8.124277269960179e-05,
+      "loss": 0.087,
+      "step": 3210
+    },
+    {
+      "epoch": 12.432432432432432,
+      "grad_norm": 3.415940999984741,
+      "learning_rate": 8.111350902376023e-05,
+      "loss": 0.1167,
+      "step": 3220
+    },
+    {
+      "epoch": 12.471042471042471,
+      "grad_norm": 0.5576305985450745,
+      "learning_rate": 8.098390509556883e-05,
+      "loss": 0.1188,
+      "step": 3230
+    },
+    {
+      "epoch": 12.50965250965251,
+      "grad_norm": 0.6188719868659973,
+      "learning_rate": 8.085396233235536e-05,
+      "loss": 0.0834,
+      "step": 3240
+    },
+    {
+      "epoch": 12.548262548262548,
+      "grad_norm": 0.4263285994529724,
+      "learning_rate": 8.072368215515306e-05,
+      "loss": 0.1006,
+      "step": 3250
+    },
+    {
+      "epoch": 12.586872586872587,
+      "grad_norm": 0.6959123611450195,
+      "learning_rate": 8.059306598868506e-05,
+      "loss": 0.0959,
+      "step": 3260
+    },
+    {
+      "epoch": 12.625482625482626,
+      "grad_norm": 0.5221771001815796,
+      "learning_rate": 8.046211526134888e-05,
+      "loss": 0.0953,
+      "step": 3270
+    },
+    {
+      "epoch": 12.664092664092664,
+      "grad_norm": 0.71525639295578,
+      "learning_rate": 8.033083140520065e-05,
+      "loss": 0.0808,
+      "step": 3280
+    },
+    {
+      "epoch": 12.702702702702704,
+      "grad_norm": 0.7279417514801025,
+      "learning_rate": 8.019921585593962e-05,
+      "loss": 0.0867,
+      "step": 3290
+    },
+    {
+      "epoch": 12.741312741312742,
+      "grad_norm": 0.5896912217140198,
+      "learning_rate": 8.006727005289232e-05,
+      "loss": 0.0848,
+      "step": 3300
+    },
+    {
+      "epoch": 12.77992277992278,
+      "grad_norm": 0.6318273544311523,
+      "learning_rate": 7.993499543899692e-05,
+      "loss": 0.0888,
+      "step": 3310
+    },
+    {
+      "epoch": 12.818532818532818,
+      "grad_norm": 0.5931304693222046,
+      "learning_rate": 7.980239346078742e-05,
+      "loss": 0.0892,
+      "step": 3320
+    },
+    {
+      "epoch": 12.857142857142858,
+      "grad_norm": 0.5776562094688416,
+      "learning_rate": 7.966946556837778e-05,
+      "loss": 0.0912,
+      "step": 3330
+    },
+    {
+      "epoch": 12.895752895752896,
+      "grad_norm": 0.5498374104499817,
+      "learning_rate": 7.953621321544616e-05,
+      "loss": 0.0942,
+      "step": 3340
+    },
+    {
+      "epoch": 12.934362934362934,
+      "grad_norm": 0.4531612992286682,
+      "learning_rate": 7.940263785921896e-05,
+      "loss": 0.0951,
+      "step": 3350
+    },
+    {
+      "epoch": 12.972972972972974,
+      "grad_norm": 0.5652436017990112,
+      "learning_rate": 7.926874096045482e-05,
+      "loss": 0.083,
+      "step": 3360
+    },
+    {
+      "epoch": 13.011583011583012,
+      "grad_norm": 0.5419857501983643,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.0894,
+      "step": 3370
+    },
+    {
+      "epoch": 13.05019305019305,
+      "grad_norm": 0.5802552103996277,
+      "learning_rate": 7.89999883959163e-05,
+      "loss": 0.1176,
+      "step": 3380
+    },
+    {
+      "epoch": 13.088803088803088,
+      "grad_norm": 0.6911759972572327,
+      "learning_rate": 7.886513566917687e-05,
+      "loss": 0.0793,
+      "step": 3390
+    },
+    {
+      "epoch": 13.127413127413128,
+      "grad_norm": 0.6647396087646484,
+      "learning_rate": 7.872996727793838e-05,
+      "loss": 0.0913,
+      "step": 3400
+    },
+    {
+      "epoch": 13.166023166023166,
+      "grad_norm": 0.7876801490783691,
+      "learning_rate": 7.859448470038069e-05,
+      "loss": 0.1018,
+      "step": 3410
+    },
+    {
+      "epoch": 13.204633204633204,
+      "grad_norm": 0.5604192018508911,
+      "learning_rate": 7.845868941811956e-05,
+      "loss": 0.0759,
+      "step": 3420
+    },
+    {
+      "epoch": 13.243243243243244,
+      "grad_norm": 0.529370903968811,
+      "learning_rate": 7.832258291619043e-05,
+      "loss": 0.0933,
+      "step": 3430
+    },
+    {
+      "epoch": 13.281853281853282,
+      "grad_norm": 0.49177083373069763,
+      "learning_rate": 7.81861666830322e-05,
+      "loss": 0.0975,
+      "step": 3440
+    },
+    {
+      "epoch": 13.32046332046332,
+      "grad_norm": 0.46308794617652893,
+      "learning_rate": 7.804944221047097e-05,
+      "loss": 0.0941,
+      "step": 3450
+    },
+    {
+      "epoch": 13.35907335907336,
+      "grad_norm": 0.5650749802589417,
+      "learning_rate": 7.791241099370364e-05,
+      "loss": 0.0805,
+      "step": 3460
+    },
+    {
+      "epoch": 13.397683397683398,
+      "grad_norm": 0.36478251218795776,
+      "learning_rate": 7.777507453128163e-05,
+      "loss": 0.1041,
+      "step": 3470
+    },
+    {
+      "epoch": 13.436293436293436,
+      "grad_norm": 0.43773508071899414,
+      "learning_rate": 7.763743432509451e-05,
+      "loss": 0.1053,
+      "step": 3480
+    },
+    {
+      "epoch": 13.474903474903474,
+      "grad_norm": 0.6326601505279541,
+      "learning_rate": 7.749949188035353e-05,
+      "loss": 0.0999,
+      "step": 3490
+    },
+    {
+      "epoch": 13.513513513513514,
+      "grad_norm": 0.7280349135398865,
+      "learning_rate": 7.736124870557516e-05,
+      "loss": 0.0776,
+      "step": 3500
+    },
+    {
+      "epoch": 13.552123552123552,
+      "grad_norm": 0.725627601146698,
+      "learning_rate": 7.722270631256459e-05,
+      "loss": 0.0937,
+      "step": 3510
+    },
+    {
+      "epoch": 13.59073359073359,
+      "grad_norm": 0.5294554829597473,
+      "learning_rate": 7.708386621639925e-05,
+      "loss": 0.0859,
+      "step": 3520
+    },
+    {
+      "epoch": 13.62934362934363,
+      "grad_norm": 0.5159903168678284,
+      "learning_rate": 7.694472993541219e-05,
+      "loss": 0.086,
+      "step": 3530
+    },
+    {
+      "epoch": 13.667953667953668,
+      "grad_norm": 0.5003053545951843,
+      "learning_rate": 7.680529899117547e-05,
+      "loss": 0.086,
+      "step": 3540
+    },
+    {
+      "epoch": 13.706563706563706,
+      "grad_norm": 0.5443400144577026,
+      "learning_rate": 7.666557490848358e-05,
+      "loss": 0.0833,
+      "step": 3550
+    },
+    {
+      "epoch": 13.745173745173744,
+      "grad_norm": 0.6605769395828247,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.0746,
+      "step": 3560
+    },
+    {
+      "epoch": 13.783783783783784,
+      "grad_norm": 0.5426715016365051,
+      "learning_rate": 7.638525344292402e-05,
+      "loss": 0.0875,
+      "step": 3570
+    },
+    {
+      "epoch": 13.822393822393822,
+      "grad_norm": 0.41429880261421204,
+      "learning_rate": 7.624465912560697e-05,
+      "loss": 0.0948,
+      "step": 3580
+    },
+    {
+      "epoch": 13.86100386100386,
+      "grad_norm": 0.5427739024162292,
+      "learning_rate": 7.610377780090249e-05,
+      "loss": 0.0916,
+      "step": 3590
+    },
+    {
+      "epoch": 13.8996138996139,
+      "grad_norm": 0.6028660535812378,
+      "learning_rate": 7.596261100946618e-05,
+      "loss": 0.0882,
+      "step": 3600
+    },
+    {
+      "epoch": 13.938223938223938,
+      "grad_norm": 0.6504780054092407,
+      "learning_rate": 7.582116029507542e-05,
+      "loss": 0.0821,
+      "step": 3610
+    },
+    {
+      "epoch": 13.976833976833976,
+      "grad_norm": 0.6405124068260193,
+      "learning_rate": 7.56794272046126e-05,
+      "loss": 0.0805,
+      "step": 3620
+    },
+    {
+      "epoch": 14.015444015444016,
+      "grad_norm": 0.7024744749069214,
+      "learning_rate": 7.55374132880481e-05,
+      "loss": 0.0648,
+      "step": 3630
+    },
+    {
+      "epoch": 14.054054054054054,
+      "grad_norm": 0.46351149678230286,
+      "learning_rate": 7.539512009842333e-05,
+      "loss": 0.0798,
+      "step": 3640
+    },
+    {
+      "epoch": 14.092664092664092,
+      "grad_norm": 0.5199838280677795,
+      "learning_rate": 7.525254919183382e-05,
+      "loss": 0.0856,
+      "step": 3650
+    },
+    {
+      "epoch": 14.13127413127413,
+      "grad_norm": 0.6727871298789978,
+      "learning_rate": 7.510970212741215e-05,
+      "loss": 0.0904,
+      "step": 3660
+    },
+    {
+      "epoch": 14.16988416988417,
+      "grad_norm": 0.5358639359474182,
+      "learning_rate": 7.496658046731096e-05,
+      "loss": 0.1062,
+      "step": 3670
+    },
+    {
+      "epoch": 14.208494208494209,
+      "grad_norm": 0.6725749373435974,
+      "learning_rate": 7.482318577668578e-05,
+      "loss": 0.0881,
+      "step": 3680
+    },
+    {
+      "epoch": 14.247104247104247,
+      "grad_norm": 0.6883063912391663,
+      "learning_rate": 7.467951962367796e-05,
+      "loss": 0.0736,
+      "step": 3690
+    },
+    {
+      "epoch": 14.285714285714286,
+      "grad_norm": 0.6106855869293213,
+      "learning_rate": 7.453558357939755e-05,
+      "loss": 0.0888,
+      "step": 3700
+    },
+    {
+      "epoch": 14.324324324324325,
+      "grad_norm": 0.5512902140617371,
+      "learning_rate": 7.439137921790606e-05,
+      "loss": 0.0883,
+      "step": 3710
+    },
+    {
+      "epoch": 14.362934362934363,
+      "grad_norm": 0.6435045599937439,
+      "learning_rate": 7.42469081161993e-05,
+      "loss": 0.0919,
+      "step": 3720
+    },
+    {
+      "epoch": 14.4015444015444,
+      "grad_norm": 0.6004988551139832,
+      "learning_rate": 7.410217185419006e-05,
+      "loss": 0.0677,
+      "step": 3730
+    },
+    {
+      "epoch": 14.44015444015444,
+      "grad_norm": 0.5087930560112,
+      "learning_rate": 7.395717201469095e-05,
+      "loss": 0.0679,
+      "step": 3740
+    },
+    {
+      "epoch": 14.478764478764479,
+      "grad_norm": 0.44460296630859375,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.0838,
+      "step": 3750
+    },
+    {
+      "epoch": 14.517374517374517,
+      "grad_norm": 0.372995525598526,
+      "learning_rate": 7.36663879488682e-05,
+      "loss": 0.0831,
+      "step": 3760
+    },
+    {
+      "epoch": 14.555984555984557,
+      "grad_norm": 0.6550002694129944,
+      "learning_rate": 7.352060690251254e-05,
+      "loss": 0.1051,
+      "step": 3770
+    },
+    {
+      "epoch": 14.594594594594595,
+      "grad_norm": 0.5978309512138367,
+      "learning_rate": 7.337456863856811e-05,
+      "loss": 0.096,
+      "step": 3780
+    },
+    {
+      "epoch": 14.633204633204633,
+      "grad_norm": 0.3988819420337677,
+      "learning_rate": 7.3228274754086e-05,
+      "loss": 0.0784,
+      "step": 3790
+    },
+    {
+      "epoch": 14.671814671814673,
+      "grad_norm": 0.5307171940803528,
+      "learning_rate": 7.308172684891267e-05,
+      "loss": 0.1048,
+      "step": 3800
+    },
+    {
+      "epoch": 14.71042471042471,
+      "grad_norm": 0.6089282631874084,
+      "learning_rate": 7.293492652567255e-05,
+      "loss": 0.073,
+      "step": 3810
+    },
+    {
+      "epoch": 14.749034749034749,
+      "grad_norm": 0.6973853707313538,
+      "learning_rate": 7.278787538975043e-05,
+      "loss": 0.0966,
+      "step": 3820
+    },
+    {
+      "epoch": 14.787644787644787,
+      "grad_norm": 0.39602163434028625,
+      "learning_rate": 7.2640575049274e-05,
+      "loss": 0.0708,
+      "step": 3830
+    },
+    {
+      "epoch": 14.826254826254827,
+      "grad_norm": 0.5536203384399414,
+      "learning_rate": 7.249302711509616e-05,
+      "loss": 0.0701,
+      "step": 3840
+    },
+    {
+      "epoch": 14.864864864864865,
+      "grad_norm": 0.5625492930412292,
+      "learning_rate": 7.23452332007775e-05,
+      "loss": 0.1033,
+      "step": 3850
+    },
+    {
+      "epoch": 14.903474903474903,
+      "grad_norm": 0.8248059749603271,
+      "learning_rate": 7.219719492256858e-05,
+      "loss": 0.0879,
+      "step": 3860
+    },
+    {
+      "epoch": 14.942084942084943,
+      "grad_norm": 1.0776482820510864,
+      "learning_rate": 7.20489138993923e-05,
+      "loss": 0.0788,
+      "step": 3870
+    },
+    {
+      "epoch": 14.980694980694981,
+      "grad_norm": 0.47371265292167664,
+      "learning_rate": 7.190039175282614e-05,
+      "loss": 0.0804,
+      "step": 3880
+    },
+    {
+      "epoch": 15.019305019305019,
+      "grad_norm": 0.48720476031303406,
+      "learning_rate": 7.175163010708455e-05,
+      "loss": 0.09,
+      "step": 3890
+    },
+    {
+      "epoch": 15.057915057915057,
+      "grad_norm": 0.5072534084320068,
+      "learning_rate": 7.1602630589001e-05,
+      "loss": 0.0861,
+      "step": 3900
+    },
+    {
+      "epoch": 15.096525096525097,
+      "grad_norm": 0.6805068850517273,
+      "learning_rate": 7.14533948280104e-05,
+      "loss": 0.0815,
+      "step": 3910
+    },
+    {
+      "epoch": 15.135135135135135,
+      "grad_norm": 0.6972246766090393,
+      "learning_rate": 7.130392445613109e-05,
+      "loss": 0.0951,
+      "step": 3920
+    },
+    {
+      "epoch": 15.173745173745173,
+      "grad_norm": 0.5297501087188721,
+      "learning_rate": 7.115422110794711e-05,
+      "loss": 0.0898,
+      "step": 3930
+    },
+    {
+      "epoch": 15.212355212355213,
+      "grad_norm": 0.4296838641166687,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.0783,
+      "step": 3940
+    },
+    {
+      "epoch": 15.250965250965251,
+      "grad_norm": 0.6048504114151001,
+      "learning_rate": 7.08541220337224e-05,
+      "loss": 0.0843,
+      "step": 3950
+    },
+    {
+      "epoch": 15.28957528957529,
+      "grad_norm": 0.45158666372299194,
+      "learning_rate": 7.070372958951706e-05,
+      "loss": 0.0847,
+      "step": 3960
+    },
+    {
+      "epoch": 15.328185328185327,
+      "grad_norm": 0.6273931860923767,
+      "learning_rate": 7.055311073264194e-05,
+      "loss": 0.0751,
+      "step": 3970
+    },
+    {
+      "epoch": 15.366795366795367,
+      "grad_norm": 0.4455814063549042,
+      "learning_rate": 7.040226711024077e-05,
+      "loss": 0.0821,
+      "step": 3980
+    },
+    {
+      "epoch": 15.405405405405405,
+      "grad_norm": 0.6072818636894226,
+      "learning_rate": 7.02512003719152e-05,
+      "loss": 0.0887,
+      "step": 3990
+    },
+    {
+      "epoch": 15.444015444015443,
+      "grad_norm": 0.7125831246376038,
+      "learning_rate": 7.00999121697069e-05,
+      "loss": 0.0901,
+      "step": 4000
+    },
+    {
+      "epoch": 15.482625482625483,
+      "grad_norm": 0.5886876583099365,
+      "learning_rate": 6.99484041580794e-05,
+      "loss": 0.0716,
+      "step": 4010
+    },
+    {
+      "epoch": 15.521235521235521,
+      "grad_norm": 0.4565495252609253,
+      "learning_rate": 6.979667799390004e-05,
+      "loss": 0.0769,
+      "step": 4020
+    },
+    {
+      "epoch": 15.55984555984556,
+      "grad_norm": 0.5533151030540466,
+      "learning_rate": 6.964473533642185e-05,
+      "loss": 0.086,
+      "step": 4030
+    },
+    {
+      "epoch": 15.5984555984556,
+      "grad_norm": 0.46190527081489563,
+      "learning_rate": 6.949257784726539e-05,
+      "loss": 0.0528,
+      "step": 4040
+    },
+    {
+      "epoch": 15.637065637065637,
+      "grad_norm": 1.3326513767242432,
+      "learning_rate": 6.934020719040056e-05,
+      "loss": 0.0868,
+      "step": 4050
+    },
+    {
+      "epoch": 15.675675675675675,
+      "grad_norm": 0.6459556818008423,
+      "learning_rate": 6.918762503212848e-05,
+      "loss": 0.0709,
+      "step": 4060
+    },
+    {
+      "epoch": 15.714285714285714,
+      "grad_norm": 0.4545834958553314,
+      "learning_rate": 6.903483304106319e-05,
+      "loss": 0.0609,
+      "step": 4070
+    },
+    {
+      "epoch": 15.752895752895753,
+      "grad_norm": 0.4588858187198639,
+      "learning_rate": 6.888183288811341e-05,
+      "loss": 0.0855,
+      "step": 4080
+    },
+    {
+      "epoch": 15.791505791505791,
+      "grad_norm": 0.8000584840774536,
+      "learning_rate": 6.87286262464643e-05,
+      "loss": 0.0841,
+      "step": 4090
+    },
+    {
+      "epoch": 15.83011583011583,
+      "grad_norm": 0.6274726390838623,
+      "learning_rate": 6.857521479155915e-05,
+      "loss": 0.0781,
+      "step": 4100
+    },
+    {
+      "epoch": 15.86872586872587,
+      "grad_norm": 0.5860035419464111,
+      "learning_rate": 6.842160020108104e-05,
+      "loss": 0.0789,
+      "step": 4110
+    },
+    {
+      "epoch": 15.907335907335908,
+      "grad_norm": 0.804555356502533,
+      "learning_rate": 6.826778415493455e-05,
+      "loss": 0.1013,
+      "step": 4120
+    },
+    {
+      "epoch": 15.945945945945946,
+      "grad_norm": 0.5548028945922852,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.0874,
+      "step": 4130
+    },
+    {
+      "epoch": 15.984555984555985,
+      "grad_norm": 0.43698981404304504,
+      "learning_rate": 6.795955442625159e-05,
+      "loss": 0.0866,
+      "step": 4140
+    },
+    {
+      "epoch": 16.023166023166024,
+      "grad_norm": 1.1259561777114868,
+      "learning_rate": 6.780514411446608e-05,
+      "loss": 0.089,
+      "step": 4150
+    },
+    {
+      "epoch": 16.06177606177606,
+      "grad_norm": 0.44387349486351013,
+      "learning_rate": 6.765053908847716e-05,
+      "loss": 0.0757,
+      "step": 4160
+    },
+    {
+      "epoch": 16.1003861003861,
+      "grad_norm": 0.5084031224250793,
+      "learning_rate": 6.749574103902064e-05,
+      "loss": 0.0756,
+      "step": 4170
+    },
+    {
+      "epoch": 16.138996138996138,
+      "grad_norm": 0.37173858284950256,
+      "learning_rate": 6.734075165894317e-05,
+      "loss": 0.077,
+      "step": 4180
+    },
+    {
+      "epoch": 16.177606177606176,
+      "grad_norm": 0.6248420476913452,
+      "learning_rate": 6.71855726431838e-05,
+      "loss": 0.0953,
+      "step": 4190
+    },
+    {
+      "epoch": 16.216216216216218,
+      "grad_norm": 0.3940771222114563,
+      "learning_rate": 6.703020568875538e-05,
+      "loss": 0.0671,
+      "step": 4200
+    },
+    {
+      "epoch": 16.254826254826256,
+      "grad_norm": 0.6416221857070923,
+      "learning_rate": 6.687465249472603e-05,
+      "loss": 0.0827,
+      "step": 4210
+    },
+    {
+      "epoch": 16.293436293436294,
+      "grad_norm": 0.623534619808197,
+      "learning_rate": 6.671891476220055e-05,
+      "loss": 0.1018,
+      "step": 4220
+    },
+    {
+      "epoch": 16.332046332046332,
+      "grad_norm": 0.6577959060668945,
+      "learning_rate": 6.656299419430183e-05,
+      "loss": 0.116,
+      "step": 4230
+    },
+    {
+      "epoch": 16.37065637065637,
+      "grad_norm": 0.5343630909919739,
+      "learning_rate": 6.640689249615223e-05,
+      "loss": 0.0923,
+      "step": 4240
+    },
+    {
+      "epoch": 16.409266409266408,
+      "grad_norm": 0.43712395429611206,
+      "learning_rate": 6.625061137485491e-05,
+      "loss": 0.0853,
+      "step": 4250
+    },
+    {
+      "epoch": 16.447876447876446,
+      "grad_norm": 0.5086554884910583,
+      "learning_rate": 6.609415253947517e-05,
+      "loss": 0.0849,
+      "step": 4260
+    },
+    {
+      "epoch": 16.486486486486488,
+      "grad_norm": 0.5114151239395142,
+      "learning_rate": 6.593751770102178e-05,
+      "loss": 0.0789,
+      "step": 4270
+    },
+    {
+      "epoch": 16.525096525096526,
+      "grad_norm": 0.6828848123550415,
+      "learning_rate": 6.578070857242823e-05,
+      "loss": 0.0769,
+      "step": 4280
+    },
+    {
+      "epoch": 16.563706563706564,
+      "grad_norm": 0.7185290455818176,
+      "learning_rate": 6.562372686853402e-05,
+      "loss": 0.0829,
+      "step": 4290
+    },
+    {
+      "epoch": 16.602316602316602,
+      "grad_norm": 0.5070241093635559,
+      "learning_rate": 6.546657430606593e-05,
+      "loss": 0.0558,
+      "step": 4300
+    },
+    {
+      "epoch": 16.64092664092664,
+      "grad_norm": 0.38561224937438965,
+      "learning_rate": 6.530925260361918e-05,
+      "loss": 0.0823,
+      "step": 4310
+    },
+    {
+      "epoch": 16.679536679536678,
+      "grad_norm": 0.7504719495773315,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.0918,
+      "step": 4320
+    },
+    {
+      "epoch": 16.71814671814672,
+      "grad_norm": 0.4384387135505676,
+      "learning_rate": 6.499410866240032e-05,
+      "loss": 0.071,
+      "step": 4330
+    },
+    {
+      "epoch": 16.756756756756758,
+      "grad_norm": 0.5349385738372803,
+      "learning_rate": 6.48362898699919e-05,
+      "loss": 0.0596,
+      "step": 4340
+    },
+    {
+      "epoch": 16.795366795366796,
+      "grad_norm": 0.5113857388496399,
+      "learning_rate": 6.467830883029443e-05,
+      "loss": 0.0576,
+      "step": 4350
+    },
+    {
+      "epoch": 16.833976833976834,
+      "grad_norm": 0.5070815086364746,
+      "learning_rate": 6.452016727096326e-05,
+      "loss": 0.0739,
+      "step": 4360
+    },
+    {
+      "epoch": 16.872586872586872,
+      "grad_norm": 0.4332325756549835,
+      "learning_rate": 6.436186692140916e-05,
+      "loss": 0.0808,
+      "step": 4370
+    },
+    {
+      "epoch": 16.91119691119691,
+      "grad_norm": 0.4109137952327728,
+      "learning_rate": 6.420340951277938e-05,
+      "loss": 0.0822,
+      "step": 4380
+    },
+    {
+      "epoch": 16.94980694980695,
+      "grad_norm": 0.5654485821723938,
+      "learning_rate": 6.404479677793874e-05,
+      "loss": 0.0659,
+      "step": 4390
+    },
+    {
+      "epoch": 16.98841698841699,
+      "grad_norm": 0.5623957514762878,
+      "learning_rate": 6.388603045145075e-05,
+      "loss": 0.0886,
+      "step": 4400
+    },
+    {
+      "epoch": 17.027027027027028,
+      "grad_norm": 0.5168842077255249,
+      "learning_rate": 6.372711226955843e-05,
+      "loss": 0.0737,
+      "step": 4410
+    },
+    {
+      "epoch": 17.065637065637066,
+      "grad_norm": 0.4881226122379303,
+      "learning_rate": 6.356804397016564e-05,
+      "loss": 0.0983,
+      "step": 4420
+    },
+    {
+      "epoch": 17.104247104247104,
+      "grad_norm": 0.5036429166793823,
+      "learning_rate": 6.340882729281779e-05,
+      "loss": 0.0842,
+      "step": 4430
+    },
+    {
+      "epoch": 17.142857142857142,
+      "grad_norm": 0.5281261205673218,
+      "learning_rate": 6.324946397868294e-05,
+      "loss": 0.0827,
+      "step": 4440
+    },
+    {
+      "epoch": 17.18146718146718,
+      "grad_norm": 0.6240306496620178,
+      "learning_rate": 6.308995577053276e-05,
+      "loss": 0.1004,
+      "step": 4450
+    },
+    {
+      "epoch": 17.22007722007722,
+      "grad_norm": 0.5073460340499878,
+      "learning_rate": 6.293030441272347e-05,
+      "loss": 0.0693,
+      "step": 4460
+    },
+    {
+      "epoch": 17.25868725868726,
+      "grad_norm": 0.4580148756504059,
+      "learning_rate": 6.277051165117677e-05,
+      "loss": 0.0734,
+      "step": 4470
+    },
+    {
+      "epoch": 17.2972972972973,
+      "grad_norm": 0.6077268719673157,
+      "learning_rate": 6.261057923336064e-05,
+      "loss": 0.0826,
+      "step": 4480
+    },
+    {
+      "epoch": 17.335907335907336,
+      "grad_norm": 0.5811083316802979,
+      "learning_rate": 6.245050890827042e-05,
+      "loss": 0.0883,
+      "step": 4490
+    },
+    {
+      "epoch": 17.374517374517374,
+      "grad_norm": 0.6039600372314453,
+      "learning_rate": 6.229030242640952e-05,
+      "loss": 0.0817,
+      "step": 4500
+    },
+    {
+      "epoch": 17.413127413127413,
+      "grad_norm": 0.3778926730155945,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.0911,
+      "step": 4510
+    },
+    {
+      "epoch": 17.45173745173745,
+      "grad_norm": 0.5251861810684204,
+      "learning_rate": 6.196948800181523e-05,
+      "loss": 0.0708,
+      "step": 4520
+    },
+    {
+      "epoch": 17.49034749034749,
+      "grad_norm": 0.5521923303604126,
+      "learning_rate": 6.180888356745695e-05,
+      "loss": 0.0656,
+      "step": 4530
+    },
+    {
+      "epoch": 17.52895752895753,
+      "grad_norm": 0.598902702331543,
+      "learning_rate": 6.164814999303995e-05,
+      "loss": 0.1051,
+      "step": 4540
+    },
+    {
+      "epoch": 17.56756756756757,
+      "grad_norm": 0.3929014205932617,
+      "learning_rate": 6.148728903632081e-05,
+      "loss": 0.0784,
+      "step": 4550
+    },
+    {
+      "epoch": 17.606177606177607,
+      "grad_norm": 0.5811739563941956,
+      "learning_rate": 6.132630245644921e-05,
+      "loss": 0.06,
+      "step": 4560
+    },
+    {
+      "epoch": 17.644787644787645,
+      "grad_norm": 0.6186365485191345,
+      "learning_rate": 6.116519201394857e-05,
+      "loss": 0.0916,
+      "step": 4570
+    },
+    {
+      "epoch": 17.683397683397683,
+      "grad_norm": 0.6466165781021118,
+      "learning_rate": 6.10039594706969e-05,
+      "loss": 0.0634,
+      "step": 4580
+    },
+    {
+      "epoch": 17.72200772200772,
+      "grad_norm": 0.6939656734466553,
+      "learning_rate": 6.084260658990744e-05,
+      "loss": 0.0772,
+      "step": 4590
+    },
+    {
+      "epoch": 17.760617760617762,
+      "grad_norm": 0.6492602825164795,
+      "learning_rate": 6.068113513610943e-05,
+      "loss": 0.0697,
+      "step": 4600
+    },
+    {
+      "epoch": 17.7992277992278,
+      "grad_norm": 0.6554388403892517,
+      "learning_rate": 6.0519546875128876e-05,
+      "loss": 0.0785,
+      "step": 4610
+    },
+    {
+      "epoch": 17.83783783783784,
+      "grad_norm": 0.5153865814208984,
+      "learning_rate": 6.035784357406906e-05,
+      "loss": 0.0619,
+      "step": 4620
+    },
+    {
+      "epoch": 17.876447876447877,
+      "grad_norm": 0.7237990498542786,
+      "learning_rate": 6.01960270012914e-05,
+      "loss": 0.082,
+      "step": 4630
+    },
+    {
+      "epoch": 17.915057915057915,
+      "grad_norm": 0.4947359561920166,
+      "learning_rate": 6.003409892639599e-05,
+      "loss": 0.0818,
+      "step": 4640
+    },
+    {
+      "epoch": 17.953667953667953,
+      "grad_norm": 0.5115700364112854,
+      "learning_rate": 5.9872061120202336e-05,
+      "loss": 0.0679,
+      "step": 4650
+    },
+    {
+      "epoch": 17.99227799227799,
+      "grad_norm": 0.5944989919662476,
+      "learning_rate": 5.9709915354729914e-05,
+      "loss": 0.0772,
+      "step": 4660
+    },
+    {
+      "epoch": 18.030888030888033,
+      "grad_norm": 0.5000519752502441,
+      "learning_rate": 5.9547663403178824e-05,
+      "loss": 0.0689,
+      "step": 4670
+    },
+    {
+      "epoch": 18.06949806949807,
+      "grad_norm": 0.46222320199012756,
+      "learning_rate": 5.9385307039910445e-05,
+      "loss": 0.0767,
+      "step": 4680
+    },
+    {
+      "epoch": 18.10810810810811,
+      "grad_norm": 0.5670591592788696,
+      "learning_rate": 5.922284804042792e-05,
+      "loss": 0.0865,
+      "step": 4690
+    },
+    {
+      "epoch": 18.146718146718147,
+      "grad_norm": 0.6035292148590088,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.0805,
+      "step": 4700
+    },
+    {
+      "epoch": 18.185328185328185,
+      "grad_norm": 0.6038900017738342,
+      "learning_rate": 5.889762924042585e-05,
+      "loss": 0.0782,
+      "step": 4710
+    },
+    {
+      "epoch": 18.223938223938223,
+      "grad_norm": 0.61341392993927,
+      "learning_rate": 5.873487299644699e-05,
+      "loss": 0.0634,
+      "step": 4720
+    },
+    {
+      "epoch": 18.26254826254826,
+      "grad_norm": 0.3309182822704315,
+      "learning_rate": 5.857202122929649e-05,
+      "loss": 0.074,
+      "step": 4730
+    },
+    {
+      "epoch": 18.301158301158303,
+      "grad_norm": 0.3919869065284729,
+      "learning_rate": 5.840907571989518e-05,
+      "loss": 0.0717,
+      "step": 4740
+    },
+    {
+      "epoch": 18.33976833976834,
+      "grad_norm": 0.47341981530189514,
+      "learning_rate": 5.824603825018904e-05,
+      "loss": 0.0634,
+      "step": 4750
+    },
+    {
+      "epoch": 18.37837837837838,
+      "grad_norm": 0.6534469127655029,
+      "learning_rate": 5.808291060312975e-05,
+      "loss": 0.0782,
+      "step": 4760
+    },
+    {
+      "epoch": 18.416988416988417,
+      "grad_norm": 0.567298412322998,
+      "learning_rate": 5.7919694562655083e-05,
+      "loss": 0.0773,
+      "step": 4770
+    },
+    {
+      "epoch": 18.455598455598455,
+      "grad_norm": 0.4733419120311737,
+      "learning_rate": 5.775639191366954e-05,
+      "loss": 0.0843,
+      "step": 4780
+    },
+    {
+      "epoch": 18.494208494208493,
+      "grad_norm": 0.646246075630188,
+      "learning_rate": 5.75930044420247e-05,
+      "loss": 0.0643,
+      "step": 4790
+    },
+    {
+      "epoch": 18.53281853281853,
+      "grad_norm": 0.596415638923645,
+      "learning_rate": 5.74295339344998e-05,
+      "loss": 0.0781,
+      "step": 4800
+    },
+    {
+      "epoch": 18.571428571428573,
+      "grad_norm": 0.7743390798568726,
+      "learning_rate": 5.726598217878211e-05,
+      "loss": 0.0666,
+      "step": 4810
+    },
+    {
+      "epoch": 18.61003861003861,
+      "grad_norm": 0.5262377858161926,
+      "learning_rate": 5.71023509634474e-05,
+      "loss": 0.075,
+      "step": 4820
+    },
+    {
+      "epoch": 18.64864864864865,
+      "grad_norm": 0.5140449404716492,
+      "learning_rate": 5.693864207794049e-05,
+      "loss": 0.0697,
+      "step": 4830
+    },
+    {
+      "epoch": 18.687258687258687,
+      "grad_norm": 0.5163661241531372,
+      "learning_rate": 5.677485731255545e-05,
+      "loss": 0.0625,
+      "step": 4840
+    },
+    {
+      "epoch": 18.725868725868725,
+      "grad_norm": 0.6141701340675354,
+      "learning_rate": 5.6610998458416296e-05,
+      "loss": 0.0698,
+      "step": 4850
+    },
+    {
+      "epoch": 18.764478764478763,
+      "grad_norm": 0.627004086971283,
+      "learning_rate": 5.644706730745716e-05,
+      "loss": 0.0564,
+      "step": 4860
+    },
+    {
+      "epoch": 18.8030888030888,
+      "grad_norm": 0.522399365901947,
+      "learning_rate": 5.628306565240287e-05,
+      "loss": 0.0681,
+      "step": 4870
+    },
+    {
+      "epoch": 18.841698841698843,
+      "grad_norm": 0.4054034948348999,
+      "learning_rate": 5.611899528674923e-05,
+      "loss": 0.0725,
+      "step": 4880
+    },
+    {
+      "epoch": 18.88030888030888,
+      "grad_norm": 0.5233802795410156,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.091,
+      "step": 4890
+    },
+    {
+      "epoch": 18.91891891891892,
+      "grad_norm": 0.6394193172454834,
+      "learning_rate": 5.579065560136467e-05,
+      "loss": 0.0563,
+      "step": 4900
+    },
+    {
+      "epoch": 18.957528957528957,
+      "grad_norm": 0.48193085193634033,
+      "learning_rate": 5.562638987230392e-05,
+      "loss": 0.0692,
+      "step": 4910
+    },
+    {
+      "epoch": 18.996138996138995,
+      "grad_norm": 0.5722125768661499,
+      "learning_rate": 5.546206261394498e-05,
+      "loss": 0.0779,
+      "step": 4920
+    },
+    {
+      "epoch": 19.034749034749034,
+      "grad_norm": 0.5283990502357483,
+      "learning_rate": 5.529767562334437e-05,
+      "loss": 0.0634,
+      "step": 4930
+    },
+    {
+      "epoch": 19.07335907335907,
+      "grad_norm": 0.3909274935722351,
+      "learning_rate": 5.5133230698211926e-05,
+      "loss": 0.0513,
+      "step": 4940
+    },
+    {
+      "epoch": 19.111969111969113,
+      "grad_norm": 0.44363778829574585,
+      "learning_rate": 5.496872963689096e-05,
+      "loss": 0.0715,
+      "step": 4950
+    },
+    {
+      "epoch": 19.15057915057915,
+      "grad_norm": 0.750079333782196,
+      "learning_rate": 5.4804174238338756e-05,
+      "loss": 0.075,
+      "step": 4960
+    },
+    {
+      "epoch": 19.18918918918919,
+      "grad_norm": 0.42553237080574036,
+      "learning_rate": 5.463956630210678e-05,
+      "loss": 0.0837,
+      "step": 4970
+    },
+    {
+      "epoch": 19.227799227799228,
+      "grad_norm": 0.5493687987327576,
+      "learning_rate": 5.4474907628321046e-05,
+      "loss": 0.0748,
+      "step": 4980
+    },
+    {
+      "epoch": 19.266409266409266,
+      "grad_norm": 0.4793163239955902,
+      "learning_rate": 5.431020001766244e-05,
+      "loss": 0.0763,
+      "step": 4990
+    },
+    {
+      "epoch": 19.305019305019304,
+      "grad_norm": 0.4897761344909668,
+      "learning_rate": 5.4145445271346986e-05,
+      "loss": 0.0827,
+      "step": 5000
+    },
+    {
+      "epoch": 19.343629343629345,
+      "grad_norm": 0.6924915313720703,
+      "learning_rate": 5.398064519110622e-05,
+      "loss": 0.0819,
+      "step": 5010
+    },
+    {
+      "epoch": 19.382239382239383,
+      "grad_norm": 0.69071364402771,
+      "learning_rate": 5.3815801579167394e-05,
+      "loss": 0.0736,
+      "step": 5020
+    },
+    {
+      "epoch": 19.42084942084942,
+      "grad_norm": 0.7378138303756714,
+      "learning_rate": 5.365091623823382e-05,
+      "loss": 0.0587,
+      "step": 5030
+    },
+    {
+      "epoch": 19.45945945945946,
+      "grad_norm": 0.7582262754440308,
+      "learning_rate": 5.348599097146521e-05,
+      "loss": 0.0649,
+      "step": 5040
+    },
+    {
+      "epoch": 19.498069498069498,
+      "grad_norm": 0.5765745639801025,
+      "learning_rate": 5.3321027582457836e-05,
+      "loss": 0.0731,
+      "step": 5050
+    },
+    {
+      "epoch": 19.536679536679536,
+      "grad_norm": 0.47059738636016846,
+      "learning_rate": 5.315602787522491e-05,
+      "loss": 0.0477,
+      "step": 5060
+    },
+    {
+      "epoch": 19.575289575289574,
+      "grad_norm": 0.6599562168121338,
+      "learning_rate": 5.299099365417678e-05,
+      "loss": 0.0912,
+      "step": 5070
+    },
+    {
+      "epoch": 19.613899613899616,
+      "grad_norm": 0.4605643153190613,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.0718,
+      "step": 5080
+    },
+    {
+      "epoch": 19.652509652509654,
+      "grad_norm": 0.5245002508163452,
+      "learning_rate": 5.26608288901438e-05,
+      "loss": 0.0598,
+      "step": 5090
+    },
+    {
+      "epoch": 19.69111969111969,
+      "grad_norm": 0.4711160361766815,
+      "learning_rate": 5.24957019577879e-05,
+      "loss": 0.0575,
+      "step": 5100
+    },
+    {
+      "epoch": 19.72972972972973,
+      "grad_norm": 0.4068230986595154,
+      "learning_rate": 5.2330547732835266e-05,
+      "loss": 0.0659,
+      "step": 5110
+    },
+    {
+      "epoch": 19.768339768339768,
+      "grad_norm": 0.40332168340682983,
+      "learning_rate": 5.2165368021385996e-05,
+      "loss": 0.0796,
+      "step": 5120
+    },
+    {
+      "epoch": 19.806949806949806,
+      "grad_norm": 0.5021756887435913,
+      "learning_rate": 5.200016462981897e-05,
+      "loss": 0.0699,
+      "step": 5130
+    },
+    {
+      "epoch": 19.845559845559844,
+      "grad_norm": 0.6182597279548645,
+      "learning_rate": 5.1834939364772015e-05,
+      "loss": 0.0712,
+      "step": 5140
+    },
+    {
+      "epoch": 19.884169884169886,
+      "grad_norm": 0.499675452709198,
+      "learning_rate": 5.166969403312214e-05,
+      "loss": 0.0626,
+      "step": 5150
+    },
+    {
+      "epoch": 19.922779922779924,
+      "grad_norm": 0.5602287650108337,
+      "learning_rate": 5.1504430441965844e-05,
+      "loss": 0.0869,
+      "step": 5160
+    },
+    {
+      "epoch": 19.961389961389962,
+      "grad_norm": 0.5517196655273438,
+      "learning_rate": 5.133915039859923e-05,
+      "loss": 0.0715,
+      "step": 5170
+    },
+    {
+      "epoch": 20.0,
+      "grad_norm": 0.3541000485420227,
+      "learning_rate": 5.1173855710498444e-05,
+      "loss": 0.0529,
+      "step": 5180
+    },
+    {
+      "epoch": 20.038610038610038,
+      "grad_norm": 0.40408554673194885,
+      "learning_rate": 5.100854818529967e-05,
+      "loss": 0.0846,
+      "step": 5190
+    },
+    {
+      "epoch": 20.077220077220076,
+      "grad_norm": 0.49958834052085876,
+      "learning_rate": 5.084322963077951e-05,
+      "loss": 0.0501,
+      "step": 5200
+    },
+    {
+      "epoch": 20.115830115830114,
+      "grad_norm": 0.4283312261104584,
+      "learning_rate": 5.067790185483522e-05,
+      "loss": 0.0738,
+      "step": 5210
+    },
+    {
+      "epoch": 20.154440154440156,
+      "grad_norm": 0.6037362813949585,
+      "learning_rate": 5.0512566665464844e-05,
+      "loss": 0.0698,
+      "step": 5220
+    },
+    {
+      "epoch": 20.193050193050194,
+      "grad_norm": 0.4373033046722412,
+      "learning_rate": 5.034722587074755e-05,
+      "loss": 0.0613,
+      "step": 5230
+    },
+    {
+      "epoch": 20.231660231660232,
+      "grad_norm": 0.3218035101890564,
+      "learning_rate": 5.018188127882375e-05,
+      "loss": 0.0603,
+      "step": 5240
+    },
+    {
+      "epoch": 20.27027027027027,
+      "grad_norm": 0.6302839517593384,
+      "learning_rate": 5.0016534697875417e-05,
+      "loss": 0.082,
+      "step": 5250
+    },
+    {
+      "epoch": 20.30888030888031,
+      "grad_norm": 0.580318033695221,
+      "learning_rate": 4.9851187936106294e-05,
+      "loss": 0.0647,
+      "step": 5260
+    },
+    {
+      "epoch": 20.347490347490346,
+      "grad_norm": 0.7693837881088257,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.073,
+      "step": 5270
+    },
+    {
+      "epoch": 20.386100386100384,
+      "grad_norm": 0.437863826751709,
+      "learning_rate": 4.95205011029106e-05,
+      "loss": 0.0761,
+      "step": 5280
+    },
+    {
+      "epoch": 20.424710424710426,
+      "grad_norm": 0.5829715728759766,
+      "learning_rate": 4.935516464782227e-05,
+      "loss": 0.0636,
+      "step": 5290
+    },
+    {
+      "epoch": 20.463320463320464,
+      "grad_norm": 0.652866780757904,
+      "learning_rate": 4.918983524455003e-05,
+      "loss": 0.0543,
+      "step": 5300
+    },
+    {
+      "epoch": 20.501930501930502,
+      "grad_norm": 0.6355207562446594,
+      "learning_rate": 4.9024514701109766e-05,
+      "loss": 0.0552,
+      "step": 5310
+    },
+    {
+      "epoch": 20.54054054054054,
+      "grad_norm": 0.6155616641044617,
+      "learning_rate": 4.885920482542043e-05,
+      "loss": 0.0681,
+      "step": 5320
+    },
+    {
+      "epoch": 20.57915057915058,
+      "grad_norm": 0.5970525145530701,
+      "learning_rate": 4.869390742528438e-05,
+      "loss": 0.0639,
+      "step": 5330
+    },
+    {
+      "epoch": 20.617760617760617,
+      "grad_norm": 0.3544915020465851,
+      "learning_rate": 4.852862430836744e-05,
+      "loss": 0.0583,
+      "step": 5340
+    },
+    {
+      "epoch": 20.656370656370655,
+      "grad_norm": 0.3478948175907135,
+      "learning_rate": 4.836335728217933e-05,
+      "loss": 0.0491,
+      "step": 5350
+    },
+    {
+      "epoch": 20.694980694980696,
+      "grad_norm": 0.42770978808403015,
+      "learning_rate": 4.819810815405379e-05,
+      "loss": 0.0635,
+      "step": 5360
+    },
+    {
+      "epoch": 20.733590733590734,
+      "grad_norm": 0.49829915165901184,
+      "learning_rate": 4.803287873112877e-05,
+      "loss": 0.0494,
+      "step": 5370
+    },
+    {
+      "epoch": 20.772200772200772,
+      "grad_norm": 0.6188846826553345,
+      "learning_rate": 4.786767082032681e-05,
+      "loss": 0.0536,
+      "step": 5380
+    },
+    {
+      "epoch": 20.81081081081081,
+      "grad_norm": 0.49089109897613525,
+      "learning_rate": 4.77024862283351e-05,
+      "loss": 0.058,
+      "step": 5390
+    },
+    {
+      "epoch": 20.84942084942085,
+      "grad_norm": 0.5045559406280518,
+      "learning_rate": 4.753732676158593e-05,
+      "loss": 0.0693,
+      "step": 5400
+    },
+    {
+      "epoch": 20.888030888030887,
+      "grad_norm": 0.4698314964771271,
+      "learning_rate": 4.737219422623672e-05,
+      "loss": 0.0632,
+      "step": 5410
+    },
+    {
+      "epoch": 20.92664092664093,
+      "grad_norm": 0.37180498242378235,
+      "learning_rate": 4.720709042815044e-05,
+      "loss": 0.0576,
+      "step": 5420
+    },
+    {
+      "epoch": 20.965250965250966,
+      "grad_norm": 0.5959806442260742,
+      "learning_rate": 4.704201717287578e-05,
+      "loss": 0.0604,
+      "step": 5430
+    },
+    {
+      "epoch": 21.003861003861005,
+      "grad_norm": 0.3700093626976013,
+      "learning_rate": 4.6876976265627404e-05,
+      "loss": 0.0819,
+      "step": 5440
+    },
+    {
+      "epoch": 21.042471042471043,
+      "grad_norm": 0.5958386659622192,
+      "learning_rate": 4.671196951126626e-05,
+      "loss": 0.0427,
+      "step": 5450
+    },
+    {
+      "epoch": 21.08108108108108,
+      "grad_norm": 0.47188055515289307,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.0518,
+      "step": 5460
+    },
+    {
+      "epoch": 21.11969111969112,
+      "grad_norm": 0.3665406107902527,
+      "learning_rate": 4.6382065678762034e-05,
+      "loss": 0.0557,
+      "step": 5470
+    },
+    {
+      "epoch": 21.158301158301157,
+      "grad_norm": 0.49395236372947693,
+      "learning_rate": 4.6217172208394424e-05,
+      "loss": 0.0634,
+      "step": 5480
+    },
+    {
+      "epoch": 21.1969111969112,
+      "grad_norm": 0.46031197905540466,
+      "learning_rate": 4.605232010642549e-05,
+      "loss": 0.0617,
+      "step": 5490
+    },
+    {
+      "epoch": 21.235521235521237,
+      "grad_norm": 0.46037164330482483,
+      "learning_rate": 4.588751117565142e-05,
+      "loss": 0.0635,
+      "step": 5500
+    },
+    {
+      "epoch": 21.274131274131275,
+      "grad_norm": 0.604911744594574,
+      "learning_rate": 4.5722747218396214e-05,
+      "loss": 0.0651,
+      "step": 5510
+    },
+    {
+      "epoch": 21.312741312741313,
+      "grad_norm": 0.4139896333217621,
+      "learning_rate": 4.5558030036492194e-05,
+      "loss": 0.0607,
+      "step": 5520
+    },
+    {
+      "epoch": 21.35135135135135,
+      "grad_norm": 0.3025991916656494,
+      "learning_rate": 4.539336143125999e-05,
+      "loss": 0.062,
+      "step": 5530
+    },
+    {
+      "epoch": 21.38996138996139,
+      "grad_norm": 0.32330894470214844,
+      "learning_rate": 4.522874320348916e-05,
+      "loss": 0.0513,
+      "step": 5540
+    },
+    {
+      "epoch": 21.428571428571427,
+      "grad_norm": 0.5037228465080261,
+      "learning_rate": 4.506417715341821e-05,
+      "loss": 0.0568,
+      "step": 5550
+    },
+    {
+      "epoch": 21.46718146718147,
+      "grad_norm": 0.5318952798843384,
+      "learning_rate": 4.489966508071511e-05,
+      "loss": 0.0523,
+      "step": 5560
+    },
+    {
+      "epoch": 21.505791505791507,
+      "grad_norm": 0.6271405816078186,
+      "learning_rate": 4.4735208784457575e-05,
+      "loss": 0.0713,
+      "step": 5570
+    },
+    {
+      "epoch": 21.544401544401545,
+      "grad_norm": 0.5892740488052368,
+      "learning_rate": 4.457081006311325e-05,
+      "loss": 0.0797,
+      "step": 5580
+    },
+    {
+      "epoch": 21.583011583011583,
+      "grad_norm": 0.516455352306366,
+      "learning_rate": 4.440647071452027e-05,
+      "loss": 0.0577,
+      "step": 5590
+    },
+    {
+      "epoch": 21.62162162162162,
+      "grad_norm": 0.5120159983634949,
+      "learning_rate": 4.424219253586737e-05,
+      "loss": 0.0648,
+      "step": 5600
+    },
+    {
+      "epoch": 21.66023166023166,
+      "grad_norm": 0.37961456179618835,
+      "learning_rate": 4.407797732367443e-05,
+      "loss": 0.0609,
+      "step": 5610
+    },
+    {
+      "epoch": 21.698841698841697,
+      "grad_norm": 0.48860493302345276,
+      "learning_rate": 4.391382687377268e-05,
+      "loss": 0.0488,
+      "step": 5620
+    },
+    {
+      "epoch": 21.73745173745174,
+      "grad_norm": 0.24185198545455933,
+      "learning_rate": 4.374974298128512e-05,
+      "loss": 0.0523,
+      "step": 5630
+    },
+    {
+      "epoch": 21.776061776061777,
+      "grad_norm": 0.5195246934890747,
+      "learning_rate": 4.358572744060699e-05,
+      "loss": 0.0579,
+      "step": 5640
+    },
+    {
+      "epoch": 21.814671814671815,
+      "grad_norm": 0.2884815037250519,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.0711,
+      "step": 5650
+    },
+    {
+      "epoch": 21.853281853281853,
+      "grad_norm": 0.507343053817749,
+      "learning_rate": 4.325790858850241e-05,
+      "loss": 0.0621,
+      "step": 5660
+    },
+    {
+      "epoch": 21.89189189189189,
+      "grad_norm": 0.5438768863677979,
+      "learning_rate": 4.309410886205043e-05,
+      "loss": 0.0608,
+      "step": 5670
+    },
+    {
+      "epoch": 21.93050193050193,
+      "grad_norm": 0.4898863434791565,
+      "learning_rate": 4.293038465731752e-05,
+      "loss": 0.0642,
+      "step": 5680
+    },
+    {
+      "epoch": 21.969111969111967,
+      "grad_norm": 0.4932737946510315,
+      "learning_rate": 4.276673776476533e-05,
+      "loss": 0.0634,
+      "step": 5690
+    },
+    {
+      "epoch": 22.00772200772201,
+      "grad_norm": 0.32997044920921326,
+      "learning_rate": 4.260316997401007e-05,
+      "loss": 0.0469,
+      "step": 5700
+    },
+    {
+      "epoch": 22.046332046332047,
+      "grad_norm": 0.338538259267807,
+      "learning_rate": 4.243968307380293e-05,
+      "loss": 0.0617,
+      "step": 5710
+    },
+    {
+      "epoch": 22.084942084942085,
+      "grad_norm": 0.49940410256385803,
+      "learning_rate": 4.22762788520104e-05,
+      "loss": 0.0522,
+      "step": 5720
+    },
+    {
+      "epoch": 22.123552123552123,
+      "grad_norm": 0.4403421878814697,
+      "learning_rate": 4.211295909559491e-05,
+      "loss": 0.0621,
+      "step": 5730
+    },
+    {
+      "epoch": 22.16216216216216,
+      "grad_norm": 0.38823920488357544,
+      "learning_rate": 4.194972559059511e-05,
+      "loss": 0.0732,
+      "step": 5740
+    },
+    {
+      "epoch": 22.2007722007722,
+      "grad_norm": 0.36218559741973877,
+      "learning_rate": 4.178658012210651e-05,
+      "loss": 0.0517,
+      "step": 5750
+    },
+    {
+      "epoch": 22.23938223938224,
+      "grad_norm": 0.34942883253097534,
+      "learning_rate": 4.162352447426177e-05,
+      "loss": 0.058,
+      "step": 5760
+    },
+    {
+      "epoch": 22.27799227799228,
+      "grad_norm": 0.505549967288971,
+      "learning_rate": 4.146056043021135e-05,
+      "loss": 0.0633,
+      "step": 5770
+    },
+    {
+      "epoch": 22.316602316602317,
+      "grad_norm": 0.5539968013763428,
+      "learning_rate": 4.1297689772103944e-05,
+      "loss": 0.0809,
+      "step": 5780
+    },
+    {
+      "epoch": 22.355212355212355,
+      "grad_norm": 0.5466805100440979,
+      "learning_rate": 4.113491428106694e-05,
+      "loss": 0.0529,
+      "step": 5790
+    },
+    {
+      "epoch": 22.393822393822393,
+      "grad_norm": 0.38340243697166443,
+      "learning_rate": 4.0972235737187055e-05,
+      "loss": 0.0687,
+      "step": 5800
+    },
+    {
+      "epoch": 22.43243243243243,
+      "grad_norm": 0.39504724740982056,
+      "learning_rate": 4.080965591949076e-05,
+      "loss": 0.0521,
+      "step": 5810
+    },
+    {
+      "epoch": 22.47104247104247,
+      "grad_norm": 0.37831607460975647,
+      "learning_rate": 4.0647176605924924e-05,
+      "loss": 0.0546,
+      "step": 5820
+    },
+    {
+      "epoch": 22.50965250965251,
+      "grad_norm": 0.5784722566604614,
+      "learning_rate": 4.0484799573337255e-05,
+      "loss": 0.0693,
+      "step": 5830
+    },
+    {
+      "epoch": 22.54826254826255,
+      "grad_norm": 0.4978543221950531,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.0557,
+      "step": 5840
+    },
+    {
+      "epoch": 22.586872586872587,
+      "grad_norm": 0.3099800646305084,
+      "learning_rate": 4.016035945287539e-05,
+      "loss": 0.0458,
+      "step": 5850
+    },
+    {
+      "epoch": 22.625482625482626,
+      "grad_norm": 0.6736605167388916,
+      "learning_rate": 3.999829991302635e-05,
+      "loss": 0.0687,
+      "step": 5860
+    },
+    {
+      "epoch": 22.664092664092664,
+      "grad_norm": 0.45186126232147217,
+      "learning_rate": 3.983634975016707e-05,
+      "loss": 0.0646,
+      "step": 5870
+    },
+    {
+      "epoch": 22.7027027027027,
+      "grad_norm": 0.5107584595680237,
+      "learning_rate": 3.967451073535854e-05,
+      "loss": 0.0666,
+      "step": 5880
+    },
+    {
+      "epoch": 22.74131274131274,
+      "grad_norm": 0.45037683844566345,
+      "learning_rate": 3.951278463844633e-05,
+      "loss": 0.0509,
+      "step": 5890
+    },
+    {
+      "epoch": 22.77992277992278,
+      "grad_norm": 0.3888896405696869,
+      "learning_rate": 3.935117322804111e-05,
+      "loss": 0.0731,
+      "step": 5900
+    },
+    {
+      "epoch": 22.81853281853282,
+      "grad_norm": 0.3771021366119385,
+      "learning_rate": 3.918967827149938e-05,
+      "loss": 0.0702,
+      "step": 5910
+    },
+    {
+      "epoch": 22.857142857142858,
+      "grad_norm": 0.5772953033447266,
+      "learning_rate": 3.9028301534904094e-05,
+      "loss": 0.0602,
+      "step": 5920
+    },
+    {
+      "epoch": 22.895752895752896,
+      "grad_norm": 0.40094444155693054,
+      "learning_rate": 3.88670447830454e-05,
+      "loss": 0.0582,
+      "step": 5930
+    },
+    {
+      "epoch": 22.934362934362934,
+      "grad_norm": 0.3225739598274231,
+      "learning_rate": 3.870590977940132e-05,
+      "loss": 0.0542,
+      "step": 5940
+    },
+    {
+      "epoch": 22.972972972972972,
+      "grad_norm": 0.7332587242126465,
+      "learning_rate": 3.8544898286118404e-05,
+      "loss": 0.0597,
+      "step": 5950
+    },
+    {
+      "epoch": 23.01158301158301,
+      "grad_norm": 0.44024458527565,
+      "learning_rate": 3.838401206399257e-05,
+      "loss": 0.0497,
+      "step": 5960
+    },
+    {
+      "epoch": 23.05019305019305,
+      "grad_norm": 0.4732349216938019,
+      "learning_rate": 3.822325287244975e-05,
+      "loss": 0.0578,
+      "step": 5970
+    },
+    {
+      "epoch": 23.08880308880309,
+      "grad_norm": 0.4735555350780487,
+      "learning_rate": 3.8062622469526725e-05,
+      "loss": 0.0615,
+      "step": 5980
+    },
+    {
+      "epoch": 23.127413127413128,
+      "grad_norm": 0.6857978105545044,
+      "learning_rate": 3.790212261185183e-05,
+      "loss": 0.0718,
+      "step": 5990
+    },
+    {
+      "epoch": 23.166023166023166,
+      "grad_norm": 0.3587701916694641,
+      "learning_rate": 3.7741755054625794e-05,
+      "loss": 0.085,
+      "step": 6000
+    },
+    {
+      "epoch": 23.204633204633204,
+      "grad_norm": 0.40470054745674133,
+      "learning_rate": 3.758152155160255e-05,
+      "loss": 0.0608,
+      "step": 6010
+    },
+    {
+      "epoch": 23.243243243243242,
+      "grad_norm": 0.5807958841323853,
+      "learning_rate": 3.742142385506999e-05,
+      "loss": 0.0606,
+      "step": 6020
+    },
+    {
+      "epoch": 23.28185328185328,
+      "grad_norm": 0.3811424672603607,
+      "learning_rate": 3.72614637158309e-05,
+      "loss": 0.0594,
+      "step": 6030
+    },
+    {
+      "epoch": 23.320463320463322,
+      "grad_norm": 0.44823238253593445,
+      "learning_rate": 3.710164288318371e-05,
+      "loss": 0.0624,
+      "step": 6040
+    },
+    {
+      "epoch": 23.35907335907336,
+      "grad_norm": 0.5011805295944214,
+      "learning_rate": 3.694196310490345e-05,
+      "loss": 0.069,
+      "step": 6050
+    },
+    {
+      "epoch": 23.397683397683398,
+      "grad_norm": 0.5222461223602295,
+      "learning_rate": 3.678242612722259e-05,
+      "loss": 0.0462,
+      "step": 6060
+    },
+    {
+      "epoch": 23.436293436293436,
+      "grad_norm": 0.41890570521354675,
+      "learning_rate": 3.6623033694811953e-05,
+      "loss": 0.0552,
+      "step": 6070
+    },
+    {
+      "epoch": 23.474903474903474,
+      "grad_norm": 0.6486966609954834,
+      "learning_rate": 3.6463787550761665e-05,
+      "loss": 0.0627,
+      "step": 6080
+    },
+    {
+      "epoch": 23.513513513513512,
+      "grad_norm": 0.46851104497909546,
+      "learning_rate": 3.630468943656202e-05,
+      "loss": 0.0536,
+      "step": 6090
+    },
+    {
+      "epoch": 23.552123552123554,
+      "grad_norm": 0.5276613831520081,
+      "learning_rate": 3.6145741092084523e-05,
+      "loss": 0.0666,
+      "step": 6100
+    },
+    {
+      "epoch": 23.590733590733592,
+      "grad_norm": 0.341413289308548,
+      "learning_rate": 3.598694425556278e-05,
+      "loss": 0.0417,
+      "step": 6110
+    },
+    {
+      "epoch": 23.62934362934363,
+      "grad_norm": 0.5024414658546448,
+      "learning_rate": 3.58283006635736e-05,
+      "loss": 0.0511,
+      "step": 6120
+    },
+    {
+      "epoch": 23.667953667953668,
+      "grad_norm": 0.27348798513412476,
+      "learning_rate": 3.566981205101781e-05,
+      "loss": 0.0405,
+      "step": 6130
+    },
+    {
+      "epoch": 23.706563706563706,
+      "grad_norm": 0.43044814467430115,
+      "learning_rate": 3.5511480151101556e-05,
+      "loss": 0.0548,
+      "step": 6140
+    },
+    {
+      "epoch": 23.745173745173744,
+      "grad_norm": 0.28613966703414917,
+      "learning_rate": 3.5353306695317104e-05,
+      "loss": 0.0755,
+      "step": 6150
+    },
+    {
+      "epoch": 23.783783783783782,
+      "grad_norm": 0.5436369180679321,
+      "learning_rate": 3.519529341342402e-05,
+      "loss": 0.071,
+      "step": 6160
+    },
+    {
+      "epoch": 23.822393822393824,
+      "grad_norm": 0.5096984505653381,
+      "learning_rate": 3.503744203343026e-05,
+      "loss": 0.0622,
+      "step": 6170
+    },
+    {
+      "epoch": 23.861003861003862,
+      "grad_norm": 0.49167925119400024,
+      "learning_rate": 3.487975428157318e-05,
+      "loss": 0.0551,
+      "step": 6180
+    },
+    {
+      "epoch": 23.8996138996139,
+      "grad_norm": 0.35319849848747253,
+      "learning_rate": 3.472223188230083e-05,
+      "loss": 0.0672,
+      "step": 6190
+    },
+    {
+      "epoch": 23.93822393822394,
+      "grad_norm": 0.2771577537059784,
+      "learning_rate": 3.4564876558252866e-05,
+      "loss": 0.0439,
+      "step": 6200
+    },
+    {
+      "epoch": 23.976833976833976,
+      "grad_norm": 0.49233347177505493,
+      "learning_rate": 3.440769003024195e-05,
+      "loss": 0.0633,
+      "step": 6210
+    },
+    {
+      "epoch": 24.015444015444015,
+      "grad_norm": 0.5419521927833557,
+      "learning_rate": 3.425067401723477e-05,
+      "loss": 0.0461,
+      "step": 6220
+    },
+    {
+      "epoch": 24.054054054054053,
+      "grad_norm": 0.37961897253990173,
+      "learning_rate": 3.409383023633325e-05,
+      "loss": 0.0504,
+      "step": 6230
+    },
+    {
+      "epoch": 24.092664092664094,
+      "grad_norm": 0.5190012454986572,
+      "learning_rate": 3.3937160402755894e-05,
+      "loss": 0.0507,
+      "step": 6240
+    },
+    {
+      "epoch": 24.131274131274132,
+      "grad_norm": 0.38911232352256775,
+      "learning_rate": 3.378066622981885e-05,
+      "loss": 0.0677,
+      "step": 6250
+    },
+    {
+      "epoch": 24.16988416988417,
+      "grad_norm": 0.3622760474681854,
+      "learning_rate": 3.362434942891738e-05,
+      "loss": 0.0525,
+      "step": 6260
+    },
+    {
+      "epoch": 24.20849420849421,
+      "grad_norm": 0.5027076601982117,
+      "learning_rate": 3.346821170950693e-05,
+      "loss": 0.05,
+      "step": 6270
+    },
+    {
+      "epoch": 24.247104247104247,
+      "grad_norm": 0.4265000820159912,
+      "learning_rate": 3.3312254779084585e-05,
+      "loss": 0.0543,
+      "step": 6280
+    },
+    {
+      "epoch": 24.285714285714285,
+      "grad_norm": 0.34943899512290955,
+      "learning_rate": 3.315648034317039e-05,
+      "loss": 0.0539,
+      "step": 6290
+    },
+    {
+      "epoch": 24.324324324324323,
+      "grad_norm": 0.6790931224822998,
+      "learning_rate": 3.3000890105288564e-05,
+      "loss": 0.06,
+      "step": 6300
+    },
+    {
+      "epoch": 24.362934362934364,
+      "grad_norm": 0.45015794038772583,
+      "learning_rate": 3.284548576694908e-05,
+      "loss": 0.0461,
+      "step": 6310
+    },
+    {
+      "epoch": 24.401544401544403,
+      "grad_norm": 0.5135745406150818,
+      "learning_rate": 3.2690269027628815e-05,
+      "loss": 0.0569,
+      "step": 6320
+    },
+    {
+      "epoch": 24.44015444015444,
+      "grad_norm": 0.6036981344223022,
+      "learning_rate": 3.253524158475324e-05,
+      "loss": 0.05,
+      "step": 6330
+    },
+    {
+      "epoch": 24.47876447876448,
+      "grad_norm": 0.5034620761871338,
+      "learning_rate": 3.238040513367757e-05,
+      "loss": 0.0585,
+      "step": 6340
+    },
+    {
+      "epoch": 24.517374517374517,
+      "grad_norm": 0.4371446967124939,
+      "learning_rate": 3.222576136766843e-05,
+      "loss": 0.0637,
+      "step": 6350
+    },
+    {
+      "epoch": 24.555984555984555,
+      "grad_norm": 0.40289029479026794,
+      "learning_rate": 3.2071311977885324e-05,
+      "loss": 0.0559,
+      "step": 6360
+    },
+    {
+      "epoch": 24.594594594594593,
+      "grad_norm": 0.5056316256523132,
+      "learning_rate": 3.191705865336197e-05,
+      "loss": 0.0456,
+      "step": 6370
+    },
+    {
+      "epoch": 24.633204633204635,
+      "grad_norm": 0.5891019701957703,
+      "learning_rate": 3.1763003080988075e-05,
+      "loss": 0.0464,
+      "step": 6380
+    },
+    {
+      "epoch": 24.671814671814673,
+      "grad_norm": 0.6418840885162354,
+      "learning_rate": 3.160914694549063e-05,
+      "loss": 0.057,
+      "step": 6390
+    },
+    {
+      "epoch": 24.71042471042471,
+      "grad_norm": 0.34795257449150085,
+      "learning_rate": 3.145549192941573e-05,
+      "loss": 0.0554,
+      "step": 6400
+    },
+    {
+      "epoch": 24.74903474903475,
+      "grad_norm": 0.43306779861450195,
+      "learning_rate": 3.130203971310999e-05,
+      "loss": 0.0379,
+      "step": 6410
+    },
+    {
+      "epoch": 24.787644787644787,
+      "grad_norm": 0.6308071613311768,
+      "learning_rate": 3.114879197470225e-05,
+      "loss": 0.0525,
+      "step": 6420
+    },
+    {
+      "epoch": 24.826254826254825,
+      "grad_norm": 0.37014514207839966,
+      "learning_rate": 3.0995750390085285e-05,
+      "loss": 0.0462,
+      "step": 6430
+    },
+    {
+      "epoch": 24.864864864864863,
+      "grad_norm": 0.45822596549987793,
+      "learning_rate": 3.084291663289728e-05,
+      "loss": 0.0506,
+      "step": 6440
+    },
+    {
+      "epoch": 24.903474903474905,
+      "grad_norm": 0.505630612373352,
+      "learning_rate": 3.069029237450375e-05,
+      "loss": 0.047,
+      "step": 6450
+    },
+    {
+      "epoch": 24.942084942084943,
+      "grad_norm": 0.39040684700012207,
+      "learning_rate": 3.053787928397911e-05,
+      "loss": 0.0436,
+      "step": 6460
+    },
+    {
+      "epoch": 24.98069498069498,
+      "grad_norm": 0.42719799280166626,
+      "learning_rate": 3.0385679028088526e-05,
+      "loss": 0.0472,
+      "step": 6470
+    },
+    {
+      "epoch": 25.01930501930502,
+      "grad_norm": 0.3602113425731659,
+      "learning_rate": 3.023369327126959e-05,
+      "loss": 0.0374,
+      "step": 6480
+    },
+    {
+      "epoch": 25.057915057915057,
+      "grad_norm": 0.6470438838005066,
+      "learning_rate": 3.0081923675614198e-05,
+      "loss": 0.0607,
+      "step": 6490
+    },
+    {
+      "epoch": 25.096525096525095,
+      "grad_norm": 0.5232688188552856,
+      "learning_rate": 2.993037190085034e-05,
+      "loss": 0.0421,
+      "step": 6500
+    },
+    {
+      "epoch": 25.135135135135137,
+      "grad_norm": 0.34110620617866516,
+      "learning_rate": 2.977903960432392e-05,
+      "loss": 0.0414,
+      "step": 6510
+    },
+    {
+      "epoch": 25.173745173745175,
+      "grad_norm": 0.5356219410896301,
+      "learning_rate": 2.9627928440980722e-05,
+      "loss": 0.0556,
+      "step": 6520
+    },
+    {
+      "epoch": 25.212355212355213,
+      "grad_norm": 0.4498227834701538,
+      "learning_rate": 2.9477040063348183e-05,
+      "loss": 0.05,
+      "step": 6530
+    },
+    {
+      "epoch": 25.25096525096525,
+      "grad_norm": 0.41610923409461975,
+      "learning_rate": 2.9326376121517456e-05,
+      "loss": 0.0557,
+      "step": 6540
+    },
+    {
+      "epoch": 25.28957528957529,
+      "grad_norm": 0.5128653645515442,
+      "learning_rate": 2.9175938263125236e-05,
+      "loss": 0.0476,
+      "step": 6550
+    },
+    {
+      "epoch": 25.328185328185327,
+      "grad_norm": 0.5739153027534485,
+      "learning_rate": 2.9025728133335873e-05,
+      "loss": 0.0609,
+      "step": 6560
+    },
+    {
+      "epoch": 25.366795366795365,
+      "grad_norm": 0.5832862854003906,
+      "learning_rate": 2.8875747374823288e-05,
+      "loss": 0.0524,
+      "step": 6570
+    },
+    {
+      "epoch": 25.405405405405407,
+      "grad_norm": 0.7269383072853088,
+      "learning_rate": 2.872599762775298e-05,
+      "loss": 0.0458,
+      "step": 6580
+    },
+    {
+      "epoch": 25.444015444015445,
+      "grad_norm": 0.44224220514297485,
+      "learning_rate": 2.857648052976425e-05,
+      "loss": 0.0604,
+      "step": 6590
+    },
+    {
+      "epoch": 25.482625482625483,
+      "grad_norm": 0.6164863705635071,
+      "learning_rate": 2.8427197715952047e-05,
+      "loss": 0.0621,
+      "step": 6600
+    },
+    {
+      "epoch": 25.52123552123552,
+      "grad_norm": 0.6670511364936829,
+      "learning_rate": 2.8278150818849393e-05,
+      "loss": 0.0456,
+      "step": 6610
+    },
+    {
+      "epoch": 25.55984555984556,
+      "grad_norm": 0.3655743896961212,
+      "learning_rate": 2.812934146840922e-05,
+      "loss": 0.0463,
+      "step": 6620
+    },
+    {
+      "epoch": 25.598455598455597,
+      "grad_norm": 0.6558657288551331,
+      "learning_rate": 2.7980771291986764e-05,
+      "loss": 0.0405,
+      "step": 6630
+    },
+    {
+      "epoch": 25.637065637065636,
+      "grad_norm": 0.5198159217834473,
+      "learning_rate": 2.783244191432167e-05,
+      "loss": 0.0389,
+      "step": 6640
+    },
+    {
+      "epoch": 25.675675675675677,
+      "grad_norm": 0.4935913383960724,
+      "learning_rate": 2.768435495752022e-05,
+      "loss": 0.0693,
+      "step": 6650
+    },
+    {
+      "epoch": 25.714285714285715,
+      "grad_norm": 0.4455951750278473,
+      "learning_rate": 2.753651204103771e-05,
+      "loss": 0.0338,
+      "step": 6660
+    },
+    {
+      "epoch": 25.752895752895753,
+      "grad_norm": 0.33703893423080444,
+      "learning_rate": 2.7388914781660523e-05,
+      "loss": 0.0427,
+      "step": 6670
+    },
+    {
+      "epoch": 25.79150579150579,
+      "grad_norm": 0.41062918305397034,
+      "learning_rate": 2.7241564793488693e-05,
+      "loss": 0.0548,
+      "step": 6680
+    },
+    {
+      "epoch": 25.83011583011583,
+      "grad_norm": 0.42094191908836365,
+      "learning_rate": 2.7094463687918037e-05,
+      "loss": 0.059,
+      "step": 6690
+    },
+    {
+      "epoch": 25.868725868725868,
+      "grad_norm": 0.635362446308136,
+      "learning_rate": 2.694761307362268e-05,
+      "loss": 0.0473,
+      "step": 6700
+    },
+    {
+      "epoch": 25.907335907335906,
+      "grad_norm": 0.5083866715431213,
+      "learning_rate": 2.6801014556537467e-05,
+      "loss": 0.0527,
+      "step": 6710
+    },
+    {
+      "epoch": 25.945945945945947,
+      "grad_norm": 0.4608486592769623,
+      "learning_rate": 2.6654669739840243e-05,
+      "loss": 0.0644,
+      "step": 6720
+    },
+    {
+      "epoch": 25.984555984555985,
+      "grad_norm": 0.5259851813316345,
+      "learning_rate": 2.650858022393451e-05,
+      "loss": 0.0507,
+      "step": 6730
+    },
+    {
+      "epoch": 26.023166023166024,
+      "grad_norm": 0.35150691866874695,
+      "learning_rate": 2.6362747606431747e-05,
+      "loss": 0.0622,
+      "step": 6740
+    },
+    {
+      "epoch": 26.06177606177606,
+      "grad_norm": 0.4495721459388733,
+      "learning_rate": 2.6217173482134172e-05,
+      "loss": 0.0443,
+      "step": 6750
+    },
+    {
+      "epoch": 26.1003861003861,
+      "grad_norm": 0.29612305760383606,
+      "learning_rate": 2.6071859443017044e-05,
+      "loss": 0.0425,
+      "step": 6760
+    },
+    {
+      "epoch": 26.138996138996138,
+      "grad_norm": 0.37452226877212524,
+      "learning_rate": 2.5926807078211414e-05,
+      "loss": 0.043,
+      "step": 6770
+    },
+    {
+      "epoch": 26.177606177606176,
+      "grad_norm": 0.5638911128044128,
+      "learning_rate": 2.5782017973986728e-05,
+      "loss": 0.0727,
+      "step": 6780
+    },
+    {
+      "epoch": 26.216216216216218,
+      "grad_norm": 0.3986194133758545,
+      "learning_rate": 2.5637493713733374e-05,
+      "loss": 0.0346,
+      "step": 6790
+    },
+    {
+      "epoch": 26.254826254826256,
+      "grad_norm": 1.514148473739624,
+      "learning_rate": 2.549323587794559e-05,
+      "loss": 0.0596,
+      "step": 6800
+    },
+    {
+      "epoch": 26.293436293436294,
+      "grad_norm": 0.2974415123462677,
+      "learning_rate": 2.5349246044203895e-05,
+      "loss": 0.0349,
+      "step": 6810
+    },
+    {
+      "epoch": 26.332046332046332,
+      "grad_norm": 0.4633919894695282,
+      "learning_rate": 2.520552578715808e-05,
+      "loss": 0.0632,
+      "step": 6820
+    },
+    {
+      "epoch": 26.37065637065637,
+      "grad_norm": 0.5830386877059937,
+      "learning_rate": 2.506207667850981e-05,
+      "loss": 0.0486,
+      "step": 6830
+    },
+    {
+      "epoch": 26.409266409266408,
+      "grad_norm": 0.5291124582290649,
+      "learning_rate": 2.4918900286995555e-05,
+      "loss": 0.0613,
+      "step": 6840
+    },
+    {
+      "epoch": 26.447876447876446,
+      "grad_norm": 0.3293599784374237,
+      "learning_rate": 2.4775998178369458e-05,
+      "loss": 0.0446,
+      "step": 6850
+    },
+    {
+      "epoch": 26.486486486486488,
+      "grad_norm": 0.4320952892303467,
+      "learning_rate": 2.4633371915386017e-05,
+      "loss": 0.0363,
+      "step": 6860
+    },
+    {
+      "epoch": 26.525096525096526,
+      "grad_norm": 0.4750666916370392,
+      "learning_rate": 2.4491023057783235e-05,
+      "loss": 0.0655,
+      "step": 6870
+    },
+    {
+      "epoch": 26.563706563706564,
+      "grad_norm": 0.47334954142570496,
+      "learning_rate": 2.4348953162265375e-05,
+      "loss": 0.0547,
+      "step": 6880
+    },
+    {
+      "epoch": 26.602316602316602,
+      "grad_norm": 0.35568034648895264,
+      "learning_rate": 2.420716378248607e-05,
+      "loss": 0.0424,
+      "step": 6890
+    },
+    {
+      "epoch": 26.64092664092664,
+      "grad_norm": 0.5133526921272278,
+      "learning_rate": 2.4065656469031266e-05,
+      "loss": 0.0598,
+      "step": 6900
+    },
+    {
+      "epoch": 26.679536679536678,
+      "grad_norm": 0.5015846490859985,
+      "learning_rate": 2.3924432769402268e-05,
+      "loss": 0.0381,
+      "step": 6910
+    },
+    {
+      "epoch": 26.71814671814672,
+      "grad_norm": 0.5141738057136536,
+      "learning_rate": 2.3783494227998844e-05,
+      "loss": 0.0587,
+      "step": 6920
+    },
+    {
+      "epoch": 26.756756756756758,
+      "grad_norm": 0.399676114320755,
+      "learning_rate": 2.3642842386102264e-05,
+      "loss": 0.0483,
+      "step": 6930
+    },
+    {
+      "epoch": 26.795366795366796,
+      "grad_norm": 0.6744953989982605,
+      "learning_rate": 2.3502478781858567e-05,
+      "loss": 0.0374,
+      "step": 6940
+    },
+    {
+      "epoch": 26.833976833976834,
+      "grad_norm": 0.44455859065055847,
+      "learning_rate": 2.3362404950261628e-05,
+      "loss": 0.0381,
+      "step": 6950
+    },
+    {
+      "epoch": 26.872586872586872,
+      "grad_norm": 0.4645119905471802,
+      "learning_rate": 2.3222622423136458e-05,
+      "loss": 0.0485,
+      "step": 6960
+    },
+    {
+      "epoch": 26.91119691119691,
+      "grad_norm": 0.35686275362968445,
+      "learning_rate": 2.3083132729122332e-05,
+      "loss": 0.0501,
+      "step": 6970
+    },
+    {
+      "epoch": 26.94980694980695,
+      "grad_norm": 0.6142265200614929,
+      "learning_rate": 2.294393739365621e-05,
+      "loss": 0.051,
+      "step": 6980
+    },
+    {
+      "epoch": 26.98841698841699,
+      "grad_norm": 0.5292761325836182,
+      "learning_rate": 2.2805037938956e-05,
+      "loss": 0.0373,
+      "step": 6990
+    },
+    {
+      "epoch": 27.027027027027028,
+      "grad_norm": 0.45108911395072937,
+      "learning_rate": 2.266643588400386e-05,
+      "loss": 0.0453,
+      "step": 7000
+    },
+    {
+      "epoch": 27.065637065637066,
+      "grad_norm": 0.4141083061695099,
+      "learning_rate": 2.252813274452969e-05,
+      "loss": 0.0583,
+      "step": 7010
+    },
+    {
+      "epoch": 27.104247104247104,
+      "grad_norm": 0.4025471806526184,
+      "learning_rate": 2.2390130032994427e-05,
+      "loss": 0.0479,
+      "step": 7020
+    },
+    {
+      "epoch": 27.142857142857142,
+      "grad_norm": 0.5654422640800476,
+      "learning_rate": 2.2252429258573633e-05,
+      "loss": 0.0686,
+      "step": 7030
+    },
+    {
+      "epoch": 27.18146718146718,
+      "grad_norm": 0.2999004125595093,
+      "learning_rate": 2.2115031927140904e-05,
+      "loss": 0.0506,
+      "step": 7040
+    },
+    {
+      "epoch": 27.22007722007722,
+      "grad_norm": 0.5117769837379456,
+      "learning_rate": 2.1977939541251463e-05,
+      "loss": 0.037,
+      "step": 7050
+    },
+    {
+      "epoch": 27.25868725868726,
+      "grad_norm": 0.39674946665763855,
+      "learning_rate": 2.1841153600125684e-05,
+      "loss": 0.0416,
+      "step": 7060
+    },
+    {
+      "epoch": 27.2972972972973,
+      "grad_norm": 0.514123260974884,
+      "learning_rate": 2.170467559963267e-05,
+      "loss": 0.0497,
+      "step": 7070
+    },
+    {
+      "epoch": 27.335907335907336,
+      "grad_norm": 0.49236828088760376,
+      "learning_rate": 2.1568507032273982e-05,
+      "loss": 0.0759,
+      "step": 7080
+    },
+    {
+      "epoch": 27.374517374517374,
+      "grad_norm": 0.7245543003082275,
+      "learning_rate": 2.1432649387167264e-05,
+      "loss": 0.0586,
+      "step": 7090
+    },
+    {
+      "epoch": 27.413127413127413,
+      "grad_norm": 0.6243254542350769,
+      "learning_rate": 2.1297104150029973e-05,
+      "loss": 0.033,
+      "step": 7100
+    },
+    {
+      "epoch": 27.45173745173745,
+      "grad_norm": 0.389565646648407,
+      "learning_rate": 2.116187280316307e-05,
+      "loss": 0.0678,
+      "step": 7110
+    },
+    {
+      "epoch": 27.49034749034749,
+      "grad_norm": 0.5558807849884033,
+      "learning_rate": 2.1026956825434908e-05,
+      "loss": 0.0341,
+      "step": 7120
+    },
+    {
+      "epoch": 27.52895752895753,
+      "grad_norm": 0.5637540817260742,
+      "learning_rate": 2.0892357692265017e-05,
+      "loss": 0.043,
+      "step": 7130
+    },
+    {
+      "epoch": 27.56756756756757,
+      "grad_norm": 0.4027647078037262,
+      "learning_rate": 2.0758076875607947e-05,
+      "loss": 0.0668,
+      "step": 7140
+    },
+    {
+      "epoch": 27.606177606177607,
+      "grad_norm": 0.42507487535476685,
+      "learning_rate": 2.0624115843937207e-05,
+      "loss": 0.0532,
+      "step": 7150
+    },
+    {
+      "epoch": 27.644787644787645,
+      "grad_norm": 0.5454823970794678,
+      "learning_rate": 2.0490476062229157e-05,
+      "loss": 0.0564,
+      "step": 7160
+    },
+    {
+      "epoch": 27.683397683397683,
+      "grad_norm": 0.5077855587005615,
+      "learning_rate": 2.035715899194704e-05,
+      "loss": 0.0483,
+      "step": 7170
+    },
+    {
+      "epoch": 27.72200772200772,
+      "grad_norm": 0.7553500533103943,
+      "learning_rate": 2.022416609102499e-05,
+      "loss": 0.0396,
+      "step": 7180
+    },
+    {
+      "epoch": 27.760617760617762,
+      "grad_norm": 0.4641231596469879,
+      "learning_rate": 2.009149881385205e-05,
+      "loss": 0.0472,
+      "step": 7190
+    },
+    {
+      "epoch": 27.7992277992278,
+      "grad_norm": 0.5995585918426514,
+      "learning_rate": 1.995915861125634e-05,
+      "loss": 0.0478,
+      "step": 7200
+    },
+    {
+      "epoch": 27.83783783783784,
+      "grad_norm": 0.370569109916687,
+      "learning_rate": 1.9827146930489065e-05,
+      "loss": 0.0475,
+      "step": 7210
+    },
+    {
+      "epoch": 27.876447876447877,
+      "grad_norm": 0.8217543363571167,
+      "learning_rate": 1.9695465215208848e-05,
+      "loss": 0.0509,
+      "step": 7220
+    },
+    {
+      "epoch": 27.915057915057915,
+      "grad_norm": 0.38071209192276,
+      "learning_rate": 1.9564114905465813e-05,
+      "loss": 0.0393,
+      "step": 7230
+    },
+    {
+      "epoch": 27.953667953667953,
+      "grad_norm": 0.7309401631355286,
+      "learning_rate": 1.9433097437685936e-05,
+      "loss": 0.0496,
+      "step": 7240
+    },
+    {
+      "epoch": 27.99227799227799,
+      "grad_norm": 0.46595320105552673,
+      "learning_rate": 1.930241424465521e-05,
+      "loss": 0.0529,
+      "step": 7250
+    },
+    {
+      "epoch": 28.030888030888033,
+      "grad_norm": 0.5158835053443909,
+      "learning_rate": 1.9172066755504115e-05,
+      "loss": 0.036,
+      "step": 7260
+    },
+    {
+      "epoch": 28.06949806949807,
+      "grad_norm": 0.2889738976955414,
+      "learning_rate": 1.9042056395691914e-05,
+      "loss": 0.031,
+      "step": 7270
+    },
+    {
+      "epoch": 28.10810810810811,
+      "grad_norm": 0.5699809789657593,
+      "learning_rate": 1.8912384586991066e-05,
+      "loss": 0.0409,
+      "step": 7280
+    },
+    {
+      "epoch": 28.146718146718147,
+      "grad_norm": 0.30492979288101196,
+      "learning_rate": 1.8783052747471717e-05,
+      "loss": 0.0533,
+      "step": 7290
+    },
+    {
+      "epoch": 28.185328185328185,
+      "grad_norm": 0.858198344707489,
+      "learning_rate": 1.865406229148611e-05,
+      "loss": 0.0536,
+      "step": 7300
+    },
+    {
+      "epoch": 28.223938223938223,
+      "grad_norm": 0.531909704208374,
+      "learning_rate": 1.8525414629653233e-05,
+      "loss": 0.0424,
+      "step": 7310
+    },
+    {
+      "epoch": 28.26254826254826,
+      "grad_norm": 1.3592851161956787,
+      "learning_rate": 1.8397111168843255e-05,
+      "loss": 0.0361,
+      "step": 7320
+    },
+    {
+      "epoch": 28.301158301158303,
+      "grad_norm": 0.42610982060432434,
+      "learning_rate": 1.8269153312162323e-05,
+      "loss": 0.0452,
+      "step": 7330
+    },
+    {
+      "epoch": 28.33976833976834,
+      "grad_norm": 0.33658021688461304,
+      "learning_rate": 1.8141542458937054e-05,
+      "loss": 0.0461,
+      "step": 7340
+    },
+    {
+      "epoch": 28.37837837837838,
+      "grad_norm": 0.506962239742279,
+      "learning_rate": 1.8014280004699268e-05,
+      "loss": 0.0381,
+      "step": 7350
+    },
+    {
+      "epoch": 28.416988416988417,
+      "grad_norm": 0.4352479577064514,
+      "learning_rate": 1.788736734117078e-05,
+      "loss": 0.0382,
+      "step": 7360
+    },
+    {
+      "epoch": 28.455598455598455,
+      "grad_norm": 0.4303007125854492,
+      "learning_rate": 1.7760805856248152e-05,
+      "loss": 0.0482,
+      "step": 7370
+    },
+    {
+      "epoch": 28.494208494208493,
+      "grad_norm": 0.40790390968322754,
+      "learning_rate": 1.7634596933987518e-05,
+      "loss": 0.0411,
+      "step": 7380
+    },
+    {
+      "epoch": 28.53281853281853,
+      "grad_norm": 0.4553315341472626,
+      "learning_rate": 1.7508741954589404e-05,
+      "loss": 0.0485,
+      "step": 7390
+    },
+    {
+      "epoch": 28.571428571428573,
+      "grad_norm": 0.47120049595832825,
+      "learning_rate": 1.7383242294383717e-05,
+      "loss": 0.0433,
+      "step": 7400
+    },
+    {
+      "epoch": 28.61003861003861,
+      "grad_norm": 0.5280634164810181,
+      "learning_rate": 1.7258099325814632e-05,
+      "loss": 0.0438,
+      "step": 7410
+    },
+    {
+      "epoch": 28.64864864864865,
+      "grad_norm": 0.43494147062301636,
+      "learning_rate": 1.7133314417425594e-05,
+      "loss": 0.0649,
+      "step": 7420
+    },
+    {
+      "epoch": 28.687258687258687,
+      "grad_norm": 0.31125617027282715,
+      "learning_rate": 1.7008888933844408e-05,
+      "loss": 0.049,
+      "step": 7430
+    },
+    {
+      "epoch": 28.725868725868725,
+      "grad_norm": 0.49624380469322205,
+      "learning_rate": 1.6884824235768172e-05,
+      "loss": 0.0387,
+      "step": 7440
+    },
+    {
+      "epoch": 28.764478764478763,
+      "grad_norm": 0.641543447971344,
+      "learning_rate": 1.6761121679948592e-05,
+      "loss": 0.0433,
+      "step": 7450
+    },
+    {
+      "epoch": 28.8030888030888,
+      "grad_norm": 0.3550838530063629,
+      "learning_rate": 1.663778261917695e-05,
+      "loss": 0.0528,
+      "step": 7460
+    },
+    {
+      "epoch": 28.841698841698843,
+      "grad_norm": 0.44131937623023987,
+      "learning_rate": 1.651480840226952e-05,
+      "loss": 0.0436,
+      "step": 7470
+    },
+    {
+      "epoch": 28.88030888030888,
+      "grad_norm": 1.983064889907837,
+      "learning_rate": 1.639220037405258e-05,
+      "loss": 0.057,
+      "step": 7480
+    },
+    {
+      "epoch": 28.91891891891892,
+      "grad_norm": 0.5690191984176636,
+      "learning_rate": 1.6269959875347906e-05,
+      "loss": 0.0486,
+      "step": 7490
+    },
+    {
+      "epoch": 28.957528957528957,
+      "grad_norm": 0.37668266892433167,
+      "learning_rate": 1.614808824295802e-05,
+      "loss": 0.0331,
+      "step": 7500
+    },
+    {
+      "epoch": 28.996138996138995,
+      "grad_norm": 0.5062463283538818,
+      "learning_rate": 1.602658680965152e-05,
+      "loss": 0.034,
+      "step": 7510
+    },
+    {
+      "epoch": 29.034749034749034,
+      "grad_norm": 0.26328518986701965,
+      "learning_rate": 1.5905456904148686e-05,
+      "loss": 0.0299,
+      "step": 7520
+    },
+    {
+      "epoch": 29.07335907335907,
+      "grad_norm": 0.38508641719818115,
+      "learning_rate": 1.57846998511067e-05,
+      "loss": 0.0351,
+      "step": 7530
+    },
+    {
+      "epoch": 29.111969111969113,
+      "grad_norm": 0.3345756232738495,
+      "learning_rate": 1.566431697110538e-05,
+      "loss": 0.043,
+      "step": 7540
+    },
+    {
+      "epoch": 29.15057915057915,
+      "grad_norm": 0.505284309387207,
+      "learning_rate": 1.554430958063259e-05,
+      "loss": 0.0424,
+      "step": 7550
+    },
+    {
+      "epoch": 29.18918918918919,
+      "grad_norm": 0.5906921029090881,
+      "learning_rate": 1.5424678992069912e-05,
+      "loss": 0.0454,
+      "step": 7560
+    },
+    {
+      "epoch": 29.227799227799228,
+      "grad_norm": 0.44763216376304626,
+      "learning_rate": 1.5305426513678362e-05,
+      "loss": 0.0478,
+      "step": 7570
+    },
+    {
+      "epoch": 29.266409266409266,
+      "grad_norm": 0.581728994846344,
+      "learning_rate": 1.518655344958388e-05,
+      "loss": 0.0576,
+      "step": 7580
+    },
+    {
+      "epoch": 29.305019305019304,
+      "grad_norm": 0.6709556579589844,
+      "learning_rate": 1.5068061099763275e-05,
+      "loss": 0.0459,
+      "step": 7590
+    },
+    {
+      "epoch": 29.343629343629345,
+      "grad_norm": 0.4280284345149994,
+      "learning_rate": 1.494995076002988e-05,
+      "loss": 0.0564,
+      "step": 7600
+    },
+    {
+      "epoch": 29.382239382239383,
+      "grad_norm": 0.3624938130378723,
+      "learning_rate": 1.4832223722019456e-05,
+      "loss": 0.0525,
+      "step": 7610
+    },
+    {
+      "epoch": 29.42084942084942,
+      "grad_norm": 0.5082321763038635,
+      "learning_rate": 1.4714881273176035e-05,
+      "loss": 0.0408,
+      "step": 7620
+    },
+    {
+      "epoch": 29.45945945945946,
+      "grad_norm": 0.469899445772171,
+      "learning_rate": 1.4597924696737835e-05,
+      "loss": 0.0497,
+      "step": 7630
+    },
+    {
+      "epoch": 29.498069498069498,
+      "grad_norm": 0.6883857846260071,
+      "learning_rate": 1.4481355271723252e-05,
+      "loss": 0.0433,
+      "step": 7640
+    },
+    {
+      "epoch": 29.536679536679536,
+      "grad_norm": 0.5980105996131897,
+      "learning_rate": 1.4365174272916809e-05,
+      "loss": 0.0352,
+      "step": 7650
+    },
+    {
+      "epoch": 29.575289575289574,
+      "grad_norm": 0.5563459992408752,
+      "learning_rate": 1.4249382970855319e-05,
+      "loss": 0.0325,
+      "step": 7660
+    },
+    {
+      "epoch": 29.613899613899616,
+      "grad_norm": 0.38310402631759644,
+      "learning_rate": 1.4133982631813903e-05,
+      "loss": 0.0433,
+      "step": 7670
+    },
+    {
+      "epoch": 29.652509652509654,
+      "grad_norm": 0.6001343727111816,
+      "learning_rate": 1.4018974517792194e-05,
+      "loss": 0.0499,
+      "step": 7680
+    },
+    {
+      "epoch": 29.69111969111969,
+      "grad_norm": 0.36467060446739197,
+      "learning_rate": 1.390435988650048e-05,
+      "loss": 0.0367,
+      "step": 7690
+    },
+    {
+      "epoch": 29.72972972972973,
+      "grad_norm": 0.44829538464546204,
+      "learning_rate": 1.3790139991346006e-05,
+      "loss": 0.0318,
+      "step": 7700
+    },
+    {
+      "epoch": 29.768339768339768,
+      "grad_norm": 0.563353419303894,
+      "learning_rate": 1.367631608141926e-05,
+      "loss": 0.048,
+      "step": 7710
+    },
+    {
+      "epoch": 29.806949806949806,
+      "grad_norm": 0.3578076958656311,
+      "learning_rate": 1.3562889401480278e-05,
+      "loss": 0.0485,
+      "step": 7720
+    },
+    {
+      "epoch": 29.845559845559844,
+      "grad_norm": 0.4575956165790558,
+      "learning_rate": 1.3449861191945074e-05,
+      "loss": 0.0439,
+      "step": 7730
+    },
+    {
+      "epoch": 29.884169884169886,
+      "grad_norm": 0.549997866153717,
+      "learning_rate": 1.3337232688872009e-05,
+      "loss": 0.0443,
+      "step": 7740
+    },
+    {
+      "epoch": 29.922779922779924,
+      "grad_norm": 0.7762444019317627,
+      "learning_rate": 1.3225005123948364e-05,
+      "loss": 0.034,
+      "step": 7750
+    },
+    {
+      "epoch": 29.961389961389962,
+      "grad_norm": 0.4942251741886139,
+      "learning_rate": 1.311317972447681e-05,
+      "loss": 0.047,
+      "step": 7760
+    },
+    {
+      "epoch": 30.0,
+      "grad_norm": 0.3711082935333252,
+      "learning_rate": 1.3001757713361996e-05,
+      "loss": 0.0515,
+      "step": 7770
+    },
+    {
+      "epoch": 30.038610038610038,
+      "grad_norm": 0.5000542402267456,
+      "learning_rate": 1.2890740309097204e-05,
+      "loss": 0.0403,
+      "step": 7780
+    },
+    {
+      "epoch": 30.077220077220076,
+      "grad_norm": 0.675679087638855,
+      "learning_rate": 1.2780128725750944e-05,
+      "loss": 0.0471,
+      "step": 7790
+    },
+    {
+      "epoch": 30.115830115830114,
+      "grad_norm": 0.5804657936096191,
+      "learning_rate": 1.266992417295379e-05,
+      "loss": 0.0434,
+      "step": 7800
+    },
+    {
+      "epoch": 30.154440154440156,
+      "grad_norm": 0.40390142798423767,
+      "learning_rate": 1.2560127855885073e-05,
+      "loss": 0.0469,
+      "step": 7810
+    },
+    {
+      "epoch": 30.193050193050194,
+      "grad_norm": 0.2912136912345886,
+      "learning_rate": 1.2450740975259745e-05,
+      "loss": 0.0293,
+      "step": 7820
+    },
+    {
+      "epoch": 30.231660231660232,
+      "grad_norm": 0.49417203664779663,
+      "learning_rate": 1.234176472731517e-05,
+      "loss": 0.0398,
+      "step": 7830
+    },
+    {
+      "epoch": 30.27027027027027,
+      "grad_norm": 0.4873831570148468,
+      "learning_rate": 1.2233200303798158e-05,
+      "loss": 0.0309,
+      "step": 7840
+    },
+    {
+      "epoch": 30.30888030888031,
+      "grad_norm": 0.640707790851593,
+      "learning_rate": 1.2125048891951846e-05,
+      "loss": 0.0428,
+      "step": 7850
+    },
+    {
+      "epoch": 30.347490347490346,
+      "grad_norm": 0.44811660051345825,
+      "learning_rate": 1.2017311674502745e-05,
+      "loss": 0.0469,
+      "step": 7860
+    },
+    {
+      "epoch": 30.386100386100384,
+      "grad_norm": 0.4043349027633667,
+      "learning_rate": 1.1909989829647822e-05,
+      "loss": 0.0499,
+      "step": 7870
+    },
+    {
+      "epoch": 30.424710424710426,
+      "grad_norm": 0.4621827006340027,
+      "learning_rate": 1.1803084531041553e-05,
+      "loss": 0.0462,
+      "step": 7880
+    },
+    {
+      "epoch": 30.463320463320464,
+      "grad_norm": 0.35089004039764404,
+      "learning_rate": 1.1696596947783162e-05,
+      "loss": 0.0421,
+      "step": 7890
+    },
+    {
+      "epoch": 30.501930501930502,
+      "grad_norm": 0.3911316394805908,
+      "learning_rate": 1.1590528244403803e-05,
+      "loss": 0.0518,
+      "step": 7900
+    },
+    {
+      "epoch": 30.54054054054054,
+      "grad_norm": 0.24839645624160767,
+      "learning_rate": 1.148487958085382e-05,
+      "loss": 0.0502,
+      "step": 7910
+    },
+    {
+      "epoch": 30.57915057915058,
+      "grad_norm": 0.4422091245651245,
+      "learning_rate": 1.1379652112490086e-05,
+      "loss": 0.0579,
+      "step": 7920
+    },
+    {
+      "epoch": 30.617760617760617,
+      "grad_norm": 0.40578359365463257,
+      "learning_rate": 1.1274846990063315e-05,
+      "loss": 0.0391,
+      "step": 7930
+    },
+    {
+      "epoch": 30.656370656370655,
+      "grad_norm": 0.47507625818252563,
+      "learning_rate": 1.117046535970554e-05,
+      "loss": 0.0458,
+      "step": 7940
+    },
+    {
+      "epoch": 30.694980694980696,
+      "grad_norm": 0.3613755702972412,
+      "learning_rate": 1.106650836291755e-05,
+      "loss": 0.0318,
+      "step": 7950
+    },
+    {
+      "epoch": 30.733590733590734,
+      "grad_norm": 0.3082832992076874,
+      "learning_rate": 1.0962977136556418e-05,
+      "loss": 0.036,
+      "step": 7960
+    },
+    {
+      "epoch": 30.772200772200772,
+      "grad_norm": 0.4993591904640198,
+      "learning_rate": 1.0859872812823024e-05,
+      "loss": 0.0312,
+      "step": 7970
+    },
+    {
+      "epoch": 30.81081081081081,
+      "grad_norm": 0.4567050635814667,
+      "learning_rate": 1.0757196519249747e-05,
+      "loss": 0.0366,
+      "step": 7980
+    },
+    {
+      "epoch": 30.84942084942085,
+      "grad_norm": 0.36421293020248413,
+      "learning_rate": 1.0654949378688077e-05,
+      "loss": 0.034,
+      "step": 7990
+    },
+    {
+      "epoch": 30.888030888030887,
+      "grad_norm": 0.7462463974952698,
+      "learning_rate": 1.0553132509296376e-05,
+      "loss": 0.0332,
+      "step": 8000
+    },
+    {
+      "epoch": 30.92664092664093,
+      "grad_norm": 0.4237750470638275,
+      "learning_rate": 1.0451747024527613e-05,
+      "loss": 0.0333,
+      "step": 8010
+    },
+    {
+      "epoch": 30.965250965250966,
+      "grad_norm": 0.7147374153137207,
+      "learning_rate": 1.0350794033117189e-05,
+      "loss": 0.0452,
+      "step": 8020
+    },
+    {
+      "epoch": 31.003861003861005,
+      "grad_norm": 0.23559176921844482,
+      "learning_rate": 1.0250274639070856e-05,
+      "loss": 0.0364,
+      "step": 8030
+    },
+    {
+      "epoch": 31.042471042471043,
+      "grad_norm": 0.5721344351768494,
+      "learning_rate": 1.0150189941652599e-05,
+      "loss": 0.0343,
+      "step": 8040
+    },
+    {
+      "epoch": 31.08108108108108,
+      "grad_norm": 0.38582655787467957,
+      "learning_rate": 1.0050541035372635e-05,
+      "loss": 0.0451,
+      "step": 8050
+    },
+    {
+      "epoch": 31.11969111969112,
+      "grad_norm": 0.19999687373638153,
+      "learning_rate": 9.951329009975458e-06,
+      "loss": 0.0284,
+      "step": 8060
+    },
+    {
+      "epoch": 31.158301158301157,
+      "grad_norm": 0.5094717144966125,
+      "learning_rate": 9.852554950427845e-06,
+      "loss": 0.0356,
+      "step": 8070
+    },
+    {
+      "epoch": 31.1969111969112,
+      "grad_norm": 0.46238014101982117,
+      "learning_rate": 9.754219936907105e-06,
+      "loss": 0.0508,
+      "step": 8080
+    },
+    {
+      "epoch": 31.235521235521237,
+      "grad_norm": 0.3049578070640564,
+      "learning_rate": 9.656325044789194e-06,
+      "loss": 0.0372,
+      "step": 8090
+    },
+    {
+      "epoch": 31.274131274131275,
+      "grad_norm": 0.5652112364768982,
+      "learning_rate": 9.55887134463697e-06,
+      "loss": 0.0449,
+      "step": 8100
+    },
+    {
+      "epoch": 31.312741312741313,
+      "grad_norm": 0.47629275918006897,
+      "learning_rate": 9.461859902188475e-06,
+      "loss": 0.0424,
+      "step": 8110
+    },
+    {
+      "epoch": 31.35135135135135,
+      "grad_norm": 0.34560078382492065,
+      "learning_rate": 9.365291778345303e-06,
+      "loss": 0.0379,
+      "step": 8120
+    },
+    {
+      "epoch": 31.38996138996139,
+      "grad_norm": 0.4096907079219818,
+      "learning_rate": 9.269168029160991e-06,
+      "loss": 0.0566,
+      "step": 8130
+    },
+    {
+      "epoch": 31.428571428571427,
+      "grad_norm": 0.4206790328025818,
+      "learning_rate": 9.173489705829447e-06,
+      "loss": 0.0416,
+      "step": 8140
+    },
+    {
+      "epoch": 31.46718146718147,
+      "grad_norm": 0.4509584307670593,
+      "learning_rate": 9.078257854673516e-06,
+      "loss": 0.0364,
+      "step": 8150
+    },
+    {
+      "epoch": 31.505791505791507,
+      "grad_norm": 0.35847026109695435,
+      "learning_rate": 8.983473517133429e-06,
+      "loss": 0.0466,
+      "step": 8160
+    },
+    {
+      "epoch": 31.544401544401545,
+      "grad_norm": 0.21652787923812866,
+      "learning_rate": 8.889137729755537e-06,
+      "loss": 0.0418,
+      "step": 8170
+    },
+    {
+      "epoch": 31.583011583011583,
+      "grad_norm": 0.504115879535675,
+      "learning_rate": 8.79525152418087e-06,
+      "loss": 0.0342,
+      "step": 8180
+    },
+    {
+      "epoch": 31.62162162162162,
+      "grad_norm": 0.5375120639801025,
+      "learning_rate": 8.701815927133961e-06,
+      "loss": 0.0432,
+      "step": 8190
+    },
+    {
+      "epoch": 31.66023166023166,
+      "grad_norm": 0.5640621185302734,
+      "learning_rate": 8.608831960411534e-06,
+      "loss": 0.0351,
+      "step": 8200
+    },
+    {
+      "epoch": 31.698841698841697,
+      "grad_norm": 0.3777157664299011,
+      "learning_rate": 8.516300640871321e-06,
+      "loss": 0.0268,
+      "step": 8210
+    },
+    {
+      "epoch": 31.73745173745174,
+      "grad_norm": 0.2588464021682739,
+      "learning_rate": 8.424222980421038e-06,
+      "loss": 0.0364,
+      "step": 8220
+    },
+    {
+      "epoch": 31.776061776061777,
+      "grad_norm": 0.3287317454814911,
+      "learning_rate": 8.332599986007184e-06,
+      "loss": 0.0356,
+      "step": 8230
+    },
+    {
+      "epoch": 31.814671814671815,
+      "grad_norm": 0.2392064929008484,
+      "learning_rate": 8.241432659604203e-06,
+      "loss": 0.0352,
+      "step": 8240
+    },
+    {
+      "epoch": 31.853281853281853,
+      "grad_norm": 0.3696592152118683,
+      "learning_rate": 8.150721998203331e-06,
+      "loss": 0.0428,
+      "step": 8250
+    },
+    {
+      "epoch": 31.89189189189189,
+      "grad_norm": 0.4237655997276306,
+      "learning_rate": 8.06046899380184e-06,
+      "loss": 0.0279,
+      "step": 8260
+    },
+    {
+      "epoch": 31.93050193050193,
+      "grad_norm": 0.4059978127479553,
+      "learning_rate": 7.970674633392133e-06,
+      "loss": 0.0284,
+      "step": 8270
+    },
+    {
+      "epoch": 31.969111969111967,
+      "grad_norm": 0.5361632704734802,
+      "learning_rate": 7.881339898950924e-06,
+      "loss": 0.0389,
+      "step": 8280
+    },
+    {
+      "epoch": 32.00772200772201,
+      "grad_norm": 0.3683216869831085,
+      "learning_rate": 7.792465767428597e-06,
+      "loss": 0.0284,
+      "step": 8290
+    },
+    {
+      "epoch": 32.04633204633205,
+      "grad_norm": 0.41536685824394226,
+      "learning_rate": 7.704053210738376e-06,
+      "loss": 0.0414,
+      "step": 8300
+    },
+    {
+      "epoch": 32.084942084942085,
+      "grad_norm": 0.6054041385650635,
+      "learning_rate": 7.6161031957458494e-06,
+      "loss": 0.0345,
+      "step": 8310
+    },
+    {
+      "epoch": 32.12355212355212,
+      "grad_norm": 0.4750712513923645,
+      "learning_rate": 7.5286166842582605e-06,
+      "loss": 0.0252,
+      "step": 8320
+    },
+    {
+      "epoch": 32.16216216216216,
+      "grad_norm": 0.6487170457839966,
+      "learning_rate": 7.4415946330140814e-06,
+      "loss": 0.0362,
+      "step": 8330
+    },
+    {
+      "epoch": 32.2007722007722,
+      "grad_norm": 0.2815876305103302,
+      "learning_rate": 7.3550379936725644e-06,
+      "loss": 0.0318,
+      "step": 8340
+    },
+    {
+      "epoch": 32.23938223938224,
+      "grad_norm": 0.4896615743637085,
+      "learning_rate": 7.2689477128032035e-06,
+      "loss": 0.0391,
+      "step": 8350
+    },
+    {
+      "epoch": 32.277992277992276,
+      "grad_norm": 0.36150723695755005,
+      "learning_rate": 7.183324731875551e-06,
+      "loss": 0.0273,
+      "step": 8360
+    },
+    {
+      "epoch": 32.316602316602314,
+      "grad_norm": 0.39038848876953125,
+      "learning_rate": 7.098169987248782e-06,
+      "loss": 0.0302,
+      "step": 8370
+    },
+    {
+      "epoch": 32.35521235521235,
+      "grad_norm": 0.5400213003158569,
+      "learning_rate": 7.013484410161553e-06,
+      "loss": 0.0436,
+      "step": 8380
+    },
+    {
+      "epoch": 32.3938223938224,
+      "grad_norm": 0.41952595114707947,
+      "learning_rate": 6.92926892672176e-06,
+      "loss": 0.0313,
+      "step": 8390
+    },
+    {
+      "epoch": 32.432432432432435,
+      "grad_norm": 0.4260327219963074,
+      "learning_rate": 6.845524457896446e-06,
+      "loss": 0.0305,
+      "step": 8400
+    },
+    {
+      "epoch": 32.47104247104247,
+      "grad_norm": 0.6441579461097717,
+      "learning_rate": 6.7622519195017165e-06,
+      "loss": 0.0314,
+      "step": 8410
+    },
+    {
+      "epoch": 32.50965250965251,
+      "grad_norm": 0.3133586049079895,
+      "learning_rate": 6.679452222192684e-06,
+      "loss": 0.0292,
+      "step": 8420
+    },
+    {
+      "epoch": 32.54826254826255,
+      "grad_norm": 0.49595722556114197,
+      "learning_rate": 6.597126271453579e-06,
+      "loss": 0.0358,
+      "step": 8430
+    },
+    {
+      "epoch": 32.58687258687259,
+      "grad_norm": 0.3354673385620117,
+      "learning_rate": 6.51527496758782e-06,
+      "loss": 0.0313,
+      "step": 8440
+    },
+    {
+      "epoch": 32.625482625482626,
+      "grad_norm": 0.5179545879364014,
+      "learning_rate": 6.433899205708155e-06,
+      "loss": 0.036,
+      "step": 8450
+    },
+    {
+      "epoch": 32.664092664092664,
+      "grad_norm": 0.36871305108070374,
+      "learning_rate": 6.352999875726856e-06,
+      "loss": 0.0316,
+      "step": 8460
+    },
+    {
+      "epoch": 32.7027027027027,
+      "grad_norm": 0.324504017829895,
+      "learning_rate": 6.272577862346052e-06,
+      "loss": 0.03,
+      "step": 8470
+    },
+    {
+      "epoch": 32.74131274131274,
+      "grad_norm": 0.4725330173969269,
+      "learning_rate": 6.192634045047996e-06,
+      "loss": 0.029,
+      "step": 8480
+    },
+    {
+      "epoch": 32.77992277992278,
+      "grad_norm": 0.5182196497917175,
+      "learning_rate": 6.113169298085458e-06,
+      "loss": 0.0325,
+      "step": 8490
+    },
+    {
+      "epoch": 32.818532818532816,
+      "grad_norm": 0.5111414194107056,
+      "learning_rate": 6.034184490472195e-06,
+      "loss": 0.0534,
+      "step": 8500
+    },
+    {
+      "epoch": 32.857142857142854,
+      "grad_norm": 0.8664683103561401,
+      "learning_rate": 5.955680485973386e-06,
+      "loss": 0.0393,
+      "step": 8510
+    },
+    {
+      "epoch": 32.89575289575289,
+      "grad_norm": 0.4252089262008667,
+      "learning_rate": 5.877658143096265e-06,
+      "loss": 0.0268,
+      "step": 8520
+    },
+    {
+      "epoch": 32.93436293436294,
+      "grad_norm": 0.3136313855648041,
+      "learning_rate": 5.800118315080661e-06,
+      "loss": 0.0379,
+      "step": 8530
+    },
+    {
+      "epoch": 32.972972972972975,
+      "grad_norm": 0.5326577425003052,
+      "learning_rate": 5.723061849889716e-06,
+      "loss": 0.0379,
+      "step": 8540
+    },
+    {
+      "epoch": 33.011583011583014,
+      "grad_norm": 0.3791945278644562,
+      "learning_rate": 5.646489590200604e-06,
+      "loss": 0.036,
+      "step": 8550
+    },
+    {
+      "epoch": 33.05019305019305,
+      "grad_norm": 0.4695245325565338,
+      "learning_rate": 5.570402373395256e-06,
+      "loss": 0.0379,
+      "step": 8560
+    },
+    {
+      "epoch": 33.08880308880309,
+      "grad_norm": 0.5764884948730469,
+      "learning_rate": 5.494801031551305e-06,
+      "loss": 0.0662,
+      "step": 8570
+    },
+    {
+      "epoch": 33.12741312741313,
+      "grad_norm": 0.672602653503418,
+      "learning_rate": 5.41968639143291e-06,
+      "loss": 0.0504,
+      "step": 8580
+    },
+    {
+      "epoch": 33.166023166023166,
+      "grad_norm": 1.0906554460525513,
+      "learning_rate": 5.345059274481751e-06,
+      "loss": 0.0463,
+      "step": 8590
+    },
+    {
+      "epoch": 33.204633204633204,
+      "grad_norm": 0.4323593080043793,
+      "learning_rate": 5.270920496808002e-06,
+      "loss": 0.0279,
+      "step": 8600
+    },
+    {
+      "epoch": 33.24324324324324,
+      "grad_norm": 0.5862171053886414,
+      "learning_rate": 5.1972708691814695e-06,
+      "loss": 0.0281,
+      "step": 8610
+    },
+    {
+      "epoch": 33.28185328185328,
+      "grad_norm": 0.6559461355209351,
+      "learning_rate": 5.124111197022674e-06,
+      "loss": 0.0371,
+      "step": 8620
+    },
+    {
+      "epoch": 33.32046332046332,
+      "grad_norm": 0.47020477056503296,
+      "learning_rate": 5.051442280394081e-06,
+      "loss": 0.0325,
+      "step": 8630
+    },
+    {
+      "epoch": 33.359073359073356,
+      "grad_norm": 0.5820066928863525,
+      "learning_rate": 4.979264913991322e-06,
+      "loss": 0.0387,
+      "step": 8640
+    },
+    {
+      "epoch": 33.397683397683394,
+      "grad_norm": 0.5221219658851624,
+      "learning_rate": 4.907579887134489e-06,
+      "loss": 0.0512,
+      "step": 8650
+    },
+    {
+      "epoch": 33.43629343629344,
+      "grad_norm": 0.7113184928894043,
+      "learning_rate": 4.836387983759572e-06,
+      "loss": 0.0363,
+      "step": 8660
+    },
+    {
+      "epoch": 33.47490347490348,
+      "grad_norm": 0.22089901566505432,
+      "learning_rate": 4.765689982409816e-06,
+      "loss": 0.0282,
+      "step": 8670
+    },
+    {
+      "epoch": 33.513513513513516,
+      "grad_norm": 0.38262519240379333,
+      "learning_rate": 4.695486656227233e-06,
+      "loss": 0.022,
+      "step": 8680
+    },
+    {
+      "epoch": 33.552123552123554,
+      "grad_norm": 0.5200034379959106,
+      "learning_rate": 4.625778772944156e-06,
+      "loss": 0.0291,
+      "step": 8690
+    },
+    {
+      "epoch": 33.59073359073359,
+      "grad_norm": 0.20605699717998505,
+      "learning_rate": 4.556567094874825e-06,
+      "loss": 0.0279,
+      "step": 8700
+    },
+    {
+      "epoch": 33.62934362934363,
+      "grad_norm": 0.5793619751930237,
+      "learning_rate": 4.487852378907059e-06,
+      "loss": 0.0413,
+      "step": 8710
+    },
+    {
+      "epoch": 33.66795366795367,
+      "grad_norm": 0.5311439633369446,
+      "learning_rate": 4.419635376493986e-06,
+      "loss": 0.0451,
+      "step": 8720
+    },
+    {
+      "epoch": 33.706563706563706,
+      "grad_norm": 0.298466295003891,
+      "learning_rate": 4.351916833645825e-06,
+      "loss": 0.0257,
+      "step": 8730
+    },
+    {
+      "epoch": 33.745173745173744,
+      "grad_norm": 0.332709401845932,
+      "learning_rate": 4.284697490921691e-06,
+      "loss": 0.0291,
+      "step": 8740
+    },
+    {
+      "epoch": 33.78378378378378,
+      "grad_norm": 0.34630295634269714,
+      "learning_rate": 4.2179780834215585e-06,
+      "loss": 0.0386,
+      "step": 8750
+    },
+    {
+      "epoch": 33.82239382239382,
+      "grad_norm": 0.4650475084781647,
+      "learning_rate": 4.151759340778178e-06,
+      "loss": 0.0313,
+      "step": 8760
+    },
+    {
+      "epoch": 33.86100386100386,
+      "grad_norm": 0.5013532042503357,
+      "learning_rate": 4.086041987149109e-06,
+      "loss": 0.0374,
+      "step": 8770
+    },
+    {
+      "epoch": 33.8996138996139,
+      "grad_norm": 0.33382976055145264,
+      "learning_rate": 4.020826741208811e-06,
+      "loss": 0.0432,
+      "step": 8780
+    },
+    {
+      "epoch": 33.938223938223935,
+      "grad_norm": 0.2519553303718567,
+      "learning_rate": 3.956114316140746e-06,
+      "loss": 0.0314,
+      "step": 8790
+    },
+    {
+      "epoch": 33.97683397683398,
+      "grad_norm": 0.44277212023735046,
+      "learning_rate": 3.891905419629643e-06,
+      "loss": 0.0254,
+      "step": 8800
+    },
+    {
+      "epoch": 34.01544401544402,
+      "grad_norm": 0.5330662131309509,
+      "learning_rate": 3.8282007538536946e-06,
+      "loss": 0.0301,
+      "step": 8810
+    },
+    {
+      "epoch": 34.054054054054056,
+      "grad_norm": 0.6141893863677979,
+      "learning_rate": 3.7650010154769265e-06,
+      "loss": 0.032,
+      "step": 8820
+    },
+    {
+      "epoch": 34.092664092664094,
+      "grad_norm": 0.5341497659683228,
+      "learning_rate": 3.7023068956415608e-06,
+      "loss": 0.0283,
+      "step": 8830
+    },
+    {
+      "epoch": 34.13127413127413,
+      "grad_norm": 0.5615679621696472,
+      "learning_rate": 3.6401190799604303e-06,
+      "loss": 0.0347,
+      "step": 8840
+    },
+    {
+      "epoch": 34.16988416988417,
+      "grad_norm": 0.2762456238269806,
+      "learning_rate": 3.578438248509536e-06,
+      "loss": 0.0266,
+      "step": 8850
+    },
+    {
+      "epoch": 34.20849420849421,
+      "grad_norm": 0.6467777490615845,
+      "learning_rate": 3.5172650758205583e-06,
+      "loss": 0.0359,
+      "step": 8860
+    },
+    {
+      "epoch": 34.24710424710425,
+      "grad_norm": 0.6442224383354187,
+      "learning_rate": 3.45660023087353e-06,
+      "loss": 0.0316,
+      "step": 8870
+    },
+    {
+      "epoch": 34.285714285714285,
+      "grad_norm": 0.4205634295940399,
+      "learning_rate": 3.3964443770894528e-06,
+      "loss": 0.0342,
+      "step": 8880
+    },
+    {
+      "epoch": 34.32432432432432,
+      "grad_norm": 0.5289274454116821,
+      "learning_rate": 3.3367981723231245e-06,
+      "loss": 0.0356,
+      "step": 8890
+    },
+    {
+      "epoch": 34.36293436293436,
+      "grad_norm": 0.4894663691520691,
+      "learning_rate": 3.2776622688558746e-06,
+      "loss": 0.0309,
+      "step": 8900
+    },
+    {
+      "epoch": 34.4015444015444,
+      "grad_norm": 0.45998555421829224,
+      "learning_rate": 3.2190373133884677e-06,
+      "loss": 0.035,
+      "step": 8910
+    },
+    {
+      "epoch": 34.44015444015444,
+      "grad_norm": 0.7644488215446472,
+      "learning_rate": 3.1609239470340446e-06,
+      "loss": 0.0457,
+      "step": 8920
+    },
+    {
+      "epoch": 34.47876447876448,
+      "grad_norm": 0.5515457987785339,
+      "learning_rate": 3.1033228053110373e-06,
+      "loss": 0.0303,
+      "step": 8930
+    },
+    {
+      "epoch": 34.51737451737452,
+      "grad_norm": 0.31745579838752747,
+      "learning_rate": 3.0462345181363314e-06,
+      "loss": 0.0372,
+      "step": 8940
+    },
+    {
+      "epoch": 34.55598455598456,
+      "grad_norm": 0.4758061468601227,
+      "learning_rate": 2.9896597098182654e-06,
+      "loss": 0.0283,
+      "step": 8950
+    },
+    {
+      "epoch": 34.5945945945946,
+      "grad_norm": 0.25310981273651123,
+      "learning_rate": 2.933598999049891e-06,
+      "loss": 0.0199,
+      "step": 8960
+    },
+    {
+      "epoch": 34.633204633204635,
+      "grad_norm": 0.26150646805763245,
+      "learning_rate": 2.8780529989021697e-06,
+      "loss": 0.0252,
+      "step": 8970
+    },
+    {
+      "epoch": 34.67181467181467,
+      "grad_norm": 0.24514614045619965,
+      "learning_rate": 2.823022316817242e-06,
+      "loss": 0.0319,
+      "step": 8980
+    },
+    {
+      "epoch": 34.71042471042471,
+      "grad_norm": 0.6540687680244446,
+      "learning_rate": 2.7685075546018456e-06,
+      "loss": 0.0524,
+      "step": 8990
+    },
+    {
+      "epoch": 34.74903474903475,
+      "grad_norm": 0.3403765559196472,
+      "learning_rate": 2.7145093084206598e-06,
+      "loss": 0.037,
+      "step": 9000
+    },
+    {
+      "epoch": 34.78764478764479,
+      "grad_norm": 0.46332094073295593,
+      "learning_rate": 2.661028168789892e-06,
+      "loss": 0.0238,
+      "step": 9010
+    },
+    {
+      "epoch": 34.826254826254825,
+      "grad_norm": 0.5999321341514587,
+      "learning_rate": 2.6080647205706855e-06,
+      "loss": 0.0405,
+      "step": 9020
+    },
+    {
+      "epoch": 34.86486486486486,
+      "grad_norm": 0.35631102323532104,
+      "learning_rate": 2.555619542962834e-06,
+      "loss": 0.0497,
+      "step": 9030
+    },
+    {
+      "epoch": 34.9034749034749,
+      "grad_norm": 0.33276018500328064,
+      "learning_rate": 2.503693209498409e-06,
+      "loss": 0.0385,
+      "step": 9040
+    },
+    {
+      "epoch": 34.94208494208494,
+      "grad_norm": 0.32796597480773926,
+      "learning_rate": 2.452286288035449e-06,
+      "loss": 0.0305,
+      "step": 9050
+    },
+    {
+      "epoch": 34.98069498069498,
+      "grad_norm": 0.38111504912376404,
+      "learning_rate": 2.4013993407518363e-06,
+      "loss": 0.0368,
+      "step": 9060
+    },
+    {
+      "epoch": 35.01930501930502,
+      "grad_norm": 0.3923816680908203,
+      "learning_rate": 2.351032924139063e-06,
+      "loss": 0.0191,
+      "step": 9070
+    },
+    {
+      "epoch": 35.05791505791506,
+      "grad_norm": 0.2567480206489563,
+      "learning_rate": 2.30118758899619e-06,
+      "loss": 0.028,
+      "step": 9080
+    },
+    {
+      "epoch": 35.0965250965251,
+      "grad_norm": 0.48664647340774536,
+      "learning_rate": 2.2518638804238157e-06,
+      "loss": 0.0361,
+      "step": 9090
+    },
+    {
+      "epoch": 35.13513513513514,
+      "grad_norm": 0.4539981186389923,
+      "learning_rate": 2.203062337818118e-06,
+      "loss": 0.0419,
+      "step": 9100
+    },
+    {
+      "epoch": 35.173745173745175,
+      "grad_norm": 0.8542642593383789,
+      "learning_rate": 2.1547834948649483e-06,
+      "loss": 0.0446,
+      "step": 9110
+    },
+    {
+      "epoch": 35.21235521235521,
+      "grad_norm": 0.5568941235542297,
+      "learning_rate": 2.1070278795340017e-06,
+      "loss": 0.0351,
+      "step": 9120
+    },
+    {
+      "epoch": 35.25096525096525,
+      "grad_norm": 0.600436270236969,
+      "learning_rate": 2.059796014073029e-06,
+      "loss": 0.0442,
+      "step": 9130
+    },
+    {
+      "epoch": 35.28957528957529,
+      "grad_norm": 0.5558466911315918,
+      "learning_rate": 2.01308841500214e-06,
+      "loss": 0.0257,
+      "step": 9140
+    },
+    {
+      "epoch": 35.32818532818533,
+      "grad_norm": 0.6728957295417786,
+      "learning_rate": 1.9669055931081704e-06,
+      "loss": 0.0514,
+      "step": 9150
+    },
+    {
+      "epoch": 35.366795366795365,
+      "grad_norm": 0.4663664996623993,
+      "learning_rate": 1.9212480534390507e-06,
+      "loss": 0.0261,
+      "step": 9160
+    },
+    {
+      "epoch": 35.4054054054054,
+      "grad_norm": 0.562464714050293,
+      "learning_rate": 1.8761162952983246e-06,
+      "loss": 0.0307,
+      "step": 9170
+    },
+    {
+      "epoch": 35.44401544401544,
+      "grad_norm": 0.4802253544330597,
+      "learning_rate": 1.8315108122396618e-06,
+      "loss": 0.029,
+      "step": 9180
+    },
+    {
+      "epoch": 35.48262548262548,
+      "grad_norm": 0.32845574617385864,
+      "learning_rate": 1.787432092061475e-06,
+      "loss": 0.0217,
+      "step": 9190
+    },
+    {
+      "epoch": 35.52123552123552,
+      "grad_norm": 0.6012297868728638,
+      "learning_rate": 1.743880616801602e-06,
+      "loss": 0.0391,
+      "step": 9200
+    },
+    {
+      "epoch": 35.55984555984556,
+      "grad_norm": 0.3478630781173706,
+      "learning_rate": 1.7008568627319865e-06,
+      "loss": 0.0419,
+      "step": 9210
+    },
+    {
+      "epoch": 35.5984555984556,
+      "grad_norm": 0.29113438725471497,
+      "learning_rate": 1.6583613003535226e-06,
+      "loss": 0.035,
+      "step": 9220
+    },
+    {
+      "epoch": 35.63706563706564,
+      "grad_norm": 0.7811099886894226,
+      "learning_rate": 1.6163943943908522e-06,
+      "loss": 0.0357,
+      "step": 9230
+    },
+    {
+      "epoch": 35.67567567567568,
+      "grad_norm": 0.2843995690345764,
+      "learning_rate": 1.5749566037873476e-06,
+      "loss": 0.0404,
+      "step": 9240
+    },
+    {
+      "epoch": 35.714285714285715,
+      "grad_norm": 0.3492255210876465,
+      "learning_rate": 1.5340483817000428e-06,
+      "loss": 0.0207,
+      "step": 9250
+    },
+    {
+      "epoch": 35.75289575289575,
+      "grad_norm": 0.23177163302898407,
+      "learning_rate": 1.4936701754947101e-06,
+      "loss": 0.0256,
+      "step": 9260
+    },
+    {
+      "epoch": 35.79150579150579,
+      "grad_norm": 0.4552899897098541,
+      "learning_rate": 1.4538224267409361e-06,
+      "loss": 0.0285,
+      "step": 9270
+    },
+    {
+      "epoch": 35.83011583011583,
+      "grad_norm": 0.4950767159461975,
+      "learning_rate": 1.414505571207314e-06,
+      "loss": 0.031,
+      "step": 9280
+    },
+    {
+      "epoch": 35.86872586872587,
+      "grad_norm": 0.7315733432769775,
+      "learning_rate": 1.3757200388566816e-06,
+      "loss": 0.0388,
+      "step": 9290
+    },
+    {
+      "epoch": 35.907335907335906,
+      "grad_norm": 0.4851663112640381,
+      "learning_rate": 1.3374662538414074e-06,
+      "loss": 0.044,
+      "step": 9300
+    },
+    {
+      "epoch": 35.945945945945944,
+      "grad_norm": 0.4416690468788147,
+      "learning_rate": 1.2997446344987617e-06,
+      "loss": 0.0285,
+      "step": 9310
+    },
+    {
+      "epoch": 35.98455598455598,
+      "grad_norm": 0.3902464509010315,
+      "learning_rate": 1.262555593346315e-06,
+      "loss": 0.0523,
+      "step": 9320
+    },
+    {
+      "epoch": 36.02316602316602,
+      "grad_norm": 0.5559689998626709,
+      "learning_rate": 1.2258995370774685e-06,
+      "loss": 0.0327,
+      "step": 9330
+    },
+    {
+      "epoch": 36.061776061776065,
+      "grad_norm": 0.3863861560821533,
+      "learning_rate": 1.1897768665569798e-06,
+      "loss": 0.0371,
+      "step": 9340
+    },
+    {
+      "epoch": 36.1003861003861,
+      "grad_norm": 0.4584692418575287,
+      "learning_rate": 1.1541879768165954e-06,
+      "loss": 0.0299,
+      "step": 9350
+    },
+    {
+      "epoch": 36.13899613899614,
+      "grad_norm": 0.5071645379066467,
+      "learning_rate": 1.1191332570507085e-06,
+      "loss": 0.0293,
+      "step": 9360
+    },
+    {
+      "epoch": 36.17760617760618,
+      "grad_norm": 0.6098319888114929,
+      "learning_rate": 1.0846130906121132e-06,
+      "loss": 0.0498,
+      "step": 9370
+    },
+    {
+      "epoch": 36.21621621621622,
+      "grad_norm": 0.4361838698387146,
+      "learning_rate": 1.0506278550078131e-06,
+      "loss": 0.0309,
+      "step": 9380
+    },
+    {
+      "epoch": 36.254826254826256,
+      "grad_norm": 0.5391000509262085,
+      "learning_rate": 1.0171779218949185e-06,
+      "loss": 0.038,
+      "step": 9390
+    },
+    {
+      "epoch": 36.293436293436294,
+      "grad_norm": 0.5742671489715576,
+      "learning_rate": 9.842636570765174e-07,
+      "loss": 0.045,
+      "step": 9400
+    },
+    {
+      "epoch": 36.33204633204633,
+      "grad_norm": 0.40355831384658813,
+      "learning_rate": 9.518854204977612e-07,
+      "loss": 0.0405,
+      "step": 9410
+    },
+    {
+      "epoch": 36.37065637065637,
+      "grad_norm": 0.549338698387146,
+      "learning_rate": 9.200435662418349e-07,
+      "loss": 0.0351,
+      "step": 9420
+    },
+    {
+      "epoch": 36.40926640926641,
+      "grad_norm": 0.22841085493564606,
+      "learning_rate": 8.887384425261658e-07,
+      "loss": 0.0275,
+      "step": 9430
+    },
+    {
+      "epoch": 36.447876447876446,
+      "grad_norm": 0.49053749442100525,
+      "learning_rate": 8.579703916985648e-07,
+      "loss": 0.0328,
+      "step": 9440
+    },
+    {
+      "epoch": 36.486486486486484,
+      "grad_norm": 0.2764204442501068,
+      "learning_rate": 8.277397502335194e-07,
+      "loss": 0.0229,
+      "step": 9450
+    },
+    {
+      "epoch": 36.52509652509652,
+      "grad_norm": 0.41354867815971375,
+      "learning_rate": 7.980468487284675e-07,
+      "loss": 0.0281,
+      "step": 9460
+    },
+    {
+      "epoch": 36.56370656370656,
+      "grad_norm": 0.2841521203517914,
+      "learning_rate": 7.688920119002297e-07,
+      "loss": 0.0479,
+      "step": 9470
+    },
+    {
+      "epoch": 36.602316602316606,
+      "grad_norm": 0.13798123598098755,
+      "learning_rate": 7.402755585814269e-07,
+      "loss": 0.0241,
+      "step": 9480
+    },
+    {
+      "epoch": 36.640926640926644,
+      "grad_norm": 0.20515076816082,
+      "learning_rate": 7.121978017170073e-07,
+      "loss": 0.029,
+      "step": 9490
+    },
+    {
+      "epoch": 36.67953667953668,
+      "grad_norm": 0.3081696629524231,
+      "learning_rate": 6.846590483608306e-07,
+      "loss": 0.0216,
+      "step": 9500
+    },
+    {
+      "epoch": 36.71814671814672,
+      "grad_norm": 0.32960227131843567,
+      "learning_rate": 6.576595996722834e-07,
+      "loss": 0.025,
+      "step": 9510
+    },
+    {
+      "epoch": 36.75675675675676,
+      "grad_norm": 0.3410300314426422,
+      "learning_rate": 6.311997509130141e-07,
+      "loss": 0.0305,
+      "step": 9520
+    },
+    {
+      "epoch": 36.795366795366796,
+      "grad_norm": 0.549889862537384,
+      "learning_rate": 6.052797914436803e-07,
+      "loss": 0.038,
+      "step": 9530
+    },
+    {
+      "epoch": 36.833976833976834,
+      "grad_norm": 0.7164877653121948,
+      "learning_rate": 5.799000047208181e-07,
+      "loss": 0.0465,
+      "step": 9540
+    },
+    {
+      "epoch": 36.87258687258687,
+      "grad_norm": 0.554248571395874,
+      "learning_rate": 5.550606682937054e-07,
+      "loss": 0.0319,
+      "step": 9550
+    },
+    {
+      "epoch": 36.91119691119691,
+      "grad_norm": 0.4524341821670532,
+      "learning_rate": 5.307620538013481e-07,
+      "loss": 0.043,
+      "step": 9560
+    },
+    {
+      "epoch": 36.94980694980695,
+      "grad_norm": 0.5518279671669006,
+      "learning_rate": 5.070044269694874e-07,
+      "loss": 0.0536,
+      "step": 9570
+    },
+    {
+      "epoch": 36.988416988416986,
+      "grad_norm": 0.6280217170715332,
+      "learning_rate": 4.837880476077417e-07,
+      "loss": 0.0353,
+      "step": 9580
+    },
+    {
+      "epoch": 37.027027027027025,
+      "grad_norm": 0.5065165162086487,
+      "learning_rate": 4.6111316960670835e-07,
+      "loss": 0.0264,
+      "step": 9590
+    },
+    {
+      "epoch": 37.06563706563706,
+      "grad_norm": 0.40988031029701233,
+      "learning_rate": 4.389800409352218e-07,
+      "loss": 0.0284,
+      "step": 9600
+    },
+    {
+      "epoch": 37.1042471042471,
+      "grad_norm": 0.5161091685295105,
+      "learning_rate": 4.173889036376277e-07,
+      "loss": 0.032,
+      "step": 9610
+    },
+    {
+      "epoch": 37.142857142857146,
+      "grad_norm": 0.5621913075447083,
+      "learning_rate": 3.963399938311463e-07,
+      "loss": 0.0375,
+      "step": 9620
+    },
+    {
+      "epoch": 37.181467181467184,
+      "grad_norm": 0.4827399253845215,
+      "learning_rate": 3.7583354170328545e-07,
+      "loss": 0.0315,
+      "step": 9630
+    },
+    {
+      "epoch": 37.22007722007722,
+      "grad_norm": 0.651488721370697,
+      "learning_rate": 3.558697715093207e-07,
+      "loss": 0.0201,
+      "step": 9640
+    },
+    {
+      "epoch": 37.25868725868726,
+      "grad_norm": 0.3131031394004822,
+      "learning_rate": 3.3644890156983576e-07,
+      "loss": 0.0245,
+      "step": 9650
+    },
+    {
+      "epoch": 37.2972972972973,
+      "grad_norm": 0.3043268322944641,
+      "learning_rate": 3.175711442683638e-07,
+      "loss": 0.0299,
+      "step": 9660
+    },
+    {
+      "epoch": 37.335907335907336,
+      "grad_norm": 0.4199933409690857,
+      "learning_rate": 2.9923670604902197e-07,
+      "loss": 0.0468,
+      "step": 9670
+    },
+    {
+      "epoch": 37.374517374517374,
+      "grad_norm": 0.28356441855430603,
+      "learning_rate": 2.814457874143028e-07,
+      "loss": 0.0255,
+      "step": 9680
+    },
+    {
+      "epoch": 37.41312741312741,
+      "grad_norm": 0.41857004165649414,
+      "learning_rate": 2.641985829228366e-07,
+      "loss": 0.0297,
+      "step": 9690
+    },
+    {
+      "epoch": 37.45173745173745,
+      "grad_norm": 0.3036278486251831,
+      "learning_rate": 2.474952811872877e-07,
+      "loss": 0.0274,
+      "step": 9700
+    },
+    {
+      "epoch": 37.49034749034749,
+      "grad_norm": 0.5052376389503479,
+      "learning_rate": 2.3133606487228397e-07,
+      "loss": 0.0276,
+      "step": 9710
+    },
+    {
+      "epoch": 37.52895752895753,
+      "grad_norm": 0.4760342836380005,
+      "learning_rate": 2.157211106924295e-07,
+      "loss": 0.034,
+      "step": 9720
+    },
+    {
+      "epoch": 37.567567567567565,
+      "grad_norm": 0.5873331427574158,
+      "learning_rate": 2.006505894103672e-07,
+      "loss": 0.0449,
+      "step": 9730
+    },
+    {
+      "epoch": 37.6061776061776,
+      "grad_norm": 0.5967539548873901,
+      "learning_rate": 1.8612466583489696e-07,
+      "loss": 0.0298,
+      "step": 9740
+    },
+    {
+      "epoch": 37.64478764478765,
+      "grad_norm": 0.33010393381118774,
+      "learning_rate": 1.7214349881918834e-07,
+      "loss": 0.0337,
+      "step": 9750
+    },
+    {
+      "epoch": 37.683397683397686,
+      "grad_norm": 0.4182218909263611,
+      "learning_rate": 1.5870724125904845e-07,
+      "loss": 0.035,
+      "step": 9760
+    },
+    {
+      "epoch": 37.722007722007724,
+      "grad_norm": 0.42870447039604187,
+      "learning_rate": 1.4581604009124006e-07,
+      "loss": 0.036,
+      "step": 9770
+    },
+    {
+      "epoch": 37.76061776061776,
+      "grad_norm": 0.5004497170448303,
+      "learning_rate": 1.334700362918717e-07,
+      "loss": 0.0332,
+      "step": 9780
+    },
+    {
+      "epoch": 37.7992277992278,
+      "grad_norm": 0.5800615549087524,
+      "learning_rate": 1.2166936487486015e-07,
+      "loss": 0.0353,
+      "step": 9790
+    },
+    {
+      "epoch": 37.83783783783784,
+      "grad_norm": 0.3894782066345215,
+      "learning_rate": 1.1041415489045914e-07,
+      "loss": 0.0274,
+      "step": 9800
+    },
+    {
+      "epoch": 37.87644787644788,
+      "grad_norm": 0.4974846839904785,
+      "learning_rate": 9.970452942384412e-08,
+      "loss": 0.036,
+      "step": 9810
+    },
+    {
+      "epoch": 37.915057915057915,
+      "grad_norm": 0.23191875219345093,
+      "learning_rate": 8.954060559375754e-08,
+      "loss": 0.0337,
+      "step": 9820
+    },
+    {
+      "epoch": 37.95366795366795,
+      "grad_norm": 0.4186077415943146,
+      "learning_rate": 7.99224945512489e-08,
+      "loss": 0.0352,
+      "step": 9830
+    },
+    {
+      "epoch": 37.99227799227799,
+      "grad_norm": 0.4193514585494995,
+      "learning_rate": 7.085030147843675e-08,
+      "loss": 0.043,
+      "step": 9840
+    },
+    {
+      "epoch": 38.03088803088803,
+      "grad_norm": 0.4409830868244171,
+      "learning_rate": 6.232412558736523e-08,
+      "loss": 0.0399,
+      "step": 9850
+    },
+    {
+      "epoch": 38.06949806949807,
+      "grad_norm": 0.5391518473625183,
+      "learning_rate": 5.434406011893822e-08,
+      "loss": 0.0236,
+      "step": 9860
+    },
+    {
+      "epoch": 38.108108108108105,
+      "grad_norm": 0.4023591876029968,
+      "learning_rate": 4.6910192341864664e-08,
+      "loss": 0.0281,
+      "step": 9870
+    },
+    {
+      "epoch": 38.14671814671814,
+      "grad_norm": 0.39196234941482544,
+      "learning_rate": 4.0022603551737035e-08,
+      "loss": 0.0408,
+      "step": 9880
+    },
+    {
+      "epoch": 38.18532818532819,
+      "grad_norm": 0.4986127018928528,
+      "learning_rate": 3.3681369070120985e-08,
+      "loss": 0.027,
+      "step": 9890
+    },
+    {
+      "epoch": 38.22393822393823,
+      "grad_norm": 0.6418594717979431,
+      "learning_rate": 2.7886558243744866e-08,
+      "loss": 0.0354,
+      "step": 9900
+    },
+    {
+      "epoch": 38.262548262548265,
+      "grad_norm": 0.324028879404068,
+      "learning_rate": 2.2638234443722596e-08,
+      "loss": 0.0377,
+      "step": 9910
+    },
+    {
+      "epoch": 38.3011583011583,
+      "grad_norm": 0.30436280369758606,
+      "learning_rate": 1.7936455064887504e-08,
+      "loss": 0.0422,
+      "step": 9920
+    },
+    {
+      "epoch": 38.33976833976834,
+      "grad_norm": 0.5513790249824524,
+      "learning_rate": 1.378127152514841e-08,
+      "loss": 0.0288,
+      "step": 9930
+    },
+    {
+      "epoch": 38.37837837837838,
+      "grad_norm": 0.3686314821243286,
+      "learning_rate": 1.0172729264917857e-08,
+      "loss": 0.0329,
+      "step": 9940
+    },
+    {
+      "epoch": 38.41698841698842,
+      "grad_norm": 0.33329352736473083,
+      "learning_rate": 7.1108677466458215e-09,
+      "loss": 0.0195,
+      "step": 9950
+    },
+    {
+      "epoch": 38.455598455598455,
+      "grad_norm": 0.4355686604976654,
+      "learning_rate": 4.595720454353414e-09,
+      "loss": 0.0265,
+      "step": 9960
+    },
+    {
+      "epoch": 38.49420849420849,
+      "grad_norm": 0.3596385419368744,
+      "learning_rate": 2.627314893294264e-09,
+      "loss": 0.0319,
+      "step": 9970
+    },
+    {
+      "epoch": 38.53281853281853,
+      "grad_norm": 0.4059530198574066,
+      "learning_rate": 1.2056725896270048e-09,
+      "loss": 0.0335,
+      "step": 9980
+    },
+    {
+      "epoch": 38.57142857142857,
+      "grad_norm": 0.6819077730178833,
+      "learning_rate": 3.308090902098826e-10,
+      "loss": 0.0307,
+      "step": 9990
+    },
+    {
+      "epoch": 38.61003861003861,
+      "grad_norm": 0.4189014136791229,
+      "learning_rate": 2.7339624120159555e-12,
+      "loss": 0.0401,
+      "step": 10000
+    },
+    {
+      "epoch": 38.61003861003861,
+      "step": 10000,
+      "total_flos": 0.0,
+      "train_loss": 0.08432228177487851,
+      "train_runtime": 8918.6677,
+      "train_samples_per_second": 17.94,
+      "train_steps_per_second": 1.121
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 39,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}