diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,21273 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 3033,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009891196834817012,
+      "grad_norm": 47.47864436307981,
+      "learning_rate": 1.6447368421052632e-07,
+      "loss": 11.4052,
+      "step": 1
+    },
+    {
+      "epoch": 0.0019782393669634025,
+      "grad_norm": 45.91292076188422,
+      "learning_rate": 3.2894736842105264e-07,
+      "loss": 11.2815,
+      "step": 2
+    },
+    {
+      "epoch": 0.002967359050445104,
+      "grad_norm": 47.726852516408236,
+      "learning_rate": 4.934210526315789e-07,
+      "loss": 11.3935,
+      "step": 3
+    },
+    {
+      "epoch": 0.003956478733926805,
+      "grad_norm": 45.10574020271687,
+      "learning_rate": 6.578947368421053e-07,
+      "loss": 11.5897,
+      "step": 4
+    },
+    {
+      "epoch": 0.004945598417408506,
+      "grad_norm": 50.8689323614659,
+      "learning_rate": 8.223684210526316e-07,
+      "loss": 11.2181,
+      "step": 5
+    },
+    {
+      "epoch": 0.005934718100890208,
+      "grad_norm": 44.83636067584137,
+      "learning_rate": 9.868421052631579e-07,
+      "loss": 11.543,
+      "step": 6
+    },
+    {
+      "epoch": 0.006923837784371909,
+      "grad_norm": 44.619692790934934,
+      "learning_rate": 1.1513157894736842e-06,
+      "loss": 11.364,
+      "step": 7
+    },
+    {
+      "epoch": 0.00791295746785361,
+      "grad_norm": 46.17981710645455,
+      "learning_rate": 1.3157894736842106e-06,
+      "loss": 11.372,
+      "step": 8
+    },
+    {
+      "epoch": 0.008902077151335312,
+      "grad_norm": 49.73917366470631,
+      "learning_rate": 1.480263157894737e-06,
+      "loss": 11.0869,
+      "step": 9
+    },
+    {
+      "epoch": 0.009891196834817012,
+      "grad_norm": 52.40342778684296,
+      "learning_rate": 1.6447368421052632e-06,
+      "loss": 11.054,
+      "step": 10
+    },
+    {
+      "epoch": 0.010880316518298714,
+      "grad_norm": 48.69950321668828,
+      "learning_rate": 1.8092105263157896e-06,
+      "loss": 11.0688,
+      "step": 11
+    },
+    {
+      "epoch": 0.011869436201780416,
+      "grad_norm": 63.65227677037272,
+      "learning_rate": 1.9736842105263157e-06,
+      "loss": 10.3072,
+      "step": 12
+    },
+    {
+      "epoch": 0.012858555885262116,
+      "grad_norm": 79.55368321861843,
+      "learning_rate": 2.138157894736842e-06,
+      "loss": 9.7245,
+      "step": 13
+    },
+    {
+      "epoch": 0.013847675568743818,
+      "grad_norm": 82.8103382216083,
+      "learning_rate": 2.3026315789473684e-06,
+      "loss": 9.4617,
+      "step": 14
+    },
+    {
+      "epoch": 0.01483679525222552,
+      "grad_norm": 84.84620253448091,
+      "learning_rate": 2.4671052631578948e-06,
+      "loss": 9.3283,
+      "step": 15
+    },
+    {
+      "epoch": 0.01582591493570722,
+      "grad_norm": 72.7809868244949,
+      "learning_rate": 2.631578947368421e-06,
+      "loss": 4.3421,
+      "step": 16
+    },
+    {
+      "epoch": 0.016815034619188922,
+      "grad_norm": 68.45135162785618,
+      "learning_rate": 2.7960526315789475e-06,
+      "loss": 4.768,
+      "step": 17
+    },
+    {
+      "epoch": 0.017804154302670624,
+      "grad_norm": 65.03039569474328,
+      "learning_rate": 2.960526315789474e-06,
+      "loss": 4.418,
+      "step": 18
+    },
+    {
+      "epoch": 0.018793273986152326,
+      "grad_norm": 45.947045347522504,
+      "learning_rate": 3.125e-06,
+      "loss": 3.3595,
+      "step": 19
+    },
+    {
+      "epoch": 0.019782393669634024,
+      "grad_norm": 43.95974652547691,
+      "learning_rate": 3.2894736842105265e-06,
+      "loss": 3.1867,
+      "step": 20
+    },
+    {
+      "epoch": 0.020771513353115726,
+      "grad_norm": 9.308745952351916,
+      "learning_rate": 3.4539473684210533e-06,
+      "loss": 1.9175,
+      "step": 21
+    },
+    {
+      "epoch": 0.021760633036597428,
+      "grad_norm": 5.755918880001644,
+      "learning_rate": 3.618421052631579e-06,
+      "loss": 1.536,
+      "step": 22
+    },
+    {
+      "epoch": 0.02274975272007913,
+      "grad_norm": 4.930173474585826,
+      "learning_rate": 3.7828947368421055e-06,
+      "loss": 1.4667,
+      "step": 23
+    },
+    {
+      "epoch": 0.02373887240356083,
+      "grad_norm": 4.28733293354817,
+      "learning_rate": 3.9473684210526315e-06,
+      "loss": 1.4006,
+      "step": 24
+    },
+    {
+      "epoch": 0.024727992087042534,
+      "grad_norm": 3.829716127893356,
+      "learning_rate": 4.111842105263159e-06,
+      "loss": 1.532,
+      "step": 25
+    },
+    {
+      "epoch": 0.025717111770524232,
+      "grad_norm": 3.0324967863470813,
+      "learning_rate": 4.276315789473684e-06,
+      "loss": 1.4976,
+      "step": 26
+    },
+    {
+      "epoch": 0.026706231454005934,
+      "grad_norm": 2.5586232440965344,
+      "learning_rate": 4.4407894736842105e-06,
+      "loss": 1.3703,
+      "step": 27
+    },
+    {
+      "epoch": 0.027695351137487636,
+      "grad_norm": 1.9156066610224773,
+      "learning_rate": 4.605263157894737e-06,
+      "loss": 1.1345,
+      "step": 28
+    },
+    {
+      "epoch": 0.028684470820969338,
+      "grad_norm": 1.74939443115538,
+      "learning_rate": 4.769736842105264e-06,
+      "loss": 1.2478,
+      "step": 29
+    },
+    {
+      "epoch": 0.02967359050445104,
+      "grad_norm": 1.5831108352532686,
+      "learning_rate": 4.9342105263157895e-06,
+      "loss": 1.1836,
+      "step": 30
+    },
+    {
+      "epoch": 0.03066271018793274,
+      "grad_norm": 2.2356207014234997,
+      "learning_rate": 5.098684210526316e-06,
+      "loss": 1.3763,
+      "step": 31
+    },
+    {
+      "epoch": 0.03165182987141444,
+      "grad_norm": 1.1026657175125092,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.0571,
+      "step": 32
+    },
+    {
+      "epoch": 0.032640949554896145,
+      "grad_norm": 1.0672135675760805,
+      "learning_rate": 5.4276315789473686e-06,
+      "loss": 1.1693,
+      "step": 33
+    },
+    {
+      "epoch": 0.033630069238377844,
+      "grad_norm": 0.9532916815841181,
+      "learning_rate": 5.592105263157895e-06,
+      "loss": 0.9262,
+      "step": 34
+    },
+    {
+      "epoch": 0.03461918892185954,
+      "grad_norm": 0.8351774057987832,
+      "learning_rate": 5.756578947368421e-06,
+      "loss": 0.9843,
+      "step": 35
+    },
+    {
+      "epoch": 0.03560830860534125,
+      "grad_norm": 0.7152560683189684,
+      "learning_rate": 5.921052631578948e-06,
+      "loss": 0.8934,
+      "step": 36
+    },
+    {
+      "epoch": 0.036597428288822946,
+      "grad_norm": 0.8605762164117818,
+      "learning_rate": 6.085526315789474e-06,
+      "loss": 1.0897,
+      "step": 37
+    },
+    {
+      "epoch": 0.03758654797230465,
+      "grad_norm": 0.8760202419017974,
+      "learning_rate": 6.25e-06,
+      "loss": 1.1443,
+      "step": 38
+    },
+    {
+      "epoch": 0.03857566765578635,
+      "grad_norm": 0.6624437855685319,
+      "learning_rate": 6.4144736842105275e-06,
+      "loss": 0.8699,
+      "step": 39
+    },
+    {
+      "epoch": 0.03956478733926805,
+      "grad_norm": 0.7049615396906419,
+      "learning_rate": 6.578947368421053e-06,
+      "loss": 0.9151,
+      "step": 40
+    },
+    {
+      "epoch": 0.040553907022749754,
+      "grad_norm": 0.7854506183295861,
+      "learning_rate": 6.743421052631579e-06,
+      "loss": 0.9542,
+      "step": 41
+    },
+    {
+      "epoch": 0.04154302670623145,
+      "grad_norm": 0.6223807683273548,
+      "learning_rate": 6.9078947368421065e-06,
+      "loss": 0.9371,
+      "step": 42
+    },
+    {
+      "epoch": 0.04253214638971316,
+      "grad_norm": 0.6248927306368205,
+      "learning_rate": 7.072368421052632e-06,
+      "loss": 0.9072,
+      "step": 43
+    },
+    {
+      "epoch": 0.043521266073194856,
+      "grad_norm": 0.647479183993099,
+      "learning_rate": 7.236842105263158e-06,
+      "loss": 0.9054,
+      "step": 44
+    },
+    {
+      "epoch": 0.04451038575667656,
+      "grad_norm": 0.5889797054704718,
+      "learning_rate": 7.401315789473684e-06,
+      "loss": 1.0402,
+      "step": 45
+    },
+    {
+      "epoch": 0.04549950544015826,
+      "grad_norm": 0.4967497474491158,
+      "learning_rate": 7.565789473684211e-06,
+      "loss": 0.9069,
+      "step": 46
+    },
+    {
+      "epoch": 0.04648862512363996,
+      "grad_norm": 0.5661932222672871,
+      "learning_rate": 7.730263157894737e-06,
+      "loss": 0.9583,
+      "step": 47
+    },
+    {
+      "epoch": 0.04747774480712166,
+      "grad_norm": 6.156382243862993,
+      "learning_rate": 7.894736842105263e-06,
+      "loss": 0.977,
+      "step": 48
+    },
+    {
+      "epoch": 0.04846686449060336,
+      "grad_norm": 0.6197962682604914,
+      "learning_rate": 8.05921052631579e-06,
+      "loss": 0.7968,
+      "step": 49
+    },
+    {
+      "epoch": 0.04945598417408507,
+      "grad_norm": 0.5935105377105369,
+      "learning_rate": 8.223684210526317e-06,
+      "loss": 0.8727,
+      "step": 50
+    },
+    {
+      "epoch": 0.050445103857566766,
+      "grad_norm": 0.5139477978028617,
+      "learning_rate": 8.388157894736843e-06,
+      "loss": 0.9035,
+      "step": 51
+    },
+    {
+      "epoch": 0.051434223541048464,
+      "grad_norm": 0.48831574166447717,
+      "learning_rate": 8.552631578947368e-06,
+      "loss": 0.7875,
+      "step": 52
+    },
+    {
+      "epoch": 0.05242334322453017,
+      "grad_norm": 0.5106264570172955,
+      "learning_rate": 8.717105263157894e-06,
+      "loss": 0.9089,
+      "step": 53
+    },
+    {
+      "epoch": 0.05341246290801187,
+      "grad_norm": 0.48126592930194784,
+      "learning_rate": 8.881578947368421e-06,
+      "loss": 0.7222,
+      "step": 54
+    },
+    {
+      "epoch": 0.05440158259149357,
+      "grad_norm": 0.3816418172871715,
+      "learning_rate": 9.046052631578948e-06,
+      "loss": 0.7726,
+      "step": 55
+    },
+    {
+      "epoch": 0.05539070227497527,
+      "grad_norm": 0.3805932483187457,
+      "learning_rate": 9.210526315789474e-06,
+      "loss": 0.7265,
+      "step": 56
+    },
+    {
+      "epoch": 0.05637982195845697,
+      "grad_norm": 0.35478918543818216,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.7009,
+      "step": 57
+    },
+    {
+      "epoch": 0.057368941641938676,
+      "grad_norm": 0.34092363071107723,
+      "learning_rate": 9.539473684210528e-06,
+      "loss": 0.7749,
+      "step": 58
+    },
+    {
+      "epoch": 0.058358061325420374,
+      "grad_norm": 0.36162518512491465,
+      "learning_rate": 9.703947368421054e-06,
+      "loss": 0.8226,
+      "step": 59
+    },
+    {
+      "epoch": 0.05934718100890208,
+      "grad_norm": 0.40209857091787987,
+      "learning_rate": 9.868421052631579e-06,
+      "loss": 0.836,
+      "step": 60
+    },
+    {
+      "epoch": 0.06033630069238378,
+      "grad_norm": 0.3549385133528555,
+      "learning_rate": 1.0032894736842106e-05,
+      "loss": 0.7356,
+      "step": 61
+    },
+    {
+      "epoch": 0.06132542037586548,
+      "grad_norm": 0.3578780388034424,
+      "learning_rate": 1.0197368421052632e-05,
+      "loss": 0.8599,
+      "step": 62
+    },
+    {
+      "epoch": 0.06231454005934718,
+      "grad_norm": 0.3669356895971538,
+      "learning_rate": 1.0361842105263159e-05,
+      "loss": 0.8198,
+      "step": 63
+    },
+    {
+      "epoch": 0.06330365974282888,
+      "grad_norm": 0.3543644351832402,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 0.8604,
+      "step": 64
+    },
+    {
+      "epoch": 0.06429277942631058,
+      "grad_norm": 0.3713938433998872,
+      "learning_rate": 1.0690789473684212e-05,
+      "loss": 0.8382,
+      "step": 65
+    },
+    {
+      "epoch": 0.06528189910979229,
+      "grad_norm": 0.3340439100059536,
+      "learning_rate": 1.0855263157894737e-05,
+      "loss": 0.7577,
+      "step": 66
+    },
+    {
+      "epoch": 0.06627101879327399,
+      "grad_norm": 0.31244239116416433,
+      "learning_rate": 1.1019736842105263e-05,
+      "loss": 0.714,
+      "step": 67
+    },
+    {
+      "epoch": 0.06726013847675569,
+      "grad_norm": 0.36393336511648067,
+      "learning_rate": 1.118421052631579e-05,
+      "loss": 0.8557,
+      "step": 68
+    },
+    {
+      "epoch": 0.06824925816023739,
+      "grad_norm": 0.39874922213238695,
+      "learning_rate": 1.1348684210526317e-05,
+      "loss": 0.7143,
+      "step": 69
+    },
+    {
+      "epoch": 0.06923837784371908,
+      "grad_norm": 0.3354909696602849,
+      "learning_rate": 1.1513157894736843e-05,
+      "loss": 0.8087,
+      "step": 70
+    },
+    {
+      "epoch": 0.0702274975272008,
+      "grad_norm": 0.3195954665583774,
+      "learning_rate": 1.167763157894737e-05,
+      "loss": 0.7187,
+      "step": 71
+    },
+    {
+      "epoch": 0.0712166172106825,
+      "grad_norm": 0.3134655600001495,
+      "learning_rate": 1.1842105263157895e-05,
+      "loss": 0.8406,
+      "step": 72
+    },
+    {
+      "epoch": 0.0722057368941642,
+      "grad_norm": 0.3026614417415018,
+      "learning_rate": 1.200657894736842e-05,
+      "loss": 0.8183,
+      "step": 73
+    },
+    {
+      "epoch": 0.07319485657764589,
+      "grad_norm": 0.3216818743924892,
+      "learning_rate": 1.2171052631578948e-05,
+      "loss": 0.849,
+      "step": 74
+    },
+    {
+      "epoch": 0.07418397626112759,
+      "grad_norm": 0.3667927953864343,
+      "learning_rate": 1.2335526315789473e-05,
+      "loss": 0.8876,
+      "step": 75
+    },
+    {
+      "epoch": 0.0751730959446093,
+      "grad_norm": 0.31846281371038865,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7137,
+      "step": 76
+    },
+    {
+      "epoch": 0.076162215628091,
+      "grad_norm": 0.3649775475138143,
+      "learning_rate": 1.2664473684210526e-05,
+      "loss": 0.9686,
+      "step": 77
+    },
+    {
+      "epoch": 0.0771513353115727,
+      "grad_norm": 0.3552395073907018,
+      "learning_rate": 1.2828947368421055e-05,
+      "loss": 0.9255,
+      "step": 78
+    },
+    {
+      "epoch": 0.0781404549950544,
+      "grad_norm": 0.3362980375908929,
+      "learning_rate": 1.299342105263158e-05,
+      "loss": 0.7663,
+      "step": 79
+    },
+    {
+      "epoch": 0.0791295746785361,
+      "grad_norm": 0.3332486864873228,
+      "learning_rate": 1.3157894736842106e-05,
+      "loss": 0.8464,
+      "step": 80
+    },
+    {
+      "epoch": 0.08011869436201781,
+      "grad_norm": 0.3563264366987691,
+      "learning_rate": 1.3322368421052633e-05,
+      "loss": 0.7251,
+      "step": 81
+    },
+    {
+      "epoch": 0.08110781404549951,
+      "grad_norm": 0.32718462995919856,
+      "learning_rate": 1.3486842105263159e-05,
+      "loss": 0.8337,
+      "step": 82
+    },
+    {
+      "epoch": 0.0820969337289812,
+      "grad_norm": 0.3098690267315848,
+      "learning_rate": 1.3651315789473684e-05,
+      "loss": 0.7089,
+      "step": 83
+    },
+    {
+      "epoch": 0.0830860534124629,
+      "grad_norm": 0.33231060025498604,
+      "learning_rate": 1.3815789473684213e-05,
+      "loss": 0.6336,
+      "step": 84
+    },
+    {
+      "epoch": 0.0840751730959446,
+      "grad_norm": 0.28062844248423013,
+      "learning_rate": 1.3980263157894739e-05,
+      "loss": 0.6994,
+      "step": 85
+    },
+    {
+      "epoch": 0.08506429277942631,
+      "grad_norm": 0.29411686322877856,
+      "learning_rate": 1.4144736842105264e-05,
+      "loss": 0.6726,
+      "step": 86
+    },
+    {
+      "epoch": 0.08605341246290801,
+      "grad_norm": 0.34261251457783337,
+      "learning_rate": 1.430921052631579e-05,
+      "loss": 0.8291,
+      "step": 87
+    },
+    {
+      "epoch": 0.08704253214638971,
+      "grad_norm": 0.36178717911541053,
+      "learning_rate": 1.4473684210526317e-05,
+      "loss": 0.8385,
+      "step": 88
+    },
+    {
+      "epoch": 0.08803165182987141,
+      "grad_norm": 0.3461353549466678,
+      "learning_rate": 1.4638157894736842e-05,
+      "loss": 0.7122,
+      "step": 89
+    },
+    {
+      "epoch": 0.08902077151335312,
+      "grad_norm": 0.35500836343925757,
+      "learning_rate": 1.4802631578947368e-05,
+      "loss": 0.7191,
+      "step": 90
+    },
+    {
+      "epoch": 0.09000989119683482,
+      "grad_norm": 0.3361253921537601,
+      "learning_rate": 1.4967105263157897e-05,
+      "loss": 0.8086,
+      "step": 91
+    },
+    {
+      "epoch": 0.09099901088031652,
+      "grad_norm": 0.33504758656742495,
+      "learning_rate": 1.5131578947368422e-05,
+      "loss": 0.7375,
+      "step": 92
+    },
+    {
+      "epoch": 0.09198813056379822,
+      "grad_norm": 0.3334568733991201,
+      "learning_rate": 1.5296052631578946e-05,
+      "loss": 0.7225,
+      "step": 93
+    },
+    {
+      "epoch": 0.09297725024727992,
+      "grad_norm": 0.3350306861891048,
+      "learning_rate": 1.5460526315789475e-05,
+      "loss": 0.7998,
+      "step": 94
+    },
+    {
+      "epoch": 0.09396636993076163,
+      "grad_norm": 0.38462200040273103,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.8569,
+      "step": 95
+    },
+    {
+      "epoch": 0.09495548961424333,
+      "grad_norm": 0.34279551157272875,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 0.6453,
+      "step": 96
+    },
+    {
+      "epoch": 0.09594460929772503,
+      "grad_norm": 0.33073887008527225,
+      "learning_rate": 1.5953947368421055e-05,
+      "loss": 0.7839,
+      "step": 97
+    },
+    {
+      "epoch": 0.09693372898120672,
+      "grad_norm": 0.2931675479755555,
+      "learning_rate": 1.611842105263158e-05,
+      "loss": 0.7174,
+      "step": 98
+    },
+    {
+      "epoch": 0.09792284866468842,
+      "grad_norm": 0.3169582507658582,
+      "learning_rate": 1.6282894736842106e-05,
+      "loss": 0.7605,
+      "step": 99
+    },
+    {
+      "epoch": 0.09891196834817013,
+      "grad_norm": 0.42763487156637653,
+      "learning_rate": 1.6447368421052635e-05,
+      "loss": 0.7602,
+      "step": 100
+    },
+    {
+      "epoch": 0.09990108803165183,
+      "grad_norm": 0.29074916324227895,
+      "learning_rate": 1.661184210526316e-05,
+      "loss": 0.6866,
+      "step": 101
+    },
+    {
+      "epoch": 0.10089020771513353,
+      "grad_norm": 0.352348279384898,
+      "learning_rate": 1.6776315789473686e-05,
+      "loss": 0.8076,
+      "step": 102
+    },
+    {
+      "epoch": 0.10187932739861523,
+      "grad_norm": 0.3273893679465929,
+      "learning_rate": 1.694078947368421e-05,
+      "loss": 0.7077,
+      "step": 103
+    },
+    {
+      "epoch": 0.10286844708209693,
+      "grad_norm": 0.3571307875845898,
+      "learning_rate": 1.7105263157894737e-05,
+      "loss": 0.7014,
+      "step": 104
+    },
+    {
+      "epoch": 0.10385756676557864,
+      "grad_norm": 0.33000827078733097,
+      "learning_rate": 1.7269736842105262e-05,
+      "loss": 0.6654,
+      "step": 105
+    },
+    {
+      "epoch": 0.10484668644906034,
+      "grad_norm": 0.4595448328039601,
+      "learning_rate": 1.7434210526315788e-05,
+      "loss": 0.7122,
+      "step": 106
+    },
+    {
+      "epoch": 0.10583580613254204,
+      "grad_norm": 0.35389814780232176,
+      "learning_rate": 1.7598684210526316e-05,
+      "loss": 0.6255,
+      "step": 107
+    },
+    {
+      "epoch": 0.10682492581602374,
+      "grad_norm": 0.35214588487585297,
+      "learning_rate": 1.7763157894736842e-05,
+      "loss": 0.6463,
+      "step": 108
+    },
+    {
+      "epoch": 0.10781404549950543,
+      "grad_norm": 0.37496257339354294,
+      "learning_rate": 1.7927631578947367e-05,
+      "loss": 0.6374,
+      "step": 109
+    },
+    {
+      "epoch": 0.10880316518298715,
+      "grad_norm": 0.3901229505469495,
+      "learning_rate": 1.8092105263157896e-05,
+      "loss": 0.8409,
+      "step": 110
+    },
+    {
+      "epoch": 0.10979228486646884,
+      "grad_norm": 0.36503555388396497,
+      "learning_rate": 1.8256578947368422e-05,
+      "loss": 0.7027,
+      "step": 111
+    },
+    {
+      "epoch": 0.11078140454995054,
+      "grad_norm": 0.39058659212806174,
+      "learning_rate": 1.8421052631578947e-05,
+      "loss": 0.666,
+      "step": 112
+    },
+    {
+      "epoch": 0.11177052423343224,
+      "grad_norm": 0.3393334155039262,
+      "learning_rate": 1.8585526315789476e-05,
+      "loss": 0.6489,
+      "step": 113
+    },
+    {
+      "epoch": 0.11275964391691394,
+      "grad_norm": 0.3641878610374112,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.5997,
+      "step": 114
+    },
+    {
+      "epoch": 0.11374876360039565,
+      "grad_norm": 0.35954765537088884,
+      "learning_rate": 1.8914473684210527e-05,
+      "loss": 0.6643,
+      "step": 115
+    },
+    {
+      "epoch": 0.11473788328387735,
+      "grad_norm": 0.332353400970844,
+      "learning_rate": 1.9078947368421056e-05,
+      "loss": 0.6558,
+      "step": 116
+    },
+    {
+      "epoch": 0.11572700296735905,
+      "grad_norm": 0.3658649526637821,
+      "learning_rate": 1.924342105263158e-05,
+      "loss": 0.748,
+      "step": 117
+    },
+    {
+      "epoch": 0.11671612265084075,
+      "grad_norm": 0.4110068913472326,
+      "learning_rate": 1.9407894736842107e-05,
+      "loss": 0.7679,
+      "step": 118
+    },
+    {
+      "epoch": 0.11770524233432245,
+      "grad_norm": 0.312188988598259,
+      "learning_rate": 1.9572368421052633e-05,
+      "loss": 0.7806,
+      "step": 119
+    },
+    {
+      "epoch": 0.11869436201780416,
+      "grad_norm": 0.39415290875961995,
+      "learning_rate": 1.9736842105263158e-05,
+      "loss": 0.6902,
+      "step": 120
+    },
+    {
+      "epoch": 0.11968348170128586,
+      "grad_norm": 0.30258592751919755,
+      "learning_rate": 1.9901315789473684e-05,
+      "loss": 0.6828,
+      "step": 121
+    },
+    {
+      "epoch": 0.12067260138476756,
+      "grad_norm": 0.32157745711913,
+      "learning_rate": 2.0065789473684213e-05,
+      "loss": 0.7062,
+      "step": 122
+    },
+    {
+      "epoch": 0.12166172106824925,
+      "grad_norm": 0.38984990793746416,
+      "learning_rate": 2.0230263157894738e-05,
+      "loss": 0.7383,
+      "step": 123
+    },
+    {
+      "epoch": 0.12265084075173097,
+      "grad_norm": 0.4111544139628007,
+      "learning_rate": 2.0394736842105264e-05,
+      "loss": 0.6269,
+      "step": 124
+    },
+    {
+      "epoch": 0.12363996043521266,
+      "grad_norm": 0.39448247544238374,
+      "learning_rate": 2.055921052631579e-05,
+      "loss": 0.7416,
+      "step": 125
+    },
+    {
+      "epoch": 0.12462908011869436,
+      "grad_norm": 0.3444637293280687,
+      "learning_rate": 2.0723684210526318e-05,
+      "loss": 0.6733,
+      "step": 126
+    },
+    {
+      "epoch": 0.12561819980217606,
+      "grad_norm": 0.4001985495233771,
+      "learning_rate": 2.0888157894736843e-05,
+      "loss": 0.6515,
+      "step": 127
+    },
+    {
+      "epoch": 0.12660731948565776,
+      "grad_norm": 0.3794186330915447,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 0.662,
+      "step": 128
+    },
+    {
+      "epoch": 0.12759643916913946,
+      "grad_norm": 0.38824093039999913,
+      "learning_rate": 2.1217105263157898e-05,
+      "loss": 0.7238,
+      "step": 129
+    },
+    {
+      "epoch": 0.12858555885262116,
+      "grad_norm": 0.4556524964885303,
+      "learning_rate": 2.1381578947368423e-05,
+      "loss": 0.6779,
+      "step": 130
+    },
+    {
+      "epoch": 0.12957467853610286,
+      "grad_norm": 0.3643359091192952,
+      "learning_rate": 2.154605263157895e-05,
+      "loss": 0.673,
+      "step": 131
+    },
+    {
+      "epoch": 0.13056379821958458,
+      "grad_norm": 0.491430142042037,
+      "learning_rate": 2.1710526315789474e-05,
+      "loss": 0.6031,
+      "step": 132
+    },
+    {
+      "epoch": 0.13155291790306628,
+      "grad_norm": 0.3644642242254432,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.6054,
+      "step": 133
+    },
+    {
+      "epoch": 0.13254203758654798,
+      "grad_norm": 0.4391260070830592,
+      "learning_rate": 2.2039473684210525e-05,
+      "loss": 0.569,
+      "step": 134
+    },
+    {
+      "epoch": 0.13353115727002968,
+      "grad_norm": 0.3275145794011892,
+      "learning_rate": 2.2203947368421054e-05,
+      "loss": 0.6724,
+      "step": 135
+    },
+    {
+      "epoch": 0.13452027695351138,
+      "grad_norm": 0.4672558471628718,
+      "learning_rate": 2.236842105263158e-05,
+      "loss": 0.6973,
+      "step": 136
+    },
+    {
+      "epoch": 0.13550939663699307,
+      "grad_norm": 0.37484287127842214,
+      "learning_rate": 2.2532894736842105e-05,
+      "loss": 0.6024,
+      "step": 137
+    },
+    {
+      "epoch": 0.13649851632047477,
+      "grad_norm": 0.3741616001117786,
+      "learning_rate": 2.2697368421052634e-05,
+      "loss": 0.6354,
+      "step": 138
+    },
+    {
+      "epoch": 0.13748763600395647,
+      "grad_norm": 0.40705218782451663,
+      "learning_rate": 2.286184210526316e-05,
+      "loss": 0.7532,
+      "step": 139
+    },
+    {
+      "epoch": 0.13847675568743817,
+      "grad_norm": 0.3677599289459387,
+      "learning_rate": 2.3026315789473685e-05,
+      "loss": 0.7191,
+      "step": 140
+    },
+    {
+      "epoch": 0.1394658753709199,
+      "grad_norm": 0.3999859660311767,
+      "learning_rate": 2.3190789473684214e-05,
+      "loss": 0.6836,
+      "step": 141
+    },
+    {
+      "epoch": 0.1404549950544016,
+      "grad_norm": 0.38230856108223893,
+      "learning_rate": 2.335526315789474e-05,
+      "loss": 0.6404,
+      "step": 142
+    },
+    {
+      "epoch": 0.1414441147378833,
+      "grad_norm": 0.3628560484729775,
+      "learning_rate": 2.3519736842105265e-05,
+      "loss": 0.654,
+      "step": 143
+    },
+    {
+      "epoch": 0.142433234421365,
+      "grad_norm": 0.3448007883826662,
+      "learning_rate": 2.368421052631579e-05,
+      "loss": 0.6698,
+      "step": 144
+    },
+    {
+      "epoch": 0.1434223541048467,
+      "grad_norm": 0.5552741489616734,
+      "learning_rate": 2.3848684210526316e-05,
+      "loss": 0.7492,
+      "step": 145
+    },
+    {
+      "epoch": 0.1444114737883284,
+      "grad_norm": 0.3342617236256358,
+      "learning_rate": 2.401315789473684e-05,
+      "loss": 0.5378,
+      "step": 146
+    },
+    {
+      "epoch": 0.14540059347181009,
+      "grad_norm": 0.3994524276771364,
+      "learning_rate": 2.4177631578947367e-05,
+      "loss": 0.5827,
+      "step": 147
+    },
+    {
+      "epoch": 0.14638971315529178,
+      "grad_norm": 0.40544863606069553,
+      "learning_rate": 2.4342105263157896e-05,
+      "loss": 0.6543,
+      "step": 148
+    },
+    {
+      "epoch": 0.14737883283877348,
+      "grad_norm": 0.36633930457542324,
+      "learning_rate": 2.450657894736842e-05,
+      "loss": 0.6954,
+      "step": 149
+    },
+    {
+      "epoch": 0.14836795252225518,
+      "grad_norm": 0.36860483728084853,
+      "learning_rate": 2.4671052631578947e-05,
+      "loss": 0.6117,
+      "step": 150
+    },
+    {
+      "epoch": 0.1493570722057369,
+      "grad_norm": 0.40299273023116766,
+      "learning_rate": 2.4835526315789476e-05,
+      "loss": 0.6677,
+      "step": 151
+    },
+    {
+      "epoch": 0.1503461918892186,
+      "grad_norm": 0.46292039106861615,
+      "learning_rate": 2.5e-05,
+      "loss": 0.7554,
+      "step": 152
+    },
+    {
+      "epoch": 0.1513353115727003,
+      "grad_norm": 0.3858429666089543,
+      "learning_rate": 2.5164473684210527e-05,
+      "loss": 0.6302,
+      "step": 153
+    },
+    {
+      "epoch": 0.152324431256182,
+      "grad_norm": 0.5074668120955147,
+      "learning_rate": 2.5328947368421052e-05,
+      "loss": 0.6442,
+      "step": 154
+    },
+    {
+      "epoch": 0.1533135509396637,
+      "grad_norm": 0.37588795715965484,
+      "learning_rate": 2.5493421052631578e-05,
+      "loss": 0.6647,
+      "step": 155
+    },
+    {
+      "epoch": 0.1543026706231454,
+      "grad_norm": 0.563121630811762,
+      "learning_rate": 2.565789473684211e-05,
+      "loss": 0.6455,
+      "step": 156
+    },
+    {
+      "epoch": 0.1552917903066271,
+      "grad_norm": 0.48379869834031464,
+      "learning_rate": 2.5822368421052635e-05,
+      "loss": 0.723,
+      "step": 157
+    },
+    {
+      "epoch": 0.1562809099901088,
+      "grad_norm": 0.3916557190918107,
+      "learning_rate": 2.598684210526316e-05,
+      "loss": 0.5728,
+      "step": 158
+    },
+    {
+      "epoch": 0.1572700296735905,
+      "grad_norm": 0.4241188982146622,
+      "learning_rate": 2.6151315789473686e-05,
+      "loss": 0.617,
+      "step": 159
+    },
+    {
+      "epoch": 0.1582591493570722,
+      "grad_norm": 0.45540106668453567,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 0.6104,
+      "step": 160
+    },
+    {
+      "epoch": 0.15924826904055392,
+      "grad_norm": 0.4247150549586619,
+      "learning_rate": 2.6480263157894737e-05,
+      "loss": 0.6001,
+      "step": 161
+    },
+    {
+      "epoch": 0.16023738872403562,
+      "grad_norm": 0.4460228840551562,
+      "learning_rate": 2.6644736842105266e-05,
+      "loss": 0.5465,
+      "step": 162
+    },
+    {
+      "epoch": 0.16122650840751732,
+      "grad_norm": 0.38351087890774094,
+      "learning_rate": 2.6809210526315792e-05,
+      "loss": 0.6591,
+      "step": 163
+    },
+    {
+      "epoch": 0.16221562809099901,
+      "grad_norm": 0.46934047916659694,
+      "learning_rate": 2.6973684210526317e-05,
+      "loss": 0.7279,
+      "step": 164
+    },
+    {
+      "epoch": 0.1632047477744807,
+      "grad_norm": 0.4222690324413176,
+      "learning_rate": 2.7138157894736843e-05,
+      "loss": 0.6264,
+      "step": 165
+    },
+    {
+      "epoch": 0.1641938674579624,
+      "grad_norm": 0.4697752681692279,
+      "learning_rate": 2.730263157894737e-05,
+      "loss": 0.7027,
+      "step": 166
+    },
+    {
+      "epoch": 0.1651829871414441,
+      "grad_norm": 0.4702974530466207,
+      "learning_rate": 2.7467105263157894e-05,
+      "loss": 0.6994,
+      "step": 167
+    },
+    {
+      "epoch": 0.1661721068249258,
+      "grad_norm": 0.541483976312347,
+      "learning_rate": 2.7631578947368426e-05,
+      "loss": 0.6525,
+      "step": 168
+    },
+    {
+      "epoch": 0.1671612265084075,
+      "grad_norm": 0.3428171204640418,
+      "learning_rate": 2.779605263157895e-05,
+      "loss": 0.6431,
+      "step": 169
+    },
+    {
+      "epoch": 0.1681503461918892,
+      "grad_norm": 0.472693449324751,
+      "learning_rate": 2.7960526315789477e-05,
+      "loss": 0.6693,
+      "step": 170
+    },
+    {
+      "epoch": 0.16913946587537093,
+      "grad_norm": 0.4554753159492845,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.6924,
+      "step": 171
+    },
+    {
+      "epoch": 0.17012858555885263,
+      "grad_norm": 0.4616962159670467,
+      "learning_rate": 2.8289473684210528e-05,
+      "loss": 0.6082,
+      "step": 172
+    },
+    {
+      "epoch": 0.17111770524233433,
+      "grad_norm": 0.42013241457970946,
+      "learning_rate": 2.8453947368421054e-05,
+      "loss": 0.5976,
+      "step": 173
+    },
+    {
+      "epoch": 0.17210682492581603,
+      "grad_norm": 0.49647836517813876,
+      "learning_rate": 2.861842105263158e-05,
+      "loss": 0.6976,
+      "step": 174
+    },
+    {
+      "epoch": 0.17309594460929772,
+      "grad_norm": 0.4656047888630097,
+      "learning_rate": 2.8782894736842108e-05,
+      "loss": 0.6689,
+      "step": 175
+    },
+    {
+      "epoch": 0.17408506429277942,
+      "grad_norm": 0.62173274981684,
+      "learning_rate": 2.8947368421052634e-05,
+      "loss": 0.5864,
+      "step": 176
+    },
+    {
+      "epoch": 0.17507418397626112,
+      "grad_norm": 0.44514460495733604,
+      "learning_rate": 2.911184210526316e-05,
+      "loss": 0.5959,
+      "step": 177
+    },
+    {
+      "epoch": 0.17606330365974282,
+      "grad_norm": 0.6575269476526132,
+      "learning_rate": 2.9276315789473684e-05,
+      "loss": 0.6128,
+      "step": 178
+    },
+    {
+      "epoch": 0.17705242334322452,
+      "grad_norm": 0.4280454874654953,
+      "learning_rate": 2.944078947368421e-05,
+      "loss": 0.7002,
+      "step": 179
+    },
+    {
+      "epoch": 0.17804154302670624,
+      "grad_norm": 0.455273030279468,
+      "learning_rate": 2.9605263157894735e-05,
+      "loss": 0.682,
+      "step": 180
+    },
+    {
+      "epoch": 0.17903066271018794,
+      "grad_norm": 0.4558325615984045,
+      "learning_rate": 2.9769736842105268e-05,
+      "loss": 0.7028,
+      "step": 181
+    },
+    {
+      "epoch": 0.18001978239366964,
+      "grad_norm": 0.5877708520421765,
+      "learning_rate": 2.9934210526315793e-05,
+      "loss": 0.6836,
+      "step": 182
+    },
+    {
+      "epoch": 0.18100890207715134,
+      "grad_norm": 0.4596794024140408,
+      "learning_rate": 3.009868421052632e-05,
+      "loss": 0.6797,
+      "step": 183
+    },
+    {
+      "epoch": 0.18199802176063304,
+      "grad_norm": 0.4319051245779275,
+      "learning_rate": 3.0263157894736844e-05,
+      "loss": 0.6282,
+      "step": 184
+    },
+    {
+      "epoch": 0.18298714144411474,
+      "grad_norm": 0.5318817072015158,
+      "learning_rate": 3.042763157894737e-05,
+      "loss": 0.6233,
+      "step": 185
+    },
+    {
+      "epoch": 0.18397626112759644,
+      "grad_norm": 0.4322921072671814,
+      "learning_rate": 3.059210526315789e-05,
+      "loss": 0.6573,
+      "step": 186
+    },
+    {
+      "epoch": 0.18496538081107813,
+      "grad_norm": 0.4885053114909321,
+      "learning_rate": 3.075657894736843e-05,
+      "loss": 0.6871,
+      "step": 187
+    },
+    {
+      "epoch": 0.18595450049455983,
+      "grad_norm": 0.4473265190512735,
+      "learning_rate": 3.092105263157895e-05,
+      "loss": 0.6173,
+      "step": 188
+    },
+    {
+      "epoch": 0.18694362017804153,
+      "grad_norm": 0.5440786287668955,
+      "learning_rate": 3.108552631578948e-05,
+      "loss": 0.6141,
+      "step": 189
+    },
+    {
+      "epoch": 0.18793273986152326,
+      "grad_norm": 0.35769667117178866,
+      "learning_rate": 3.125e-05,
+      "loss": 0.694,
+      "step": 190
+    },
+    {
+      "epoch": 0.18892185954500496,
+      "grad_norm": 0.5059649250810946,
+      "learning_rate": 3.141447368421053e-05,
+      "loss": 0.6728,
+      "step": 191
+    },
+    {
+      "epoch": 0.18991097922848665,
+      "grad_norm": 0.38729560510609135,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.5709,
+      "step": 192
+    },
+    {
+      "epoch": 0.19090009891196835,
+      "grad_norm": 0.5371616873491071,
+      "learning_rate": 3.174342105263158e-05,
+      "loss": 0.7155,
+      "step": 193
+    },
+    {
+      "epoch": 0.19188921859545005,
+      "grad_norm": 0.6053710744464506,
+      "learning_rate": 3.190789473684211e-05,
+      "loss": 0.6955,
+      "step": 194
+    },
+    {
+      "epoch": 0.19287833827893175,
+      "grad_norm": 0.413136441297711,
+      "learning_rate": 3.207236842105263e-05,
+      "loss": 0.5837,
+      "step": 195
+    },
+    {
+      "epoch": 0.19386745796241345,
+      "grad_norm": 0.6398949332164223,
+      "learning_rate": 3.223684210526316e-05,
+      "loss": 0.7342,
+      "step": 196
+    },
+    {
+      "epoch": 0.19485657764589515,
+      "grad_norm": 0.61809961884169,
+      "learning_rate": 3.240131578947368e-05,
+      "loss": 0.6147,
+      "step": 197
+    },
+    {
+      "epoch": 0.19584569732937684,
+      "grad_norm": 0.44467152661669995,
+      "learning_rate": 3.256578947368421e-05,
+      "loss": 0.5646,
+      "step": 198
+    },
+    {
+      "epoch": 0.19683481701285854,
+      "grad_norm": 0.4851641919265251,
+      "learning_rate": 3.2730263157894734e-05,
+      "loss": 0.6334,
+      "step": 199
+    },
+    {
+      "epoch": 0.19782393669634027,
+      "grad_norm": 0.5062493576014657,
+      "learning_rate": 3.289473684210527e-05,
+      "loss": 0.5911,
+      "step": 200
+    },
+    {
+      "epoch": 0.19881305637982197,
+      "grad_norm": 0.5527966709235791,
+      "learning_rate": 3.305921052631579e-05,
+      "loss": 0.6273,
+      "step": 201
+    },
+    {
+      "epoch": 0.19980217606330367,
+      "grad_norm": 0.633334866950127,
+      "learning_rate": 3.322368421052632e-05,
+      "loss": 0.7025,
+      "step": 202
+    },
+    {
+      "epoch": 0.20079129574678536,
+      "grad_norm": 0.48779295740663486,
+      "learning_rate": 3.338815789473684e-05,
+      "loss": 0.5861,
+      "step": 203
+    },
+    {
+      "epoch": 0.20178041543026706,
+      "grad_norm": 0.6070005751744674,
+      "learning_rate": 3.355263157894737e-05,
+      "loss": 0.7428,
+      "step": 204
+    },
+    {
+      "epoch": 0.20276953511374876,
+      "grad_norm": 0.534136315876146,
+      "learning_rate": 3.371710526315789e-05,
+      "loss": 0.5728,
+      "step": 205
+    },
+    {
+      "epoch": 0.20375865479723046,
+      "grad_norm": 0.48792763692000546,
+      "learning_rate": 3.388157894736842e-05,
+      "loss": 0.586,
+      "step": 206
+    },
+    {
+      "epoch": 0.20474777448071216,
+      "grad_norm": 0.52009416122117,
+      "learning_rate": 3.404605263157895e-05,
+      "loss": 0.4851,
+      "step": 207
+    },
+    {
+      "epoch": 0.20573689416419386,
+      "grad_norm": 0.5698103555541044,
+      "learning_rate": 3.421052631578947e-05,
+      "loss": 0.6331,
+      "step": 208
+    },
+    {
+      "epoch": 0.20672601384767555,
+      "grad_norm": 0.509391746916777,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.5839,
+      "step": 209
+    },
+    {
+      "epoch": 0.20771513353115728,
+      "grad_norm": 0.6184628148195033,
+      "learning_rate": 3.4539473684210524e-05,
+      "loss": 0.7618,
+      "step": 210
+    },
+    {
+      "epoch": 0.20870425321463898,
+      "grad_norm": 0.55512739516661,
+      "learning_rate": 3.470394736842105e-05,
+      "loss": 0.6801,
+      "step": 211
+    },
+    {
+      "epoch": 0.20969337289812068,
+      "grad_norm": 0.6380611857299497,
+      "learning_rate": 3.4868421052631575e-05,
+      "loss": 0.6445,
+      "step": 212
+    },
+    {
+      "epoch": 0.21068249258160238,
+      "grad_norm": 0.48106021749776834,
+      "learning_rate": 3.503289473684211e-05,
+      "loss": 0.704,
+      "step": 213
+    },
+    {
+      "epoch": 0.21167161226508407,
+      "grad_norm": 0.5155510274611654,
+      "learning_rate": 3.519736842105263e-05,
+      "loss": 0.603,
+      "step": 214
+    },
+    {
+      "epoch": 0.21266073194856577,
+      "grad_norm": 0.41115306939250096,
+      "learning_rate": 3.536184210526316e-05,
+      "loss": 0.6204,
+      "step": 215
+    },
+    {
+      "epoch": 0.21364985163204747,
+      "grad_norm": 0.42847524914883034,
+      "learning_rate": 3.5526315789473684e-05,
+      "loss": 0.6293,
+      "step": 216
+    },
+    {
+      "epoch": 0.21463897131552917,
+      "grad_norm": 0.40362083266366067,
+      "learning_rate": 3.569078947368421e-05,
+      "loss": 0.6025,
+      "step": 217
+    },
+    {
+      "epoch": 0.21562809099901087,
+      "grad_norm": 0.36583415568507316,
+      "learning_rate": 3.5855263157894735e-05,
+      "loss": 0.6429,
+      "step": 218
+    },
+    {
+      "epoch": 0.2166172106824926,
+      "grad_norm": 0.3792237531903649,
+      "learning_rate": 3.6019736842105264e-05,
+      "loss": 0.6304,
+      "step": 219
+    },
+    {
+      "epoch": 0.2176063303659743,
+      "grad_norm": 0.5016218500071241,
+      "learning_rate": 3.618421052631579e-05,
+      "loss": 0.6072,
+      "step": 220
+    },
+    {
+      "epoch": 0.218595450049456,
+      "grad_norm": 0.5195805824452137,
+      "learning_rate": 3.6348684210526315e-05,
+      "loss": 0.6955,
+      "step": 221
+    },
+    {
+      "epoch": 0.2195845697329377,
+      "grad_norm": 0.40646432999857474,
+      "learning_rate": 3.6513157894736844e-05,
+      "loss": 0.5941,
+      "step": 222
+    },
+    {
+      "epoch": 0.2205736894164194,
+      "grad_norm": 0.48614279865705684,
+      "learning_rate": 3.6677631578947366e-05,
+      "loss": 0.7392,
+      "step": 223
+    },
+    {
+      "epoch": 0.2215628090999011,
+      "grad_norm": 0.4833284553690217,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.7144,
+      "step": 224
+    },
+    {
+      "epoch": 0.22255192878338279,
+      "grad_norm": 0.4123480628971663,
+      "learning_rate": 3.7006578947368424e-05,
+      "loss": 0.5997,
+      "step": 225
+    },
+    {
+      "epoch": 0.22354104846686448,
+      "grad_norm": 0.58482567880444,
+      "learning_rate": 3.717105263157895e-05,
+      "loss": 0.625,
+      "step": 226
+    },
+    {
+      "epoch": 0.22453016815034618,
+      "grad_norm": 0.4839150358758274,
+      "learning_rate": 3.7335526315789475e-05,
+      "loss": 0.5947,
+      "step": 227
+    },
+    {
+      "epoch": 0.22551928783382788,
+      "grad_norm": 0.5821490050298971,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.63,
+      "step": 228
+    },
+    {
+      "epoch": 0.2265084075173096,
+      "grad_norm": 0.4025831144151843,
+      "learning_rate": 3.7664473684210526e-05,
+      "loss": 0.5721,
+      "step": 229
+    },
+    {
+      "epoch": 0.2274975272007913,
+      "grad_norm": 0.5564325010312093,
+      "learning_rate": 3.7828947368421054e-05,
+      "loss": 0.6745,
+      "step": 230
+    },
+    {
+      "epoch": 0.228486646884273,
+      "grad_norm": 0.559427271458402,
+      "learning_rate": 3.7993421052631577e-05,
+      "loss": 0.6608,
+      "step": 231
+    },
+    {
+      "epoch": 0.2294757665677547,
+      "grad_norm": 0.4961691077638613,
+      "learning_rate": 3.815789473684211e-05,
+      "loss": 0.673,
+      "step": 232
+    },
+    {
+      "epoch": 0.2304648862512364,
+      "grad_norm": 0.4451618595226397,
+      "learning_rate": 3.8322368421052634e-05,
+      "loss": 0.6879,
+      "step": 233
+    },
+    {
+      "epoch": 0.2314540059347181,
+      "grad_norm": 0.4764748813476515,
+      "learning_rate": 3.848684210526316e-05,
+      "loss": 0.6593,
+      "step": 234
+    },
+    {
+      "epoch": 0.2324431256181998,
+      "grad_norm": 0.4781168643440134,
+      "learning_rate": 3.8651315789473685e-05,
+      "loss": 0.6521,
+      "step": 235
+    },
+    {
+      "epoch": 0.2334322453016815,
+      "grad_norm": 0.4317515889033491,
+      "learning_rate": 3.8815789473684214e-05,
+      "loss": 0.7187,
+      "step": 236
+    },
+    {
+      "epoch": 0.2344213649851632,
+      "grad_norm": 0.5010679062349306,
+      "learning_rate": 3.8980263157894736e-05,
+      "loss": 0.6075,
+      "step": 237
+    },
+    {
+      "epoch": 0.2354104846686449,
+      "grad_norm": 0.48215015664396305,
+      "learning_rate": 3.9144736842105265e-05,
+      "loss": 0.6168,
+      "step": 238
+    },
+    {
+      "epoch": 0.23639960435212662,
+      "grad_norm": 0.4522478488377238,
+      "learning_rate": 3.9309210526315794e-05,
+      "loss": 0.6396,
+      "step": 239
+    },
+    {
+      "epoch": 0.23738872403560832,
+      "grad_norm": 0.4577399461188642,
+      "learning_rate": 3.9473684210526316e-05,
+      "loss": 0.7265,
+      "step": 240
+    },
+    {
+      "epoch": 0.23837784371909002,
+      "grad_norm": 2.1998133857858337,
+      "learning_rate": 3.9638157894736845e-05,
+      "loss": 0.6542,
+      "step": 241
+    },
+    {
+      "epoch": 0.23936696340257171,
+      "grad_norm": 0.4803483208680404,
+      "learning_rate": 3.980263157894737e-05,
+      "loss": 0.6325,
+      "step": 242
+    },
+    {
+      "epoch": 0.2403560830860534,
+      "grad_norm": 0.42001637823181315,
+      "learning_rate": 3.9967105263157896e-05,
+      "loss": 0.6449,
+      "step": 243
+    },
+    {
+      "epoch": 0.2413452027695351,
+      "grad_norm": 0.6051649931482518,
+      "learning_rate": 4.0131578947368425e-05,
+      "loss": 0.6015,
+      "step": 244
+    },
+    {
+      "epoch": 0.2423343224530168,
+      "grad_norm": 0.462743464454851,
+      "learning_rate": 4.0296052631578954e-05,
+      "loss": 0.6353,
+      "step": 245
+    },
+    {
+      "epoch": 0.2433234421364985,
+      "grad_norm": 0.5965079725135056,
+      "learning_rate": 4.0460526315789476e-05,
+      "loss": 0.6813,
+      "step": 246
+    },
+    {
+      "epoch": 0.2443125618199802,
+      "grad_norm": 0.515915006211205,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.6897,
+      "step": 247
+    },
+    {
+      "epoch": 0.24530168150346193,
+      "grad_norm": 0.5532763347341395,
+      "learning_rate": 4.078947368421053e-05,
+      "loss": 0.7042,
+      "step": 248
+    },
+    {
+      "epoch": 0.24629080118694363,
+      "grad_norm": 0.46137687834271673,
+      "learning_rate": 4.0953947368421056e-05,
+      "loss": 0.5839,
+      "step": 249
+    },
+    {
+      "epoch": 0.24727992087042533,
+      "grad_norm": 0.47745663163187557,
+      "learning_rate": 4.111842105263158e-05,
+      "loss": 0.6566,
+      "step": 250
+    },
+    {
+      "epoch": 0.24826904055390703,
+      "grad_norm": 0.44493193669631287,
+      "learning_rate": 4.128289473684211e-05,
+      "loss": 0.6223,
+      "step": 251
+    },
+    {
+      "epoch": 0.24925816023738873,
+      "grad_norm": 0.4169310337329136,
+      "learning_rate": 4.1447368421052636e-05,
+      "loss": 0.5799,
+      "step": 252
+    },
+    {
+      "epoch": 0.2502472799208704,
+      "grad_norm": 0.5324378545388998,
+      "learning_rate": 4.161184210526316e-05,
+      "loss": 0.586,
+      "step": 253
+    },
+    {
+      "epoch": 0.2512363996043521,
+      "grad_norm": 0.4713752675672079,
+      "learning_rate": 4.177631578947369e-05,
+      "loss": 0.5844,
+      "step": 254
+    },
+    {
+      "epoch": 0.2522255192878338,
+      "grad_norm": 0.4246792763221131,
+      "learning_rate": 4.194078947368421e-05,
+      "loss": 0.6691,
+      "step": 255
+    },
+    {
+      "epoch": 0.2532146389713155,
+      "grad_norm": 0.5996820122986719,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.6205,
+      "step": 256
+    },
+    {
+      "epoch": 0.2542037586547972,
+      "grad_norm": 0.47226024365214625,
+      "learning_rate": 4.226973684210527e-05,
+      "loss": 0.6654,
+      "step": 257
+    },
+    {
+      "epoch": 0.2551928783382789,
+      "grad_norm": 0.4718080623174236,
+      "learning_rate": 4.2434210526315796e-05,
+      "loss": 0.5252,
+      "step": 258
+    },
+    {
+      "epoch": 0.2561819980217606,
+      "grad_norm": 0.6350779709078309,
+      "learning_rate": 4.259868421052632e-05,
+      "loss": 0.6221,
+      "step": 259
+    },
+    {
+      "epoch": 0.2571711177052423,
+      "grad_norm": 0.5362489170689355,
+      "learning_rate": 4.2763157894736847e-05,
+      "loss": 0.59,
+      "step": 260
+    },
+    {
+      "epoch": 0.258160237388724,
+      "grad_norm": 0.5037993125750314,
+      "learning_rate": 4.292763157894737e-05,
+      "loss": 0.5977,
+      "step": 261
+    },
+    {
+      "epoch": 0.2591493570722057,
+      "grad_norm": 0.7119819210503119,
+      "learning_rate": 4.30921052631579e-05,
+      "loss": 0.5687,
+      "step": 262
+    },
+    {
+      "epoch": 0.26013847675568746,
+      "grad_norm": 0.48734181172443314,
+      "learning_rate": 4.3256578947368426e-05,
+      "loss": 0.5818,
+      "step": 263
+    },
+    {
+      "epoch": 0.26112759643916916,
+      "grad_norm": 0.5307408330986926,
+      "learning_rate": 4.342105263157895e-05,
+      "loss": 0.5485,
+      "step": 264
+    },
+    {
+      "epoch": 0.26211671612265086,
+      "grad_norm": 0.6102212182790211,
+      "learning_rate": 4.358552631578948e-05,
+      "loss": 0.6235,
+      "step": 265
+    },
+    {
+      "epoch": 0.26310583580613256,
+      "grad_norm": 0.5928998977874212,
+      "learning_rate": 4.375e-05,
+      "loss": 0.6044,
+      "step": 266
+    },
+    {
+      "epoch": 0.26409495548961426,
+      "grad_norm": 0.6688550804132813,
+      "learning_rate": 4.391447368421053e-05,
+      "loss": 0.6217,
+      "step": 267
+    },
+    {
+      "epoch": 0.26508407517309596,
+      "grad_norm": 0.7818753906542993,
+      "learning_rate": 4.407894736842105e-05,
+      "loss": 0.5282,
+      "step": 268
+    },
+    {
+      "epoch": 0.26607319485657766,
+      "grad_norm": 1.0091627489834931,
+      "learning_rate": 4.424342105263158e-05,
+      "loss": 0.5692,
+      "step": 269
+    },
+    {
+      "epoch": 0.26706231454005935,
+      "grad_norm": 0.6011818743170259,
+      "learning_rate": 4.440789473684211e-05,
+      "loss": 0.6119,
+      "step": 270
+    },
+    {
+      "epoch": 0.26805143422354105,
+      "grad_norm": 0.8838865925562805,
+      "learning_rate": 4.457236842105264e-05,
+      "loss": 0.6043,
+      "step": 271
+    },
+    {
+      "epoch": 0.26904055390702275,
+      "grad_norm": 0.6314911229429284,
+      "learning_rate": 4.473684210526316e-05,
+      "loss": 0.7673,
+      "step": 272
+    },
+    {
+      "epoch": 0.27002967359050445,
+      "grad_norm": 0.6804637239015356,
+      "learning_rate": 4.490131578947369e-05,
+      "loss": 0.5926,
+      "step": 273
+    },
+    {
+      "epoch": 0.27101879327398615,
+      "grad_norm": 0.547035473768601,
+      "learning_rate": 4.506578947368421e-05,
+      "loss": 0.5431,
+      "step": 274
+    },
+    {
+      "epoch": 0.27200791295746785,
+      "grad_norm": 0.49151905906223226,
+      "learning_rate": 4.523026315789474e-05,
+      "loss": 0.5535,
+      "step": 275
+    },
+    {
+      "epoch": 0.27299703264094954,
+      "grad_norm": 0.4380519372013023,
+      "learning_rate": 4.539473684210527e-05,
+      "loss": 0.7109,
+      "step": 276
+    },
+    {
+      "epoch": 0.27398615232443124,
+      "grad_norm": 0.7217367910139244,
+      "learning_rate": 4.555921052631579e-05,
+      "loss": 0.6292,
+      "step": 277
+    },
+    {
+      "epoch": 0.27497527200791294,
+      "grad_norm": 0.5585776926241103,
+      "learning_rate": 4.572368421052632e-05,
+      "loss": 0.6974,
+      "step": 278
+    },
+    {
+      "epoch": 0.27596439169139464,
+      "grad_norm": 0.4555730594700509,
+      "learning_rate": 4.588815789473684e-05,
+      "loss": 0.6425,
+      "step": 279
+    },
+    {
+      "epoch": 0.27695351137487634,
+      "grad_norm": 0.6389516893536279,
+      "learning_rate": 4.605263157894737e-05,
+      "loss": 0.5537,
+      "step": 280
+    },
+    {
+      "epoch": 0.27794263105835804,
+      "grad_norm": 0.46095286770730254,
+      "learning_rate": 4.621710526315789e-05,
+      "loss": 0.6019,
+      "step": 281
+    },
+    {
+      "epoch": 0.2789317507418398,
+      "grad_norm": 0.44048312996010536,
+      "learning_rate": 4.638157894736843e-05,
+      "loss": 0.612,
+      "step": 282
+    },
+    {
+      "epoch": 0.2799208704253215,
+      "grad_norm": 1.0389287178149595,
+      "learning_rate": 4.654605263157895e-05,
+      "loss": 0.6571,
+      "step": 283
+    },
+    {
+      "epoch": 0.2809099901088032,
+      "grad_norm": 0.6903364570618536,
+      "learning_rate": 4.671052631578948e-05,
+      "loss": 0.6973,
+      "step": 284
+    },
+    {
+      "epoch": 0.2818991097922849,
+      "grad_norm": 0.4477703451666338,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.5894,
+      "step": 285
+    },
+    {
+      "epoch": 0.2828882294757666,
+      "grad_norm": 0.71253667717858,
+      "learning_rate": 4.703947368421053e-05,
+      "loss": 0.6158,
+      "step": 286
+    },
+    {
+      "epoch": 0.2838773491592483,
+      "grad_norm": 0.6605589318131366,
+      "learning_rate": 4.720394736842105e-05,
+      "loss": 0.5547,
+      "step": 287
+    },
+    {
+      "epoch": 0.28486646884273,
+      "grad_norm": 0.4142085186922097,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.6035,
+      "step": 288
+    },
+    {
+      "epoch": 0.2858555885262117,
+      "grad_norm": 0.5522552002749757,
+      "learning_rate": 4.753289473684211e-05,
+      "loss": 0.6205,
+      "step": 289
+    },
+    {
+      "epoch": 0.2868447082096934,
+      "grad_norm": 0.4553692557081208,
+      "learning_rate": 4.769736842105263e-05,
+      "loss": 0.5847,
+      "step": 290
+    },
+    {
+      "epoch": 0.2878338278931751,
+      "grad_norm": 0.4677603638371184,
+      "learning_rate": 4.786184210526316e-05,
+      "loss": 0.6153,
+      "step": 291
+    },
+    {
+      "epoch": 0.2888229475766568,
+      "grad_norm": 0.716448078885814,
+      "learning_rate": 4.802631578947368e-05,
+      "loss": 0.6088,
+      "step": 292
+    },
+    {
+      "epoch": 0.2898120672601385,
+      "grad_norm": 0.5629123910634363,
+      "learning_rate": 4.819078947368421e-05,
+      "loss": 0.7418,
+      "step": 293
+    },
+    {
+      "epoch": 0.29080118694362017,
+      "grad_norm": 0.5684632906175011,
+      "learning_rate": 4.8355263157894734e-05,
+      "loss": 0.6617,
+      "step": 294
+    },
+    {
+      "epoch": 0.29179030662710187,
+      "grad_norm": 0.7817295121483087,
+      "learning_rate": 4.851973684210527e-05,
+      "loss": 0.6587,
+      "step": 295
+    },
+    {
+      "epoch": 0.29277942631058357,
+      "grad_norm": 0.43426647960692005,
+      "learning_rate": 4.868421052631579e-05,
+      "loss": 0.5338,
+      "step": 296
+    },
+    {
+      "epoch": 0.29376854599406527,
+      "grad_norm": 0.7019981788147629,
+      "learning_rate": 4.884868421052632e-05,
+      "loss": 0.6053,
+      "step": 297
+    },
+    {
+      "epoch": 0.29475766567754697,
+      "grad_norm": 3.0951979705836496,
+      "learning_rate": 4.901315789473684e-05,
+      "loss": 0.5918,
+      "step": 298
+    },
+    {
+      "epoch": 0.29574678536102866,
+      "grad_norm": 1.0233134362024605,
+      "learning_rate": 4.917763157894737e-05,
+      "loss": 0.6081,
+      "step": 299
+    },
+    {
+      "epoch": 0.29673590504451036,
+      "grad_norm": 0.9953914245723955,
+      "learning_rate": 4.9342105263157894e-05,
+      "loss": 0.5341,
+      "step": 300
+    },
+    {
+      "epoch": 0.29772502472799206,
+      "grad_norm": 0.7092328246731249,
+      "learning_rate": 4.950657894736843e-05,
+      "loss": 0.7031,
+      "step": 301
+    },
+    {
+      "epoch": 0.2987141444114738,
+      "grad_norm": 1.1657097575552566,
+      "learning_rate": 4.967105263157895e-05,
+      "loss": 0.5386,
+      "step": 302
+    },
+    {
+      "epoch": 0.2997032640949555,
+      "grad_norm": 0.64809864753375,
+      "learning_rate": 4.983552631578948e-05,
+      "loss": 0.5822,
+      "step": 303
+    },
+    {
+      "epoch": 0.3006923837784372,
+      "grad_norm": 10.779831456434477,
+      "learning_rate": 5e-05,
+      "loss": 1.3367,
+      "step": 304
+    },
+    {
+      "epoch": 0.3016815034619189,
+      "grad_norm": 2.263953506492712,
+      "learning_rate": 4.9981678270428736e-05,
+      "loss": 0.652,
+      "step": 305
+    },
+    {
+      "epoch": 0.3026706231454006,
+      "grad_norm": 1.1026406639523274,
+      "learning_rate": 4.996335654085746e-05,
+      "loss": 0.5982,
+      "step": 306
+    },
+    {
+      "epoch": 0.3036597428288823,
+      "grad_norm": 1.5390268402197822,
+      "learning_rate": 4.994503481128619e-05,
+      "loss": 0.5838,
+      "step": 307
+    },
+    {
+      "epoch": 0.304648862512364,
+      "grad_norm": 2.325507669292494,
+      "learning_rate": 4.9926713081714915e-05,
+      "loss": 0.7096,
+      "step": 308
+    },
+    {
+      "epoch": 0.3056379821958457,
+      "grad_norm": 0.9681764004322482,
+      "learning_rate": 4.990839135214365e-05,
+      "loss": 0.5579,
+      "step": 309
+    },
+    {
+      "epoch": 0.3066271018793274,
+      "grad_norm": 1.1536956780162377,
+      "learning_rate": 4.9890069622572374e-05,
+      "loss": 0.6127,
+      "step": 310
+    },
+    {
+      "epoch": 0.3076162215628091,
+      "grad_norm": 0.934115930451099,
+      "learning_rate": 4.98717478930011e-05,
+      "loss": 0.5808,
+      "step": 311
+    },
+    {
+      "epoch": 0.3086053412462908,
+      "grad_norm": 1.1749017282382166,
+      "learning_rate": 4.985342616342983e-05,
+      "loss": 0.6422,
+      "step": 312
+    },
+    {
+      "epoch": 0.3095944609297725,
+      "grad_norm": 0.8515985999112896,
+      "learning_rate": 4.983510443385856e-05,
+      "loss": 0.6175,
+      "step": 313
+    },
+    {
+      "epoch": 0.3105835806132542,
+      "grad_norm": 6.518960373233818,
+      "learning_rate": 4.981678270428729e-05,
+      "loss": 0.8142,
+      "step": 314
+    },
+    {
+      "epoch": 0.3115727002967359,
+      "grad_norm": 2.1921818273490374,
+      "learning_rate": 4.979846097471601e-05,
+      "loss": 0.6226,
+      "step": 315
+    },
+    {
+      "epoch": 0.3125618199802176,
+      "grad_norm": 1.6420624734489004,
+      "learning_rate": 4.9780139245144747e-05,
+      "loss": 0.6004,
+      "step": 316
+    },
+    {
+      "epoch": 0.3135509396636993,
+      "grad_norm": 0.7686264715968514,
+      "learning_rate": 4.976181751557347e-05,
+      "loss": 0.587,
+      "step": 317
+    },
+    {
+      "epoch": 0.314540059347181,
+      "grad_norm": 1.6301825964138899,
+      "learning_rate": 4.9743495786002206e-05,
+      "loss": 0.5783,
+      "step": 318
+    },
+    {
+      "epoch": 0.3155291790306627,
+      "grad_norm": 1.7471449432767963,
+      "learning_rate": 4.9725174056430926e-05,
+      "loss": 0.6721,
+      "step": 319
+    },
+    {
+      "epoch": 0.3165182987141444,
+      "grad_norm": 1.1192919506428751,
+      "learning_rate": 4.970685232685966e-05,
+      "loss": 0.6192,
+      "step": 320
+    },
+    {
+      "epoch": 0.31750741839762614,
+      "grad_norm": 1.269040128114316,
+      "learning_rate": 4.9688530597288385e-05,
+      "loss": 0.6129,
+      "step": 321
+    },
+    {
+      "epoch": 0.31849653808110784,
+      "grad_norm": 0.828446461431732,
+      "learning_rate": 4.967020886771712e-05,
+      "loss": 0.6483,
+      "step": 322
+    },
+    {
+      "epoch": 0.31948565776458954,
+      "grad_norm": 0.8666239492089782,
+      "learning_rate": 4.9651887138145845e-05,
+      "loss": 0.5733,
+      "step": 323
+    },
+    {
+      "epoch": 0.32047477744807124,
+      "grad_norm": 0.7549123620762485,
+      "learning_rate": 4.963356540857457e-05,
+      "loss": 0.6196,
+      "step": 324
+    },
+    {
+      "epoch": 0.32146389713155293,
+      "grad_norm": 1.8217639567133426,
+      "learning_rate": 4.96152436790033e-05,
+      "loss": 0.6356,
+      "step": 325
+    },
+    {
+      "epoch": 0.32245301681503463,
+      "grad_norm": 2.292729397977442,
+      "learning_rate": 4.959692194943203e-05,
+      "loss": 0.6607,
+      "step": 326
+    },
+    {
+      "epoch": 0.32344213649851633,
+      "grad_norm": 0.83818756217215,
+      "learning_rate": 4.957860021986076e-05,
+      "loss": 0.6217,
+      "step": 327
+    },
+    {
+      "epoch": 0.32443125618199803,
+      "grad_norm": 0.676734652693397,
+      "learning_rate": 4.9560278490289484e-05,
+      "loss": 0.592,
+      "step": 328
+    },
+    {
+      "epoch": 0.3254203758654797,
+      "grad_norm": 0.5902774697246586,
+      "learning_rate": 4.954195676071822e-05,
+      "loss": 0.6153,
+      "step": 329
+    },
+    {
+      "epoch": 0.3264094955489614,
+      "grad_norm": 0.800723034946388,
+      "learning_rate": 4.9523635031146943e-05,
+      "loss": 0.664,
+      "step": 330
+    },
+    {
+      "epoch": 0.3273986152324431,
+      "grad_norm": 0.5767708938288155,
+      "learning_rate": 4.950531330157567e-05,
+      "loss": 0.5934,
+      "step": 331
+    },
+    {
+      "epoch": 0.3283877349159248,
+      "grad_norm": 0.7493746467171817,
+      "learning_rate": 4.9486991572004396e-05,
+      "loss": 0.633,
+      "step": 332
+    },
+    {
+      "epoch": 0.3293768545994065,
+      "grad_norm": 0.8168408769910189,
+      "learning_rate": 4.946866984243313e-05,
+      "loss": 0.5523,
+      "step": 333
+    },
+    {
+      "epoch": 0.3303659742828882,
+      "grad_norm": 0.6357012806203373,
+      "learning_rate": 4.9450348112861856e-05,
+      "loss": 0.5584,
+      "step": 334
+    },
+    {
+      "epoch": 0.3313550939663699,
+      "grad_norm": 0.9570738821630661,
+      "learning_rate": 4.943202638329059e-05,
+      "loss": 0.5602,
+      "step": 335
+    },
+    {
+      "epoch": 0.3323442136498516,
+      "grad_norm": 0.6271482411746069,
+      "learning_rate": 4.941370465371931e-05,
+      "loss": 0.6037,
+      "step": 336
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.8342569793379687,
+      "learning_rate": 4.939538292414804e-05,
+      "loss": 0.6013,
+      "step": 337
+    },
+    {
+      "epoch": 0.334322453016815,
+      "grad_norm": 0.8429515515235894,
+      "learning_rate": 4.937706119457677e-05,
+      "loss": 0.5744,
+      "step": 338
+    },
+    {
+      "epoch": 0.3353115727002967,
+      "grad_norm": 0.41480878493392426,
+      "learning_rate": 4.93587394650055e-05,
+      "loss": 0.533,
+      "step": 339
+    },
+    {
+      "epoch": 0.3363006923837784,
+      "grad_norm": 1.0762372950869659,
+      "learning_rate": 4.934041773543423e-05,
+      "loss": 0.6092,
+      "step": 340
+    },
+    {
+      "epoch": 0.33728981206726016,
+      "grad_norm": 0.5352412089968016,
+      "learning_rate": 4.9322096005862954e-05,
+      "loss": 0.5744,
+      "step": 341
+    },
+    {
+      "epoch": 0.33827893175074186,
+      "grad_norm": 11.795603000530546,
+      "learning_rate": 4.930377427629169e-05,
+      "loss": 0.8007,
+      "step": 342
+    },
+    {
+      "epoch": 0.33926805143422356,
+      "grad_norm": 1.6654251993458582,
+      "learning_rate": 4.9285452546720414e-05,
+      "loss": 0.6833,
+      "step": 343
+    },
+    {
+      "epoch": 0.34025717111770526,
+      "grad_norm": 1.023245256880444,
+      "learning_rate": 4.926713081714914e-05,
+      "loss": 0.5568,
+      "step": 344
+    },
+    {
+      "epoch": 0.34124629080118696,
+      "grad_norm": 0.9874560723539015,
+      "learning_rate": 4.924880908757787e-05,
+      "loss": 0.6252,
+      "step": 345
+    },
+    {
+      "epoch": 0.34223541048466866,
+      "grad_norm": 1.1346945125596772,
+      "learning_rate": 4.92304873580066e-05,
+      "loss": 0.667,
+      "step": 346
+    },
+    {
+      "epoch": 0.34322453016815035,
+      "grad_norm": 0.7734482167051114,
+      "learning_rate": 4.9212165628435326e-05,
+      "loss": 0.5809,
+      "step": 347
+    },
+    {
+      "epoch": 0.34421364985163205,
+      "grad_norm": 0.8763032330208194,
+      "learning_rate": 4.919384389886405e-05,
+      "loss": 0.6812,
+      "step": 348
+    },
+    {
+      "epoch": 0.34520276953511375,
+      "grad_norm": 0.8014512357084477,
+      "learning_rate": 4.917552216929278e-05,
+      "loss": 0.5391,
+      "step": 349
+    },
+    {
+      "epoch": 0.34619188921859545,
+      "grad_norm": 0.6557118095626164,
+      "learning_rate": 4.915720043972151e-05,
+      "loss": 0.5202,
+      "step": 350
+    },
+    {
+      "epoch": 0.34718100890207715,
+      "grad_norm": 0.6296821730209303,
+      "learning_rate": 4.913887871015024e-05,
+      "loss": 0.4884,
+      "step": 351
+    },
+    {
+      "epoch": 0.34817012858555885,
+      "grad_norm": 0.7422194964288009,
+      "learning_rate": 4.912055698057897e-05,
+      "loss": 0.5452,
+      "step": 352
+    },
+    {
+      "epoch": 0.34915924826904055,
+      "grad_norm": 0.5946076118500246,
+      "learning_rate": 4.91022352510077e-05,
+      "loss": 0.6166,
+      "step": 353
+    },
+    {
+      "epoch": 0.35014836795252224,
+      "grad_norm": 0.732570052375268,
+      "learning_rate": 4.9083913521436425e-05,
+      "loss": 0.6611,
+      "step": 354
+    },
+    {
+      "epoch": 0.35113748763600394,
+      "grad_norm": 0.4698106032909168,
+      "learning_rate": 4.906559179186516e-05,
+      "loss": 0.5601,
+      "step": 355
+    },
+    {
+      "epoch": 0.35212660731948564,
+      "grad_norm": 1.2453092526176186,
+      "learning_rate": 4.9047270062293885e-05,
+      "loss": 0.6009,
+      "step": 356
+    },
+    {
+      "epoch": 0.35311572700296734,
+      "grad_norm": 0.4418805854104305,
+      "learning_rate": 4.902894833272261e-05,
+      "loss": 0.6021,
+      "step": 357
+    },
+    {
+      "epoch": 0.35410484668644904,
+      "grad_norm": 0.5326586341210466,
+      "learning_rate": 4.901062660315134e-05,
+      "loss": 0.5839,
+      "step": 358
+    },
+    {
+      "epoch": 0.35509396636993074,
+      "grad_norm": 0.41554111643426483,
+      "learning_rate": 4.899230487358007e-05,
+      "loss": 0.5615,
+      "step": 359
+    },
+    {
+      "epoch": 0.3560830860534125,
+      "grad_norm": 0.534276978725234,
+      "learning_rate": 4.89739831440088e-05,
+      "loss": 0.6854,
+      "step": 360
+    },
+    {
+      "epoch": 0.3570722057368942,
+      "grad_norm": 0.41429157493399854,
+      "learning_rate": 4.8955661414437523e-05,
+      "loss": 0.5839,
+      "step": 361
+    },
+    {
+      "epoch": 0.3580613254203759,
+      "grad_norm": 1.8567649357537324,
+      "learning_rate": 4.893733968486625e-05,
+      "loss": 0.5924,
+      "step": 362
+    },
+    {
+      "epoch": 0.3590504451038576,
+      "grad_norm": 0.8369610972802223,
+      "learning_rate": 4.891901795529498e-05,
+      "loss": 0.6518,
+      "step": 363
+    },
+    {
+      "epoch": 0.3600395647873393,
+      "grad_norm": 0.5759620691607789,
+      "learning_rate": 4.8900696225723716e-05,
+      "loss": 0.5869,
+      "step": 364
+    },
+    {
+      "epoch": 0.361028684470821,
+      "grad_norm": 0.8489394500046676,
+      "learning_rate": 4.8882374496152436e-05,
+      "loss": 0.621,
+      "step": 365
+    },
+    {
+      "epoch": 0.3620178041543027,
+      "grad_norm": 0.6560658008069779,
+      "learning_rate": 4.886405276658117e-05,
+      "loss": 0.6094,
+      "step": 366
+    },
+    {
+      "epoch": 0.3630069238377844,
+      "grad_norm": 0.7865989416550275,
+      "learning_rate": 4.8845731037009895e-05,
+      "loss": 0.5788,
+      "step": 367
+    },
+    {
+      "epoch": 0.3639960435212661,
+      "grad_norm": 0.6537131935874835,
+      "learning_rate": 4.882740930743863e-05,
+      "loss": 0.5797,
+      "step": 368
+    },
+    {
+      "epoch": 0.3649851632047478,
+      "grad_norm": 0.6909493769793963,
+      "learning_rate": 4.8809087577867355e-05,
+      "loss": 0.5842,
+      "step": 369
+    },
+    {
+      "epoch": 0.3659742828882295,
+      "grad_norm": 0.4769547461410241,
+      "learning_rate": 4.879076584829608e-05,
+      "loss": 0.6062,
+      "step": 370
+    },
+    {
+      "epoch": 0.3669634025717112,
+      "grad_norm": 0.8200716144700505,
+      "learning_rate": 4.877244411872481e-05,
+      "loss": 0.6748,
+      "step": 371
+    },
+    {
+      "epoch": 0.36795252225519287,
+      "grad_norm": 0.45770566187660844,
+      "learning_rate": 4.875412238915354e-05,
+      "loss": 0.527,
+      "step": 372
+    },
+    {
+      "epoch": 0.36894164193867457,
+      "grad_norm": 0.6555897440490205,
+      "learning_rate": 4.873580065958227e-05,
+      "loss": 0.5222,
+      "step": 373
+    },
+    {
+      "epoch": 0.36993076162215627,
+      "grad_norm": 0.7580354956119526,
+      "learning_rate": 4.8717478930010994e-05,
+      "loss": 0.6309,
+      "step": 374
+    },
+    {
+      "epoch": 0.37091988130563797,
+      "grad_norm": 0.5439082223239077,
+      "learning_rate": 4.869915720043972e-05,
+      "loss": 0.5571,
+      "step": 375
+    },
+    {
+      "epoch": 0.37190900098911966,
+      "grad_norm": 0.8331347227307716,
+      "learning_rate": 4.8680835470868454e-05,
+      "loss": 0.5607,
+      "step": 376
+    },
+    {
+      "epoch": 0.37289812067260136,
+      "grad_norm": 0.5141516865612674,
+      "learning_rate": 4.866251374129718e-05,
+      "loss": 0.6567,
+      "step": 377
+    },
+    {
+      "epoch": 0.37388724035608306,
+      "grad_norm": 0.7141235700823007,
+      "learning_rate": 4.8644192011725906e-05,
+      "loss": 0.5669,
+      "step": 378
+    },
+    {
+      "epoch": 0.37487636003956476,
+      "grad_norm": 0.5560006726258888,
+      "learning_rate": 4.862587028215464e-05,
+      "loss": 0.6324,
+      "step": 379
+    },
+    {
+      "epoch": 0.3758654797230465,
+      "grad_norm": 0.7409978312950718,
+      "learning_rate": 4.8607548552583366e-05,
+      "loss": 0.7242,
+      "step": 380
+    },
+    {
+      "epoch": 0.3768545994065282,
+      "grad_norm": 0.5440930687616704,
+      "learning_rate": 4.85892268230121e-05,
+      "loss": 0.6254,
+      "step": 381
+    },
+    {
+      "epoch": 0.3778437190900099,
+      "grad_norm": 0.6478252439008088,
+      "learning_rate": 4.857090509344082e-05,
+      "loss": 0.5264,
+      "step": 382
+    },
+    {
+      "epoch": 0.3788328387734916,
+      "grad_norm": 0.4709599342774292,
+      "learning_rate": 4.855258336386955e-05,
+      "loss": 0.5647,
+      "step": 383
+    },
+    {
+      "epoch": 0.3798219584569733,
+      "grad_norm": 1.18640771551092,
+      "learning_rate": 4.853426163429828e-05,
+      "loss": 0.7512,
+      "step": 384
+    },
+    {
+      "epoch": 0.380811078140455,
+      "grad_norm": 0.3923235190086449,
+      "learning_rate": 4.851593990472701e-05,
+      "loss": 0.6049,
+      "step": 385
+    },
+    {
+      "epoch": 0.3818001978239367,
+      "grad_norm": 0.4781166108075703,
+      "learning_rate": 4.849761817515573e-05,
+      "loss": 0.5453,
+      "step": 386
+    },
+    {
+      "epoch": 0.3827893175074184,
+      "grad_norm": 0.3725786214381448,
+      "learning_rate": 4.8479296445584464e-05,
+      "loss": 0.5387,
+      "step": 387
+    },
+    {
+      "epoch": 0.3837784371909001,
+      "grad_norm": 0.4987148506610239,
+      "learning_rate": 4.84609747160132e-05,
+      "loss": 0.547,
+      "step": 388
+    },
+    {
+      "epoch": 0.3847675568743818,
+      "grad_norm": 0.35932327262892194,
+      "learning_rate": 4.8442652986441924e-05,
+      "loss": 0.5214,
+      "step": 389
+    },
+    {
+      "epoch": 0.3857566765578635,
+      "grad_norm": 0.690564658045055,
+      "learning_rate": 4.842433125687065e-05,
+      "loss": 0.591,
+      "step": 390
+    },
+    {
+      "epoch": 0.3867457962413452,
+      "grad_norm": 0.3463972013191391,
+      "learning_rate": 4.840600952729938e-05,
+      "loss": 0.6385,
+      "step": 391
+    },
+    {
+      "epoch": 0.3877349159248269,
+      "grad_norm": 0.3543061291829264,
+      "learning_rate": 4.838768779772811e-05,
+      "loss": 0.517,
+      "step": 392
+    },
+    {
+      "epoch": 0.3887240356083086,
+      "grad_norm": 0.3660023274107496,
+      "learning_rate": 4.8369366068156837e-05,
+      "loss": 0.5781,
+      "step": 393
+    },
+    {
+      "epoch": 0.3897131552917903,
+      "grad_norm": 0.35899772143056496,
+      "learning_rate": 4.835104433858556e-05,
+      "loss": 0.6156,
+      "step": 394
+    },
+    {
+      "epoch": 0.390702274975272,
+      "grad_norm": 0.4516133836964091,
+      "learning_rate": 4.833272260901429e-05,
+      "loss": 0.555,
+      "step": 395
+    },
+    {
+      "epoch": 0.3916913946587537,
+      "grad_norm": 4.069658753953614,
+      "learning_rate": 4.831440087944302e-05,
+      "loss": 0.6786,
+      "step": 396
+    },
+    {
+      "epoch": 0.3926805143422354,
+      "grad_norm": 0.5512608577859164,
+      "learning_rate": 4.829607914987175e-05,
+      "loss": 0.5819,
+      "step": 397
+    },
+    {
+      "epoch": 0.3936696340257171,
+      "grad_norm": 0.6001718577059937,
+      "learning_rate": 4.827775742030048e-05,
+      "loss": 0.5866,
+      "step": 398
+    },
+    {
+      "epoch": 0.39465875370919884,
+      "grad_norm": 0.43723220327977047,
+      "learning_rate": 4.82594356907292e-05,
+      "loss": 0.6223,
+      "step": 399
+    },
+    {
+      "epoch": 0.39564787339268054,
+      "grad_norm": 0.4091883159043,
+      "learning_rate": 4.8241113961157935e-05,
+      "loss": 0.5937,
+      "step": 400
+    },
+    {
+      "epoch": 0.39663699307616224,
+      "grad_norm": 0.3627182581990674,
+      "learning_rate": 4.822279223158667e-05,
+      "loss": 0.4927,
+      "step": 401
+    },
+    {
+      "epoch": 0.39762611275964393,
+      "grad_norm": 3.395118475745649,
+      "learning_rate": 4.8204470502015395e-05,
+      "loss": 0.6651,
+      "step": 402
+    },
+    {
+      "epoch": 0.39861523244312563,
+      "grad_norm": 0.503406651457647,
+      "learning_rate": 4.818614877244412e-05,
+      "loss": 0.6462,
+      "step": 403
+    },
+    {
+      "epoch": 0.39960435212660733,
+      "grad_norm": 0.41164305421638864,
+      "learning_rate": 4.816782704287285e-05,
+      "loss": 0.5303,
+      "step": 404
+    },
+    {
+      "epoch": 0.40059347181008903,
+      "grad_norm": 0.44166637125698777,
+      "learning_rate": 4.814950531330158e-05,
+      "loss": 0.5474,
+      "step": 405
+    },
+    {
+      "epoch": 0.40158259149357073,
+      "grad_norm": 0.48008127596420697,
+      "learning_rate": 4.813118358373031e-05,
+      "loss": 0.6364,
+      "step": 406
+    },
+    {
+      "epoch": 0.4025717111770524,
+      "grad_norm": 0.46798165410172304,
+      "learning_rate": 4.8112861854159033e-05,
+      "loss": 0.6383,
+      "step": 407
+    },
+    {
+      "epoch": 0.4035608308605341,
+      "grad_norm": 0.44125266116511586,
+      "learning_rate": 4.809454012458776e-05,
+      "loss": 0.5955,
+      "step": 408
+    },
+    {
+      "epoch": 0.4045499505440158,
+      "grad_norm": 0.4395464902377213,
+      "learning_rate": 4.807621839501649e-05,
+      "loss": 0.7118,
+      "step": 409
+    },
+    {
+      "epoch": 0.4055390702274975,
+      "grad_norm": 0.3816754087372529,
+      "learning_rate": 4.805789666544522e-05,
+      "loss": 0.5148,
+      "step": 410
+    },
+    {
+      "epoch": 0.4065281899109792,
+      "grad_norm": 0.41359029720451895,
+      "learning_rate": 4.8039574935873946e-05,
+      "loss": 0.588,
+      "step": 411
+    },
+    {
+      "epoch": 0.4075173095944609,
+      "grad_norm": 0.4693298663552246,
+      "learning_rate": 4.802125320630268e-05,
+      "loss": 0.6457,
+      "step": 412
+    },
+    {
+      "epoch": 0.4085064292779426,
+      "grad_norm": 0.4067116282641899,
+      "learning_rate": 4.8002931476731406e-05,
+      "loss": 0.5495,
+      "step": 413
+    },
+    {
+      "epoch": 0.4094955489614243,
+      "grad_norm": 0.4272277873309838,
+      "learning_rate": 4.798460974716014e-05,
+      "loss": 0.6337,
+      "step": 414
+    },
+    {
+      "epoch": 0.410484668644906,
+      "grad_norm": 0.4811569091836984,
+      "learning_rate": 4.7966288017588865e-05,
+      "loss": 0.6149,
+      "step": 415
+    },
+    {
+      "epoch": 0.4114737883283877,
+      "grad_norm": 0.38122687491339136,
+      "learning_rate": 4.794796628801759e-05,
+      "loss": 0.5846,
+      "step": 416
+    },
+    {
+      "epoch": 0.4124629080118694,
+      "grad_norm": 0.4862215332358342,
+      "learning_rate": 4.792964455844632e-05,
+      "loss": 0.6186,
+      "step": 417
+    },
+    {
+      "epoch": 0.4134520276953511,
+      "grad_norm": 0.47460277960742403,
+      "learning_rate": 4.791132282887505e-05,
+      "loss": 0.5941,
+      "step": 418
+    },
+    {
+      "epoch": 0.41444114737883286,
+      "grad_norm": 0.40971075817407676,
+      "learning_rate": 4.789300109930378e-05,
+      "loss": 0.5409,
+      "step": 419
+    },
+    {
+      "epoch": 0.41543026706231456,
+      "grad_norm": 0.40503923340944115,
+      "learning_rate": 4.7874679369732504e-05,
+      "loss": 0.5656,
+      "step": 420
+    },
+    {
+      "epoch": 0.41641938674579626,
+      "grad_norm": 0.47166101201631216,
+      "learning_rate": 4.785635764016123e-05,
+      "loss": 0.5734,
+      "step": 421
+    },
+    {
+      "epoch": 0.41740850642927796,
+      "grad_norm": 0.4074965319088034,
+      "learning_rate": 4.7838035910589964e-05,
+      "loss": 0.552,
+      "step": 422
+    },
+    {
+      "epoch": 0.41839762611275966,
+      "grad_norm": 0.39208549972043516,
+      "learning_rate": 4.781971418101869e-05,
+      "loss": 0.6738,
+      "step": 423
+    },
+    {
+      "epoch": 0.41938674579624136,
+      "grad_norm": 0.6011637351858017,
+      "learning_rate": 4.7801392451447416e-05,
+      "loss": 0.5603,
+      "step": 424
+    },
+    {
+      "epoch": 0.42037586547972305,
+      "grad_norm": 0.3726260226014429,
+      "learning_rate": 4.778307072187615e-05,
+      "loss": 0.5816,
+      "step": 425
+    },
+    {
+      "epoch": 0.42136498516320475,
+      "grad_norm": 0.6236845405373085,
+      "learning_rate": 4.7764748992304876e-05,
+      "loss": 0.6482,
+      "step": 426
+    },
+    {
+      "epoch": 0.42235410484668645,
+      "grad_norm": 3.2608951858084922,
+      "learning_rate": 4.774642726273361e-05,
+      "loss": 0.6115,
+      "step": 427
+    },
+    {
+      "epoch": 0.42334322453016815,
+      "grad_norm": 0.5292345408652966,
+      "learning_rate": 4.772810553316233e-05,
+      "loss": 0.5683,
+      "step": 428
+    },
+    {
+      "epoch": 0.42433234421364985,
+      "grad_norm": 0.7800091939799507,
+      "learning_rate": 4.770978380359106e-05,
+      "loss": 0.5404,
+      "step": 429
+    },
+    {
+      "epoch": 0.42532146389713155,
+      "grad_norm": 0.3947675558178534,
+      "learning_rate": 4.769146207401979e-05,
+      "loss": 0.5609,
+      "step": 430
+    },
+    {
+      "epoch": 0.42631058358061324,
+      "grad_norm": 8.27438656210951,
+      "learning_rate": 4.767314034444852e-05,
+      "loss": 0.8391,
+      "step": 431
+    },
+    {
+      "epoch": 0.42729970326409494,
+      "grad_norm": 0.8743079722440797,
+      "learning_rate": 4.765481861487724e-05,
+      "loss": 0.5703,
+      "step": 432
+    },
+    {
+      "epoch": 0.42828882294757664,
+      "grad_norm": 1.903800162095679,
+      "learning_rate": 4.7636496885305975e-05,
+      "loss": 0.5864,
+      "step": 433
+    },
+    {
+      "epoch": 0.42927794263105834,
+      "grad_norm": 0.6629417966412571,
+      "learning_rate": 4.76181751557347e-05,
+      "loss": 0.5367,
+      "step": 434
+    },
+    {
+      "epoch": 0.43026706231454004,
+      "grad_norm": 0.7219662282705351,
+      "learning_rate": 4.7599853426163434e-05,
+      "loss": 0.6087,
+      "step": 435
+    },
+    {
+      "epoch": 0.43125618199802174,
+      "grad_norm": 0.4530515462755643,
+      "learning_rate": 4.758153169659216e-05,
+      "loss": 0.5095,
+      "step": 436
+    },
+    {
+      "epoch": 0.43224530168150344,
+      "grad_norm": 0.7726116565074944,
+      "learning_rate": 4.756320996702089e-05,
+      "loss": 0.6424,
+      "step": 437
+    },
+    {
+      "epoch": 0.4332344213649852,
+      "grad_norm": 0.5383329629673866,
+      "learning_rate": 4.754488823744962e-05,
+      "loss": 0.6031,
+      "step": 438
+    },
+    {
+      "epoch": 0.4342235410484669,
+      "grad_norm": 0.6766939106280287,
+      "learning_rate": 4.7526566507878347e-05,
+      "loss": 0.6031,
+      "step": 439
+    },
+    {
+      "epoch": 0.4352126607319486,
+      "grad_norm": 0.6940726364507884,
+      "learning_rate": 4.750824477830707e-05,
+      "loss": 0.5308,
+      "step": 440
+    },
+    {
+      "epoch": 0.4362017804154303,
+      "grad_norm": 0.5386095591035055,
+      "learning_rate": 4.74899230487358e-05,
+      "loss": 0.5361,
+      "step": 441
+    },
+    {
+      "epoch": 0.437190900098912,
+      "grad_norm": 0.6810339477078869,
+      "learning_rate": 4.747160131916453e-05,
+      "loss": 0.5189,
+      "step": 442
+    },
+    {
+      "epoch": 0.4381800197823937,
+      "grad_norm": 0.4935297438689495,
+      "learning_rate": 4.745327958959326e-05,
+      "loss": 0.5648,
+      "step": 443
+    },
+    {
+      "epoch": 0.4391691394658754,
+      "grad_norm": 0.5426668934382847,
+      "learning_rate": 4.743495786002199e-05,
+      "loss": 0.5745,
+      "step": 444
+    },
+    {
+      "epoch": 0.4401582591493571,
+      "grad_norm": 0.5200388242525367,
+      "learning_rate": 4.741663613045071e-05,
+      "loss": 0.5863,
+      "step": 445
+    },
+    {
+      "epoch": 0.4411473788328388,
+      "grad_norm": 1.3830554463390863,
+      "learning_rate": 4.7398314400879445e-05,
+      "loss": 0.5881,
+      "step": 446
+    },
+    {
+      "epoch": 0.4421364985163205,
+      "grad_norm": 0.6948346194161272,
+      "learning_rate": 4.737999267130817e-05,
+      "loss": 0.5239,
+      "step": 447
+    },
+    {
+      "epoch": 0.4431256181998022,
+      "grad_norm": 0.6601938762365581,
+      "learning_rate": 4.7361670941736905e-05,
+      "loss": 0.5487,
+      "step": 448
+    },
+    {
+      "epoch": 0.44411473788328387,
+      "grad_norm": 0.5617566077718212,
+      "learning_rate": 4.734334921216563e-05,
+      "loss": 0.6214,
+      "step": 449
+    },
+    {
+      "epoch": 0.44510385756676557,
+      "grad_norm": 0.732252290442145,
+      "learning_rate": 4.732502748259436e-05,
+      "loss": 0.5628,
+      "step": 450
+    },
+    {
+      "epoch": 0.44609297725024727,
+      "grad_norm": 0.44306508264201677,
+      "learning_rate": 4.730670575302309e-05,
+      "loss": 0.6078,
+      "step": 451
+    },
+    {
+      "epoch": 0.44708209693372897,
+      "grad_norm": 0.5296732642576735,
+      "learning_rate": 4.728838402345182e-05,
+      "loss": 0.5147,
+      "step": 452
+    },
+    {
+      "epoch": 0.44807121661721067,
+      "grad_norm": 1.652456677052837,
+      "learning_rate": 4.7270062293880544e-05,
+      "loss": 0.5598,
+      "step": 453
+    },
+    {
+      "epoch": 0.44906033630069236,
+      "grad_norm": 0.6144592738271647,
+      "learning_rate": 4.725174056430927e-05,
+      "loss": 0.51,
+      "step": 454
+    },
+    {
+      "epoch": 0.45004945598417406,
+      "grad_norm": 0.4613783043555175,
+      "learning_rate": 4.7233418834738e-05,
+      "loss": 0.6202,
+      "step": 455
+    },
+    {
+      "epoch": 0.45103857566765576,
+      "grad_norm": 0.5213448885231209,
+      "learning_rate": 4.721509710516673e-05,
+      "loss": 0.5308,
+      "step": 456
+    },
+    {
+      "epoch": 0.4520276953511375,
+      "grad_norm": 0.46820252742016644,
+      "learning_rate": 4.7196775375595456e-05,
+      "loss": 0.5832,
+      "step": 457
+    },
+    {
+      "epoch": 0.4530168150346192,
+      "grad_norm": 0.42474287967898067,
+      "learning_rate": 4.717845364602418e-05,
+      "loss": 0.5684,
+      "step": 458
+    },
+    {
+      "epoch": 0.4540059347181009,
+      "grad_norm": 0.39392310575419126,
+      "learning_rate": 4.7160131916452916e-05,
+      "loss": 0.624,
+      "step": 459
+    },
+    {
+      "epoch": 0.4549950544015826,
+      "grad_norm": 0.4175763227925975,
+      "learning_rate": 4.714181018688165e-05,
+      "loss": 0.6065,
+      "step": 460
+    },
+    {
+      "epoch": 0.4559841740850643,
+      "grad_norm": 0.3800322746234243,
+      "learning_rate": 4.7123488457310375e-05,
+      "loss": 0.6242,
+      "step": 461
+    },
+    {
+      "epoch": 0.456973293768546,
+      "grad_norm": 0.37365387127732047,
+      "learning_rate": 4.71051667277391e-05,
+      "loss": 0.6967,
+      "step": 462
+    },
+    {
+      "epoch": 0.4579624134520277,
+      "grad_norm": 0.46031496253337684,
+      "learning_rate": 4.708684499816783e-05,
+      "loss": 0.5547,
+      "step": 463
+    },
+    {
+      "epoch": 0.4589515331355094,
+      "grad_norm": 0.39156501494001045,
+      "learning_rate": 4.706852326859656e-05,
+      "loss": 0.5693,
+      "step": 464
+    },
+    {
+      "epoch": 0.4599406528189911,
+      "grad_norm": 0.37353175638665764,
+      "learning_rate": 4.705020153902529e-05,
+      "loss": 0.587,
+      "step": 465
+    },
+    {
+      "epoch": 0.4609297725024728,
+      "grad_norm": 0.43060709388709534,
+      "learning_rate": 4.7031879809454014e-05,
+      "loss": 0.5674,
+      "step": 466
+    },
+    {
+      "epoch": 0.4619188921859545,
+      "grad_norm": 0.38735425009204794,
+      "learning_rate": 4.701355807988274e-05,
+      "loss": 0.6058,
+      "step": 467
+    },
+    {
+      "epoch": 0.4629080118694362,
+      "grad_norm": 0.46266944891118184,
+      "learning_rate": 4.6995236350311474e-05,
+      "loss": 0.5652,
+      "step": 468
+    },
+    {
+      "epoch": 0.4638971315529179,
+      "grad_norm": 0.36247469180107994,
+      "learning_rate": 4.69769146207402e-05,
+      "loss": 0.6043,
+      "step": 469
+    },
+    {
+      "epoch": 0.4648862512363996,
+      "grad_norm": 0.4030557538436561,
+      "learning_rate": 4.6958592891168927e-05,
+      "loss": 0.6128,
+      "step": 470
+    },
+    {
+      "epoch": 0.4658753709198813,
+      "grad_norm": 0.6033825885688812,
+      "learning_rate": 4.694027116159766e-05,
+      "loss": 0.5646,
+      "step": 471
+    },
+    {
+      "epoch": 0.466864490603363,
+      "grad_norm": 0.37720757452451065,
+      "learning_rate": 4.6921949432026386e-05,
+      "loss": 0.5317,
+      "step": 472
+    },
+    {
+      "epoch": 0.4678536102868447,
+      "grad_norm": 0.5333495762992678,
+      "learning_rate": 4.690362770245512e-05,
+      "loss": 0.6658,
+      "step": 473
+    },
+    {
+      "epoch": 0.4688427299703264,
+      "grad_norm": 0.5133974898384014,
+      "learning_rate": 4.688530597288384e-05,
+      "loss": 0.5035,
+      "step": 474
+    },
+    {
+      "epoch": 0.4698318496538081,
+      "grad_norm": 0.4539945208932187,
+      "learning_rate": 4.686698424331257e-05,
+      "loss": 0.5747,
+      "step": 475
+    },
+    {
+      "epoch": 0.4708209693372898,
+      "grad_norm": 0.41564631661574264,
+      "learning_rate": 4.68486625137413e-05,
+      "loss": 0.4843,
+      "step": 476
+    },
+    {
+      "epoch": 0.47181008902077154,
+      "grad_norm": 0.38138121789633134,
+      "learning_rate": 4.683034078417003e-05,
+      "loss": 0.5409,
+      "step": 477
+    },
+    {
+      "epoch": 0.47279920870425324,
+      "grad_norm": 0.44871468038998025,
+      "learning_rate": 4.681201905459875e-05,
+      "loss": 0.5109,
+      "step": 478
+    },
+    {
+      "epoch": 0.47378832838773494,
+      "grad_norm": 0.3177694456049873,
+      "learning_rate": 4.6793697325027485e-05,
+      "loss": 0.602,
+      "step": 479
+    },
+    {
+      "epoch": 0.47477744807121663,
+      "grad_norm": 0.46257643128201287,
+      "learning_rate": 4.677537559545621e-05,
+      "loss": 0.6047,
+      "step": 480
+    },
+    {
+      "epoch": 0.47576656775469833,
+      "grad_norm": 4.67269769452667,
+      "learning_rate": 4.6757053865884944e-05,
+      "loss": 0.8503,
+      "step": 481
+    },
+    {
+      "epoch": 0.47675568743818003,
+      "grad_norm": 0.39836333155328146,
+      "learning_rate": 4.673873213631367e-05,
+      "loss": 0.6113,
+      "step": 482
+    },
+    {
+      "epoch": 0.47774480712166173,
+      "grad_norm": 2.353788660145488,
+      "learning_rate": 4.67204104067424e-05,
+      "loss": 0.7275,
+      "step": 483
+    },
+    {
+      "epoch": 0.47873392680514343,
+      "grad_norm": 0.601097344141485,
+      "learning_rate": 4.670208867717113e-05,
+      "loss": 0.6335,
+      "step": 484
+    },
+    {
+      "epoch": 0.4797230464886251,
+      "grad_norm": 0.36883129234828055,
+      "learning_rate": 4.668376694759986e-05,
+      "loss": 0.6315,
+      "step": 485
+    },
+    {
+      "epoch": 0.4807121661721068,
+      "grad_norm": 1.8909305038276623,
+      "learning_rate": 4.666544521802858e-05,
+      "loss": 0.7504,
+      "step": 486
+    },
+    {
+      "epoch": 0.4817012858555885,
+      "grad_norm": 0.4591587176811992,
+      "learning_rate": 4.664712348845731e-05,
+      "loss": 0.5666,
+      "step": 487
+    },
+    {
+      "epoch": 0.4826904055390702,
+      "grad_norm": 0.4144642652670824,
+      "learning_rate": 4.662880175888604e-05,
+      "loss": 0.561,
+      "step": 488
+    },
+    {
+      "epoch": 0.4836795252225519,
+      "grad_norm": 5.896448976676767,
+      "learning_rate": 4.661048002931477e-05,
+      "loss": 0.9015,
+      "step": 489
+    },
+    {
+      "epoch": 0.4846686449060336,
+      "grad_norm": 0.5720698016034343,
+      "learning_rate": 4.65921582997435e-05,
+      "loss": 0.6114,
+      "step": 490
+    },
+    {
+      "epoch": 0.4856577645895153,
+      "grad_norm": 0.3679119804465383,
+      "learning_rate": 4.657383657017222e-05,
+      "loss": 0.5466,
+      "step": 491
+    },
+    {
+      "epoch": 0.486646884272997,
+      "grad_norm": 0.44989689849502357,
+      "learning_rate": 4.6555514840600955e-05,
+      "loss": 0.5291,
+      "step": 492
+    },
+    {
+      "epoch": 0.4876360039564787,
+      "grad_norm": 0.5024872015252905,
+      "learning_rate": 4.653719311102968e-05,
+      "loss": 0.6009,
+      "step": 493
+    },
+    {
+      "epoch": 0.4886251236399604,
+      "grad_norm": 0.464562771119027,
+      "learning_rate": 4.6518871381458415e-05,
+      "loss": 0.5875,
+      "step": 494
+    },
+    {
+      "epoch": 0.4896142433234421,
+      "grad_norm": 0.48167751471277875,
+      "learning_rate": 4.650054965188714e-05,
+      "loss": 0.4983,
+      "step": 495
+    },
+    {
+      "epoch": 0.49060336300692386,
+      "grad_norm": 1.9507530661852923,
+      "learning_rate": 4.648222792231587e-05,
+      "loss": 0.6099,
+      "step": 496
+    },
+    {
+      "epoch": 0.49159248269040556,
+      "grad_norm": 0.606821397983761,
+      "learning_rate": 4.64639061927446e-05,
+      "loss": 0.5899,
+      "step": 497
+    },
+    {
+      "epoch": 0.49258160237388726,
+      "grad_norm": 0.533867569738737,
+      "learning_rate": 4.644558446317333e-05,
+      "loss": 0.6233,
+      "step": 498
+    },
+    {
+      "epoch": 0.49357072205736896,
+      "grad_norm": 0.529229125116131,
+      "learning_rate": 4.6427262733602054e-05,
+      "loss": 0.5453,
+      "step": 499
+    },
+    {
+      "epoch": 0.49455984174085066,
+      "grad_norm": 0.4916159848073325,
+      "learning_rate": 4.640894100403078e-05,
+      "loss": 0.5724,
+      "step": 500
+    },
+    {
+      "epoch": 0.49554896142433236,
+      "grad_norm": 0.4818563677803172,
+      "learning_rate": 4.639061927445951e-05,
+      "loss": 0.5693,
+      "step": 501
+    },
+    {
+      "epoch": 0.49653808110781406,
+      "grad_norm": 4.7169140468910316,
+      "learning_rate": 4.637229754488824e-05,
+      "loss": 0.6424,
+      "step": 502
+    },
+    {
+      "epoch": 0.49752720079129575,
+      "grad_norm": 0.6812953196471956,
+      "learning_rate": 4.6353975815316966e-05,
+      "loss": 0.5695,
+      "step": 503
+    },
+    {
+      "epoch": 0.49851632047477745,
+      "grad_norm": 0.4430846469861221,
+      "learning_rate": 4.633565408574569e-05,
+      "loss": 0.5638,
+      "step": 504
+    },
+    {
+      "epoch": 0.49950544015825915,
+      "grad_norm": 0.5867096506625531,
+      "learning_rate": 4.6317332356174426e-05,
+      "loss": 0.577,
+      "step": 505
+    },
+    {
+      "epoch": 0.5004945598417408,
+      "grad_norm": 0.9442304457268548,
+      "learning_rate": 4.629901062660315e-05,
+      "loss": 0.5029,
+      "step": 506
+    },
+    {
+      "epoch": 0.5014836795252225,
+      "grad_norm": 0.5146057080865236,
+      "learning_rate": 4.6280688897031885e-05,
+      "loss": 0.5866,
+      "step": 507
+    },
+    {
+      "epoch": 0.5024727992087042,
+      "grad_norm": 0.40718135381499504,
+      "learning_rate": 4.626236716746061e-05,
+      "loss": 0.5325,
+      "step": 508
+    },
+    {
+      "epoch": 0.503461918892186,
+      "grad_norm": 0.5519103208111765,
+      "learning_rate": 4.624404543788934e-05,
+      "loss": 0.6358,
+      "step": 509
+    },
+    {
+      "epoch": 0.5044510385756676,
+      "grad_norm": 0.43699948345174316,
+      "learning_rate": 4.622572370831807e-05,
+      "loss": 0.608,
+      "step": 510
+    },
+    {
+      "epoch": 0.5054401582591493,
+      "grad_norm": 0.521982987910703,
+      "learning_rate": 4.62074019787468e-05,
+      "loss": 0.6314,
+      "step": 511
+    },
+    {
+      "epoch": 0.506429277942631,
+      "grad_norm": 0.3869957567924524,
+      "learning_rate": 4.6189080249175524e-05,
+      "loss": 0.4741,
+      "step": 512
+    },
+    {
+      "epoch": 0.5074183976261127,
+      "grad_norm": 0.4939706787894611,
+      "learning_rate": 4.617075851960425e-05,
+      "loss": 0.562,
+      "step": 513
+    },
+    {
+      "epoch": 0.5084075173095944,
+      "grad_norm": 0.4173292639174069,
+      "learning_rate": 4.6152436790032984e-05,
+      "loss": 0.5002,
+      "step": 514
+    },
+    {
+      "epoch": 0.5093966369930761,
+      "grad_norm": 0.4096494198363875,
+      "learning_rate": 4.613411506046171e-05,
+      "loss": 0.4885,
+      "step": 515
+    },
+    {
+      "epoch": 0.5103857566765578,
+      "grad_norm": 9.858981341609425,
+      "learning_rate": 4.6115793330890437e-05,
+      "loss": 0.685,
+      "step": 516
+    },
+    {
+      "epoch": 0.5113748763600395,
+      "grad_norm": 0.5195002022290981,
+      "learning_rate": 4.609747160131916e-05,
+      "loss": 0.5369,
+      "step": 517
+    },
+    {
+      "epoch": 0.5123639960435212,
+      "grad_norm": 0.3934474506050803,
+      "learning_rate": 4.6079149871747896e-05,
+      "loss": 0.5853,
+      "step": 518
+    },
+    {
+      "epoch": 0.5133531157270029,
+      "grad_norm": 0.4134221147318054,
+      "learning_rate": 4.606082814217663e-05,
+      "loss": 0.5617,
+      "step": 519
+    },
+    {
+      "epoch": 0.5143422354104846,
+      "grad_norm": 0.42829816283403555,
+      "learning_rate": 4.604250641260535e-05,
+      "loss": 0.5531,
+      "step": 520
+    },
+    {
+      "epoch": 0.5153313550939663,
+      "grad_norm": 0.48752740238717335,
+      "learning_rate": 4.602418468303408e-05,
+      "loss": 0.606,
+      "step": 521
+    },
+    {
+      "epoch": 0.516320474777448,
+      "grad_norm": 0.38538760178318904,
+      "learning_rate": 4.600586295346281e-05,
+      "loss": 0.5335,
+      "step": 522
+    },
+    {
+      "epoch": 0.5173095944609297,
+      "grad_norm": 0.3936884719789974,
+      "learning_rate": 4.598754122389154e-05,
+      "loss": 0.5447,
+      "step": 523
+    },
+    {
+      "epoch": 0.5182987141444114,
+      "grad_norm": 0.4059453713559687,
+      "learning_rate": 4.596921949432026e-05,
+      "loss": 0.4842,
+      "step": 524
+    },
+    {
+      "epoch": 0.5192878338278932,
+      "grad_norm": 1.5542942915947051,
+      "learning_rate": 4.5950897764748995e-05,
+      "loss": 0.6171,
+      "step": 525
+    },
+    {
+      "epoch": 0.5202769535113749,
+      "grad_norm": 1.5720532397067744,
+      "learning_rate": 4.593257603517772e-05,
+      "loss": 0.5877,
+      "step": 526
+    },
+    {
+      "epoch": 0.5212660731948566,
+      "grad_norm": 2.9326347862637765,
+      "learning_rate": 4.5914254305606454e-05,
+      "loss": 0.6602,
+      "step": 527
+    },
+    {
+      "epoch": 0.5222551928783383,
+      "grad_norm": 0.40876849501610224,
+      "learning_rate": 4.589593257603518e-05,
+      "loss": 0.4963,
+      "step": 528
+    },
+    {
+      "epoch": 0.52324431256182,
+      "grad_norm": 0.43764445066833524,
+      "learning_rate": 4.587761084646391e-05,
+      "loss": 0.5417,
+      "step": 529
+    },
+    {
+      "epoch": 0.5242334322453017,
+      "grad_norm": 0.40565909079834533,
+      "learning_rate": 4.5859289116892634e-05,
+      "loss": 0.576,
+      "step": 530
+    },
+    {
+      "epoch": 0.5252225519287834,
+      "grad_norm": 0.4217327834248698,
+      "learning_rate": 4.584096738732137e-05,
+      "loss": 0.6196,
+      "step": 531
+    },
+    {
+      "epoch": 0.5262116716122651,
+      "grad_norm": 0.46160735689946497,
+      "learning_rate": 4.582264565775009e-05,
+      "loss": 0.5486,
+      "step": 532
+    },
+    {
+      "epoch": 0.5272007912957468,
+      "grad_norm": 0.4032237658743589,
+      "learning_rate": 4.580432392817882e-05,
+      "loss": 0.5626,
+      "step": 533
+    },
+    {
+      "epoch": 0.5281899109792285,
+      "grad_norm": 0.5635928214859514,
+      "learning_rate": 4.578600219860755e-05,
+      "loss": 0.6173,
+      "step": 534
+    },
+    {
+      "epoch": 0.5291790306627102,
+      "grad_norm": 0.6342796371814479,
+      "learning_rate": 4.576768046903628e-05,
+      "loss": 0.6277,
+      "step": 535
+    },
+    {
+      "epoch": 0.5301681503461919,
+      "grad_norm": 0.4677423139160351,
+      "learning_rate": 4.574935873946501e-05,
+      "loss": 0.5694,
+      "step": 536
+    },
+    {
+      "epoch": 0.5311572700296736,
+      "grad_norm": 0.40238302891813643,
+      "learning_rate": 4.573103700989373e-05,
+      "loss": 0.6217,
+      "step": 537
+    },
+    {
+      "epoch": 0.5321463897131553,
+      "grad_norm": 0.504563589844548,
+      "learning_rate": 4.5712715280322465e-05,
+      "loss": 0.6222,
+      "step": 538
+    },
+    {
+      "epoch": 0.533135509396637,
+      "grad_norm": 0.42246468004279764,
+      "learning_rate": 4.569439355075119e-05,
+      "loss": 0.5684,
+      "step": 539
+    },
+    {
+      "epoch": 0.5341246290801187,
+      "grad_norm": 0.40403558987941973,
+      "learning_rate": 4.5676071821179925e-05,
+      "loss": 0.558,
+      "step": 540
+    },
+    {
+      "epoch": 0.5351137487636004,
+      "grad_norm": 0.44549737299183073,
+      "learning_rate": 4.5657750091608644e-05,
+      "loss": 0.5609,
+      "step": 541
+    },
+    {
+      "epoch": 0.5361028684470821,
+      "grad_norm": 20.81218570435005,
+      "learning_rate": 4.563942836203738e-05,
+      "loss": 0.9656,
+      "step": 542
+    },
+    {
+      "epoch": 0.5370919881305638,
+      "grad_norm": 0.4727615722395577,
+      "learning_rate": 4.562110663246611e-05,
+      "loss": 0.4987,
+      "step": 543
+    },
+    {
+      "epoch": 0.5380811078140455,
+      "grad_norm": 0.3679816340175357,
+      "learning_rate": 4.560278490289484e-05,
+      "loss": 0.5986,
+      "step": 544
+    },
+    {
+      "epoch": 0.5390702274975272,
+      "grad_norm": 0.4198111058431937,
+      "learning_rate": 4.5584463173323564e-05,
+      "loss": 0.6406,
+      "step": 545
+    },
+    {
+      "epoch": 0.5400593471810089,
+      "grad_norm": 0.381705361347438,
+      "learning_rate": 4.556614144375229e-05,
+      "loss": 0.5868,
+      "step": 546
+    },
+    {
+      "epoch": 0.5410484668644906,
+      "grad_norm": 0.36128173218405524,
+      "learning_rate": 4.554781971418102e-05,
+      "loss": 0.5606,
+      "step": 547
+    },
+    {
+      "epoch": 0.5420375865479723,
+      "grad_norm": 0.36624104089112686,
+      "learning_rate": 4.552949798460975e-05,
+      "loss": 0.5441,
+      "step": 548
+    },
+    {
+      "epoch": 0.543026706231454,
+      "grad_norm": 2.643110350633743,
+      "learning_rate": 4.5511176255038476e-05,
+      "loss": 0.6998,
+      "step": 549
+    },
+    {
+      "epoch": 0.5440158259149357,
+      "grad_norm": 0.45802173595768614,
+      "learning_rate": 4.54928545254672e-05,
+      "loss": 0.6603,
+      "step": 550
+    },
+    {
+      "epoch": 0.5450049455984174,
+      "grad_norm": 0.39506056655492705,
+      "learning_rate": 4.5474532795895936e-05,
+      "loss": 0.5296,
+      "step": 551
+    },
+    {
+      "epoch": 0.5459940652818991,
+      "grad_norm": 0.45358696386112984,
+      "learning_rate": 4.545621106632466e-05,
+      "loss": 0.5913,
+      "step": 552
+    },
+    {
+      "epoch": 0.5469831849653808,
+      "grad_norm": 0.8759613406131889,
+      "learning_rate": 4.5437889336753395e-05,
+      "loss": 0.5732,
+      "step": 553
+    },
+    {
+      "epoch": 0.5479723046488625,
+      "grad_norm": 0.5560376705190068,
+      "learning_rate": 4.5419567607182115e-05,
+      "loss": 0.5513,
+      "step": 554
+    },
+    {
+      "epoch": 0.5489614243323442,
+      "grad_norm": 0.46334900525439865,
+      "learning_rate": 4.540124587761085e-05,
+      "loss": 0.5847,
+      "step": 555
+    },
+    {
+      "epoch": 0.5499505440158259,
+      "grad_norm": 0.5044951628643652,
+      "learning_rate": 4.538292414803958e-05,
+      "loss": 0.6368,
+      "step": 556
+    },
+    {
+      "epoch": 0.5509396636993076,
+      "grad_norm": 0.44658793634543775,
+      "learning_rate": 4.536460241846831e-05,
+      "loss": 0.5482,
+      "step": 557
+    },
+    {
+      "epoch": 0.5519287833827893,
+      "grad_norm": 0.4668038982374279,
+      "learning_rate": 4.5346280688897034e-05,
+      "loss": 0.6158,
+      "step": 558
+    },
+    {
+      "epoch": 0.552917903066271,
+      "grad_norm": 0.48096355582962474,
+      "learning_rate": 4.532795895932576e-05,
+      "loss": 0.57,
+      "step": 559
+    },
+    {
+      "epoch": 0.5539070227497527,
+      "grad_norm": 0.4897334267462889,
+      "learning_rate": 4.5309637229754494e-05,
+      "loss": 0.552,
+      "step": 560
+    },
+    {
+      "epoch": 0.5548961424332344,
+      "grad_norm": 0.4185534474944087,
+      "learning_rate": 4.529131550018322e-05,
+      "loss": 0.5191,
+      "step": 561
+    },
+    {
+      "epoch": 0.5558852621167161,
+      "grad_norm": 0.5365259911263918,
+      "learning_rate": 4.527299377061195e-05,
+      "loss": 0.5084,
+      "step": 562
+    },
+    {
+      "epoch": 0.5568743818001978,
+      "grad_norm": 0.4215213687101347,
+      "learning_rate": 4.525467204104067e-05,
+      "loss": 0.6025,
+      "step": 563
+    },
+    {
+      "epoch": 0.5578635014836796,
+      "grad_norm": 0.4740697064883748,
+      "learning_rate": 4.5236350311469406e-05,
+      "loss": 0.4966,
+      "step": 564
+    },
+    {
+      "epoch": 0.5588526211671613,
+      "grad_norm": 0.44324661986533054,
+      "learning_rate": 4.521802858189813e-05,
+      "loss": 0.566,
+      "step": 565
+    },
+    {
+      "epoch": 0.559841740850643,
+      "grad_norm": 2.2285546984735856,
+      "learning_rate": 4.519970685232686e-05,
+      "loss": 0.7091,
+      "step": 566
+    },
+    {
+      "epoch": 0.5608308605341247,
+      "grad_norm": 0.5556074562961142,
+      "learning_rate": 4.518138512275559e-05,
+      "loss": 0.594,
+      "step": 567
+    },
+    {
+      "epoch": 0.5618199802176064,
+      "grad_norm": 0.4500464488699885,
+      "learning_rate": 4.516306339318432e-05,
+      "loss": 0.5656,
+      "step": 568
+    },
+    {
+      "epoch": 0.5628090999010881,
+      "grad_norm": 0.3512247907513221,
+      "learning_rate": 4.514474166361305e-05,
+      "loss": 0.5336,
+      "step": 569
+    },
+    {
+      "epoch": 0.5637982195845698,
+      "grad_norm": 0.4304687263152149,
+      "learning_rate": 4.512641993404177e-05,
+      "loss": 0.5865,
+      "step": 570
+    },
+    {
+      "epoch": 0.5647873392680515,
+      "grad_norm": 0.4521665914106063,
+      "learning_rate": 4.5108098204470505e-05,
+      "loss": 0.5873,
+      "step": 571
+    },
+    {
+      "epoch": 0.5657764589515332,
+      "grad_norm": 0.4159456344130283,
+      "learning_rate": 4.508977647489923e-05,
+      "loss": 0.5975,
+      "step": 572
+    },
+    {
+      "epoch": 0.5667655786350149,
+      "grad_norm": 0.5296905855219289,
+      "learning_rate": 4.5071454745327964e-05,
+      "loss": 0.5955,
+      "step": 573
+    },
+    {
+      "epoch": 0.5677546983184966,
+      "grad_norm": 0.4014808388631957,
+      "learning_rate": 4.505313301575669e-05,
+      "loss": 0.5433,
+      "step": 574
+    },
+    {
+      "epoch": 0.5687438180019783,
+      "grad_norm": 0.4319671517552426,
+      "learning_rate": 4.503481128618542e-05,
+      "loss": 0.4904,
+      "step": 575
+    },
+    {
+      "epoch": 0.56973293768546,
+      "grad_norm": 3.6175278656140404,
+      "learning_rate": 4.5016489556614144e-05,
+      "loss": 0.6217,
+      "step": 576
+    },
+    {
+      "epoch": 0.5707220573689417,
+      "grad_norm": 0.42678930653486047,
+      "learning_rate": 4.499816782704288e-05,
+      "loss": 0.5212,
+      "step": 577
+    },
+    {
+      "epoch": 0.5717111770524234,
+      "grad_norm": 0.4066859848927131,
+      "learning_rate": 4.49798460974716e-05,
+      "loss": 0.542,
+      "step": 578
+    },
+    {
+      "epoch": 0.5727002967359051,
+      "grad_norm": 0.4150317023868254,
+      "learning_rate": 4.496152436790033e-05,
+      "loss": 0.5509,
+      "step": 579
+    },
+    {
+      "epoch": 0.5736894164193868,
+      "grad_norm": 0.39696777501880237,
+      "learning_rate": 4.494320263832906e-05,
+      "loss": 0.6286,
+      "step": 580
+    },
+    {
+      "epoch": 0.5746785361028685,
+      "grad_norm": 0.42488826730568463,
+      "learning_rate": 4.492488090875779e-05,
+      "loss": 0.5245,
+      "step": 581
+    },
+    {
+      "epoch": 0.5756676557863502,
+      "grad_norm": 0.3832259003808749,
+      "learning_rate": 4.490655917918652e-05,
+      "loss": 0.513,
+      "step": 582
+    },
+    {
+      "epoch": 0.5766567754698319,
+      "grad_norm": 0.37765307913240526,
+      "learning_rate": 4.488823744961524e-05,
+      "loss": 0.5797,
+      "step": 583
+    },
+    {
+      "epoch": 0.5776458951533135,
+      "grad_norm": 0.5243524846267854,
+      "learning_rate": 4.4869915720043975e-05,
+      "loss": 0.5494,
+      "step": 584
+    },
+    {
+      "epoch": 0.5786350148367952,
+      "grad_norm": 0.35608241349020836,
+      "learning_rate": 4.48515939904727e-05,
+      "loss": 0.5169,
+      "step": 585
+    },
+    {
+      "epoch": 0.579624134520277,
+      "grad_norm": 0.5072400729075438,
+      "learning_rate": 4.4833272260901435e-05,
+      "loss": 0.5375,
+      "step": 586
+    },
+    {
+      "epoch": 0.5806132542037586,
+      "grad_norm": 0.3887888599949883,
+      "learning_rate": 4.4814950531330155e-05,
+      "loss": 0.5928,
+      "step": 587
+    },
+    {
+      "epoch": 0.5816023738872403,
+      "grad_norm": 0.38637277802524667,
+      "learning_rate": 4.479662880175889e-05,
+      "loss": 0.5631,
+      "step": 588
+    },
+    {
+      "epoch": 0.582591493570722,
+      "grad_norm": 0.4544815503207846,
+      "learning_rate": 4.4778307072187614e-05,
+      "loss": 0.5284,
+      "step": 589
+    },
+    {
+      "epoch": 0.5835806132542037,
+      "grad_norm": 0.44860638139409253,
+      "learning_rate": 4.475998534261635e-05,
+      "loss": 0.5568,
+      "step": 590
+    },
+    {
+      "epoch": 0.5845697329376854,
+      "grad_norm": 0.34783304628715694,
+      "learning_rate": 4.4741663613045074e-05,
+      "loss": 0.4736,
+      "step": 591
+    },
+    {
+      "epoch": 0.5855588526211671,
+      "grad_norm": 0.46635755164532483,
+      "learning_rate": 4.47233418834738e-05,
+      "loss": 0.5721,
+      "step": 592
+    },
+    {
+      "epoch": 0.5865479723046488,
+      "grad_norm": 0.35836770919240934,
+      "learning_rate": 4.4705020153902533e-05,
+      "loss": 0.5742,
+      "step": 593
+    },
+    {
+      "epoch": 0.5875370919881305,
+      "grad_norm": 2.1646391480973386,
+      "learning_rate": 4.468669842433126e-05,
+      "loss": 0.7749,
+      "step": 594
+    },
+    {
+      "epoch": 0.5885262116716122,
+      "grad_norm": 0.6322154073368694,
+      "learning_rate": 4.4668376694759986e-05,
+      "loss": 0.5222,
+      "step": 595
+    },
+    {
+      "epoch": 0.5895153313550939,
+      "grad_norm": 0.3447318173329335,
+      "learning_rate": 4.465005496518871e-05,
+      "loss": 0.5024,
+      "step": 596
+    },
+    {
+      "epoch": 0.5905044510385756,
+      "grad_norm": 0.5950854440009333,
+      "learning_rate": 4.4631733235617446e-05,
+      "loss": 0.6533,
+      "step": 597
+    },
+    {
+      "epoch": 0.5914935707220573,
+      "grad_norm": 0.521884110600529,
+      "learning_rate": 4.461341150604617e-05,
+      "loss": 0.5968,
+      "step": 598
+    },
+    {
+      "epoch": 0.592482690405539,
+      "grad_norm": 0.3392872051376154,
+      "learning_rate": 4.45950897764749e-05,
+      "loss": 0.5776,
+      "step": 599
+    },
+    {
+      "epoch": 0.5934718100890207,
+      "grad_norm": 0.5377239708646002,
+      "learning_rate": 4.4576768046903625e-05,
+      "loss": 0.6145,
+      "step": 600
+    },
+    {
+      "epoch": 0.5944609297725024,
+      "grad_norm": 0.4318047502673512,
+      "learning_rate": 4.455844631733236e-05,
+      "loss": 0.6059,
+      "step": 601
+    },
+    {
+      "epoch": 0.5954500494559841,
+      "grad_norm": 2.3631173625344735,
+      "learning_rate": 4.454012458776109e-05,
+      "loss": 0.779,
+      "step": 602
+    },
+    {
+      "epoch": 0.5964391691394659,
+      "grad_norm": 0.5336870304548778,
+      "learning_rate": 4.452180285818982e-05,
+      "loss": 0.5058,
+      "step": 603
+    },
+    {
+      "epoch": 0.5974282888229476,
+      "grad_norm": 0.4034031794946735,
+      "learning_rate": 4.4503481128618544e-05,
+      "loss": 0.5717,
+      "step": 604
+    },
+    {
+      "epoch": 0.5984174085064293,
+      "grad_norm": 0.37891278714404253,
+      "learning_rate": 4.448515939904727e-05,
+      "loss": 0.5299,
+      "step": 605
+    },
+    {
+      "epoch": 0.599406528189911,
+      "grad_norm": 1.6328876825145693,
+      "learning_rate": 4.4466837669476004e-05,
+      "loss": 0.7685,
+      "step": 606
+    },
+    {
+      "epoch": 0.6003956478733927,
+      "grad_norm": 0.4889387354780191,
+      "learning_rate": 4.444851593990473e-05,
+      "loss": 0.5767,
+      "step": 607
+    },
+    {
+      "epoch": 0.6013847675568744,
+      "grad_norm": 0.39614179421575757,
+      "learning_rate": 4.443019421033346e-05,
+      "loss": 0.5488,
+      "step": 608
+    },
+    {
+      "epoch": 0.6023738872403561,
+      "grad_norm": 0.4660736811721914,
+      "learning_rate": 4.441187248076218e-05,
+      "loss": 0.5036,
+      "step": 609
+    },
+    {
+      "epoch": 0.6033630069238378,
+      "grad_norm": 0.4305830100351165,
+      "learning_rate": 4.4393550751190916e-05,
+      "loss": 0.5586,
+      "step": 610
+    },
+    {
+      "epoch": 0.6043521266073195,
+      "grad_norm": 0.42857086912608233,
+      "learning_rate": 4.437522902161964e-05,
+      "loss": 0.5373,
+      "step": 611
+    },
+    {
+      "epoch": 0.6053412462908012,
+      "grad_norm": 0.4259812924670851,
+      "learning_rate": 4.435690729204837e-05,
+      "loss": 0.6364,
+      "step": 612
+    },
+    {
+      "epoch": 0.6063303659742829,
+      "grad_norm": 0.35384845137297105,
+      "learning_rate": 4.4338585562477096e-05,
+      "loss": 0.5686,
+      "step": 613
+    },
+    {
+      "epoch": 0.6073194856577646,
+      "grad_norm": 0.43840991923734457,
+      "learning_rate": 4.432026383290583e-05,
+      "loss": 0.5664,
+      "step": 614
+    },
+    {
+      "epoch": 0.6083086053412463,
+      "grad_norm": 0.4160175071465533,
+      "learning_rate": 4.430194210333456e-05,
+      "loss": 0.5562,
+      "step": 615
+    },
+    {
+      "epoch": 0.609297725024728,
+      "grad_norm": 0.4076622693219068,
+      "learning_rate": 4.428362037376328e-05,
+      "loss": 0.5819,
+      "step": 616
+    },
+    {
+      "epoch": 0.6102868447082097,
+      "grad_norm": 0.43491508340384,
+      "learning_rate": 4.4265298644192015e-05,
+      "loss": 0.6215,
+      "step": 617
+    },
+    {
+      "epoch": 0.6112759643916914,
+      "grad_norm": 0.9665467383337245,
+      "learning_rate": 4.424697691462074e-05,
+      "loss": 0.6142,
+      "step": 618
+    },
+    {
+      "epoch": 0.6122650840751731,
+      "grad_norm": 0.3867966415220155,
+      "learning_rate": 4.4228655185049474e-05,
+      "loss": 0.5207,
+      "step": 619
+    },
+    {
+      "epoch": 0.6132542037586548,
+      "grad_norm": 0.3384128186488838,
+      "learning_rate": 4.42103334554782e-05,
+      "loss": 0.5391,
+      "step": 620
+    },
+    {
+      "epoch": 0.6142433234421365,
+      "grad_norm": 0.37030112577139734,
+      "learning_rate": 4.419201172590693e-05,
+      "loss": 0.5357,
+      "step": 621
+    },
+    {
+      "epoch": 0.6152324431256182,
+      "grad_norm": 0.33532983716976167,
+      "learning_rate": 4.4173689996335654e-05,
+      "loss": 0.5596,
+      "step": 622
+    },
+    {
+      "epoch": 0.6162215628090999,
+      "grad_norm": 2.1912897019885156,
+      "learning_rate": 4.415536826676439e-05,
+      "loss": 0.663,
+      "step": 623
+    },
+    {
+      "epoch": 0.6172106824925816,
+      "grad_norm": 0.4108780719007618,
+      "learning_rate": 4.413704653719311e-05,
+      "loss": 0.5381,
+      "step": 624
+    },
+    {
+      "epoch": 0.6181998021760633,
+      "grad_norm": 0.345525682169045,
+      "learning_rate": 4.411872480762184e-05,
+      "loss": 0.6086,
+      "step": 625
+    },
+    {
+      "epoch": 0.619188921859545,
+      "grad_norm": 0.32145894628431493,
+      "learning_rate": 4.410040307805057e-05,
+      "loss": 0.5079,
+      "step": 626
+    },
+    {
+      "epoch": 0.6201780415430267,
+      "grad_norm": 0.3400074427817454,
+      "learning_rate": 4.40820813484793e-05,
+      "loss": 0.556,
+      "step": 627
+    },
+    {
+      "epoch": 0.6211671612265084,
+      "grad_norm": 0.336784464756408,
+      "learning_rate": 4.406375961890803e-05,
+      "loss": 0.5383,
+      "step": 628
+    },
+    {
+      "epoch": 0.6221562809099901,
+      "grad_norm": 0.36068226884339905,
+      "learning_rate": 4.404543788933675e-05,
+      "loss": 0.5838,
+      "step": 629
+    },
+    {
+      "epoch": 0.6231454005934718,
+      "grad_norm": 0.35409438086231915,
+      "learning_rate": 4.4027116159765485e-05,
+      "loss": 0.4777,
+      "step": 630
+    },
+    {
+      "epoch": 0.6241345202769535,
+      "grad_norm": 0.3726791740087236,
+      "learning_rate": 4.400879443019421e-05,
+      "loss": 0.6175,
+      "step": 631
+    },
+    {
+      "epoch": 0.6251236399604352,
+      "grad_norm": 0.4023806669048874,
+      "learning_rate": 4.3990472700622945e-05,
+      "loss": 0.6292,
+      "step": 632
+    },
+    {
+      "epoch": 0.6261127596439169,
+      "grad_norm": 0.34330582894277495,
+      "learning_rate": 4.3972150971051665e-05,
+      "loss": 0.4987,
+      "step": 633
+    },
+    {
+      "epoch": 0.6271018793273986,
+      "grad_norm": 0.3305730900775558,
+      "learning_rate": 4.39538292414804e-05,
+      "loss": 0.5235,
+      "step": 634
+    },
+    {
+      "epoch": 0.6280909990108803,
+      "grad_norm": 0.36092394055577376,
+      "learning_rate": 4.3935507511909124e-05,
+      "loss": 0.5585,
+      "step": 635
+    },
+    {
+      "epoch": 0.629080118694362,
+      "grad_norm": 0.4093047909780486,
+      "learning_rate": 4.391718578233786e-05,
+      "loss": 0.5418,
+      "step": 636
+    },
+    {
+      "epoch": 0.6300692383778437,
+      "grad_norm": 0.33645740331275986,
+      "learning_rate": 4.3898864052766584e-05,
+      "loss": 0.4971,
+      "step": 637
+    },
+    {
+      "epoch": 0.6310583580613254,
+      "grad_norm": 0.3611452319329633,
+      "learning_rate": 4.388054232319531e-05,
+      "loss": 0.5567,
+      "step": 638
+    },
+    {
+      "epoch": 0.6320474777448071,
+      "grad_norm": 0.4930946827145663,
+      "learning_rate": 4.3862220593624043e-05,
+      "loss": 0.5986,
+      "step": 639
+    },
+    {
+      "epoch": 0.6330365974282888,
+      "grad_norm": 0.3231177239659706,
+      "learning_rate": 4.384389886405277e-05,
+      "loss": 0.5975,
+      "step": 640
+    },
+    {
+      "epoch": 0.6340257171117705,
+      "grad_norm": 0.3827959704241396,
+      "learning_rate": 4.3825577134481496e-05,
+      "loss": 0.5177,
+      "step": 641
+    },
+    {
+      "epoch": 0.6350148367952523,
+      "grad_norm": 0.41977721218388675,
+      "learning_rate": 4.380725540491022e-05,
+      "loss": 0.6479,
+      "step": 642
+    },
+    {
+      "epoch": 0.636003956478734,
+      "grad_norm": 0.3662099933871658,
+      "learning_rate": 4.3788933675338956e-05,
+      "loss": 0.6163,
+      "step": 643
+    },
+    {
+      "epoch": 0.6369930761622157,
+      "grad_norm": 0.3712913474912623,
+      "learning_rate": 4.377061194576768e-05,
+      "loss": 0.4442,
+      "step": 644
+    },
+    {
+      "epoch": 0.6379821958456974,
+      "grad_norm": 0.5071826500684371,
+      "learning_rate": 4.375229021619641e-05,
+      "loss": 0.5606,
+      "step": 645
+    },
+    {
+      "epoch": 0.6389713155291791,
+      "grad_norm": 0.35640208864347855,
+      "learning_rate": 4.3733968486625135e-05,
+      "loss": 0.505,
+      "step": 646
+    },
+    {
+      "epoch": 0.6399604352126608,
+      "grad_norm": 0.4223414729847192,
+      "learning_rate": 4.371564675705387e-05,
+      "loss": 0.5597,
+      "step": 647
+    },
+    {
+      "epoch": 0.6409495548961425,
+      "grad_norm": 0.524278275869879,
+      "learning_rate": 4.3697325027482595e-05,
+      "loss": 0.5428,
+      "step": 648
+    },
+    {
+      "epoch": 0.6419386745796242,
+      "grad_norm": 0.34090519496656335,
+      "learning_rate": 4.367900329791133e-05,
+      "loss": 0.5068,
+      "step": 649
+    },
+    {
+      "epoch": 0.6429277942631059,
+      "grad_norm": 0.4974539026917071,
+      "learning_rate": 4.3660681568340054e-05,
+      "loss": 0.5449,
+      "step": 650
+    },
+    {
+      "epoch": 0.6439169139465876,
+      "grad_norm": 0.4263296924702846,
+      "learning_rate": 4.364235983876878e-05,
+      "loss": 0.5863,
+      "step": 651
+    },
+    {
+      "epoch": 0.6449060336300693,
+      "grad_norm": 0.36550014682885573,
+      "learning_rate": 4.3624038109197514e-05,
+      "loss": 0.5139,
+      "step": 652
+    },
+    {
+      "epoch": 0.645895153313551,
+      "grad_norm": 0.4420805717206251,
+      "learning_rate": 4.360571637962624e-05,
+      "loss": 0.4919,
+      "step": 653
+    },
+    {
+      "epoch": 0.6468842729970327,
+      "grad_norm": 3.5175564338439544,
+      "learning_rate": 4.358739465005497e-05,
+      "loss": 0.6138,
+      "step": 654
+    },
+    {
+      "epoch": 0.6478733926805144,
+      "grad_norm": 0.44353550687361537,
+      "learning_rate": 4.356907292048369e-05,
+      "loss": 0.5678,
+      "step": 655
+    },
+    {
+      "epoch": 0.6488625123639961,
+      "grad_norm": 0.42907631403450497,
+      "learning_rate": 4.3550751190912426e-05,
+      "loss": 0.5126,
+      "step": 656
+    },
+    {
+      "epoch": 0.6498516320474778,
+      "grad_norm": 0.5224325181644863,
+      "learning_rate": 4.353242946134115e-05,
+      "loss": 0.567,
+      "step": 657
+    },
+    {
+      "epoch": 0.6508407517309595,
+      "grad_norm": 0.36393145929326026,
+      "learning_rate": 4.351410773176988e-05,
+      "loss": 0.5758,
+      "step": 658
+    },
+    {
+      "epoch": 0.6518298714144412,
+      "grad_norm": 0.43831662678367533,
+      "learning_rate": 4.3495786002198606e-05,
+      "loss": 0.4662,
+      "step": 659
+    },
+    {
+      "epoch": 0.6528189910979229,
+      "grad_norm": 0.44731721791336854,
+      "learning_rate": 4.347746427262734e-05,
+      "loss": 0.5162,
+      "step": 660
+    },
+    {
+      "epoch": 0.6538081107814046,
+      "grad_norm": 0.3625805902395787,
+      "learning_rate": 4.3459142543056065e-05,
+      "loss": 0.5866,
+      "step": 661
+    },
+    {
+      "epoch": 0.6547972304648862,
+      "grad_norm": 0.374836185068099,
+      "learning_rate": 4.344082081348479e-05,
+      "loss": 0.5622,
+      "step": 662
+    },
+    {
+      "epoch": 0.655786350148368,
+      "grad_norm": 0.3636250559467318,
+      "learning_rate": 4.3422499083913525e-05,
+      "loss": 0.4704,
+      "step": 663
+    },
+    {
+      "epoch": 0.6567754698318496,
+      "grad_norm": 0.368172907844344,
+      "learning_rate": 4.340417735434225e-05,
+      "loss": 0.4582,
+      "step": 664
+    },
+    {
+      "epoch": 0.6577645895153313,
+      "grad_norm": 0.3781176871717141,
+      "learning_rate": 4.3385855624770985e-05,
+      "loss": 0.5608,
+      "step": 665
+    },
+    {
+      "epoch": 0.658753709198813,
+      "grad_norm": 0.3472496247475107,
+      "learning_rate": 4.336753389519971e-05,
+      "loss": 0.5267,
+      "step": 666
+    },
+    {
+      "epoch": 0.6597428288822947,
+      "grad_norm": 0.35710749282494153,
+      "learning_rate": 4.334921216562844e-05,
+      "loss": 0.5912,
+      "step": 667
+    },
+    {
+      "epoch": 0.6607319485657764,
+      "grad_norm": 0.34177842883688003,
+      "learning_rate": 4.3330890436057164e-05,
+      "loss": 0.5239,
+      "step": 668
+    },
+    {
+      "epoch": 0.6617210682492581,
+      "grad_norm": 0.45206446669886513,
+      "learning_rate": 4.33125687064859e-05,
+      "loss": 0.6048,
+      "step": 669
+    },
+    {
+      "epoch": 0.6627101879327398,
+      "grad_norm": 0.43477515611045725,
+      "learning_rate": 4.3294246976914623e-05,
+      "loss": 0.5396,
+      "step": 670
+    },
+    {
+      "epoch": 0.6636993076162215,
+      "grad_norm": 0.4495761387137212,
+      "learning_rate": 4.327592524734335e-05,
+      "loss": 0.5174,
+      "step": 671
+    },
+    {
+      "epoch": 0.6646884272997032,
+      "grad_norm": 1.8122069786779942,
+      "learning_rate": 4.3257603517772076e-05,
+      "loss": 0.5415,
+      "step": 672
+    },
+    {
+      "epoch": 0.6656775469831849,
+      "grad_norm": 0.54700387793728,
+      "learning_rate": 4.323928178820081e-05,
+      "loss": 0.5466,
+      "step": 673
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.3497883391162035,
+      "learning_rate": 4.322096005862954e-05,
+      "loss": 0.5633,
+      "step": 674
+    },
+    {
+      "epoch": 0.6676557863501483,
+      "grad_norm": 0.45126911871908887,
+      "learning_rate": 4.320263832905826e-05,
+      "loss": 0.5531,
+      "step": 675
+    },
+    {
+      "epoch": 0.66864490603363,
+      "grad_norm": 0.3951817778962431,
+      "learning_rate": 4.3184316599486995e-05,
+      "loss": 0.5225,
+      "step": 676
+    },
+    {
+      "epoch": 0.6696340257171117,
+      "grad_norm": 0.5240280173507978,
+      "learning_rate": 4.316599486991572e-05,
+      "loss": 0.5278,
+      "step": 677
+    },
+    {
+      "epoch": 0.6706231454005934,
+      "grad_norm": 0.3375650680362749,
+      "learning_rate": 4.3147673140344455e-05,
+      "loss": 0.5464,
+      "step": 678
+    },
+    {
+      "epoch": 0.6716122650840751,
+      "grad_norm": 0.41081775932232456,
+      "learning_rate": 4.3129351410773175e-05,
+      "loss": 0.5289,
+      "step": 679
+    },
+    {
+      "epoch": 0.6726013847675568,
+      "grad_norm": 0.42912334923955686,
+      "learning_rate": 4.311102968120191e-05,
+      "loss": 0.5685,
+      "step": 680
+    },
+    {
+      "epoch": 0.6735905044510386,
+      "grad_norm": 0.44709522287251285,
+      "learning_rate": 4.3092707951630634e-05,
+      "loss": 0.5861,
+      "step": 681
+    },
+    {
+      "epoch": 0.6745796241345203,
+      "grad_norm": 8.932319607989578,
+      "learning_rate": 4.307438622205937e-05,
+      "loss": 0.5878,
+      "step": 682
+    },
+    {
+      "epoch": 0.675568743818002,
+      "grad_norm": 0.6163038315017687,
+      "learning_rate": 4.3056064492488094e-05,
+      "loss": 0.552,
+      "step": 683
+    },
+    {
+      "epoch": 0.6765578635014837,
+      "grad_norm": 0.4130993321464594,
+      "learning_rate": 4.303774276291682e-05,
+      "loss": 0.484,
+      "step": 684
+    },
+    {
+      "epoch": 0.6775469831849654,
+      "grad_norm": 0.44149481672035984,
+      "learning_rate": 4.301942103334555e-05,
+      "loss": 0.487,
+      "step": 685
+    },
+    {
+      "epoch": 0.6785361028684471,
+      "grad_norm": 0.44270474817714917,
+      "learning_rate": 4.300109930377428e-05,
+      "loss": 0.4829,
+      "step": 686
+    },
+    {
+      "epoch": 0.6795252225519288,
+      "grad_norm": 1.778964998297047,
+      "learning_rate": 4.2982777574203006e-05,
+      "loss": 0.6008,
+      "step": 687
+    },
+    {
+      "epoch": 0.6805143422354105,
+      "grad_norm": 0.5042046907833749,
+      "learning_rate": 4.296445584463173e-05,
+      "loss": 0.5508,
+      "step": 688
+    },
+    {
+      "epoch": 0.6815034619188922,
+      "grad_norm": 1.2364847410894697,
+      "learning_rate": 4.2946134115060466e-05,
+      "loss": 0.4864,
+      "step": 689
+    },
+    {
+      "epoch": 0.6824925816023739,
+      "grad_norm": 5.675707936422887,
+      "learning_rate": 4.292781238548919e-05,
+      "loss": 0.5267,
+      "step": 690
+    },
+    {
+      "epoch": 0.6834817012858556,
+      "grad_norm": 1.0113014216945861,
+      "learning_rate": 4.290949065591792e-05,
+      "loss": 0.6003,
+      "step": 691
+    },
+    {
+      "epoch": 0.6844708209693373,
+      "grad_norm": 0.7307635089929869,
+      "learning_rate": 4.2891168926346645e-05,
+      "loss": 0.5435,
+      "step": 692
+    },
+    {
+      "epoch": 0.685459940652819,
+      "grad_norm": 1.005536324793941,
+      "learning_rate": 4.287284719677538e-05,
+      "loss": 0.5953,
+      "step": 693
+    },
+    {
+      "epoch": 0.6864490603363007,
+      "grad_norm": 0.8795150549550624,
+      "learning_rate": 4.2854525467204105e-05,
+      "loss": 0.5776,
+      "step": 694
+    },
+    {
+      "epoch": 0.6874381800197824,
+      "grad_norm": 0.9217585885833345,
+      "learning_rate": 4.283620373763284e-05,
+      "loss": 0.4713,
+      "step": 695
+    },
+    {
+      "epoch": 0.6884272997032641,
+      "grad_norm": 0.5750818575992418,
+      "learning_rate": 4.281788200806156e-05,
+      "loss": 0.5646,
+      "step": 696
+    },
+    {
+      "epoch": 0.6894164193867458,
+      "grad_norm": 1.1839342901013068,
+      "learning_rate": 4.279956027849029e-05,
+      "loss": 0.5912,
+      "step": 697
+    },
+    {
+      "epoch": 0.6904055390702275,
+      "grad_norm": 0.40639325595326,
+      "learning_rate": 4.2781238548919024e-05,
+      "loss": 0.5001,
+      "step": 698
+    },
+    {
+      "epoch": 0.6913946587537092,
+      "grad_norm": 0.9197610387321684,
+      "learning_rate": 4.276291681934775e-05,
+      "loss": 0.5174,
+      "step": 699
+    },
+    {
+      "epoch": 0.6923837784371909,
+      "grad_norm": 2.2435633788044482,
+      "learning_rate": 4.274459508977648e-05,
+      "loss": 0.6464,
+      "step": 700
+    },
+    {
+      "epoch": 0.6933728981206726,
+      "grad_norm": 2.2636363371000616,
+      "learning_rate": 4.27262733602052e-05,
+      "loss": 0.6616,
+      "step": 701
+    },
+    {
+      "epoch": 0.6943620178041543,
+      "grad_norm": 1.0957925072649075,
+      "learning_rate": 4.2707951630633937e-05,
+      "loss": 0.5103,
+      "step": 702
+    },
+    {
+      "epoch": 0.695351137487636,
+      "grad_norm": 0.44136767500918594,
+      "learning_rate": 4.268962990106266e-05,
+      "loss": 0.5773,
+      "step": 703
+    },
+    {
+      "epoch": 0.6963402571711177,
+      "grad_norm": 1.2119093160665309,
+      "learning_rate": 4.267130817149139e-05,
+      "loss": 0.5676,
+      "step": 704
+    },
+    {
+      "epoch": 0.6973293768545994,
+      "grad_norm": 0.9844741497724914,
+      "learning_rate": 4.2652986441920116e-05,
+      "loss": 0.5346,
+      "step": 705
+    },
+    {
+      "epoch": 0.6983184965380811,
+      "grad_norm": 0.9825434500331776,
+      "learning_rate": 4.263466471234885e-05,
+      "loss": 0.579,
+      "step": 706
+    },
+    {
+      "epoch": 0.6993076162215628,
+      "grad_norm": 1.0080629257558236,
+      "learning_rate": 4.2616342982777575e-05,
+      "loss": 0.5322,
+      "step": 707
+    },
+    {
+      "epoch": 0.7002967359050445,
+      "grad_norm": 0.6306084604291936,
+      "learning_rate": 4.25980212532063e-05,
+      "loss": 0.6135,
+      "step": 708
+    },
+    {
+      "epoch": 0.7012858555885262,
+      "grad_norm": 0.7935850704567368,
+      "learning_rate": 4.257969952363503e-05,
+      "loss": 0.4881,
+      "step": 709
+    },
+    {
+      "epoch": 0.7022749752720079,
+      "grad_norm": 0.42031411509137817,
+      "learning_rate": 4.256137779406376e-05,
+      "loss": 0.5627,
+      "step": 710
+    },
+    {
+      "epoch": 0.7032640949554896,
+      "grad_norm": 0.6431022906724274,
+      "learning_rate": 4.2543056064492495e-05,
+      "loss": 0.5521,
+      "step": 711
+    },
+    {
+      "epoch": 0.7042532146389713,
+      "grad_norm": 0.4051950629961147,
+      "learning_rate": 4.252473433492122e-05,
+      "loss": 0.539,
+      "step": 712
+    },
+    {
+      "epoch": 0.705242334322453,
+      "grad_norm": 0.5504001902905677,
+      "learning_rate": 4.250641260534995e-05,
+      "loss": 0.6014,
+      "step": 713
+    },
+    {
+      "epoch": 0.7062314540059347,
+      "grad_norm": 0.42943927228102546,
+      "learning_rate": 4.2488090875778674e-05,
+      "loss": 0.5071,
+      "step": 714
+    },
+    {
+      "epoch": 0.7072205736894164,
+      "grad_norm": 0.4564346346810455,
+      "learning_rate": 4.246976914620741e-05,
+      "loss": 0.5798,
+      "step": 715
+    },
+    {
+      "epoch": 0.7082096933728981,
+      "grad_norm": 0.41971345322735765,
+      "learning_rate": 4.2451447416636133e-05,
+      "loss": 0.4746,
+      "step": 716
+    },
+    {
+      "epoch": 0.7091988130563798,
+      "grad_norm": 0.4767124000008028,
+      "learning_rate": 4.243312568706486e-05,
+      "loss": 0.5221,
+      "step": 717
+    },
+    {
+      "epoch": 0.7101879327398615,
+      "grad_norm": 0.473920208531501,
+      "learning_rate": 4.2414803957493586e-05,
+      "loss": 0.6092,
+      "step": 718
+    },
+    {
+      "epoch": 0.7111770524233432,
+      "grad_norm": 0.4008182380340919,
+      "learning_rate": 4.239648222792232e-05,
+      "loss": 0.5633,
+      "step": 719
+    },
+    {
+      "epoch": 0.712166172106825,
+      "grad_norm": 0.4572650059707555,
+      "learning_rate": 4.2378160498351046e-05,
+      "loss": 0.4485,
+      "step": 720
+    },
+    {
+      "epoch": 0.7131552917903067,
+      "grad_norm": 0.3982981440820982,
+      "learning_rate": 4.235983876877977e-05,
+      "loss": 0.5882,
+      "step": 721
+    },
+    {
+      "epoch": 0.7141444114737884,
+      "grad_norm": 0.4445315839865244,
+      "learning_rate": 4.2341517039208506e-05,
+      "loss": 0.499,
+      "step": 722
+    },
+    {
+      "epoch": 0.7151335311572701,
+      "grad_norm": 0.45327248020817545,
+      "learning_rate": 4.232319530963723e-05,
+      "loss": 0.5828,
+      "step": 723
+    },
+    {
+      "epoch": 0.7161226508407518,
+      "grad_norm": 0.45072469495117606,
+      "learning_rate": 4.2304873580065965e-05,
+      "loss": 0.5684,
+      "step": 724
+    },
+    {
+      "epoch": 0.7171117705242335,
+      "grad_norm": 0.5030033920719464,
+      "learning_rate": 4.2286551850494685e-05,
+      "loss": 0.5462,
+      "step": 725
+    },
+    {
+      "epoch": 0.7181008902077152,
+      "grad_norm": 0.41612647044276724,
+      "learning_rate": 4.226823012092342e-05,
+      "loss": 0.5931,
+      "step": 726
+    },
+    {
+      "epoch": 0.7190900098911969,
+      "grad_norm": 0.48469080484439975,
+      "learning_rate": 4.2249908391352144e-05,
+      "loss": 0.5663,
+      "step": 727
+    },
+    {
+      "epoch": 0.7200791295746786,
+      "grad_norm": 0.3288997249334829,
+      "learning_rate": 4.223158666178088e-05,
+      "loss": 0.5097,
+      "step": 728
+    },
+    {
+      "epoch": 0.7210682492581603,
+      "grad_norm": 0.4306446848182276,
+      "learning_rate": 4.2213264932209604e-05,
+      "loss": 0.5869,
+      "step": 729
+    },
+    {
+      "epoch": 0.722057368941642,
+      "grad_norm": 4.7698964565308675,
+      "learning_rate": 4.219494320263833e-05,
+      "loss": 0.6767,
+      "step": 730
+    },
+    {
+      "epoch": 0.7230464886251237,
+      "grad_norm": 0.5127544163487205,
+      "learning_rate": 4.217662147306706e-05,
+      "loss": 0.6094,
+      "step": 731
+    },
+    {
+      "epoch": 0.7240356083086054,
+      "grad_norm": 0.41334473621761253,
+      "learning_rate": 4.215829974349579e-05,
+      "loss": 0.5082,
+      "step": 732
+    },
+    {
+      "epoch": 0.7250247279920871,
+      "grad_norm": 0.36240790003835044,
+      "learning_rate": 4.2139978013924516e-05,
+      "loss": 0.5948,
+      "step": 733
+    },
+    {
+      "epoch": 0.7260138476755688,
+      "grad_norm": 0.36783320263349717,
+      "learning_rate": 4.212165628435324e-05,
+      "loss": 0.4404,
+      "step": 734
+    },
+    {
+      "epoch": 0.7270029673590505,
+      "grad_norm": 0.42754353298521264,
+      "learning_rate": 4.2103334554781976e-05,
+      "loss": 0.5593,
+      "step": 735
+    },
+    {
+      "epoch": 0.7279920870425322,
+      "grad_norm": 0.334115716001555,
+      "learning_rate": 4.20850128252107e-05,
+      "loss": 0.5471,
+      "step": 736
+    },
+    {
+      "epoch": 0.7289812067260139,
+      "grad_norm": 0.3964403729349191,
+      "learning_rate": 4.206669109563943e-05,
+      "loss": 0.5169,
+      "step": 737
+    },
+    {
+      "epoch": 0.7299703264094956,
+      "grad_norm": 0.3524084819698355,
+      "learning_rate": 4.2048369366068155e-05,
+      "loss": 0.5195,
+      "step": 738
+    },
+    {
+      "epoch": 0.7309594460929772,
+      "grad_norm": 0.8826038613244607,
+      "learning_rate": 4.203004763649689e-05,
+      "loss": 0.4875,
+      "step": 739
+    },
+    {
+      "epoch": 0.731948565776459,
+      "grad_norm": 0.39827011655841327,
+      "learning_rate": 4.2011725906925615e-05,
+      "loss": 0.5543,
+      "step": 740
+    },
+    {
+      "epoch": 0.7329376854599406,
+      "grad_norm": 0.42818287781464126,
+      "learning_rate": 4.199340417735435e-05,
+      "loss": 0.5792,
+      "step": 741
+    },
+    {
+      "epoch": 0.7339268051434223,
+      "grad_norm": 1.6645913416124833,
+      "learning_rate": 4.197508244778307e-05,
+      "loss": 0.5845,
+      "step": 742
+    },
+    {
+      "epoch": 0.734915924826904,
+      "grad_norm": 0.41168909047511215,
+      "learning_rate": 4.19567607182118e-05,
+      "loss": 0.5943,
+      "step": 743
+    },
+    {
+      "epoch": 0.7359050445103857,
+      "grad_norm": 0.48966326888171574,
+      "learning_rate": 4.193843898864053e-05,
+      "loss": 0.5834,
+      "step": 744
+    },
+    {
+      "epoch": 0.7368941641938674,
+      "grad_norm": 0.7123014415902675,
+      "learning_rate": 4.192011725906926e-05,
+      "loss": 0.5631,
+      "step": 745
+    },
+    {
+      "epoch": 0.7378832838773491,
+      "grad_norm": 0.44472269484910176,
+      "learning_rate": 4.190179552949799e-05,
+      "loss": 0.6197,
+      "step": 746
+    },
+    {
+      "epoch": 0.7388724035608308,
+      "grad_norm": 0.5054408180848112,
+      "learning_rate": 4.1883473799926713e-05,
+      "loss": 0.5104,
+      "step": 747
+    },
+    {
+      "epoch": 0.7398615232443125,
+      "grad_norm": 0.5692571400825903,
+      "learning_rate": 4.1865152070355447e-05,
+      "loss": 0.5738,
+      "step": 748
+    },
+    {
+      "epoch": 0.7408506429277942,
+      "grad_norm": 0.45821840458928037,
+      "learning_rate": 4.184683034078417e-05,
+      "loss": 0.5037,
+      "step": 749
+    },
+    {
+      "epoch": 0.7418397626112759,
+      "grad_norm": 1.64441512213632,
+      "learning_rate": 4.18285086112129e-05,
+      "loss": 0.5396,
+      "step": 750
+    },
+    {
+      "epoch": 0.7428288822947576,
+      "grad_norm": 0.45495252031691974,
+      "learning_rate": 4.1810186881641626e-05,
+      "loss": 0.5177,
+      "step": 751
+    },
+    {
+      "epoch": 0.7438180019782393,
+      "grad_norm": 0.49653489147704893,
+      "learning_rate": 4.179186515207036e-05,
+      "loss": 0.6201,
+      "step": 752
+    },
+    {
+      "epoch": 0.744807121661721,
+      "grad_norm": 0.3524624469672887,
+      "learning_rate": 4.1773543422499085e-05,
+      "loss": 0.492,
+      "step": 753
+    },
+    {
+      "epoch": 0.7457962413452027,
+      "grad_norm": 0.4437060967033037,
+      "learning_rate": 4.175522169292781e-05,
+      "loss": 0.5195,
+      "step": 754
+    },
+    {
+      "epoch": 0.7467853610286844,
+      "grad_norm": 0.5587440781800124,
+      "learning_rate": 4.173689996335654e-05,
+      "loss": 0.5915,
+      "step": 755
+    },
+    {
+      "epoch": 0.7477744807121661,
+      "grad_norm": 0.3503317806200955,
+      "learning_rate": 4.171857823378527e-05,
+      "loss": 0.5469,
+      "step": 756
+    },
+    {
+      "epoch": 0.7487636003956478,
+      "grad_norm": 0.6127018887132037,
+      "learning_rate": 4.1700256504214005e-05,
+      "loss": 0.593,
+      "step": 757
+    },
+    {
+      "epoch": 0.7497527200791295,
+      "grad_norm": 0.37713529443389476,
+      "learning_rate": 4.168193477464273e-05,
+      "loss": 0.5149,
+      "step": 758
+    },
+    {
+      "epoch": 0.7507418397626113,
+      "grad_norm": 0.4418061035198679,
+      "learning_rate": 4.166361304507146e-05,
+      "loss": 0.5793,
+      "step": 759
+    },
+    {
+      "epoch": 0.751730959446093,
+      "grad_norm": 0.45422082453440427,
+      "learning_rate": 4.1645291315500184e-05,
+      "loss": 0.5298,
+      "step": 760
+    },
+    {
+      "epoch": 0.7527200791295747,
+      "grad_norm": 0.3609291523640555,
+      "learning_rate": 4.162696958592892e-05,
+      "loss": 0.5442,
+      "step": 761
+    },
+    {
+      "epoch": 0.7537091988130564,
+      "grad_norm": 0.3952489577476288,
+      "learning_rate": 4.1608647856357644e-05,
+      "loss": 0.4714,
+      "step": 762
+    },
+    {
+      "epoch": 0.7546983184965381,
+      "grad_norm": 0.4370926922917512,
+      "learning_rate": 4.159032612678637e-05,
+      "loss": 0.5229,
+      "step": 763
+    },
+    {
+      "epoch": 0.7556874381800198,
+      "grad_norm": 0.37870613078543963,
+      "learning_rate": 4.1572004397215096e-05,
+      "loss": 0.4551,
+      "step": 764
+    },
+    {
+      "epoch": 0.7566765578635015,
+      "grad_norm": 0.43124183371724567,
+      "learning_rate": 4.155368266764383e-05,
+      "loss": 0.5467,
+      "step": 765
+    },
+    {
+      "epoch": 0.7576656775469832,
+      "grad_norm": 0.3746671658838302,
+      "learning_rate": 4.1535360938072556e-05,
+      "loss": 0.5807,
+      "step": 766
+    },
+    {
+      "epoch": 0.7586547972304649,
+      "grad_norm": 0.46058945803907064,
+      "learning_rate": 4.151703920850128e-05,
+      "loss": 0.5734,
+      "step": 767
+    },
+    {
+      "epoch": 0.7596439169139466,
+      "grad_norm": 0.3766410341280779,
+      "learning_rate": 4.149871747893001e-05,
+      "loss": 0.531,
+      "step": 768
+    },
+    {
+      "epoch": 0.7606330365974283,
+      "grad_norm": 0.4917230854366619,
+      "learning_rate": 4.148039574935874e-05,
+      "loss": 0.529,
+      "step": 769
+    },
+    {
+      "epoch": 0.76162215628091,
+      "grad_norm": 0.3601762759017485,
+      "learning_rate": 4.1462074019787475e-05,
+      "loss": 0.508,
+      "step": 770
+    },
+    {
+      "epoch": 0.7626112759643917,
+      "grad_norm": 1.0927135392910616,
+      "learning_rate": 4.1443752290216195e-05,
+      "loss": 0.5042,
+      "step": 771
+    },
+    {
+      "epoch": 0.7636003956478734,
+      "grad_norm": 0.564538713331542,
+      "learning_rate": 4.142543056064493e-05,
+      "loss": 0.5492,
+      "step": 772
+    },
+    {
+      "epoch": 0.7645895153313551,
+      "grad_norm": 0.378038419799154,
+      "learning_rate": 4.1407108831073654e-05,
+      "loss": 0.5775,
+      "step": 773
+    },
+    {
+      "epoch": 0.7655786350148368,
+      "grad_norm": 0.5132969793247761,
+      "learning_rate": 4.138878710150239e-05,
+      "loss": 0.543,
+      "step": 774
+    },
+    {
+      "epoch": 0.7665677546983185,
+      "grad_norm": 0.4657837803710372,
+      "learning_rate": 4.1370465371931114e-05,
+      "loss": 0.599,
+      "step": 775
+    },
+    {
+      "epoch": 0.7675568743818002,
+      "grad_norm": 0.41364334636794253,
+      "learning_rate": 4.135214364235984e-05,
+      "loss": 0.4887,
+      "step": 776
+    },
+    {
+      "epoch": 0.7685459940652819,
+      "grad_norm": 0.48494156005877015,
+      "learning_rate": 4.133382191278857e-05,
+      "loss": 0.5205,
+      "step": 777
+    },
+    {
+      "epoch": 0.7695351137487636,
+      "grad_norm": 0.5890553667830905,
+      "learning_rate": 4.13155001832173e-05,
+      "loss": 0.5641,
+      "step": 778
+    },
+    {
+      "epoch": 0.7705242334322453,
+      "grad_norm": 0.4152911192832434,
+      "learning_rate": 4.1297178453646027e-05,
+      "loss": 0.5257,
+      "step": 779
+    },
+    {
+      "epoch": 0.771513353115727,
+      "grad_norm": 1.6781124035661552,
+      "learning_rate": 4.127885672407475e-05,
+      "loss": 0.602,
+      "step": 780
+    },
+    {
+      "epoch": 0.7725024727992087,
+      "grad_norm": 0.5975859652214398,
+      "learning_rate": 4.1260534994503486e-05,
+      "loss": 0.5373,
+      "step": 781
+    },
+    {
+      "epoch": 0.7734915924826904,
+      "grad_norm": 0.6170145683827101,
+      "learning_rate": 4.124221326493221e-05,
+      "loss": 0.5326,
+      "step": 782
+    },
+    {
+      "epoch": 0.7744807121661721,
+      "grad_norm": 0.612440557441423,
+      "learning_rate": 4.122389153536094e-05,
+      "loss": 0.5741,
+      "step": 783
+    },
+    {
+      "epoch": 0.7754698318496538,
+      "grad_norm": 0.43061617125576557,
+      "learning_rate": 4.1205569805789665e-05,
+      "loss": 0.5607,
+      "step": 784
+    },
+    {
+      "epoch": 0.7764589515331355,
+      "grad_norm": 0.5036215670510789,
+      "learning_rate": 4.11872480762184e-05,
+      "loss": 0.558,
+      "step": 785
+    },
+    {
+      "epoch": 0.7774480712166172,
+      "grad_norm": 0.44972944661661773,
+      "learning_rate": 4.1168926346647125e-05,
+      "loss": 0.5524,
+      "step": 786
+    },
+    {
+      "epoch": 0.7784371909000989,
+      "grad_norm": 0.5179361442529129,
+      "learning_rate": 4.115060461707586e-05,
+      "loss": 0.5389,
+      "step": 787
+    },
+    {
+      "epoch": 0.7794263105835806,
+      "grad_norm": 0.40648583153660667,
+      "learning_rate": 4.113228288750458e-05,
+      "loss": 0.5839,
+      "step": 788
+    },
+    {
+      "epoch": 0.7804154302670623,
+      "grad_norm": 0.4831293697463915,
+      "learning_rate": 4.111396115793331e-05,
+      "loss": 0.571,
+      "step": 789
+    },
+    {
+      "epoch": 0.781404549950544,
+      "grad_norm": 0.8247754938487293,
+      "learning_rate": 4.109563942836204e-05,
+      "loss": 0.5338,
+      "step": 790
+    },
+    {
+      "epoch": 0.7823936696340257,
+      "grad_norm": 0.4025900842786157,
+      "learning_rate": 4.107731769879077e-05,
+      "loss": 0.5634,
+      "step": 791
+    },
+    {
+      "epoch": 0.7833827893175074,
+      "grad_norm": 0.40749518524574807,
+      "learning_rate": 4.105899596921949e-05,
+      "loss": 0.4944,
+      "step": 792
+    },
+    {
+      "epoch": 0.7843719090009891,
+      "grad_norm": 0.5169892850825292,
+      "learning_rate": 4.1040674239648223e-05,
+      "loss": 0.5161,
+      "step": 793
+    },
+    {
+      "epoch": 0.7853610286844708,
+      "grad_norm": 0.35450257355535614,
+      "learning_rate": 4.102235251007696e-05,
+      "loss": 0.5398,
+      "step": 794
+    },
+    {
+      "epoch": 0.7863501483679525,
+      "grad_norm": 0.4013582591042465,
+      "learning_rate": 4.100403078050568e-05,
+      "loss": 0.5003,
+      "step": 795
+    },
+    {
+      "epoch": 0.7873392680514342,
+      "grad_norm": 0.37673552582477354,
+      "learning_rate": 4.098570905093441e-05,
+      "loss": 0.4951,
+      "step": 796
+    },
+    {
+      "epoch": 0.7883283877349159,
+      "grad_norm": 0.33875066525744113,
+      "learning_rate": 4.0967387321363136e-05,
+      "loss": 0.4806,
+      "step": 797
+    },
+    {
+      "epoch": 0.7893175074183977,
+      "grad_norm": 0.476408908388567,
+      "learning_rate": 4.094906559179187e-05,
+      "loss": 0.5316,
+      "step": 798
+    },
+    {
+      "epoch": 0.7903066271018794,
+      "grad_norm": 0.3860532691526498,
+      "learning_rate": 4.0930743862220596e-05,
+      "loss": 0.5714,
+      "step": 799
+    },
+    {
+      "epoch": 0.7912957467853611,
+      "grad_norm": 0.30426399871981497,
+      "learning_rate": 4.091242213264932e-05,
+      "loss": 0.548,
+      "step": 800
+    },
+    {
+      "epoch": 0.7922848664688428,
+      "grad_norm": 0.43022889880235965,
+      "learning_rate": 4.089410040307805e-05,
+      "loss": 0.4889,
+      "step": 801
+    },
+    {
+      "epoch": 0.7932739861523245,
+      "grad_norm": 0.4014383758314632,
+      "learning_rate": 4.087577867350678e-05,
+      "loss": 0.5388,
+      "step": 802
+    },
+    {
+      "epoch": 0.7942631058358062,
+      "grad_norm": 0.358329442879335,
+      "learning_rate": 4.085745694393551e-05,
+      "loss": 0.6311,
+      "step": 803
+    },
+    {
+      "epoch": 0.7952522255192879,
+      "grad_norm": 0.5003987855371286,
+      "learning_rate": 4.083913521436424e-05,
+      "loss": 0.5505,
+      "step": 804
+    },
+    {
+      "epoch": 0.7962413452027696,
+      "grad_norm": 0.39761394658298255,
+      "learning_rate": 4.082081348479297e-05,
+      "loss": 0.5423,
+      "step": 805
+    },
+    {
+      "epoch": 0.7972304648862513,
+      "grad_norm": 0.39424324806445543,
+      "learning_rate": 4.0802491755221694e-05,
+      "loss": 0.5376,
+      "step": 806
+    },
+    {
+      "epoch": 0.798219584569733,
+      "grad_norm": 0.4111741063989657,
+      "learning_rate": 4.078417002565043e-05,
+      "loss": 0.5637,
+      "step": 807
+    },
+    {
+      "epoch": 0.7992087042532147,
+      "grad_norm": 0.4089042750817777,
+      "learning_rate": 4.0765848296079154e-05,
+      "loss": 0.6027,
+      "step": 808
+    },
+    {
+      "epoch": 0.8001978239366964,
+      "grad_norm": 0.35649158749683113,
+      "learning_rate": 4.074752656650788e-05,
+      "loss": 0.5493,
+      "step": 809
+    },
+    {
+      "epoch": 0.8011869436201781,
+      "grad_norm": 0.34390908785069546,
+      "learning_rate": 4.0729204836936606e-05,
+      "loss": 0.4582,
+      "step": 810
+    },
+    {
+      "epoch": 0.8021760633036598,
+      "grad_norm": 0.46108022004907095,
+      "learning_rate": 4.071088310736534e-05,
+      "loss": 0.589,
+      "step": 811
+    },
+    {
+      "epoch": 0.8031651829871415,
+      "grad_norm": 0.4470976570889435,
+      "learning_rate": 4.0692561377794066e-05,
+      "loss": 0.5783,
+      "step": 812
+    },
+    {
+      "epoch": 0.8041543026706232,
+      "grad_norm": 0.41289681603112355,
+      "learning_rate": 4.067423964822279e-05,
+      "loss": 0.5335,
+      "step": 813
+    },
+    {
+      "epoch": 0.8051434223541049,
+      "grad_norm": 0.3897455999305961,
+      "learning_rate": 4.065591791865152e-05,
+      "loss": 0.5395,
+      "step": 814
+    },
+    {
+      "epoch": 0.8061325420375866,
+      "grad_norm": 0.5398949072700544,
+      "learning_rate": 4.063759618908025e-05,
+      "loss": 0.5419,
+      "step": 815
+    },
+    {
+      "epoch": 0.8071216617210683,
+      "grad_norm": 0.40262994039410505,
+      "learning_rate": 4.061927445950898e-05,
+      "loss": 0.5674,
+      "step": 816
+    },
+    {
+      "epoch": 0.80811078140455,
+      "grad_norm": 8.042258970876679,
+      "learning_rate": 4.0600952729937705e-05,
+      "loss": 1.1958,
+      "step": 817
+    },
+    {
+      "epoch": 0.8090999010880316,
+      "grad_norm": 0.7150914848705251,
+      "learning_rate": 4.058263100036644e-05,
+      "loss": 0.4752,
+      "step": 818
+    },
+    {
+      "epoch": 0.8100890207715133,
+      "grad_norm": 0.37450227733321634,
+      "learning_rate": 4.0564309270795165e-05,
+      "loss": 0.519,
+      "step": 819
+    },
+    {
+      "epoch": 0.811078140454995,
+      "grad_norm": 0.7771184561273069,
+      "learning_rate": 4.05459875412239e-05,
+      "loss": 0.5667,
+      "step": 820
+    },
+    {
+      "epoch": 0.8120672601384767,
+      "grad_norm": 0.39320567398584594,
+      "learning_rate": 4.0527665811652624e-05,
+      "loss": 0.5189,
+      "step": 821
+    },
+    {
+      "epoch": 0.8130563798219584,
+      "grad_norm": 0.6164066236372652,
+      "learning_rate": 4.050934408208135e-05,
+      "loss": 0.5956,
+      "step": 822
+    },
+    {
+      "epoch": 0.8140454995054401,
+      "grad_norm": 0.43784506001787876,
+      "learning_rate": 4.049102235251008e-05,
+      "loss": 0.5005,
+      "step": 823
+    },
+    {
+      "epoch": 0.8150346191889218,
+      "grad_norm": 0.650347294511098,
+      "learning_rate": 4.047270062293881e-05,
+      "loss": 0.5911,
+      "step": 824
+    },
+    {
+      "epoch": 0.8160237388724035,
+      "grad_norm": 0.3924537077442831,
+      "learning_rate": 4.045437889336754e-05,
+      "loss": 0.5942,
+      "step": 825
+    },
+    {
+      "epoch": 0.8170128585558852,
+      "grad_norm": 0.5306835116575841,
+      "learning_rate": 4.043605716379626e-05,
+      "loss": 0.591,
+      "step": 826
+    },
+    {
+      "epoch": 0.8180019782393669,
+      "grad_norm": 0.47068526472918815,
+      "learning_rate": 4.041773543422499e-05,
+      "loss": 0.5701,
+      "step": 827
+    },
+    {
+      "epoch": 0.8189910979228486,
+      "grad_norm": 0.5433358297169092,
+      "learning_rate": 4.039941370465372e-05,
+      "loss": 0.5032,
+      "step": 828
+    },
+    {
+      "epoch": 0.8199802176063303,
+      "grad_norm": 0.4117742532527006,
+      "learning_rate": 4.038109197508245e-05,
+      "loss": 0.5051,
+      "step": 829
+    },
+    {
+      "epoch": 0.820969337289812,
+      "grad_norm": 0.45313367798585935,
+      "learning_rate": 4.0362770245511175e-05,
+      "loss": 0.5236,
+      "step": 830
+    },
+    {
+      "epoch": 0.8219584569732937,
+      "grad_norm": 0.4743390307289405,
+      "learning_rate": 4.034444851593991e-05,
+      "loss": 0.5661,
+      "step": 831
+    },
+    {
+      "epoch": 0.8229475766567754,
+      "grad_norm": 0.43419602719141287,
+      "learning_rate": 4.0326126786368635e-05,
+      "loss": 0.5262,
+      "step": 832
+    },
+    {
+      "epoch": 0.8239366963402571,
+      "grad_norm": 0.44595712321645226,
+      "learning_rate": 4.030780505679737e-05,
+      "loss": 0.557,
+      "step": 833
+    },
+    {
+      "epoch": 0.8249258160237388,
+      "grad_norm": 0.46065824846474296,
+      "learning_rate": 4.028948332722609e-05,
+      "loss": 0.5425,
+      "step": 834
+    },
+    {
+      "epoch": 0.8259149357072205,
+      "grad_norm": 0.47395010516783537,
+      "learning_rate": 4.027116159765482e-05,
+      "loss": 0.5432,
+      "step": 835
+    },
+    {
+      "epoch": 0.8269040553907022,
+      "grad_norm": 0.409348387036287,
+      "learning_rate": 4.025283986808355e-05,
+      "loss": 0.5371,
+      "step": 836
+    },
+    {
+      "epoch": 0.827893175074184,
+      "grad_norm": 0.49737572694928917,
+      "learning_rate": 4.023451813851228e-05,
+      "loss": 0.517,
+      "step": 837
+    },
+    {
+      "epoch": 0.8288822947576657,
+      "grad_norm": 0.3964787257419728,
+      "learning_rate": 4.0216196408941e-05,
+      "loss": 0.5921,
+      "step": 838
+    },
+    {
+      "epoch": 0.8298714144411474,
+      "grad_norm": 0.3783571743456649,
+      "learning_rate": 4.0197874679369734e-05,
+      "loss": 0.5817,
+      "step": 839
+    },
+    {
+      "epoch": 0.8308605341246291,
+      "grad_norm": 0.45317557022906846,
+      "learning_rate": 4.017955294979846e-05,
+      "loss": 0.5361,
+      "step": 840
+    },
+    {
+      "epoch": 0.8318496538081108,
+      "grad_norm": 0.3855785042525516,
+      "learning_rate": 4.016123122022719e-05,
+      "loss": 0.5561,
+      "step": 841
+    },
+    {
+      "epoch": 0.8328387734915925,
+      "grad_norm": 0.37735899597002825,
+      "learning_rate": 4.014290949065592e-05,
+      "loss": 0.4952,
+      "step": 842
+    },
+    {
+      "epoch": 0.8338278931750742,
+      "grad_norm": 0.3910395508613529,
+      "learning_rate": 4.0124587761084646e-05,
+      "loss": 0.5248,
+      "step": 843
+    },
+    {
+      "epoch": 0.8348170128585559,
+      "grad_norm": 0.41255181351776066,
+      "learning_rate": 4.010626603151338e-05,
+      "loss": 0.6143,
+      "step": 844
+    },
+    {
+      "epoch": 0.8358061325420376,
+      "grad_norm": 0.3881103870770859,
+      "learning_rate": 4.0087944301942106e-05,
+      "loss": 0.5858,
+      "step": 845
+    },
+    {
+      "epoch": 0.8367952522255193,
+      "grad_norm": 0.4610095412501324,
+      "learning_rate": 4.006962257237083e-05,
+      "loss": 0.4924,
+      "step": 846
+    },
+    {
+      "epoch": 0.837784371909001,
+      "grad_norm": 0.36829299206900973,
+      "learning_rate": 4.005130084279956e-05,
+      "loss": 0.5079,
+      "step": 847
+    },
+    {
+      "epoch": 0.8387734915924827,
+      "grad_norm": 0.40289319760057474,
+      "learning_rate": 4.003297911322829e-05,
+      "loss": 0.5482,
+      "step": 848
+    },
+    {
+      "epoch": 0.8397626112759644,
+      "grad_norm": 0.4206511643989216,
+      "learning_rate": 4.001465738365702e-05,
+      "loss": 0.5605,
+      "step": 849
+    },
+    {
+      "epoch": 0.8407517309594461,
+      "grad_norm": 0.41731945925784314,
+      "learning_rate": 3.999633565408575e-05,
+      "loss": 0.6092,
+      "step": 850
+    },
+    {
+      "epoch": 0.8417408506429278,
+      "grad_norm": 0.30257425739688576,
+      "learning_rate": 3.997801392451447e-05,
+      "loss": 0.579,
+      "step": 851
+    },
+    {
+      "epoch": 0.8427299703264095,
+      "grad_norm": 0.3864935018364275,
+      "learning_rate": 3.9959692194943204e-05,
+      "loss": 0.4771,
+      "step": 852
+    },
+    {
+      "epoch": 0.8437190900098912,
+      "grad_norm": 0.3988101514171298,
+      "learning_rate": 3.994137046537194e-05,
+      "loss": 0.5624,
+      "step": 853
+    },
+    {
+      "epoch": 0.8447082096933729,
+      "grad_norm": 0.39420755605728475,
+      "learning_rate": 3.9923048735800664e-05,
+      "loss": 0.5568,
+      "step": 854
+    },
+    {
+      "epoch": 0.8456973293768546,
+      "grad_norm": 0.4921966493944131,
+      "learning_rate": 3.990472700622939e-05,
+      "loss": 0.5038,
+      "step": 855
+    },
+    {
+      "epoch": 0.8466864490603363,
+      "grad_norm": 0.4365043760382965,
+      "learning_rate": 3.9886405276658117e-05,
+      "loss": 0.5222,
+      "step": 856
+    },
+    {
+      "epoch": 0.847675568743818,
+      "grad_norm": 0.44342345181331283,
+      "learning_rate": 3.986808354708685e-05,
+      "loss": 0.5426,
+      "step": 857
+    },
+    {
+      "epoch": 0.8486646884272997,
+      "grad_norm": 0.36474923885017907,
+      "learning_rate": 3.9849761817515576e-05,
+      "loss": 0.4866,
+      "step": 858
+    },
+    {
+      "epoch": 0.8496538081107814,
+      "grad_norm": 0.4279054783487708,
+      "learning_rate": 3.98314400879443e-05,
+      "loss": 0.4454,
+      "step": 859
+    },
+    {
+      "epoch": 0.8506429277942631,
+      "grad_norm": 0.32658261042232156,
+      "learning_rate": 3.981311835837303e-05,
+      "loss": 0.5478,
+      "step": 860
+    },
+    {
+      "epoch": 0.8516320474777448,
+      "grad_norm": 0.3632391166208969,
+      "learning_rate": 3.979479662880176e-05,
+      "loss": 0.5293,
+      "step": 861
+    },
+    {
+      "epoch": 0.8526211671612265,
+      "grad_norm": 0.361466100899412,
+      "learning_rate": 3.977647489923049e-05,
+      "loss": 0.5775,
+      "step": 862
+    },
+    {
+      "epoch": 0.8536102868447082,
+      "grad_norm": 0.40555093574101875,
+      "learning_rate": 3.9758153169659215e-05,
+      "loss": 0.4856,
+      "step": 863
+    },
+    {
+      "epoch": 0.8545994065281899,
+      "grad_norm": 0.3488107870670207,
+      "learning_rate": 3.973983144008794e-05,
+      "loss": 0.514,
+      "step": 864
+    },
+    {
+      "epoch": 0.8555885262116716,
+      "grad_norm": 0.3843960708975495,
+      "learning_rate": 3.9721509710516675e-05,
+      "loss": 0.5101,
+      "step": 865
+    },
+    {
+      "epoch": 0.8565776458951533,
+      "grad_norm": 0.41358392461083593,
+      "learning_rate": 3.970318798094541e-05,
+      "loss": 0.5516,
+      "step": 866
+    },
+    {
+      "epoch": 0.857566765578635,
+      "grad_norm": 0.32460712314978024,
+      "learning_rate": 3.9684866251374134e-05,
+      "loss": 0.5351,
+      "step": 867
+    },
+    {
+      "epoch": 0.8585558852621167,
+      "grad_norm": 0.4590823844091551,
+      "learning_rate": 3.966654452180286e-05,
+      "loss": 0.6195,
+      "step": 868
+    },
+    {
+      "epoch": 0.8595450049455984,
+      "grad_norm": 0.36553223034798327,
+      "learning_rate": 3.964822279223159e-05,
+      "loss": 0.5288,
+      "step": 869
+    },
+    {
+      "epoch": 0.8605341246290801,
+      "grad_norm": 0.42785798677542214,
+      "learning_rate": 3.962990106266032e-05,
+      "loss": 0.5341,
+      "step": 870
+    },
+    {
+      "epoch": 0.8615232443125618,
+      "grad_norm": 0.34046802166560325,
+      "learning_rate": 3.961157933308905e-05,
+      "loss": 0.4989,
+      "step": 871
+    },
+    {
+      "epoch": 0.8625123639960435,
+      "grad_norm": 0.4203148762636131,
+      "learning_rate": 3.959325760351777e-05,
+      "loss": 0.4626,
+      "step": 872
+    },
+    {
+      "epoch": 0.8635014836795252,
+      "grad_norm": 0.46638243088571246,
+      "learning_rate": 3.95749358739465e-05,
+      "loss": 0.5535,
+      "step": 873
+    },
+    {
+      "epoch": 0.8644906033630069,
+      "grad_norm": 0.3207059127416101,
+      "learning_rate": 3.955661414437523e-05,
+      "loss": 0.4701,
+      "step": 874
+    },
+    {
+      "epoch": 0.8654797230464887,
+      "grad_norm": 0.4361065262778062,
+      "learning_rate": 3.953829241480396e-05,
+      "loss": 0.5232,
+      "step": 875
+    },
+    {
+      "epoch": 0.8664688427299704,
+      "grad_norm": 0.4179015275913999,
+      "learning_rate": 3.9519970685232686e-05,
+      "loss": 0.4782,
+      "step": 876
+    },
+    {
+      "epoch": 0.8674579624134521,
+      "grad_norm": 0.39023161716199395,
+      "learning_rate": 3.950164895566142e-05,
+      "loss": 0.5666,
+      "step": 877
+    },
+    {
+      "epoch": 0.8684470820969338,
+      "grad_norm": 2.1308947045622375,
+      "learning_rate": 3.9483327226090145e-05,
+      "loss": 0.4915,
+      "step": 878
+    },
+    {
+      "epoch": 0.8694362017804155,
+      "grad_norm": 0.6355227582113737,
+      "learning_rate": 3.946500549651888e-05,
+      "loss": 0.5845,
+      "step": 879
+    },
+    {
+      "epoch": 0.8704253214638972,
+      "grad_norm": 0.4231786188850144,
+      "learning_rate": 3.94466837669476e-05,
+      "loss": 0.5186,
+      "step": 880
+    },
+    {
+      "epoch": 0.8714144411473789,
+      "grad_norm": 0.5963762480510161,
+      "learning_rate": 3.942836203737633e-05,
+      "loss": 0.5478,
+      "step": 881
+    },
+    {
+      "epoch": 0.8724035608308606,
+      "grad_norm": 0.44967258871723204,
+      "learning_rate": 3.941004030780506e-05,
+      "loss": 0.6044,
+      "step": 882
+    },
+    {
+      "epoch": 0.8733926805143423,
+      "grad_norm": 0.49666107894686895,
+      "learning_rate": 3.939171857823379e-05,
+      "loss": 0.5572,
+      "step": 883
+    },
+    {
+      "epoch": 0.874381800197824,
+      "grad_norm": 0.4511223362706553,
+      "learning_rate": 3.937339684866251e-05,
+      "loss": 0.5167,
+      "step": 884
+    },
+    {
+      "epoch": 0.8753709198813057,
+      "grad_norm": 0.39070885224778895,
+      "learning_rate": 3.9355075119091244e-05,
+      "loss": 0.5711,
+      "step": 885
+    },
+    {
+      "epoch": 0.8763600395647874,
+      "grad_norm": 0.4817666139801221,
+      "learning_rate": 3.933675338951997e-05,
+      "loss": 0.5614,
+      "step": 886
+    },
+    {
+      "epoch": 0.8773491592482691,
+      "grad_norm": 0.37439555094432153,
+      "learning_rate": 3.93184316599487e-05,
+      "loss": 0.473,
+      "step": 887
+    },
+    {
+      "epoch": 0.8783382789317508,
+      "grad_norm": 0.512933754535988,
+      "learning_rate": 3.930010993037743e-05,
+      "loss": 0.5664,
+      "step": 888
+    },
+    {
+      "epoch": 0.8793273986152325,
+      "grad_norm": 0.3861195811560788,
+      "learning_rate": 3.9281788200806156e-05,
+      "loss": 0.4614,
+      "step": 889
+    },
+    {
+      "epoch": 0.8803165182987142,
+      "grad_norm": 0.5026948257818956,
+      "learning_rate": 3.926346647123489e-05,
+      "loss": 0.5455,
+      "step": 890
+    },
+    {
+      "epoch": 0.8813056379821959,
+      "grad_norm": 0.3797918839860446,
+      "learning_rate": 3.9245144741663616e-05,
+      "loss": 0.5669,
+      "step": 891
+    },
+    {
+      "epoch": 0.8822947576656776,
+      "grad_norm": 0.3580599914264445,
+      "learning_rate": 3.922682301209234e-05,
+      "loss": 0.4969,
+      "step": 892
+    },
+    {
+      "epoch": 0.8832838773491593,
+      "grad_norm": 0.4680809277442312,
+      "learning_rate": 3.920850128252107e-05,
+      "loss": 0.53,
+      "step": 893
+    },
+    {
+      "epoch": 0.884272997032641,
+      "grad_norm": 0.3182331740958757,
+      "learning_rate": 3.91901795529498e-05,
+      "loss": 0.4916,
+      "step": 894
+    },
+    {
+      "epoch": 0.8852621167161226,
+      "grad_norm": 0.38629102830836864,
+      "learning_rate": 3.917185782337853e-05,
+      "loss": 0.4831,
+      "step": 895
+    },
+    {
+      "epoch": 0.8862512363996043,
+      "grad_norm": 0.38003929877472475,
+      "learning_rate": 3.915353609380726e-05,
+      "loss": 0.4544,
+      "step": 896
+    },
+    {
+      "epoch": 0.887240356083086,
+      "grad_norm": 0.35819047353737965,
+      "learning_rate": 3.913521436423598e-05,
+      "loss": 0.6131,
+      "step": 897
+    },
+    {
+      "epoch": 0.8882294757665677,
+      "grad_norm": 0.391106759277783,
+      "learning_rate": 3.9116892634664714e-05,
+      "loss": 0.5327,
+      "step": 898
+    },
+    {
+      "epoch": 0.8892185954500494,
+      "grad_norm": 0.44913570347970116,
+      "learning_rate": 3.909857090509344e-05,
+      "loss": 0.5853,
+      "step": 899
+    },
+    {
+      "epoch": 0.8902077151335311,
+      "grad_norm": 0.33654774158156275,
+      "learning_rate": 3.9080249175522174e-05,
+      "loss": 0.5172,
+      "step": 900
+    },
+    {
+      "epoch": 0.8911968348170128,
+      "grad_norm": 0.4143598750420622,
+      "learning_rate": 3.90619274459509e-05,
+      "loss": 0.4848,
+      "step": 901
+    },
+    {
+      "epoch": 0.8921859545004945,
+      "grad_norm": 0.524991445546096,
+      "learning_rate": 3.904360571637963e-05,
+      "loss": 0.4199,
+      "step": 902
+    },
+    {
+      "epoch": 0.8931750741839762,
+      "grad_norm": 0.3927175649035216,
+      "learning_rate": 3.902528398680836e-05,
+      "loss": 0.5881,
+      "step": 903
+    },
+    {
+      "epoch": 0.8941641938674579,
+      "grad_norm": 0.4079491608662268,
+      "learning_rate": 3.9006962257237086e-05,
+      "loss": 0.5405,
+      "step": 904
+    },
+    {
+      "epoch": 0.8951533135509396,
+      "grad_norm": 0.424142781885458,
+      "learning_rate": 3.898864052766581e-05,
+      "loss": 0.5487,
+      "step": 905
+    },
+    {
+      "epoch": 0.8961424332344213,
+      "grad_norm": 0.41683695083750005,
+      "learning_rate": 3.897031879809454e-05,
+      "loss": 0.4787,
+      "step": 906
+    },
+    {
+      "epoch": 0.897131552917903,
+      "grad_norm": 0.3547610738651701,
+      "learning_rate": 3.895199706852327e-05,
+      "loss": 0.4979,
+      "step": 907
+    },
+    {
+      "epoch": 0.8981206726013847,
+      "grad_norm": 2.3469674956937228,
+      "learning_rate": 3.8933675338952e-05,
+      "loss": 0.5794,
+      "step": 908
+    },
+    {
+      "epoch": 0.8991097922848664,
+      "grad_norm": 0.6412020698124778,
+      "learning_rate": 3.8915353609380725e-05,
+      "loss": 0.5268,
+      "step": 909
+    },
+    {
+      "epoch": 0.9000989119683481,
+      "grad_norm": 1.0383233848694802,
+      "learning_rate": 3.889703187980945e-05,
+      "loss": 0.5414,
+      "step": 910
+    },
+    {
+      "epoch": 0.9010880316518298,
+      "grad_norm": 0.43251475437881326,
+      "learning_rate": 3.8878710150238185e-05,
+      "loss": 0.4797,
+      "step": 911
+    },
+    {
+      "epoch": 0.9020771513353115,
+      "grad_norm": 0.5157606933923597,
+      "learning_rate": 3.886038842066692e-05,
+      "loss": 0.5654,
+      "step": 912
+    },
+    {
+      "epoch": 0.9030662710187932,
+      "grad_norm": 0.3743540399979926,
+      "learning_rate": 3.8842066691095644e-05,
+      "loss": 0.5119,
+      "step": 913
+    },
+    {
+      "epoch": 0.904055390702275,
+      "grad_norm": 0.42098522740414857,
+      "learning_rate": 3.882374496152437e-05,
+      "loss": 0.489,
+      "step": 914
+    },
+    {
+      "epoch": 0.9050445103857567,
+      "grad_norm": 0.5243073127176688,
+      "learning_rate": 3.88054232319531e-05,
+      "loss": 0.5039,
+      "step": 915
+    },
+    {
+      "epoch": 0.9060336300692384,
+      "grad_norm": 0.3986349967328183,
+      "learning_rate": 3.878710150238183e-05,
+      "loss": 0.5337,
+      "step": 916
+    },
+    {
+      "epoch": 0.9070227497527201,
+      "grad_norm": 0.5429828786199771,
+      "learning_rate": 3.876877977281056e-05,
+      "loss": 0.5475,
+      "step": 917
+    },
+    {
+      "epoch": 0.9080118694362018,
+      "grad_norm": 0.44994465926714916,
+      "learning_rate": 3.875045804323928e-05,
+      "loss": 0.5136,
+      "step": 918
+    },
+    {
+      "epoch": 0.9090009891196835,
+      "grad_norm": 0.3415738750821067,
+      "learning_rate": 3.873213631366801e-05,
+      "loss": 0.5346,
+      "step": 919
+    },
+    {
+      "epoch": 0.9099901088031652,
+      "grad_norm": 0.3937349524036801,
+      "learning_rate": 3.871381458409674e-05,
+      "loss": 0.466,
+      "step": 920
+    },
+    {
+      "epoch": 0.9109792284866469,
+      "grad_norm": 0.5131695235804504,
+      "learning_rate": 3.869549285452547e-05,
+      "loss": 0.6701,
+      "step": 921
+    },
+    {
+      "epoch": 0.9119683481701286,
+      "grad_norm": 0.33653367131443224,
+      "learning_rate": 3.8677171124954196e-05,
+      "loss": 0.4473,
+      "step": 922
+    },
+    {
+      "epoch": 0.9129574678536103,
+      "grad_norm": 0.5446871845384862,
+      "learning_rate": 3.865884939538292e-05,
+      "loss": 0.5528,
+      "step": 923
+    },
+    {
+      "epoch": 0.913946587537092,
+      "grad_norm": 0.5015814740270789,
+      "learning_rate": 3.8640527665811655e-05,
+      "loss": 0.5226,
+      "step": 924
+    },
+    {
+      "epoch": 0.9149357072205737,
+      "grad_norm": 0.7983585338010668,
+      "learning_rate": 3.862220593624039e-05,
+      "loss": 0.4371,
+      "step": 925
+    },
+    {
+      "epoch": 0.9159248269040554,
+      "grad_norm": 0.48904182262923174,
+      "learning_rate": 3.860388420666911e-05,
+      "loss": 0.5594,
+      "step": 926
+    },
+    {
+      "epoch": 0.9169139465875371,
+      "grad_norm": 0.4503935788610765,
+      "learning_rate": 3.858556247709784e-05,
+      "loss": 0.6133,
+      "step": 927
+    },
+    {
+      "epoch": 0.9179030662710188,
+      "grad_norm": 0.3165947004747236,
+      "learning_rate": 3.856724074752657e-05,
+      "loss": 0.4772,
+      "step": 928
+    },
+    {
+      "epoch": 0.9188921859545005,
+      "grad_norm": 0.5586077488303228,
+      "learning_rate": 3.85489190179553e-05,
+      "loss": 0.5199,
+      "step": 929
+    },
+    {
+      "epoch": 0.9198813056379822,
+      "grad_norm": 0.5085374651276036,
+      "learning_rate": 3.853059728838402e-05,
+      "loss": 0.5306,
+      "step": 930
+    },
+    {
+      "epoch": 0.9208704253214639,
+      "grad_norm": 0.4129244011177407,
+      "learning_rate": 3.8512275558812754e-05,
+      "loss": 0.4806,
+      "step": 931
+    },
+    {
+      "epoch": 0.9218595450049456,
+      "grad_norm": 0.4861459748388243,
+      "learning_rate": 3.849395382924148e-05,
+      "loss": 0.4908,
+      "step": 932
+    },
+    {
+      "epoch": 0.9228486646884273,
+      "grad_norm": 0.41526131109253767,
+      "learning_rate": 3.847563209967021e-05,
+      "loss": 0.5276,
+      "step": 933
+    },
+    {
+      "epoch": 0.923837784371909,
+      "grad_norm": 0.4009408084973653,
+      "learning_rate": 3.845731037009894e-05,
+      "loss": 0.5562,
+      "step": 934
+    },
+    {
+      "epoch": 0.9248269040553907,
+      "grad_norm": 0.5745910621076383,
+      "learning_rate": 3.8438988640527666e-05,
+      "loss": 0.5774,
+      "step": 935
+    },
+    {
+      "epoch": 0.9258160237388724,
+      "grad_norm": 0.4637760415387979,
+      "learning_rate": 3.84206669109564e-05,
+      "loss": 0.5435,
+      "step": 936
+    },
+    {
+      "epoch": 0.9268051434223541,
+      "grad_norm": 0.41864980359230897,
+      "learning_rate": 3.8402345181385126e-05,
+      "loss": 0.5391,
+      "step": 937
+    },
+    {
+      "epoch": 0.9277942631058358,
+      "grad_norm": 1.1869062780052408,
+      "learning_rate": 3.838402345181385e-05,
+      "loss": 0.5023,
+      "step": 938
+    },
+    {
+      "epoch": 0.9287833827893175,
+      "grad_norm": 0.43776779645633507,
+      "learning_rate": 3.836570172224258e-05,
+      "loss": 0.5096,
+      "step": 939
+    },
+    {
+      "epoch": 0.9297725024727992,
+      "grad_norm": 0.5198208217050955,
+      "learning_rate": 3.834737999267131e-05,
+      "loss": 0.5335,
+      "step": 940
+    },
+    {
+      "epoch": 0.9307616221562809,
+      "grad_norm": 0.5235285409308027,
+      "learning_rate": 3.832905826310004e-05,
+      "loss": 0.5847,
+      "step": 941
+    },
+    {
+      "epoch": 0.9317507418397626,
+      "grad_norm": 0.44126732924600726,
+      "learning_rate": 3.831073653352877e-05,
+      "loss": 0.4719,
+      "step": 942
+    },
+    {
+      "epoch": 0.9327398615232443,
+      "grad_norm": 0.5054573511058064,
+      "learning_rate": 3.829241480395749e-05,
+      "loss": 0.5096,
+      "step": 943
+    },
+    {
+      "epoch": 0.933728981206726,
+      "grad_norm": 0.5805833518404947,
+      "learning_rate": 3.8274093074386224e-05,
+      "loss": 0.5011,
+      "step": 944
+    },
+    {
+      "epoch": 0.9347181008902077,
+      "grad_norm": 2.2314124854517408,
+      "learning_rate": 3.825577134481495e-05,
+      "loss": 0.5527,
+      "step": 945
+    },
+    {
+      "epoch": 0.9357072205736894,
+      "grad_norm": 0.5483697457861862,
+      "learning_rate": 3.8237449615243684e-05,
+      "loss": 0.5321,
+      "step": 946
+    },
+    {
+      "epoch": 0.9366963402571711,
+      "grad_norm": 0.47552103451002065,
+      "learning_rate": 3.8219127885672404e-05,
+      "loss": 0.5372,
+      "step": 947
+    },
+    {
+      "epoch": 0.9376854599406528,
+      "grad_norm": 0.31286512825007917,
+      "learning_rate": 3.820080615610114e-05,
+      "loss": 0.4985,
+      "step": 948
+    },
+    {
+      "epoch": 0.9386745796241345,
+      "grad_norm": 1.7242082421798055,
+      "learning_rate": 3.818248442652987e-05,
+      "loss": 0.5405,
+      "step": 949
+    },
+    {
+      "epoch": 0.9396636993076162,
+      "grad_norm": 0.4700031516218264,
+      "learning_rate": 3.8164162696958596e-05,
+      "loss": 0.4871,
+      "step": 950
+    },
+    {
+      "epoch": 0.9406528189910979,
+      "grad_norm": 0.38084464993511985,
+      "learning_rate": 3.814584096738732e-05,
+      "loss": 0.4736,
+      "step": 951
+    },
+    {
+      "epoch": 0.9416419386745796,
+      "grad_norm": 0.4483615144248822,
+      "learning_rate": 3.812751923781605e-05,
+      "loss": 0.5035,
+      "step": 952
+    },
+    {
+      "epoch": 0.9426310583580614,
+      "grad_norm": 0.3741684709974548,
+      "learning_rate": 3.810919750824478e-05,
+      "loss": 0.5079,
+      "step": 953
+    },
+    {
+      "epoch": 0.9436201780415431,
+      "grad_norm": 0.3159720304753471,
+      "learning_rate": 3.809087577867351e-05,
+      "loss": 0.5318,
+      "step": 954
+    },
+    {
+      "epoch": 0.9446092977250248,
+      "grad_norm": 0.4215113817807427,
+      "learning_rate": 3.8072554049102235e-05,
+      "loss": 0.5136,
+      "step": 955
+    },
+    {
+      "epoch": 0.9455984174085065,
+      "grad_norm": 0.37425060046032715,
+      "learning_rate": 3.805423231953096e-05,
+      "loss": 0.5254,
+      "step": 956
+    },
+    {
+      "epoch": 0.9465875370919882,
+      "grad_norm": 0.3257675406702953,
+      "learning_rate": 3.8035910589959695e-05,
+      "loss": 0.5079,
+      "step": 957
+    },
+    {
+      "epoch": 0.9475766567754699,
+      "grad_norm": 0.45095100461312454,
+      "learning_rate": 3.801758886038842e-05,
+      "loss": 0.512,
+      "step": 958
+    },
+    {
+      "epoch": 0.9485657764589516,
+      "grad_norm": 0.4090749619120724,
+      "learning_rate": 3.7999267130817154e-05,
+      "loss": 0.5396,
+      "step": 959
+    },
+    {
+      "epoch": 0.9495548961424333,
+      "grad_norm": 0.39265990057296607,
+      "learning_rate": 3.798094540124588e-05,
+      "loss": 0.4991,
+      "step": 960
+    },
+    {
+      "epoch": 0.950544015825915,
+      "grad_norm": 0.4063218758435763,
+      "learning_rate": 3.796262367167461e-05,
+      "loss": 0.489,
+      "step": 961
+    },
+    {
+      "epoch": 0.9515331355093967,
+      "grad_norm": 0.4829638146436522,
+      "learning_rate": 3.794430194210334e-05,
+      "loss": 0.5859,
+      "step": 962
+    },
+    {
+      "epoch": 0.9525222551928784,
+      "grad_norm": 0.33683234605221707,
+      "learning_rate": 3.792598021253207e-05,
+      "loss": 0.5402,
+      "step": 963
+    },
+    {
+      "epoch": 0.9535113748763601,
+      "grad_norm": 0.4136710761072618,
+      "learning_rate": 3.790765848296079e-05,
+      "loss": 0.6248,
+      "step": 964
+    },
+    {
+      "epoch": 0.9545004945598418,
+      "grad_norm": 0.44419096696115934,
+      "learning_rate": 3.788933675338952e-05,
+      "loss": 0.5687,
+      "step": 965
+    },
+    {
+      "epoch": 0.9554896142433235,
+      "grad_norm": 0.314815697263912,
+      "learning_rate": 3.787101502381825e-05,
+      "loss": 0.5257,
+      "step": 966
+    },
+    {
+      "epoch": 0.9564787339268052,
+      "grad_norm": 0.3767859758997778,
+      "learning_rate": 3.785269329424698e-05,
+      "loss": 0.4821,
+      "step": 967
+    },
+    {
+      "epoch": 0.9574678536102869,
+      "grad_norm": 0.3695256393459967,
+      "learning_rate": 3.7834371564675706e-05,
+      "loss": 0.5113,
+      "step": 968
+    },
+    {
+      "epoch": 0.9584569732937686,
+      "grad_norm": 0.29897151173130265,
+      "learning_rate": 3.781604983510443e-05,
+      "loss": 0.5127,
+      "step": 969
+    },
+    {
+      "epoch": 0.9594460929772503,
+      "grad_norm": 0.3406098944483567,
+      "learning_rate": 3.7797728105533165e-05,
+      "loss": 0.4787,
+      "step": 970
+    },
+    {
+      "epoch": 0.960435212660732,
+      "grad_norm": 0.3727822915676798,
+      "learning_rate": 3.777940637596189e-05,
+      "loss": 0.5348,
+      "step": 971
+    },
+    {
+      "epoch": 0.9614243323442137,
+      "grad_norm": 0.3140702134023394,
+      "learning_rate": 3.776108464639062e-05,
+      "loss": 0.4602,
+      "step": 972
+    },
+    {
+      "epoch": 0.9624134520276953,
+      "grad_norm": 0.3848508421746314,
+      "learning_rate": 3.774276291681935e-05,
+      "loss": 0.5356,
+      "step": 973
+    },
+    {
+      "epoch": 0.963402571711177,
+      "grad_norm": 0.32214698366653616,
+      "learning_rate": 3.772444118724808e-05,
+      "loss": 0.5184,
+      "step": 974
+    },
+    {
+      "epoch": 0.9643916913946587,
+      "grad_norm": 0.36714549493193244,
+      "learning_rate": 3.770611945767681e-05,
+      "loss": 0.4755,
+      "step": 975
+    },
+    {
+      "epoch": 0.9653808110781404,
+      "grad_norm": 0.34869975067509607,
+      "learning_rate": 3.768779772810553e-05,
+      "loss": 0.6225,
+      "step": 976
+    },
+    {
+      "epoch": 0.9663699307616221,
+      "grad_norm": 0.3820217670798009,
+      "learning_rate": 3.7669475998534264e-05,
+      "loss": 0.4952,
+      "step": 977
+    },
+    {
+      "epoch": 0.9673590504451038,
+      "grad_norm": 0.3985817080513296,
+      "learning_rate": 3.765115426896299e-05,
+      "loss": 0.5567,
+      "step": 978
+    },
+    {
+      "epoch": 0.9683481701285855,
+      "grad_norm": 0.31651201835869763,
+      "learning_rate": 3.7632832539391723e-05,
+      "loss": 0.5074,
+      "step": 979
+    },
+    {
+      "epoch": 0.9693372898120672,
+      "grad_norm": 0.37619456746828517,
+      "learning_rate": 3.761451080982045e-05,
+      "loss": 0.5589,
+      "step": 980
+    },
+    {
+      "epoch": 0.9703264094955489,
+      "grad_norm": 0.3623920056918156,
+      "learning_rate": 3.7596189080249176e-05,
+      "loss": 0.4757,
+      "step": 981
+    },
+    {
+      "epoch": 0.9713155291790306,
+      "grad_norm": 0.3420143952848393,
+      "learning_rate": 3.75778673506779e-05,
+      "loss": 0.4884,
+      "step": 982
+    },
+    {
+      "epoch": 0.9723046488625123,
+      "grad_norm": 0.32751462019753536,
+      "learning_rate": 3.7559545621106636e-05,
+      "loss": 0.4947,
+      "step": 983
+    },
+    {
+      "epoch": 0.973293768545994,
+      "grad_norm": 0.31318281028088973,
+      "learning_rate": 3.754122389153536e-05,
+      "loss": 0.4836,
+      "step": 984
+    },
+    {
+      "epoch": 0.9742828882294757,
+      "grad_norm": 0.38127965484424997,
+      "learning_rate": 3.752290216196409e-05,
+      "loss": 0.5345,
+      "step": 985
+    },
+    {
+      "epoch": 0.9752720079129574,
+      "grad_norm": 0.37819403084012576,
+      "learning_rate": 3.750458043239282e-05,
+      "loss": 0.4964,
+      "step": 986
+    },
+    {
+      "epoch": 0.9762611275964391,
+      "grad_norm": 0.3560452300150932,
+      "learning_rate": 3.748625870282155e-05,
+      "loss": 0.5693,
+      "step": 987
+    },
+    {
+      "epoch": 0.9772502472799208,
+      "grad_norm": 0.40812765024892683,
+      "learning_rate": 3.746793697325028e-05,
+      "loss": 0.5594,
+      "step": 988
+    },
+    {
+      "epoch": 0.9782393669634025,
+      "grad_norm": 0.39374505335822296,
+      "learning_rate": 3.7449615243679e-05,
+      "loss": 0.5583,
+      "step": 989
+    },
+    {
+      "epoch": 0.9792284866468842,
+      "grad_norm": 0.37540173769820445,
+      "learning_rate": 3.7431293514107734e-05,
+      "loss": 0.4881,
+      "step": 990
+    },
+    {
+      "epoch": 0.9802176063303659,
+      "grad_norm": 0.443962903657654,
+      "learning_rate": 3.741297178453646e-05,
+      "loss": 0.5349,
+      "step": 991
+    },
+    {
+      "epoch": 0.9812067260138477,
+      "grad_norm": 0.3858662193153671,
+      "learning_rate": 3.7394650054965194e-05,
+      "loss": 0.5483,
+      "step": 992
+    },
+    {
+      "epoch": 0.9821958456973294,
+      "grad_norm": 0.8166142256059559,
+      "learning_rate": 3.7376328325393914e-05,
+      "loss": 0.591,
+      "step": 993
+    },
+    {
+      "epoch": 0.9831849653808111,
+      "grad_norm": 0.4159757188981098,
+      "learning_rate": 3.735800659582265e-05,
+      "loss": 0.5583,
+      "step": 994
+    },
+    {
+      "epoch": 0.9841740850642928,
+      "grad_norm": 0.8142513421373967,
+      "learning_rate": 3.733968486625137e-05,
+      "loss": 0.5662,
+      "step": 995
+    },
+    {
+      "epoch": 0.9851632047477745,
+      "grad_norm": 0.491701116849365,
+      "learning_rate": 3.7321363136680106e-05,
+      "loss": 0.4751,
+      "step": 996
+    },
+    {
+      "epoch": 0.9861523244312562,
+      "grad_norm": 0.6257416157107429,
+      "learning_rate": 3.730304140710883e-05,
+      "loss": 0.5465,
+      "step": 997
+    },
+    {
+      "epoch": 0.9871414441147379,
+      "grad_norm": 0.3803596226603389,
+      "learning_rate": 3.728471967753756e-05,
+      "loss": 0.5614,
+      "step": 998
+    },
+    {
+      "epoch": 0.9881305637982196,
+      "grad_norm": 0.48848649838986524,
+      "learning_rate": 3.726639794796629e-05,
+      "loss": 0.5025,
+      "step": 999
+    },
+    {
+      "epoch": 0.9891196834817013,
+      "grad_norm": 0.6067005301320401,
+      "learning_rate": 3.724807621839502e-05,
+      "loss": 0.5909,
+      "step": 1000
+    },
+    {
+      "epoch": 0.990108803165183,
+      "grad_norm": 0.30491012951863644,
+      "learning_rate": 3.7229754488823745e-05,
+      "loss": 0.5312,
+      "step": 1001
+    },
+    {
+      "epoch": 0.9910979228486647,
+      "grad_norm": 0.5263181916386561,
+      "learning_rate": 3.721143275925247e-05,
+      "loss": 0.5418,
+      "step": 1002
+    },
+    {
+      "epoch": 0.9920870425321464,
+      "grad_norm": 0.48669107191244604,
+      "learning_rate": 3.7193111029681205e-05,
+      "loss": 0.4754,
+      "step": 1003
+    },
+    {
+      "epoch": 0.9930761622156281,
+      "grad_norm": 0.32683564491319556,
+      "learning_rate": 3.717478930010993e-05,
+      "loss": 0.4993,
+      "step": 1004
+    },
+    {
+      "epoch": 0.9940652818991098,
+      "grad_norm": 0.41176858367306507,
+      "learning_rate": 3.7156467570538664e-05,
+      "loss": 0.5087,
+      "step": 1005
+    },
+    {
+      "epoch": 0.9950544015825915,
+      "grad_norm": 0.47830883072176333,
+      "learning_rate": 3.7138145840967384e-05,
+      "loss": 0.5465,
+      "step": 1006
+    },
+    {
+      "epoch": 0.9960435212660732,
+      "grad_norm": 0.3751711273770695,
+      "learning_rate": 3.711982411139612e-05,
+      "loss": 0.5465,
+      "step": 1007
+    },
+    {
+      "epoch": 0.9970326409495549,
+      "grad_norm": 0.3863008248417529,
+      "learning_rate": 3.710150238182485e-05,
+      "loss": 0.5453,
+      "step": 1008
+    },
+    {
+      "epoch": 0.9980217606330366,
+      "grad_norm": 0.35255489104921495,
+      "learning_rate": 3.708318065225358e-05,
+      "loss": 0.4655,
+      "step": 1009
+    },
+    {
+      "epoch": 0.9990108803165183,
+      "grad_norm": 0.3485913687147211,
+      "learning_rate": 3.70648589226823e-05,
+      "loss": 0.5097,
+      "step": 1010
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.4578352977667433,
+      "learning_rate": 3.704653719311103e-05,
+      "loss": 0.5795,
+      "step": 1011
+    },
+    {
+      "epoch": 1.0009891196834817,
+      "grad_norm": 0.38768311863290106,
+      "learning_rate": 3.702821546353976e-05,
+      "loss": 0.401,
+      "step": 1012
+    },
+    {
+      "epoch": 1.0019782393669634,
+      "grad_norm": 0.3714367929643285,
+      "learning_rate": 3.700989373396849e-05,
+      "loss": 0.43,
+      "step": 1013
+    },
+    {
+      "epoch": 1.002967359050445,
+      "grad_norm": 0.39615149577959957,
+      "learning_rate": 3.6991572004397216e-05,
+      "loss": 0.5277,
+      "step": 1014
+    },
+    {
+      "epoch": 1.0039564787339268,
+      "grad_norm": 0.2867304423631312,
+      "learning_rate": 3.697325027482594e-05,
+      "loss": 0.4223,
+      "step": 1015
+    },
+    {
+      "epoch": 1.0049455984174085,
+      "grad_norm": 0.3693181837334904,
+      "learning_rate": 3.6954928545254675e-05,
+      "loss": 0.4469,
+      "step": 1016
+    },
+    {
+      "epoch": 1.0059347181008902,
+      "grad_norm": 0.34055845537716634,
+      "learning_rate": 3.69366068156834e-05,
+      "loss": 0.4902,
+      "step": 1017
+    },
+    {
+      "epoch": 1.006923837784372,
+      "grad_norm": 0.3429833347762005,
+      "learning_rate": 3.691828508611213e-05,
+      "loss": 0.4396,
+      "step": 1018
+    },
+    {
+      "epoch": 1.0079129574678536,
+      "grad_norm": 0.3693590550132719,
+      "learning_rate": 3.6899963356540855e-05,
+      "loss": 0.425,
+      "step": 1019
+    },
+    {
+      "epoch": 1.0089020771513353,
+      "grad_norm": 3.0499675675684976,
+      "learning_rate": 3.688164162696959e-05,
+      "loss": 0.5673,
+      "step": 1020
+    },
+    {
+      "epoch": 1.009891196834817,
+      "grad_norm": 3.45703439958464,
+      "learning_rate": 3.686331989739832e-05,
+      "loss": 0.4238,
+      "step": 1021
+    },
+    {
+      "epoch": 1.0108803165182987,
+      "grad_norm": 1.1262732708801717,
+      "learning_rate": 3.684499816782704e-05,
+      "loss": 0.4644,
+      "step": 1022
+    },
+    {
+      "epoch": 1.0118694362017804,
+      "grad_norm": 0.35784290295603,
+      "learning_rate": 3.6826676438255774e-05,
+      "loss": 0.489,
+      "step": 1023
+    },
+    {
+      "epoch": 1.012858555885262,
+      "grad_norm": 0.4608616029921624,
+      "learning_rate": 3.68083547086845e-05,
+      "loss": 0.4936,
+      "step": 1024
+    },
+    {
+      "epoch": 1.0138476755687438,
+      "grad_norm": 0.424887693355373,
+      "learning_rate": 3.6790032979113233e-05,
+      "loss": 0.5167,
+      "step": 1025
+    },
+    {
+      "epoch": 1.0148367952522255,
+      "grad_norm": 0.447495609689506,
+      "learning_rate": 3.677171124954196e-05,
+      "loss": 0.4492,
+      "step": 1026
+    },
+    {
+      "epoch": 1.0158259149357072,
+      "grad_norm": 0.36530975855838116,
+      "learning_rate": 3.6753389519970686e-05,
+      "loss": 0.4778,
+      "step": 1027
+    },
+    {
+      "epoch": 1.0168150346191889,
+      "grad_norm": 1.9814451602515812,
+      "learning_rate": 3.673506779039941e-05,
+      "loss": 0.4786,
+      "step": 1028
+    },
+    {
+      "epoch": 1.0178041543026706,
+      "grad_norm": 0.5630016104667595,
+      "learning_rate": 3.6716746060828146e-05,
+      "loss": 0.4327,
+      "step": 1029
+    },
+    {
+      "epoch": 1.0187932739861523,
+      "grad_norm": 0.36111632730607574,
+      "learning_rate": 3.669842433125687e-05,
+      "loss": 0.4235,
+      "step": 1030
+    },
+    {
+      "epoch": 1.019782393669634,
+      "grad_norm": 5.436341015640222,
+      "learning_rate": 3.66801026016856e-05,
+      "loss": 0.5393,
+      "step": 1031
+    },
+    {
+      "epoch": 1.0207715133531157,
+      "grad_norm": 0.7375907758295829,
+      "learning_rate": 3.666178087211433e-05,
+      "loss": 0.4285,
+      "step": 1032
+    },
+    {
+      "epoch": 1.0217606330365974,
+      "grad_norm": 0.3037209911593098,
+      "learning_rate": 3.664345914254306e-05,
+      "loss": 0.4599,
+      "step": 1033
+    },
+    {
+      "epoch": 1.022749752720079,
+      "grad_norm": 0.7465649356485863,
+      "learning_rate": 3.662513741297179e-05,
+      "loss": 0.4379,
+      "step": 1034
+    },
+    {
+      "epoch": 1.0237388724035608,
+      "grad_norm": 0.43790277448859827,
+      "learning_rate": 3.660681568340051e-05,
+      "loss": 0.4044,
+      "step": 1035
+    },
+    {
+      "epoch": 1.0247279920870425,
+      "grad_norm": 0.6032981713354464,
+      "learning_rate": 3.6588493953829244e-05,
+      "loss": 0.4908,
+      "step": 1036
+    },
+    {
+      "epoch": 1.0257171117705242,
+      "grad_norm": 0.5761051820115305,
+      "learning_rate": 3.657017222425797e-05,
+      "loss": 0.4773,
+      "step": 1037
+    },
+    {
+      "epoch": 1.0267062314540059,
+      "grad_norm": 0.7647537299042617,
+      "learning_rate": 3.6551850494686704e-05,
+      "loss": 0.4242,
+      "step": 1038
+    },
+    {
+      "epoch": 1.0276953511374876,
+      "grad_norm": 0.5762122743965227,
+      "learning_rate": 3.6533528765115424e-05,
+      "loss": 0.4405,
+      "step": 1039
+    },
+    {
+      "epoch": 1.0286844708209693,
+      "grad_norm": 0.3009983442497535,
+      "learning_rate": 3.651520703554416e-05,
+      "loss": 0.4678,
+      "step": 1040
+    },
+    {
+      "epoch": 1.029673590504451,
+      "grad_norm": 0.562112523671321,
+      "learning_rate": 3.649688530597288e-05,
+      "loss": 0.4621,
+      "step": 1041
+    },
+    {
+      "epoch": 1.0306627101879327,
+      "grad_norm": 0.37989462590052814,
+      "learning_rate": 3.6478563576401616e-05,
+      "loss": 0.4267,
+      "step": 1042
+    },
+    {
+      "epoch": 1.0316518298714143,
+      "grad_norm": 1.0300574287245914,
+      "learning_rate": 3.646024184683034e-05,
+      "loss": 0.4476,
+      "step": 1043
+    },
+    {
+      "epoch": 1.032640949554896,
+      "grad_norm": 0.556363323634819,
+      "learning_rate": 3.644192011725907e-05,
+      "loss": 0.4468,
+      "step": 1044
+    },
+    {
+      "epoch": 1.0336300692383777,
+      "grad_norm": 0.3386788946695046,
+      "learning_rate": 3.64235983876878e-05,
+      "loss": 0.4364,
+      "step": 1045
+    },
+    {
+      "epoch": 1.0346191889218594,
+      "grad_norm": 0.5192836886115852,
+      "learning_rate": 3.640527665811653e-05,
+      "loss": 0.4294,
+      "step": 1046
+    },
+    {
+      "epoch": 1.0356083086053411,
+      "grad_norm": 0.37484641311334366,
+      "learning_rate": 3.6386954928545255e-05,
+      "loss": 0.4691,
+      "step": 1047
+    },
+    {
+      "epoch": 1.0365974282888228,
+      "grad_norm": 0.43983528468602306,
+      "learning_rate": 3.636863319897398e-05,
+      "loss": 0.4607,
+      "step": 1048
+    },
+    {
+      "epoch": 1.0375865479723045,
+      "grad_norm": 1.031961425019673,
+      "learning_rate": 3.6350311469402715e-05,
+      "loss": 0.5229,
+      "step": 1049
+    },
+    {
+      "epoch": 1.0385756676557865,
+      "grad_norm": 3.2210405891971194,
+      "learning_rate": 3.633198973983144e-05,
+      "loss": 0.5568,
+      "step": 1050
+    },
+    {
+      "epoch": 1.039564787339268,
+      "grad_norm": 0.6782949578972237,
+      "learning_rate": 3.6313668010260175e-05,
+      "loss": 0.4354,
+      "step": 1051
+    },
+    {
+      "epoch": 1.0405539070227499,
+      "grad_norm": 0.4127185908000618,
+      "learning_rate": 3.6295346280688894e-05,
+      "loss": 0.4387,
+      "step": 1052
+    },
+    {
+      "epoch": 1.0415430267062316,
+      "grad_norm": 0.490074753069455,
+      "learning_rate": 3.627702455111763e-05,
+      "loss": 0.4307,
+      "step": 1053
+    },
+    {
+      "epoch": 1.0425321463897133,
+      "grad_norm": 0.6921248034479954,
+      "learning_rate": 3.6258702821546354e-05,
+      "loss": 0.4723,
+      "step": 1054
+    },
+    {
+      "epoch": 1.043521266073195,
+      "grad_norm": 0.393153265610656,
+      "learning_rate": 3.624038109197509e-05,
+      "loss": 0.4716,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0445103857566767,
+      "grad_norm": 0.5649667904822261,
+      "learning_rate": 3.6222059362403813e-05,
+      "loss": 0.4793,
+      "step": 1056
+    },
+    {
+      "epoch": 1.0454995054401583,
+      "grad_norm": 1.8299145260790468,
+      "learning_rate": 3.620373763283254e-05,
+      "loss": 0.4621,
+      "step": 1057
+    },
+    {
+      "epoch": 1.04648862512364,
+      "grad_norm": 1.725090631699042,
+      "learning_rate": 3.618541590326127e-05,
+      "loss": 0.4317,
+      "step": 1058
+    },
+    {
+      "epoch": 1.0474777448071217,
+      "grad_norm": 0.537785509003733,
+      "learning_rate": 3.616709417369e-05,
+      "loss": 0.4513,
+      "step": 1059
+    },
+    {
+      "epoch": 1.0484668644906034,
+      "grad_norm": 0.33792548688856516,
+      "learning_rate": 3.6148772444118726e-05,
+      "loss": 0.4657,
+      "step": 1060
+    },
+    {
+      "epoch": 1.0494559841740851,
+      "grad_norm": 0.8248367654993363,
+      "learning_rate": 3.613045071454745e-05,
+      "loss": 0.4434,
+      "step": 1061
+    },
+    {
+      "epoch": 1.0504451038575668,
+      "grad_norm": 0.4266254409159955,
+      "learning_rate": 3.6112128984976185e-05,
+      "loss": 0.4681,
+      "step": 1062
+    },
+    {
+      "epoch": 1.0514342235410485,
+      "grad_norm": 0.42576334310945435,
+      "learning_rate": 3.609380725540491e-05,
+      "loss": 0.4561,
+      "step": 1063
+    },
+    {
+      "epoch": 1.0524233432245302,
+      "grad_norm": 0.5266659764246051,
+      "learning_rate": 3.607548552583364e-05,
+      "loss": 0.4802,
+      "step": 1064
+    },
+    {
+      "epoch": 1.053412462908012,
+      "grad_norm": 0.810147632482765,
+      "learning_rate": 3.6057163796262365e-05,
+      "loss": 0.4591,
+      "step": 1065
+    },
+    {
+      "epoch": 1.0544015825914936,
+      "grad_norm": 0.413804915322679,
+      "learning_rate": 3.60388420666911e-05,
+      "loss": 0.3825,
+      "step": 1066
+    },
+    {
+      "epoch": 1.0553907022749753,
+      "grad_norm": 0.44504381838193996,
+      "learning_rate": 3.602052033711983e-05,
+      "loss": 0.4876,
+      "step": 1067
+    },
+    {
+      "epoch": 1.056379821958457,
+      "grad_norm": 0.4439897157440488,
+      "learning_rate": 3.600219860754855e-05,
+      "loss": 0.4253,
+      "step": 1068
+    },
+    {
+      "epoch": 1.0573689416419387,
+      "grad_norm": 0.3500704299271919,
+      "learning_rate": 3.5983876877977284e-05,
+      "loss": 0.4336,
+      "step": 1069
+    },
+    {
+      "epoch": 1.0583580613254204,
+      "grad_norm": 0.33953086838015223,
+      "learning_rate": 3.596555514840601e-05,
+      "loss": 0.4255,
+      "step": 1070
+    },
+    {
+      "epoch": 1.0593471810089021,
+      "grad_norm": 0.48346975571239686,
+      "learning_rate": 3.5947233418834744e-05,
+      "loss": 0.4609,
+      "step": 1071
+    },
+    {
+      "epoch": 1.0603363006923838,
+      "grad_norm": 0.3187757649005785,
+      "learning_rate": 3.592891168926347e-05,
+      "loss": 0.4665,
+      "step": 1072
+    },
+    {
+      "epoch": 1.0613254203758655,
+      "grad_norm": 0.42833406748043845,
+      "learning_rate": 3.5910589959692196e-05,
+      "loss": 0.3678,
+      "step": 1073
+    },
+    {
+      "epoch": 1.0623145400593472,
+      "grad_norm": 0.4566593411257765,
+      "learning_rate": 3.589226823012092e-05,
+      "loss": 0.4284,
+      "step": 1074
+    },
+    {
+      "epoch": 1.063303659742829,
+      "grad_norm": 0.3468485072829719,
+      "learning_rate": 3.5873946500549656e-05,
+      "loss": 0.4746,
+      "step": 1075
+    },
+    {
+      "epoch": 1.0642927794263106,
+      "grad_norm": 0.510162571679684,
+      "learning_rate": 3.585562477097838e-05,
+      "loss": 0.4224,
+      "step": 1076
+    },
+    {
+      "epoch": 1.0652818991097923,
+      "grad_norm": 0.3979585894735656,
+      "learning_rate": 3.583730304140711e-05,
+      "loss": 0.4646,
+      "step": 1077
+    },
+    {
+      "epoch": 1.066271018793274,
+      "grad_norm": 0.36760884964977675,
+      "learning_rate": 3.5818981311835835e-05,
+      "loss": 0.4491,
+      "step": 1078
+    },
+    {
+      "epoch": 1.0672601384767557,
+      "grad_norm": 0.5145237916838321,
+      "learning_rate": 3.580065958226457e-05,
+      "loss": 0.49,
+      "step": 1079
+    },
+    {
+      "epoch": 1.0682492581602374,
+      "grad_norm": 0.3129933723610313,
+      "learning_rate": 3.57823378526933e-05,
+      "loss": 0.443,
+      "step": 1080
+    },
+    {
+      "epoch": 1.0692383778437191,
+      "grad_norm": 0.44086395340998813,
+      "learning_rate": 3.576401612312202e-05,
+      "loss": 0.4624,
+      "step": 1081
+    },
+    {
+      "epoch": 1.0702274975272008,
+      "grad_norm": 0.3521648112115606,
+      "learning_rate": 3.5745694393550754e-05,
+      "loss": 0.492,
+      "step": 1082
+    },
+    {
+      "epoch": 1.0712166172106825,
+      "grad_norm": 0.34957384405320735,
+      "learning_rate": 3.572737266397948e-05,
+      "loss": 0.4699,
+      "step": 1083
+    },
+    {
+      "epoch": 1.0722057368941642,
+      "grad_norm": 0.4051017779258357,
+      "learning_rate": 3.5709050934408214e-05,
+      "loss": 0.4751,
+      "step": 1084
+    },
+    {
+      "epoch": 1.073194856577646,
+      "grad_norm": 0.3301743394383747,
+      "learning_rate": 3.5690729204836934e-05,
+      "loss": 0.4765,
+      "step": 1085
+    },
+    {
+      "epoch": 1.0741839762611276,
+      "grad_norm": 0.37660803140457133,
+      "learning_rate": 3.567240747526567e-05,
+      "loss": 0.4697,
+      "step": 1086
+    },
+    {
+      "epoch": 1.0751730959446093,
+      "grad_norm": 0.33541330344202874,
+      "learning_rate": 3.565408574569439e-05,
+      "loss": 0.3767,
+      "step": 1087
+    },
+    {
+      "epoch": 1.076162215628091,
+      "grad_norm": 0.31431860272850104,
+      "learning_rate": 3.5635764016123127e-05,
+      "loss": 0.4464,
+      "step": 1088
+    },
+    {
+      "epoch": 1.0771513353115727,
+      "grad_norm": 0.3822236919960425,
+      "learning_rate": 3.561744228655185e-05,
+      "loss": 0.417,
+      "step": 1089
+    },
+    {
+      "epoch": 1.0781404549950544,
+      "grad_norm": 0.3436590951082615,
+      "learning_rate": 3.559912055698058e-05,
+      "loss": 0.4275,
+      "step": 1090
+    },
+    {
+      "epoch": 1.079129574678536,
+      "grad_norm": 0.3085576056896924,
+      "learning_rate": 3.558079882740931e-05,
+      "loss": 0.3597,
+      "step": 1091
+    },
+    {
+      "epoch": 1.0801186943620178,
+      "grad_norm": 0.27756477292049936,
+      "learning_rate": 3.556247709783804e-05,
+      "loss": 0.3784,
+      "step": 1092
+    },
+    {
+      "epoch": 1.0811078140454995,
+      "grad_norm": 0.41722858436924626,
+      "learning_rate": 3.5544155368266765e-05,
+      "loss": 0.4965,
+      "step": 1093
+    },
+    {
+      "epoch": 1.0820969337289812,
+      "grad_norm": 0.3379090288767247,
+      "learning_rate": 3.552583363869549e-05,
+      "loss": 0.4804,
+      "step": 1094
+    },
+    {
+      "epoch": 1.083086053412463,
+      "grad_norm": 0.352934575359697,
+      "learning_rate": 3.5507511909124225e-05,
+      "loss": 0.388,
+      "step": 1095
+    },
+    {
+      "epoch": 1.0840751730959446,
+      "grad_norm": 2.8132481274950853,
+      "learning_rate": 3.548919017955295e-05,
+      "loss": 0.5017,
+      "step": 1096
+    },
+    {
+      "epoch": 1.0850642927794263,
+      "grad_norm": 0.45069457909979543,
+      "learning_rate": 3.547086844998168e-05,
+      "loss": 0.4375,
+      "step": 1097
+    },
+    {
+      "epoch": 1.086053412462908,
+      "grad_norm": 0.3169100654997534,
+      "learning_rate": 3.5452546720410404e-05,
+      "loss": 0.4386,
+      "step": 1098
+    },
+    {
+      "epoch": 1.0870425321463897,
+      "grad_norm": 3.015261219455845,
+      "learning_rate": 3.543422499083914e-05,
+      "loss": 0.6241,
+      "step": 1099
+    },
+    {
+      "epoch": 1.0880316518298714,
+      "grad_norm": 0.7258230030251859,
+      "learning_rate": 3.5415903261267864e-05,
+      "loss": 0.4545,
+      "step": 1100
+    },
+    {
+      "epoch": 1.089020771513353,
+      "grad_norm": 0.3799505369853889,
+      "learning_rate": 3.53975815316966e-05,
+      "loss": 0.4984,
+      "step": 1101
+    },
+    {
+      "epoch": 1.0900098911968348,
+      "grad_norm": 0.4198479739903225,
+      "learning_rate": 3.537925980212532e-05,
+      "loss": 0.4321,
+      "step": 1102
+    },
+    {
+      "epoch": 1.0909990108803165,
+      "grad_norm": 0.44474996311910875,
+      "learning_rate": 3.536093807255405e-05,
+      "loss": 0.385,
+      "step": 1103
+    },
+    {
+      "epoch": 1.0919881305637982,
+      "grad_norm": 0.34423848805072893,
+      "learning_rate": 3.534261634298278e-05,
+      "loss": 0.3839,
+      "step": 1104
+    },
+    {
+      "epoch": 1.0929772502472799,
+      "grad_norm": 0.4385335696893382,
+      "learning_rate": 3.532429461341151e-05,
+      "loss": 0.5029,
+      "step": 1105
+    },
+    {
+      "epoch": 1.0939663699307616,
+      "grad_norm": 0.44073911702739826,
+      "learning_rate": 3.5305972883840236e-05,
+      "loss": 0.4517,
+      "step": 1106
+    },
+    {
+      "epoch": 1.0949554896142433,
+      "grad_norm": 0.3661390388337049,
+      "learning_rate": 3.528765115426896e-05,
+      "loss": 0.4409,
+      "step": 1107
+    },
+    {
+      "epoch": 1.095944609297725,
+      "grad_norm": 0.3778751285178957,
+      "learning_rate": 3.5269329424697696e-05,
+      "loss": 0.4174,
+      "step": 1108
+    },
+    {
+      "epoch": 1.0969337289812067,
+      "grad_norm": 0.3484208155588471,
+      "learning_rate": 3.525100769512642e-05,
+      "loss": 0.4595,
+      "step": 1109
+    },
+    {
+      "epoch": 1.0979228486646884,
+      "grad_norm": 0.3721575495834725,
+      "learning_rate": 3.523268596555515e-05,
+      "loss": 0.4108,
+      "step": 1110
+    },
+    {
+      "epoch": 1.09891196834817,
+      "grad_norm": 0.3260714299366415,
+      "learning_rate": 3.5214364235983875e-05,
+      "loss": 0.4622,
+      "step": 1111
+    },
+    {
+      "epoch": 1.0999010880316518,
+      "grad_norm": 0.3354899051869184,
+      "learning_rate": 3.519604250641261e-05,
+      "loss": 0.4113,
+      "step": 1112
+    },
+    {
+      "epoch": 1.1008902077151335,
+      "grad_norm": 0.2888023111039527,
+      "learning_rate": 3.5177720776841334e-05,
+      "loss": 0.3783,
+      "step": 1113
+    },
+    {
+      "epoch": 1.1018793273986152,
+      "grad_norm": 0.3379822216843918,
+      "learning_rate": 3.515939904727006e-05,
+      "loss": 0.5117,
+      "step": 1114
+    },
+    {
+      "epoch": 1.1028684470820969,
+      "grad_norm": 0.3591859612245374,
+      "learning_rate": 3.5141077317698794e-05,
+      "loss": 0.487,
+      "step": 1115
+    },
+    {
+      "epoch": 1.1038575667655786,
+      "grad_norm": 0.30210957464850935,
+      "learning_rate": 3.512275558812752e-05,
+      "loss": 0.4574,
+      "step": 1116
+    },
+    {
+      "epoch": 1.1048466864490603,
+      "grad_norm": 0.3056806916327237,
+      "learning_rate": 3.5104433858556254e-05,
+      "loss": 0.4382,
+      "step": 1117
+    },
+    {
+      "epoch": 1.105835806132542,
+      "grad_norm": 0.3406702445819911,
+      "learning_rate": 3.508611212898498e-05,
+      "loss": 0.4476,
+      "step": 1118
+    },
+    {
+      "epoch": 1.1068249258160237,
+      "grad_norm": 0.31733420049096794,
+      "learning_rate": 3.5067790399413706e-05,
+      "loss": 0.454,
+      "step": 1119
+    },
+    {
+      "epoch": 1.1078140454995054,
+      "grad_norm": 0.3577310727863552,
+      "learning_rate": 3.504946866984243e-05,
+      "loss": 0.486,
+      "step": 1120
+    },
+    {
+      "epoch": 1.108803165182987,
+      "grad_norm": 0.39669890919733924,
+      "learning_rate": 3.5031146940271166e-05,
+      "loss": 0.5032,
+      "step": 1121
+    },
+    {
+      "epoch": 1.1097922848664687,
+      "grad_norm": 0.36109756118213504,
+      "learning_rate": 3.501282521069989e-05,
+      "loss": 0.4855,
+      "step": 1122
+    },
+    {
+      "epoch": 1.1107814045499504,
+      "grad_norm": 0.41157327887209877,
+      "learning_rate": 3.499450348112862e-05,
+      "loss": 0.492,
+      "step": 1123
+    },
+    {
+      "epoch": 1.1117705242334321,
+      "grad_norm": 0.8883017039989023,
+      "learning_rate": 3.4976181751557345e-05,
+      "loss": 0.4755,
+      "step": 1124
+    },
+    {
+      "epoch": 1.1127596439169138,
+      "grad_norm": 0.4045680018041389,
+      "learning_rate": 3.495786002198608e-05,
+      "loss": 0.5405,
+      "step": 1125
+    },
+    {
+      "epoch": 1.1137487636003955,
+      "grad_norm": 0.46285838105173954,
+      "learning_rate": 3.4939538292414805e-05,
+      "loss": 0.4135,
+      "step": 1126
+    },
+    {
+      "epoch": 1.1147378832838775,
+      "grad_norm": 0.4009969222725594,
+      "learning_rate": 3.492121656284353e-05,
+      "loss": 0.4772,
+      "step": 1127
+    },
+    {
+      "epoch": 1.115727002967359,
+      "grad_norm": 0.3744825716220794,
+      "learning_rate": 3.4902894833272265e-05,
+      "loss": 0.4132,
+      "step": 1128
+    },
+    {
+      "epoch": 1.1167161226508409,
+      "grad_norm": 0.4811370559022165,
+      "learning_rate": 3.488457310370099e-05,
+      "loss": 0.4486,
+      "step": 1129
+    },
+    {
+      "epoch": 1.1177052423343223,
+      "grad_norm": 0.32402321215164476,
+      "learning_rate": 3.4866251374129724e-05,
+      "loss": 0.4404,
+      "step": 1130
+    },
+    {
+      "epoch": 1.1186943620178043,
+      "grad_norm": 0.35736672487594145,
+      "learning_rate": 3.4847929644558444e-05,
+      "loss": 0.4517,
+      "step": 1131
+    },
+    {
+      "epoch": 1.119683481701286,
+      "grad_norm": 0.43284952844544416,
+      "learning_rate": 3.482960791498718e-05,
+      "loss": 0.432,
+      "step": 1132
+    },
+    {
+      "epoch": 1.1206726013847677,
+      "grad_norm": 0.33367942093658537,
+      "learning_rate": 3.4811286185415903e-05,
+      "loss": 0.4439,
+      "step": 1133
+    },
+    {
+      "epoch": 1.1216617210682494,
+      "grad_norm": 0.34533013092078974,
+      "learning_rate": 3.479296445584464e-05,
+      "loss": 0.4537,
+      "step": 1134
+    },
+    {
+      "epoch": 1.122650840751731,
+      "grad_norm": 0.4103596180059782,
+      "learning_rate": 3.477464272627336e-05,
+      "loss": 0.4401,
+      "step": 1135
+    },
+    {
+      "epoch": 1.1236399604352127,
+      "grad_norm": 0.3508858030722054,
+      "learning_rate": 3.475632099670209e-05,
+      "loss": 0.4513,
+      "step": 1136
+    },
+    {
+      "epoch": 1.1246290801186944,
+      "grad_norm": 0.2644357297408127,
+      "learning_rate": 3.4737999267130816e-05,
+      "loss": 0.4285,
+      "step": 1137
+    },
+    {
+      "epoch": 1.1256181998021761,
+      "grad_norm": 0.8993722168956318,
+      "learning_rate": 3.471967753755955e-05,
+      "loss": 0.5024,
+      "step": 1138
+    },
+    {
+      "epoch": 1.1266073194856578,
+      "grad_norm": 0.40831707081279583,
+      "learning_rate": 3.4701355807988275e-05,
+      "loss": 0.455,
+      "step": 1139
+    },
+    {
+      "epoch": 1.1275964391691395,
+      "grad_norm": 0.3182793210526495,
+      "learning_rate": 3.4683034078417e-05,
+      "loss": 0.4162,
+      "step": 1140
+    },
+    {
+      "epoch": 1.1285855588526212,
+      "grad_norm": 0.3821840881244759,
+      "learning_rate": 3.4664712348845735e-05,
+      "loss": 0.4608,
+      "step": 1141
+    },
+    {
+      "epoch": 1.129574678536103,
+      "grad_norm": 0.4427568558577197,
+      "learning_rate": 3.464639061927446e-05,
+      "loss": 0.5133,
+      "step": 1142
+    },
+    {
+      "epoch": 1.1305637982195846,
+      "grad_norm": 0.9566817035370797,
+      "learning_rate": 3.462806888970319e-05,
+      "loss": 0.4638,
+      "step": 1143
+    },
+    {
+      "epoch": 1.1315529179030663,
+      "grad_norm": 0.6038194624696839,
+      "learning_rate": 3.4609747160131914e-05,
+      "loss": 0.3681,
+      "step": 1144
+    },
+    {
+      "epoch": 1.132542037586548,
+      "grad_norm": 0.6807613022411125,
+      "learning_rate": 3.459142543056065e-05,
+      "loss": 0.4679,
+      "step": 1145
+    },
+    {
+      "epoch": 1.1335311572700297,
+      "grad_norm": 0.4115234393020162,
+      "learning_rate": 3.4573103700989374e-05,
+      "loss": 0.4422,
+      "step": 1146
+    },
+    {
+      "epoch": 1.1345202769535114,
+      "grad_norm": 0.4341844759408288,
+      "learning_rate": 3.455478197141811e-05,
+      "loss": 0.4423,
+      "step": 1147
+    },
+    {
+      "epoch": 1.1355093966369931,
+      "grad_norm": 0.5051556302844019,
+      "learning_rate": 3.453646024184683e-05,
+      "loss": 0.4541,
+      "step": 1148
+    },
+    {
+      "epoch": 1.1364985163204748,
+      "grad_norm": 0.3758242355747177,
+      "learning_rate": 3.451813851227556e-05,
+      "loss": 0.4276,
+      "step": 1149
+    },
+    {
+      "epoch": 1.1374876360039565,
+      "grad_norm": 0.3320259649587403,
+      "learning_rate": 3.4499816782704286e-05,
+      "loss": 0.4316,
+      "step": 1150
+    },
+    {
+      "epoch": 1.1384767556874382,
+      "grad_norm": 0.45695875442585837,
+      "learning_rate": 3.448149505313302e-05,
+      "loss": 0.4378,
+      "step": 1151
+    },
+    {
+      "epoch": 1.13946587537092,
+      "grad_norm": 0.43478750136447,
+      "learning_rate": 3.4463173323561746e-05,
+      "loss": 0.5218,
+      "step": 1152
+    },
+    {
+      "epoch": 1.1404549950544016,
+      "grad_norm": 0.8782929825237592,
+      "learning_rate": 3.444485159399047e-05,
+      "loss": 0.4065,
+      "step": 1153
+    },
+    {
+      "epoch": 1.1414441147378833,
+      "grad_norm": 0.48626073683964544,
+      "learning_rate": 3.4426529864419206e-05,
+      "loss": 0.4389,
+      "step": 1154
+    },
+    {
+      "epoch": 1.142433234421365,
+      "grad_norm": 0.49062321657072966,
+      "learning_rate": 3.440820813484793e-05,
+      "loss": 0.4402,
+      "step": 1155
+    },
+    {
+      "epoch": 1.1434223541048467,
+      "grad_norm": 3.9259528695103088,
+      "learning_rate": 3.438988640527666e-05,
+      "loss": 0.5033,
+      "step": 1156
+    },
+    {
+      "epoch": 1.1444114737883284,
+      "grad_norm": 0.4908247634768901,
+      "learning_rate": 3.4371564675705385e-05,
+      "loss": 0.4815,
+      "step": 1157
+    },
+    {
+      "epoch": 1.1454005934718101,
+      "grad_norm": 0.44304978825609653,
+      "learning_rate": 3.435324294613412e-05,
+      "loss": 0.3978,
+      "step": 1158
+    },
+    {
+      "epoch": 1.1463897131552918,
+      "grad_norm": 0.32374507014467085,
+      "learning_rate": 3.4334921216562844e-05,
+      "loss": 0.4137,
+      "step": 1159
+    },
+    {
+      "epoch": 1.1473788328387735,
+      "grad_norm": 0.42036545626914656,
+      "learning_rate": 3.431659948699157e-05,
+      "loss": 0.4604,
+      "step": 1160
+    },
+    {
+      "epoch": 1.1483679525222552,
+      "grad_norm": 0.4166874208449986,
+      "learning_rate": 3.42982777574203e-05,
+      "loss": 0.4776,
+      "step": 1161
+    },
+    {
+      "epoch": 1.149357072205737,
+      "grad_norm": 0.31668752094571023,
+      "learning_rate": 3.427995602784903e-05,
+      "loss": 0.4456,
+      "step": 1162
+    },
+    {
+      "epoch": 1.1503461918892186,
+      "grad_norm": 0.43054522427624803,
+      "learning_rate": 3.4261634298277764e-05,
+      "loss": 0.443,
+      "step": 1163
+    },
+    {
+      "epoch": 1.1513353115727003,
+      "grad_norm": 0.42615533383169335,
+      "learning_rate": 3.424331256870649e-05,
+      "loss": 0.4032,
+      "step": 1164
+    },
+    {
+      "epoch": 1.152324431256182,
+      "grad_norm": 0.3387153691463948,
+      "learning_rate": 3.4224990839135217e-05,
+      "loss": 0.4377,
+      "step": 1165
+    },
+    {
+      "epoch": 1.1533135509396637,
+      "grad_norm": 0.36898697794603996,
+      "learning_rate": 3.420666910956394e-05,
+      "loss": 0.3753,
+      "step": 1166
+    },
+    {
+      "epoch": 1.1543026706231454,
+      "grad_norm": 0.3421939288510364,
+      "learning_rate": 3.4188347379992676e-05,
+      "loss": 0.4382,
+      "step": 1167
+    },
+    {
+      "epoch": 1.155291790306627,
+      "grad_norm": 0.31926334513776805,
+      "learning_rate": 3.41700256504214e-05,
+      "loss": 0.4165,
+      "step": 1168
+    },
+    {
+      "epoch": 1.1562809099901088,
+      "grad_norm": 0.40484584927634126,
+      "learning_rate": 3.415170392085013e-05,
+      "loss": 0.4012,
+      "step": 1169
+    },
+    {
+      "epoch": 1.1572700296735905,
+      "grad_norm": 0.38104083917952464,
+      "learning_rate": 3.4133382191278855e-05,
+      "loss": 0.4316,
+      "step": 1170
+    },
+    {
+      "epoch": 1.1582591493570722,
+      "grad_norm": 0.3709082882174054,
+      "learning_rate": 3.411506046170759e-05,
+      "loss": 0.4243,
+      "step": 1171
+    },
+    {
+      "epoch": 1.159248269040554,
+      "grad_norm": 0.38180703630514035,
+      "learning_rate": 3.4096738732136315e-05,
+      "loss": 0.4204,
+      "step": 1172
+    },
+    {
+      "epoch": 1.1602373887240356,
+      "grad_norm": 0.35066602597324176,
+      "learning_rate": 3.407841700256504e-05,
+      "loss": 0.4517,
+      "step": 1173
+    },
+    {
+      "epoch": 1.1612265084075173,
+      "grad_norm": 0.3409612624272448,
+      "learning_rate": 3.4060095272993775e-05,
+      "loss": 0.4681,
+      "step": 1174
+    },
+    {
+      "epoch": 1.162215628090999,
+      "grad_norm": 0.3242800885803111,
+      "learning_rate": 3.40417735434225e-05,
+      "loss": 0.5142,
+      "step": 1175
+    },
+    {
+      "epoch": 1.1632047477744807,
+      "grad_norm": 0.3999158161657895,
+      "learning_rate": 3.4023451813851234e-05,
+      "loss": 0.4721,
+      "step": 1176
+    },
+    {
+      "epoch": 1.1641938674579624,
+      "grad_norm": 0.3057813262892474,
+      "learning_rate": 3.4005130084279954e-05,
+      "loss": 0.4081,
+      "step": 1177
+    },
+    {
+      "epoch": 1.165182987141444,
+      "grad_norm": 0.29265432398856944,
+      "learning_rate": 3.398680835470869e-05,
+      "loss": 0.4741,
+      "step": 1178
+    },
+    {
+      "epoch": 1.1661721068249258,
+      "grad_norm": 0.35031812706583215,
+      "learning_rate": 3.3968486625137414e-05,
+      "loss": 0.4069,
+      "step": 1179
+    },
+    {
+      "epoch": 1.1671612265084075,
+      "grad_norm": 0.3199265954315745,
+      "learning_rate": 3.395016489556615e-05,
+      "loss": 0.4153,
+      "step": 1180
+    },
+    {
+      "epoch": 1.1681503461918892,
+      "grad_norm": 0.3322579671121011,
+      "learning_rate": 3.393184316599487e-05,
+      "loss": 0.4966,
+      "step": 1181
+    },
+    {
+      "epoch": 1.1691394658753709,
+      "grad_norm": 0.31674223480365804,
+      "learning_rate": 3.39135214364236e-05,
+      "loss": 0.4019,
+      "step": 1182
+    },
+    {
+      "epoch": 1.1701285855588526,
+      "grad_norm": 0.3452256992238975,
+      "learning_rate": 3.3895199706852326e-05,
+      "loss": 0.508,
+      "step": 1183
+    },
+    {
+      "epoch": 1.1711177052423343,
+      "grad_norm": 0.3126306353162765,
+      "learning_rate": 3.387687797728106e-05,
+      "loss": 0.4411,
+      "step": 1184
+    },
+    {
+      "epoch": 1.172106824925816,
+      "grad_norm": 0.3504430898459003,
+      "learning_rate": 3.3858556247709786e-05,
+      "loss": 0.4409,
+      "step": 1185
+    },
+    {
+      "epoch": 1.1730959446092977,
+      "grad_norm": 0.32158432591053415,
+      "learning_rate": 3.384023451813851e-05,
+      "loss": 0.4216,
+      "step": 1186
+    },
+    {
+      "epoch": 1.1740850642927794,
+      "grad_norm": 0.33575292344710606,
+      "learning_rate": 3.3821912788567245e-05,
+      "loss": 0.4653,
+      "step": 1187
+    },
+    {
+      "epoch": 1.175074183976261,
+      "grad_norm": 0.2953528980384244,
+      "learning_rate": 3.380359105899597e-05,
+      "loss": 0.49,
+      "step": 1188
+    },
+    {
+      "epoch": 1.1760633036597428,
+      "grad_norm": 0.44904372451288704,
+      "learning_rate": 3.37852693294247e-05,
+      "loss": 0.49,
+      "step": 1189
+    },
+    {
+      "epoch": 1.1770524233432245,
+      "grad_norm": 0.3985352253667881,
+      "learning_rate": 3.3766947599853424e-05,
+      "loss": 0.4351,
+      "step": 1190
+    },
+    {
+      "epoch": 1.1780415430267062,
+      "grad_norm": 0.36529013646478736,
+      "learning_rate": 3.374862587028216e-05,
+      "loss": 0.3994,
+      "step": 1191
+    },
+    {
+      "epoch": 1.1790306627101879,
+      "grad_norm": 0.4181274052477301,
+      "learning_rate": 3.3730304140710884e-05,
+      "loss": 0.461,
+      "step": 1192
+    },
+    {
+      "epoch": 1.1800197823936696,
+      "grad_norm": 0.37986297109107886,
+      "learning_rate": 3.371198241113962e-05,
+      "loss": 0.4425,
+      "step": 1193
+    },
+    {
+      "epoch": 1.1810089020771513,
+      "grad_norm": 0.38662532285644546,
+      "learning_rate": 3.369366068156834e-05,
+      "loss": 0.4557,
+      "step": 1194
+    },
+    {
+      "epoch": 1.181998021760633,
+      "grad_norm": 0.320518534529602,
+      "learning_rate": 3.367533895199707e-05,
+      "loss": 0.387,
+      "step": 1195
+    },
+    {
+      "epoch": 1.1829871414441147,
+      "grad_norm": 0.3794192278894525,
+      "learning_rate": 3.3657017222425796e-05,
+      "loss": 0.4355,
+      "step": 1196
+    },
+    {
+      "epoch": 1.1839762611275964,
+      "grad_norm": 0.3258255894383896,
+      "learning_rate": 3.363869549285453e-05,
+      "loss": 0.4617,
+      "step": 1197
+    },
+    {
+      "epoch": 1.184965380811078,
+      "grad_norm": 0.3753454976012276,
+      "learning_rate": 3.3620373763283256e-05,
+      "loss": 0.4796,
+      "step": 1198
+    },
+    {
+      "epoch": 1.1859545004945597,
+      "grad_norm": 0.31066105843247516,
+      "learning_rate": 3.360205203371198e-05,
+      "loss": 0.4841,
+      "step": 1199
+    },
+    {
+      "epoch": 1.1869436201780414,
+      "grad_norm": 0.37065175137983325,
+      "learning_rate": 3.3583730304140716e-05,
+      "loss": 0.4436,
+      "step": 1200
+    },
+    {
+      "epoch": 1.1879327398615231,
+      "grad_norm": 0.38153937820459927,
+      "learning_rate": 3.356540857456944e-05,
+      "loss": 0.4449,
+      "step": 1201
+    },
+    {
+      "epoch": 1.188921859545005,
+      "grad_norm": 0.31499452837981,
+      "learning_rate": 3.354708684499817e-05,
+      "loss": 0.3774,
+      "step": 1202
+    },
+    {
+      "epoch": 1.1899109792284865,
+      "grad_norm": 0.36683386599623874,
+      "learning_rate": 3.3528765115426895e-05,
+      "loss": 0.4489,
+      "step": 1203
+    },
+    {
+      "epoch": 1.1909000989119685,
+      "grad_norm": 0.3929092276173119,
+      "learning_rate": 3.351044338585563e-05,
+      "loss": 0.4983,
+      "step": 1204
+    },
+    {
+      "epoch": 1.19188921859545,
+      "grad_norm": 0.33361040009672116,
+      "learning_rate": 3.3492121656284355e-05,
+      "loss": 0.404,
+      "step": 1205
+    },
+    {
+      "epoch": 1.1928783382789319,
+      "grad_norm": 0.32846256380043704,
+      "learning_rate": 3.347379992671308e-05,
+      "loss": 0.4868,
+      "step": 1206
+    },
+    {
+      "epoch": 1.1938674579624133,
+      "grad_norm": 0.3340382987952063,
+      "learning_rate": 3.345547819714181e-05,
+      "loss": 0.4268,
+      "step": 1207
+    },
+    {
+      "epoch": 1.1948565776458953,
+      "grad_norm": 0.37162462421580916,
+      "learning_rate": 3.343715646757054e-05,
+      "loss": 0.4391,
+      "step": 1208
+    },
+    {
+      "epoch": 1.1958456973293767,
+      "grad_norm": 0.3365741665506994,
+      "learning_rate": 3.341883473799927e-05,
+      "loss": 0.4853,
+      "step": 1209
+    },
+    {
+      "epoch": 1.1968348170128587,
+      "grad_norm": 0.38473036798920535,
+      "learning_rate": 3.3400513008428e-05,
+      "loss": 0.4402,
+      "step": 1210
+    },
+    {
+      "epoch": 1.1978239366963404,
+      "grad_norm": 0.2943755821058674,
+      "learning_rate": 3.338219127885673e-05,
+      "loss": 0.4238,
+      "step": 1211
+    },
+    {
+      "epoch": 1.198813056379822,
+      "grad_norm": 0.3355562973386804,
+      "learning_rate": 3.336386954928545e-05,
+      "loss": 0.3917,
+      "step": 1212
+    },
+    {
+      "epoch": 1.1998021760633037,
+      "grad_norm": 0.317119955078073,
+      "learning_rate": 3.3345547819714186e-05,
+      "loss": 0.4691,
+      "step": 1213
+    },
+    {
+      "epoch": 1.2007912957467854,
+      "grad_norm": 0.33025445538947157,
+      "learning_rate": 3.332722609014291e-05,
+      "loss": 0.5038,
+      "step": 1214
+    },
+    {
+      "epoch": 1.2017804154302671,
+      "grad_norm": 0.3340396874499203,
+      "learning_rate": 3.330890436057164e-05,
+      "loss": 0.4274,
+      "step": 1215
+    },
+    {
+      "epoch": 1.2027695351137488,
+      "grad_norm": 0.34655165596355814,
+      "learning_rate": 3.3290582631000365e-05,
+      "loss": 0.4066,
+      "step": 1216
+    },
+    {
+      "epoch": 1.2037586547972305,
+      "grad_norm": 0.6174225922607776,
+      "learning_rate": 3.32722609014291e-05,
+      "loss": 0.4073,
+      "step": 1217
+    },
+    {
+      "epoch": 1.2047477744807122,
+      "grad_norm": 0.3327762274517774,
+      "learning_rate": 3.3253939171857825e-05,
+      "loss": 0.446,
+      "step": 1218
+    },
+    {
+      "epoch": 1.205736894164194,
+      "grad_norm": 0.32324569403115144,
+      "learning_rate": 3.323561744228655e-05,
+      "loss": 0.3973,
+      "step": 1219
+    },
+    {
+      "epoch": 1.2067260138476756,
+      "grad_norm": 0.3183062423357114,
+      "learning_rate": 3.321729571271528e-05,
+      "loss": 0.3924,
+      "step": 1220
+    },
+    {
+      "epoch": 1.2077151335311573,
+      "grad_norm": 0.3100295021920569,
+      "learning_rate": 3.319897398314401e-05,
+      "loss": 0.4126,
+      "step": 1221
+    },
+    {
+      "epoch": 1.208704253214639,
+      "grad_norm": 0.3373361154076598,
+      "learning_rate": 3.3180652253572744e-05,
+      "loss": 0.4281,
+      "step": 1222
+    },
+    {
+      "epoch": 1.2096933728981207,
+      "grad_norm": 0.3765180938062049,
+      "learning_rate": 3.3162330524001464e-05,
+      "loss": 0.5078,
+      "step": 1223
+    },
+    {
+      "epoch": 1.2106824925816024,
+      "grad_norm": 0.3143717459812389,
+      "learning_rate": 3.31440087944302e-05,
+      "loss": 0.4406,
+      "step": 1224
+    },
+    {
+      "epoch": 1.2116716122650841,
+      "grad_norm": 0.43796853097237903,
+      "learning_rate": 3.3125687064858924e-05,
+      "loss": 0.4843,
+      "step": 1225
+    },
+    {
+      "epoch": 1.2126607319485658,
+      "grad_norm": 0.27756885414505256,
+      "learning_rate": 3.310736533528766e-05,
+      "loss": 0.3495,
+      "step": 1226
+    },
+    {
+      "epoch": 1.2136498516320475,
+      "grad_norm": 0.34846044043895186,
+      "learning_rate": 3.308904360571638e-05,
+      "loss": 0.4714,
+      "step": 1227
+    },
+    {
+      "epoch": 1.2146389713155292,
+      "grad_norm": 0.2792141653842612,
+      "learning_rate": 3.307072187614511e-05,
+      "loss": 0.435,
+      "step": 1228
+    },
+    {
+      "epoch": 1.215628090999011,
+      "grad_norm": 0.3247383857479883,
+      "learning_rate": 3.3052400146573836e-05,
+      "loss": 0.4369,
+      "step": 1229
+    },
+    {
+      "epoch": 1.2166172106824926,
+      "grad_norm": 0.31934633509474236,
+      "learning_rate": 3.303407841700257e-05,
+      "loss": 0.4247,
+      "step": 1230
+    },
+    {
+      "epoch": 1.2176063303659743,
+      "grad_norm": 0.3117420006063186,
+      "learning_rate": 3.3015756687431296e-05,
+      "loss": 0.4221,
+      "step": 1231
+    },
+    {
+      "epoch": 1.218595450049456,
+      "grad_norm": 0.37143959870180393,
+      "learning_rate": 3.299743495786002e-05,
+      "loss": 0.4451,
+      "step": 1232
+    },
+    {
+      "epoch": 1.2195845697329377,
+      "grad_norm": 0.2747122535928208,
+      "learning_rate": 3.297911322828875e-05,
+      "loss": 0.4168,
+      "step": 1233
+    },
+    {
+      "epoch": 1.2205736894164194,
+      "grad_norm": 0.32821875211096874,
+      "learning_rate": 3.296079149871748e-05,
+      "loss": 0.4876,
+      "step": 1234
+    },
+    {
+      "epoch": 1.2215628090999011,
+      "grad_norm": 0.2848558407621968,
+      "learning_rate": 3.294246976914621e-05,
+      "loss": 0.4402,
+      "step": 1235
+    },
+    {
+      "epoch": 1.2225519287833828,
+      "grad_norm": 0.2885008544507463,
+      "learning_rate": 3.2924148039574935e-05,
+      "loss": 0.4349,
+      "step": 1236
+    },
+    {
+      "epoch": 1.2235410484668645,
+      "grad_norm": 0.33320909099352525,
+      "learning_rate": 3.290582631000367e-05,
+      "loss": 0.4433,
+      "step": 1237
+    },
+    {
+      "epoch": 1.2245301681503462,
+      "grad_norm": 0.25857871277266625,
+      "learning_rate": 3.2887504580432394e-05,
+      "loss": 0.4339,
+      "step": 1238
+    },
+    {
+      "epoch": 1.225519287833828,
+      "grad_norm": 0.32250835248537585,
+      "learning_rate": 3.286918285086113e-05,
+      "loss": 0.4378,
+      "step": 1239
+    },
+    {
+      "epoch": 1.2265084075173096,
+      "grad_norm": 0.30737128663542473,
+      "learning_rate": 3.285086112128985e-05,
+      "loss": 0.4577,
+      "step": 1240
+    },
+    {
+      "epoch": 1.2274975272007913,
+      "grad_norm": 0.2742545175524849,
+      "learning_rate": 3.283253939171858e-05,
+      "loss": 0.428,
+      "step": 1241
+    },
+    {
+      "epoch": 1.228486646884273,
+      "grad_norm": 0.28951560409808336,
+      "learning_rate": 3.2814217662147307e-05,
+      "loss": 0.3863,
+      "step": 1242
+    },
+    {
+      "epoch": 1.2294757665677547,
+      "grad_norm": 0.7199122350538452,
+      "learning_rate": 3.279589593257604e-05,
+      "loss": 0.4365,
+      "step": 1243
+    },
+    {
+      "epoch": 1.2304648862512364,
+      "grad_norm": 0.25600755117681134,
+      "learning_rate": 3.277757420300476e-05,
+      "loss": 0.4074,
+      "step": 1244
+    },
+    {
+      "epoch": 1.231454005934718,
+      "grad_norm": 0.8811383887062425,
+      "learning_rate": 3.275925247343349e-05,
+      "loss": 0.481,
+      "step": 1245
+    },
+    {
+      "epoch": 1.2324431256181998,
+      "grad_norm": 0.3023345814779684,
+      "learning_rate": 3.2740930743862226e-05,
+      "loss": 0.4768,
+      "step": 1246
+    },
+    {
+      "epoch": 1.2334322453016815,
+      "grad_norm": 0.2755925407169624,
+      "learning_rate": 3.272260901429095e-05,
+      "loss": 0.4027,
+      "step": 1247
+    },
+    {
+      "epoch": 1.2344213649851632,
+      "grad_norm": 11.463153475362521,
+      "learning_rate": 3.270428728471968e-05,
+      "loss": 0.4636,
+      "step": 1248
+    },
+    {
+      "epoch": 1.235410484668645,
+      "grad_norm": 0.45187999656619815,
+      "learning_rate": 3.2685965555148405e-05,
+      "loss": 0.4927,
+      "step": 1249
+    },
+    {
+      "epoch": 1.2363996043521266,
+      "grad_norm": 0.32892957675978074,
+      "learning_rate": 3.266764382557714e-05,
+      "loss": 0.4548,
+      "step": 1250
+    },
+    {
+      "epoch": 1.2373887240356083,
+      "grad_norm": 0.3303929130848598,
+      "learning_rate": 3.2649322096005865e-05,
+      "loss": 0.4709,
+      "step": 1251
+    },
+    {
+      "epoch": 1.23837784371909,
+      "grad_norm": 0.3373802532562506,
+      "learning_rate": 3.263100036643459e-05,
+      "loss": 0.4582,
+      "step": 1252
+    },
+    {
+      "epoch": 1.2393669634025717,
+      "grad_norm": 0.3779420909570119,
+      "learning_rate": 3.261267863686332e-05,
+      "loss": 0.4451,
+      "step": 1253
+    },
+    {
+      "epoch": 1.2403560830860534,
+      "grad_norm": 0.291216947790094,
+      "learning_rate": 3.259435690729205e-05,
+      "loss": 0.4013,
+      "step": 1254
+    },
+    {
+      "epoch": 1.241345202769535,
+      "grad_norm": 0.34985117379943337,
+      "learning_rate": 3.257603517772078e-05,
+      "loss": 0.4513,
+      "step": 1255
+    },
+    {
+      "epoch": 1.2423343224530168,
+      "grad_norm": 0.31095271457516216,
+      "learning_rate": 3.255771344814951e-05,
+      "loss": 0.4411,
+      "step": 1256
+    },
+    {
+      "epoch": 1.2433234421364985,
+      "grad_norm": 1.047735080448275,
+      "learning_rate": 3.253939171857823e-05,
+      "loss": 0.5013,
+      "step": 1257
+    },
+    {
+      "epoch": 1.2443125618199802,
+      "grad_norm": 0.27825085687486434,
+      "learning_rate": 3.252106998900696e-05,
+      "loss": 0.4161,
+      "step": 1258
+    },
+    {
+      "epoch": 1.2453016815034619,
+      "grad_norm": 0.3636283861995447,
+      "learning_rate": 3.2502748259435696e-05,
+      "loss": 0.5048,
+      "step": 1259
+    },
+    {
+      "epoch": 1.2462908011869436,
+      "grad_norm": 0.2972461553564986,
+      "learning_rate": 3.248442652986442e-05,
+      "loss": 0.4577,
+      "step": 1260
+    },
+    {
+      "epoch": 1.2472799208704253,
+      "grad_norm": 0.4663431964179042,
+      "learning_rate": 3.246610480029315e-05,
+      "loss": 0.4355,
+      "step": 1261
+    },
+    {
+      "epoch": 1.248269040553907,
+      "grad_norm": 0.7010092875578634,
+      "learning_rate": 3.2447783070721876e-05,
+      "loss": 0.4501,
+      "step": 1262
+    },
+    {
+      "epoch": 1.2492581602373887,
+      "grad_norm": 0.301261398514134,
+      "learning_rate": 3.242946134115061e-05,
+      "loss": 0.4805,
+      "step": 1263
+    },
+    {
+      "epoch": 1.2502472799208704,
+      "grad_norm": 0.3418549326196357,
+      "learning_rate": 3.2411139611579335e-05,
+      "loss": 0.4795,
+      "step": 1264
+    },
+    {
+      "epoch": 1.251236399604352,
+      "grad_norm": 0.29526440353783767,
+      "learning_rate": 3.239281788200806e-05,
+      "loss": 0.4332,
+      "step": 1265
+    },
+    {
+      "epoch": 1.2522255192878338,
+      "grad_norm": 0.2967530321345406,
+      "learning_rate": 3.237449615243679e-05,
+      "loss": 0.3992,
+      "step": 1266
+    },
+    {
+      "epoch": 1.2532146389713155,
+      "grad_norm": 0.30327817973420274,
+      "learning_rate": 3.235617442286552e-05,
+      "loss": 0.4832,
+      "step": 1267
+    },
+    {
+      "epoch": 1.2542037586547972,
+      "grad_norm": 0.28390309436771854,
+      "learning_rate": 3.233785269329425e-05,
+      "loss": 0.394,
+      "step": 1268
+    },
+    {
+      "epoch": 1.2551928783382789,
+      "grad_norm": 0.3640976735315917,
+      "learning_rate": 3.2319530963722974e-05,
+      "loss": 0.4319,
+      "step": 1269
+    },
+    {
+      "epoch": 1.2561819980217606,
+      "grad_norm": 0.3039414854494338,
+      "learning_rate": 3.230120923415171e-05,
+      "loss": 0.4777,
+      "step": 1270
+    },
+    {
+      "epoch": 1.2571711177052423,
+      "grad_norm": 0.2797669473882524,
+      "learning_rate": 3.2282887504580434e-05,
+      "loss": 0.4259,
+      "step": 1271
+    },
+    {
+      "epoch": 1.258160237388724,
+      "grad_norm": 0.3431193474424225,
+      "learning_rate": 3.226456577500917e-05,
+      "loss": 0.4484,
+      "step": 1272
+    },
+    {
+      "epoch": 1.2591493570722057,
+      "grad_norm": 0.2748018185129936,
+      "learning_rate": 3.224624404543789e-05,
+      "loss": 0.4019,
+      "step": 1273
+    },
+    {
+      "epoch": 1.2601384767556874,
+      "grad_norm": 0.3078022305195795,
+      "learning_rate": 3.222792231586662e-05,
+      "loss": 0.4357,
+      "step": 1274
+    },
+    {
+      "epoch": 1.2611275964391693,
+      "grad_norm": 0.28817276925435575,
+      "learning_rate": 3.2209600586295346e-05,
+      "loss": 0.4581,
+      "step": 1275
+    },
+    {
+      "epoch": 1.2621167161226508,
+      "grad_norm": 0.28915115293238075,
+      "learning_rate": 3.219127885672408e-05,
+      "loss": 0.4113,
+      "step": 1276
+    },
+    {
+      "epoch": 1.2631058358061327,
+      "grad_norm": 0.29431101862681536,
+      "learning_rate": 3.2172957127152806e-05,
+      "loss": 0.4844,
+      "step": 1277
+    },
+    {
+      "epoch": 1.2640949554896141,
+      "grad_norm": 0.31417644441909515,
+      "learning_rate": 3.215463539758153e-05,
+      "loss": 0.4359,
+      "step": 1278
+    },
+    {
+      "epoch": 1.265084075173096,
+      "grad_norm": 0.2932996151523536,
+      "learning_rate": 3.213631366801026e-05,
+      "loss": 0.4566,
+      "step": 1279
+    },
+    {
+      "epoch": 1.2660731948565775,
+      "grad_norm": 0.3544874502119,
+      "learning_rate": 3.211799193843899e-05,
+      "loss": 0.4528,
+      "step": 1280
+    },
+    {
+      "epoch": 1.2670623145400595,
+      "grad_norm": 0.31431737172579327,
+      "learning_rate": 3.209967020886772e-05,
+      "loss": 0.4873,
+      "step": 1281
+    },
+    {
+      "epoch": 1.268051434223541,
+      "grad_norm": 0.32083957972618893,
+      "learning_rate": 3.2081348479296445e-05,
+      "loss": 0.4405,
+      "step": 1282
+    },
+    {
+      "epoch": 1.2690405539070229,
+      "grad_norm": 0.28531911086752,
+      "learning_rate": 3.206302674972518e-05,
+      "loss": 0.4763,
+      "step": 1283
+    },
+    {
+      "epoch": 1.2700296735905043,
+      "grad_norm": 0.39038530630890933,
+      "learning_rate": 3.2044705020153904e-05,
+      "loss": 0.487,
+      "step": 1284
+    },
+    {
+      "epoch": 1.2710187932739863,
+      "grad_norm": 0.3231694114187269,
+      "learning_rate": 3.202638329058264e-05,
+      "loss": 0.4575,
+      "step": 1285
+    },
+    {
+      "epoch": 1.2720079129574677,
+      "grad_norm": 0.39483284030242527,
+      "learning_rate": 3.200806156101136e-05,
+      "loss": 0.5078,
+      "step": 1286
+    },
+    {
+      "epoch": 1.2729970326409497,
+      "grad_norm": 0.2624573775627927,
+      "learning_rate": 3.198973983144009e-05,
+      "loss": 0.4189,
+      "step": 1287
+    },
+    {
+      "epoch": 1.2739861523244311,
+      "grad_norm": 0.3539064045837323,
+      "learning_rate": 3.197141810186882e-05,
+      "loss": 0.4083,
+      "step": 1288
+    },
+    {
+      "epoch": 1.274975272007913,
+      "grad_norm": 0.41113892601872426,
+      "learning_rate": 3.195309637229755e-05,
+      "loss": 0.4359,
+      "step": 1289
+    },
+    {
+      "epoch": 1.2759643916913945,
+      "grad_norm": 0.3263835404973625,
+      "learning_rate": 3.193477464272627e-05,
+      "loss": 0.4466,
+      "step": 1290
+    },
+    {
+      "epoch": 1.2769535113748764,
+      "grad_norm": 0.3521424288990003,
+      "learning_rate": 3.1916452913155e-05,
+      "loss": 0.4346,
+      "step": 1291
+    },
+    {
+      "epoch": 1.277942631058358,
+      "grad_norm": 0.32960482803892793,
+      "learning_rate": 3.189813118358373e-05,
+      "loss": 0.4081,
+      "step": 1292
+    },
+    {
+      "epoch": 1.2789317507418398,
+      "grad_norm": 0.37621318875682574,
+      "learning_rate": 3.187980945401246e-05,
+      "loss": 0.5071,
+      "step": 1293
+    },
+    {
+      "epoch": 1.2799208704253215,
+      "grad_norm": 0.2842212763877933,
+      "learning_rate": 3.186148772444119e-05,
+      "loss": 0.4383,
+      "step": 1294
+    },
+    {
+      "epoch": 1.2809099901088032,
+      "grad_norm": 0.4185836752845537,
+      "learning_rate": 3.1843165994869915e-05,
+      "loss": 0.4605,
+      "step": 1295
+    },
+    {
+      "epoch": 1.281899109792285,
+      "grad_norm": 0.2864590173818887,
+      "learning_rate": 3.182484426529865e-05,
+      "loss": 0.4115,
+      "step": 1296
+    },
+    {
+      "epoch": 1.2828882294757666,
+      "grad_norm": 0.31251818275646565,
+      "learning_rate": 3.1806522535727375e-05,
+      "loss": 0.4749,
+      "step": 1297
+    },
+    {
+      "epoch": 1.2838773491592483,
+      "grad_norm": 0.40375308662312326,
+      "learning_rate": 3.17882008061561e-05,
+      "loss": 0.4805,
+      "step": 1298
+    },
+    {
+      "epoch": 1.28486646884273,
+      "grad_norm": 0.2787733861619942,
+      "learning_rate": 3.176987907658483e-05,
+      "loss": 0.4695,
+      "step": 1299
+    },
+    {
+      "epoch": 1.2858555885262117,
+      "grad_norm": 0.3390155103152818,
+      "learning_rate": 3.175155734701356e-05,
+      "loss": 0.4799,
+      "step": 1300
+    },
+    {
+      "epoch": 1.2868447082096934,
+      "grad_norm": 0.39915923407897047,
+      "learning_rate": 3.173323561744229e-05,
+      "loss": 0.4911,
+      "step": 1301
+    },
+    {
+      "epoch": 1.2878338278931751,
+      "grad_norm": 0.2829054197112506,
+      "learning_rate": 3.171491388787102e-05,
+      "loss": 0.3704,
+      "step": 1302
+    },
+    {
+      "epoch": 1.2888229475766568,
+      "grad_norm": 0.3354351078071799,
+      "learning_rate": 3.169659215829974e-05,
+      "loss": 0.4108,
+      "step": 1303
+    },
+    {
+      "epoch": 1.2898120672601385,
+      "grad_norm": 0.2977538229373433,
+      "learning_rate": 3.167827042872847e-05,
+      "loss": 0.4279,
+      "step": 1304
+    },
+    {
+      "epoch": 1.2908011869436202,
+      "grad_norm": 0.328567314688651,
+      "learning_rate": 3.16599486991572e-05,
+      "loss": 0.4056,
+      "step": 1305
+    },
+    {
+      "epoch": 1.291790306627102,
+      "grad_norm": 0.3262674751990915,
+      "learning_rate": 3.164162696958593e-05,
+      "loss": 0.4404,
+      "step": 1306
+    },
+    {
+      "epoch": 1.2927794263105836,
+      "grad_norm": 0.38924960367345507,
+      "learning_rate": 3.162330524001466e-05,
+      "loss": 0.4364,
+      "step": 1307
+    },
+    {
+      "epoch": 1.2937685459940653,
+      "grad_norm": 0.28787499890259116,
+      "learning_rate": 3.1604983510443386e-05,
+      "loss": 0.5439,
+      "step": 1308
+    },
+    {
+      "epoch": 1.294757665677547,
+      "grad_norm": 0.3356946307141818,
+      "learning_rate": 3.158666178087212e-05,
+      "loss": 0.4324,
+      "step": 1309
+    },
+    {
+      "epoch": 1.2957467853610287,
+      "grad_norm": 0.4879317138260975,
+      "learning_rate": 3.1568340051300845e-05,
+      "loss": 0.4807,
+      "step": 1310
+    },
+    {
+      "epoch": 1.2967359050445104,
+      "grad_norm": 0.2955688732632104,
+      "learning_rate": 3.155001832172957e-05,
+      "loss": 0.4036,
+      "step": 1311
+    },
+    {
+      "epoch": 1.2977250247279921,
+      "grad_norm": 0.4032945847602782,
+      "learning_rate": 3.15316965921583e-05,
+      "loss": 0.4401,
+      "step": 1312
+    },
+    {
+      "epoch": 1.2987141444114738,
+      "grad_norm": 0.3717861992107911,
+      "learning_rate": 3.151337486258703e-05,
+      "loss": 0.412,
+      "step": 1313
+    },
+    {
+      "epoch": 1.2997032640949555,
+      "grad_norm": 0.36230212240482274,
+      "learning_rate": 3.149505313301576e-05,
+      "loss": 0.4756,
+      "step": 1314
+    },
+    {
+      "epoch": 1.3006923837784372,
+      "grad_norm": 0.3219831393538455,
+      "learning_rate": 3.1476731403444484e-05,
+      "loss": 0.4307,
+      "step": 1315
+    },
+    {
+      "epoch": 1.301681503461919,
+      "grad_norm": 0.2980114239679098,
+      "learning_rate": 3.145840967387321e-05,
+      "loss": 0.3987,
+      "step": 1316
+    },
+    {
+      "epoch": 1.3026706231454006,
+      "grad_norm": 0.8151806056234506,
+      "learning_rate": 3.1440087944301944e-05,
+      "loss": 0.4737,
+      "step": 1317
+    },
+    {
+      "epoch": 1.3036597428288823,
+      "grad_norm": 0.4605566888952793,
+      "learning_rate": 3.142176621473068e-05,
+      "loss": 0.4641,
+      "step": 1318
+    },
+    {
+      "epoch": 1.304648862512364,
+      "grad_norm": 0.3980859249064098,
+      "learning_rate": 3.14034444851594e-05,
+      "loss": 0.459,
+      "step": 1319
+    },
+    {
+      "epoch": 1.3056379821958457,
+      "grad_norm": 0.2841614140383684,
+      "learning_rate": 3.138512275558813e-05,
+      "loss": 0.4057,
+      "step": 1320
+    },
+    {
+      "epoch": 1.3066271018793274,
+      "grad_norm": 1.9073972336742402,
+      "learning_rate": 3.1366801026016856e-05,
+      "loss": 0.4767,
+      "step": 1321
+    },
+    {
+      "epoch": 1.307616221562809,
+      "grad_norm": 0.5592709384548382,
+      "learning_rate": 3.134847929644559e-05,
+      "loss": 0.4819,
+      "step": 1322
+    },
+    {
+      "epoch": 1.3086053412462908,
+      "grad_norm": 0.3222748433424541,
+      "learning_rate": 3.1330157566874316e-05,
+      "loss": 0.4371,
+      "step": 1323
+    },
+    {
+      "epoch": 1.3095944609297725,
+      "grad_norm": 0.4650918381065197,
+      "learning_rate": 3.131183583730304e-05,
+      "loss": 0.415,
+      "step": 1324
+    },
+    {
+      "epoch": 1.3105835806132542,
+      "grad_norm": 0.4224117379256276,
+      "learning_rate": 3.129351410773177e-05,
+      "loss": 0.4566,
+      "step": 1325
+    },
+    {
+      "epoch": 1.311572700296736,
+      "grad_norm": 0.46207236337386226,
+      "learning_rate": 3.12751923781605e-05,
+      "loss": 0.5005,
+      "step": 1326
+    },
+    {
+      "epoch": 1.3125618199802176,
+      "grad_norm": 0.4683411564216293,
+      "learning_rate": 3.125687064858923e-05,
+      "loss": 0.4825,
+      "step": 1327
+    },
+    {
+      "epoch": 1.3135509396636993,
+      "grad_norm": 0.40426282504932587,
+      "learning_rate": 3.1238548919017955e-05,
+      "loss": 0.4031,
+      "step": 1328
+    },
+    {
+      "epoch": 1.314540059347181,
+      "grad_norm": 0.4937917446065795,
+      "learning_rate": 3.122022718944669e-05,
+      "loss": 0.4375,
+      "step": 1329
+    },
+    {
+      "epoch": 1.3155291790306627,
+      "grad_norm": 0.44951168903509353,
+      "learning_rate": 3.1201905459875414e-05,
+      "loss": 0.4931,
+      "step": 1330
+    },
+    {
+      "epoch": 1.3165182987141444,
+      "grad_norm": 0.391635440933602,
+      "learning_rate": 3.118358373030415e-05,
+      "loss": 0.3879,
+      "step": 1331
+    },
+    {
+      "epoch": 1.317507418397626,
+      "grad_norm": 0.5580933846109488,
+      "learning_rate": 3.116526200073287e-05,
+      "loss": 0.4064,
+      "step": 1332
+    },
+    {
+      "epoch": 1.3184965380811078,
+      "grad_norm": 0.26495237831546364,
+      "learning_rate": 3.11469402711616e-05,
+      "loss": 0.3427,
+      "step": 1333
+    },
+    {
+      "epoch": 1.3194856577645895,
+      "grad_norm": 0.4754556883385079,
+      "learning_rate": 3.112861854159033e-05,
+      "loss": 0.4223,
+      "step": 1334
+    },
+    {
+      "epoch": 1.3204747774480712,
+      "grad_norm": 0.3881819412366177,
+      "learning_rate": 3.111029681201906e-05,
+      "loss": 0.4623,
+      "step": 1335
+    },
+    {
+      "epoch": 1.3214638971315529,
+      "grad_norm": 0.27166678881531925,
+      "learning_rate": 3.109197508244778e-05,
+      "loss": 0.4131,
+      "step": 1336
+    },
+    {
+      "epoch": 1.3224530168150346,
+      "grad_norm": 0.40190887482300114,
+      "learning_rate": 3.107365335287651e-05,
+      "loss": 0.492,
+      "step": 1337
+    },
+    {
+      "epoch": 1.3234421364985163,
+      "grad_norm": 0.3539039447670423,
+      "learning_rate": 3.105533162330524e-05,
+      "loss": 0.4205,
+      "step": 1338
+    },
+    {
+      "epoch": 1.324431256181998,
+      "grad_norm": 0.35944898679004156,
+      "learning_rate": 3.103700989373397e-05,
+      "loss": 0.4791,
+      "step": 1339
+    },
+    {
+      "epoch": 1.3254203758654797,
+      "grad_norm": 0.405897222512928,
+      "learning_rate": 3.10186881641627e-05,
+      "loss": 0.5133,
+      "step": 1340
+    },
+    {
+      "epoch": 1.3264094955489614,
+      "grad_norm": 0.3149489860402367,
+      "learning_rate": 3.1000366434591425e-05,
+      "loss": 0.4603,
+      "step": 1341
+    },
+    {
+      "epoch": 1.327398615232443,
+      "grad_norm": 0.45252664436277323,
+      "learning_rate": 3.098204470502016e-05,
+      "loss": 0.5322,
+      "step": 1342
+    },
+    {
+      "epoch": 1.3283877349159248,
+      "grad_norm": 0.2880720148908769,
+      "learning_rate": 3.0963722975448885e-05,
+      "loss": 0.4097,
+      "step": 1343
+    },
+    {
+      "epoch": 1.3293768545994065,
+      "grad_norm": 0.43858458830839114,
+      "learning_rate": 3.094540124587761e-05,
+      "loss": 0.4331,
+      "step": 1344
+    },
+    {
+      "epoch": 1.3303659742828882,
+      "grad_norm": 0.2965447563557795,
+      "learning_rate": 3.092707951630634e-05,
+      "loss": 0.4524,
+      "step": 1345
+    },
+    {
+      "epoch": 1.3313550939663699,
+      "grad_norm": 0.3131743992651659,
+      "learning_rate": 3.090875778673507e-05,
+      "loss": 0.4266,
+      "step": 1346
+    },
+    {
+      "epoch": 1.3323442136498516,
+      "grad_norm": 0.35252494183868,
+      "learning_rate": 3.08904360571638e-05,
+      "loss": 0.4365,
+      "step": 1347
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.3138151302405501,
+      "learning_rate": 3.087211432759253e-05,
+      "loss": 0.4451,
+      "step": 1348
+    },
+    {
+      "epoch": 1.334322453016815,
+      "grad_norm": 0.3115085002074383,
+      "learning_rate": 3.085379259802125e-05,
+      "loss": 0.4425,
+      "step": 1349
+    },
+    {
+      "epoch": 1.3353115727002967,
+      "grad_norm": 0.30006114039678367,
+      "learning_rate": 3.083547086844998e-05,
+      "loss": 0.3935,
+      "step": 1350
+    },
+    {
+      "epoch": 1.3363006923837784,
+      "grad_norm": 0.28834125078240946,
+      "learning_rate": 3.081714913887871e-05,
+      "loss": 0.413,
+      "step": 1351
+    },
+    {
+      "epoch": 1.3372898120672603,
+      "grad_norm": 0.3561705460960808,
+      "learning_rate": 3.079882740930744e-05,
+      "loss": 0.4968,
+      "step": 1352
+    },
+    {
+      "epoch": 1.3382789317507418,
+      "grad_norm": 0.3260716628155875,
+      "learning_rate": 3.078050567973617e-05,
+      "loss": 0.4487,
+      "step": 1353
+    },
+    {
+      "epoch": 1.3392680514342237,
+      "grad_norm": 0.32982555741060104,
+      "learning_rate": 3.0762183950164896e-05,
+      "loss": 0.441,
+      "step": 1354
+    },
+    {
+      "epoch": 1.3402571711177051,
+      "grad_norm": 0.30807762347144296,
+      "learning_rate": 3.074386222059363e-05,
+      "loss": 0.4121,
+      "step": 1355
+    },
+    {
+      "epoch": 1.341246290801187,
+      "grad_norm": 0.3532020602660648,
+      "learning_rate": 3.0725540491022355e-05,
+      "loss": 0.505,
+      "step": 1356
+    },
+    {
+      "epoch": 1.3422354104846685,
+      "grad_norm": 0.2876611640436799,
+      "learning_rate": 3.070721876145108e-05,
+      "loss": 0.4096,
+      "step": 1357
+    },
+    {
+      "epoch": 1.3432245301681505,
+      "grad_norm": 0.3186642457128235,
+      "learning_rate": 3.068889703187981e-05,
+      "loss": 0.4585,
+      "step": 1358
+    },
+    {
+      "epoch": 1.344213649851632,
+      "grad_norm": 0.31673282289780236,
+      "learning_rate": 3.067057530230854e-05,
+      "loss": 0.4235,
+      "step": 1359
+    },
+    {
+      "epoch": 1.3452027695351139,
+      "grad_norm": 0.35253205951974537,
+      "learning_rate": 3.065225357273727e-05,
+      "loss": 0.4569,
+      "step": 1360
+    },
+    {
+      "epoch": 1.3461918892185953,
+      "grad_norm": 0.3349663982529888,
+      "learning_rate": 3.0633931843165994e-05,
+      "loss": 0.3936,
+      "step": 1361
+    },
+    {
+      "epoch": 1.3471810089020773,
+      "grad_norm": 0.33498838963025845,
+      "learning_rate": 3.061561011359472e-05,
+      "loss": 0.4983,
+      "step": 1362
+    },
+    {
+      "epoch": 1.3481701285855587,
+      "grad_norm": 0.3400374587506467,
+      "learning_rate": 3.0597288384023454e-05,
+      "loss": 0.4503,
+      "step": 1363
+    },
+    {
+      "epoch": 1.3491592482690407,
+      "grad_norm": 0.3203012395834267,
+      "learning_rate": 3.057896665445218e-05,
+      "loss": 0.3944,
+      "step": 1364
+    },
+    {
+      "epoch": 1.3501483679525221,
+      "grad_norm": 0.3920156373833764,
+      "learning_rate": 3.0560644924880913e-05,
+      "loss": 0.4813,
+      "step": 1365
+    },
+    {
+      "epoch": 1.351137487636004,
+      "grad_norm": 0.30315401884490906,
+      "learning_rate": 3.054232319530964e-05,
+      "loss": 0.4454,
+      "step": 1366
+    },
+    {
+      "epoch": 1.3521266073194855,
+      "grad_norm": 0.35739212453761815,
+      "learning_rate": 3.0524001465738366e-05,
+      "loss": 0.439,
+      "step": 1367
+    },
+    {
+      "epoch": 1.3531157270029674,
+      "grad_norm": 0.3417886435859495,
+      "learning_rate": 3.0505679736167096e-05,
+      "loss": 0.4665,
+      "step": 1368
+    },
+    {
+      "epoch": 1.354104846686449,
+      "grad_norm": 0.3235348071287897,
+      "learning_rate": 3.0487358006595822e-05,
+      "loss": 0.4226,
+      "step": 1369
+    },
+    {
+      "epoch": 1.3550939663699308,
+      "grad_norm": 0.3159309730114347,
+      "learning_rate": 3.0469036277024556e-05,
+      "loss": 0.4466,
+      "step": 1370
+    },
+    {
+      "epoch": 1.3560830860534125,
+      "grad_norm": 0.324193279571812,
+      "learning_rate": 3.045071454745328e-05,
+      "loss": 0.4501,
+      "step": 1371
+    },
+    {
+      "epoch": 1.3570722057368942,
+      "grad_norm": 0.3071795678407246,
+      "learning_rate": 3.0432392817882012e-05,
+      "loss": 0.4487,
+      "step": 1372
+    },
+    {
+      "epoch": 1.358061325420376,
+      "grad_norm": 0.30058020510609845,
+      "learning_rate": 3.0414071088310735e-05,
+      "loss": 0.4561,
+      "step": 1373
+    },
+    {
+      "epoch": 1.3590504451038576,
+      "grad_norm": 0.3505053014597225,
+      "learning_rate": 3.0395749358739468e-05,
+      "loss": 0.4025,
+      "step": 1374
+    },
+    {
+      "epoch": 1.3600395647873393,
+      "grad_norm": 0.3111111301895231,
+      "learning_rate": 3.037742762916819e-05,
+      "loss": 0.4098,
+      "step": 1375
+    },
+    {
+      "epoch": 1.361028684470821,
+      "grad_norm": 0.2601225263616035,
+      "learning_rate": 3.0359105899596924e-05,
+      "loss": 0.4243,
+      "step": 1376
+    },
+    {
+      "epoch": 1.3620178041543027,
+      "grad_norm": 0.316696497556371,
+      "learning_rate": 3.0340784170025654e-05,
+      "loss": 0.3969,
+      "step": 1377
+    },
+    {
+      "epoch": 1.3630069238377844,
+      "grad_norm": 0.34847389358503794,
+      "learning_rate": 3.032246244045438e-05,
+      "loss": 0.4551,
+      "step": 1378
+    },
+    {
+      "epoch": 1.3639960435212661,
+      "grad_norm": 0.2902604562577088,
+      "learning_rate": 3.030414071088311e-05,
+      "loss": 0.407,
+      "step": 1379
+    },
+    {
+      "epoch": 1.3649851632047478,
+      "grad_norm": 0.27781817505418216,
+      "learning_rate": 3.0285818981311837e-05,
+      "loss": 0.4328,
+      "step": 1380
+    },
+    {
+      "epoch": 1.3659742828882295,
+      "grad_norm": 5.942774002381965,
+      "learning_rate": 3.0267497251740567e-05,
+      "loss": 0.4695,
+      "step": 1381
+    },
+    {
+      "epoch": 1.3669634025717112,
+      "grad_norm": 0.38926658156561217,
+      "learning_rate": 3.0249175522169293e-05,
+      "loss": 0.4174,
+      "step": 1382
+    },
+    {
+      "epoch": 1.367952522255193,
+      "grad_norm": 0.4415126867453672,
+      "learning_rate": 3.0230853792598023e-05,
+      "loss": 0.4475,
+      "step": 1383
+    },
+    {
+      "epoch": 1.3689416419386746,
+      "grad_norm": 0.3120038844745485,
+      "learning_rate": 3.021253206302675e-05,
+      "loss": 0.469,
+      "step": 1384
+    },
+    {
+      "epoch": 1.3699307616221563,
+      "grad_norm": 0.41969819981367795,
+      "learning_rate": 3.019421033345548e-05,
+      "loss": 0.4923,
+      "step": 1385
+    },
+    {
+      "epoch": 1.370919881305638,
+      "grad_norm": 0.38256379980543986,
+      "learning_rate": 3.0175888603884205e-05,
+      "loss": 0.4336,
+      "step": 1386
+    },
+    {
+      "epoch": 1.3719090009891197,
+      "grad_norm": 0.3767270435686749,
+      "learning_rate": 3.0157566874312935e-05,
+      "loss": 0.4539,
+      "step": 1387
+    },
+    {
+      "epoch": 1.3728981206726014,
+      "grad_norm": 0.3157034106065137,
+      "learning_rate": 3.013924514474166e-05,
+      "loss": 0.445,
+      "step": 1388
+    },
+    {
+      "epoch": 1.3738872403560831,
+      "grad_norm": 0.36045163968750676,
+      "learning_rate": 3.0120923415170395e-05,
+      "loss": 0.4121,
+      "step": 1389
+    },
+    {
+      "epoch": 1.3748763600395648,
+      "grad_norm": 0.31479023453240734,
+      "learning_rate": 3.0102601685599125e-05,
+      "loss": 0.4216,
+      "step": 1390
+    },
+    {
+      "epoch": 1.3758654797230465,
+      "grad_norm": 0.2997878080503291,
+      "learning_rate": 3.008427995602785e-05,
+      "loss": 0.3644,
+      "step": 1391
+    },
+    {
+      "epoch": 1.3768545994065282,
+      "grad_norm": 0.3320290130778573,
+      "learning_rate": 3.006595822645658e-05,
+      "loss": 0.4513,
+      "step": 1392
+    },
+    {
+      "epoch": 1.37784371909001,
+      "grad_norm": 0.29477176122877125,
+      "learning_rate": 3.0047636496885307e-05,
+      "loss": 0.4601,
+      "step": 1393
+    },
+    {
+      "epoch": 1.3788328387734916,
+      "grad_norm": 0.3412218077170071,
+      "learning_rate": 3.0029314767314037e-05,
+      "loss": 0.4154,
+      "step": 1394
+    },
+    {
+      "epoch": 1.3798219584569733,
+      "grad_norm": 0.3275380780138223,
+      "learning_rate": 3.0010993037742764e-05,
+      "loss": 0.4176,
+      "step": 1395
+    },
+    {
+      "epoch": 1.380811078140455,
+      "grad_norm": 0.2837253095050283,
+      "learning_rate": 2.9992671308171493e-05,
+      "loss": 0.4187,
+      "step": 1396
+    },
+    {
+      "epoch": 1.3818001978239367,
+      "grad_norm": 0.4176886469982826,
+      "learning_rate": 2.997434957860022e-05,
+      "loss": 0.4724,
+      "step": 1397
+    },
+    {
+      "epoch": 1.3827893175074184,
+      "grad_norm": 0.31073598352099785,
+      "learning_rate": 2.995602784902895e-05,
+      "loss": 0.4342,
+      "step": 1398
+    },
+    {
+      "epoch": 1.3837784371909,
+      "grad_norm": 0.3543815361213989,
+      "learning_rate": 2.9937706119457676e-05,
+      "loss": 0.5104,
+      "step": 1399
+    },
+    {
+      "epoch": 1.3847675568743818,
+      "grad_norm": 0.33664475436986274,
+      "learning_rate": 2.9919384389886406e-05,
+      "loss": 0.4302,
+      "step": 1400
+    },
+    {
+      "epoch": 1.3857566765578635,
+      "grad_norm": 0.36124601508500437,
+      "learning_rate": 2.990106266031514e-05,
+      "loss": 0.4512,
+      "step": 1401
+    },
+    {
+      "epoch": 1.3867457962413452,
+      "grad_norm": 0.35574503741610786,
+      "learning_rate": 2.9882740930743862e-05,
+      "loss": 0.4337,
+      "step": 1402
+    },
+    {
+      "epoch": 1.387734915924827,
+      "grad_norm": 0.36123605275941045,
+      "learning_rate": 2.9864419201172595e-05,
+      "loss": 0.4312,
+      "step": 1403
+    },
+    {
+      "epoch": 1.3887240356083086,
+      "grad_norm": 0.3363751995869784,
+      "learning_rate": 2.9846097471601318e-05,
+      "loss": 0.4351,
+      "step": 1404
+    },
+    {
+      "epoch": 1.3897131552917903,
+      "grad_norm": 0.34547691242406064,
+      "learning_rate": 2.982777574203005e-05,
+      "loss": 0.4442,
+      "step": 1405
+    },
+    {
+      "epoch": 1.390702274975272,
+      "grad_norm": 0.364044074816971,
+      "learning_rate": 2.9809454012458778e-05,
+      "loss": 0.4708,
+      "step": 1406
+    },
+    {
+      "epoch": 1.3916913946587537,
+      "grad_norm": 0.3049021711534288,
+      "learning_rate": 2.9791132282887508e-05,
+      "loss": 0.477,
+      "step": 1407
+    },
+    {
+      "epoch": 1.3926805143422354,
+      "grad_norm": 0.38211284345526886,
+      "learning_rate": 2.9772810553316234e-05,
+      "loss": 0.43,
+      "step": 1408
+    },
+    {
+      "epoch": 1.393669634025717,
+      "grad_norm": 0.41386742466147336,
+      "learning_rate": 2.9754488823744964e-05,
+      "loss": 0.4561,
+      "step": 1409
+    },
+    {
+      "epoch": 1.3946587537091988,
+      "grad_norm": 0.3386310090552587,
+      "learning_rate": 2.973616709417369e-05,
+      "loss": 0.4387,
+      "step": 1410
+    },
+    {
+      "epoch": 1.3956478733926805,
+      "grad_norm": 0.3687161941753462,
+      "learning_rate": 2.971784536460242e-05,
+      "loss": 0.4446,
+      "step": 1411
+    },
+    {
+      "epoch": 1.3966369930761622,
+      "grad_norm": 0.3763623220861594,
+      "learning_rate": 2.9699523635031147e-05,
+      "loss": 0.4598,
+      "step": 1412
+    },
+    {
+      "epoch": 1.3976261127596439,
+      "grad_norm": 0.3938022952660516,
+      "learning_rate": 2.9681201905459876e-05,
+      "loss": 0.3538,
+      "step": 1413
+    },
+    {
+      "epoch": 1.3986152324431256,
+      "grad_norm": 0.36478061032823716,
+      "learning_rate": 2.9662880175888606e-05,
+      "loss": 0.4526,
+      "step": 1414
+    },
+    {
+      "epoch": 1.3996043521266073,
+      "grad_norm": 0.35983777332657113,
+      "learning_rate": 2.9644558446317333e-05,
+      "loss": 0.3961,
+      "step": 1415
+    },
+    {
+      "epoch": 1.400593471810089,
+      "grad_norm": 0.4361337967780688,
+      "learning_rate": 2.9626236716746066e-05,
+      "loss": 0.46,
+      "step": 1416
+    },
+    {
+      "epoch": 1.4015825914935707,
+      "grad_norm": 0.31864426898365455,
+      "learning_rate": 2.960791498717479e-05,
+      "loss": 0.4578,
+      "step": 1417
+    },
+    {
+      "epoch": 1.4025717111770524,
+      "grad_norm": 4.370523949146226,
+      "learning_rate": 2.9589593257603522e-05,
+      "loss": 0.5349,
+      "step": 1418
+    },
+    {
+      "epoch": 1.403560830860534,
+      "grad_norm": 0.7799236801274586,
+      "learning_rate": 2.9571271528032245e-05,
+      "loss": 0.4095,
+      "step": 1419
+    },
+    {
+      "epoch": 1.4045499505440158,
+      "grad_norm": 0.36466940872257664,
+      "learning_rate": 2.9552949798460978e-05,
+      "loss": 0.3791,
+      "step": 1420
+    },
+    {
+      "epoch": 1.4055390702274975,
+      "grad_norm": 0.5175092771422681,
+      "learning_rate": 2.95346280688897e-05,
+      "loss": 0.399,
+      "step": 1421
+    },
+    {
+      "epoch": 1.4065281899109792,
+      "grad_norm": 1.7636557470614664,
+      "learning_rate": 2.9516306339318434e-05,
+      "loss": 0.4801,
+      "step": 1422
+    },
+    {
+      "epoch": 1.4075173095944609,
+      "grad_norm": 0.4215106804422865,
+      "learning_rate": 2.9497984609747157e-05,
+      "loss": 0.4709,
+      "step": 1423
+    },
+    {
+      "epoch": 1.4085064292779426,
+      "grad_norm": 0.4032698691829994,
+      "learning_rate": 2.947966288017589e-05,
+      "loss": 0.4624,
+      "step": 1424
+    },
+    {
+      "epoch": 1.4094955489614243,
+      "grad_norm": 0.4050242307208399,
+      "learning_rate": 2.946134115060462e-05,
+      "loss": 0.5015,
+      "step": 1425
+    },
+    {
+      "epoch": 1.410484668644906,
+      "grad_norm": 0.31172007206763763,
+      "learning_rate": 2.9443019421033347e-05,
+      "loss": 0.442,
+      "step": 1426
+    },
+    {
+      "epoch": 1.4114737883283877,
+      "grad_norm": 0.36677022548571625,
+      "learning_rate": 2.9424697691462077e-05,
+      "loss": 0.4427,
+      "step": 1427
+    },
+    {
+      "epoch": 1.4124629080118694,
+      "grad_norm": 0.3546662288262847,
+      "learning_rate": 2.9406375961890803e-05,
+      "loss": 0.4413,
+      "step": 1428
+    },
+    {
+      "epoch": 1.413452027695351,
+      "grad_norm": 0.36917339334462623,
+      "learning_rate": 2.9388054232319533e-05,
+      "loss": 0.4743,
+      "step": 1429
+    },
+    {
+      "epoch": 1.4144411473788328,
+      "grad_norm": 0.3675002506888929,
+      "learning_rate": 2.936973250274826e-05,
+      "loss": 0.46,
+      "step": 1430
+    },
+    {
+      "epoch": 1.4154302670623147,
+      "grad_norm": 0.28465523267849413,
+      "learning_rate": 2.935141077317699e-05,
+      "loss": 0.418,
+      "step": 1431
+    },
+    {
+      "epoch": 1.4164193867457961,
+      "grad_norm": 0.4170643740008079,
+      "learning_rate": 2.9333089043605716e-05,
+      "loss": 0.4097,
+      "step": 1432
+    },
+    {
+      "epoch": 1.417408506429278,
+      "grad_norm": 0.3046886523834855,
+      "learning_rate": 2.9314767314034445e-05,
+      "loss": 0.4429,
+      "step": 1433
+    },
+    {
+      "epoch": 1.4183976261127595,
+      "grad_norm": 0.40000915818868443,
+      "learning_rate": 2.9296445584463172e-05,
+      "loss": 0.4261,
+      "step": 1434
+    },
+    {
+      "epoch": 1.4193867457962415,
+      "grad_norm": 0.38918672454921527,
+      "learning_rate": 2.9278123854891905e-05,
+      "loss": 0.4585,
+      "step": 1435
+    },
+    {
+      "epoch": 1.420375865479723,
+      "grad_norm": 0.29249532179463994,
+      "learning_rate": 2.9259802125320628e-05,
+      "loss": 0.3921,
+      "step": 1436
+    },
+    {
+      "epoch": 1.4213649851632049,
+      "grad_norm": 0.3243674617976441,
+      "learning_rate": 2.924148039574936e-05,
+      "loss": 0.511,
+      "step": 1437
+    },
+    {
+      "epoch": 1.4223541048466863,
+      "grad_norm": 0.31778105502346576,
+      "learning_rate": 2.922315866617809e-05,
+      "loss": 0.4576,
+      "step": 1438
+    },
+    {
+      "epoch": 1.4233432245301683,
+      "grad_norm": 0.3062041751866015,
+      "learning_rate": 2.9204836936606817e-05,
+      "loss": 0.4189,
+      "step": 1439
+    },
+    {
+      "epoch": 1.4243323442136497,
+      "grad_norm": 0.309264172194014,
+      "learning_rate": 2.9186515207035547e-05,
+      "loss": 0.4606,
+      "step": 1440
+    },
+    {
+      "epoch": 1.4253214638971317,
+      "grad_norm": 0.2801605149580132,
+      "learning_rate": 2.9168193477464274e-05,
+      "loss": 0.3993,
+      "step": 1441
+    },
+    {
+      "epoch": 1.4263105835806131,
+      "grad_norm": 0.30985960538013585,
+      "learning_rate": 2.9149871747893003e-05,
+      "loss": 0.4407,
+      "step": 1442
+    },
+    {
+      "epoch": 1.427299703264095,
+      "grad_norm": 0.3388775836866452,
+      "learning_rate": 2.913155001832173e-05,
+      "loss": 0.4429,
+      "step": 1443
+    },
+    {
+      "epoch": 1.4282888229475765,
+      "grad_norm": 0.2889630554202242,
+      "learning_rate": 2.911322828875046e-05,
+      "loss": 0.432,
+      "step": 1444
+    },
+    {
+      "epoch": 1.4292779426310585,
+      "grad_norm": 0.315605316728071,
+      "learning_rate": 2.9094906559179186e-05,
+      "loss": 0.4516,
+      "step": 1445
+    },
+    {
+      "epoch": 1.43026706231454,
+      "grad_norm": 0.3457337439226024,
+      "learning_rate": 2.9076584829607916e-05,
+      "loss": 0.4674,
+      "step": 1446
+    },
+    {
+      "epoch": 1.4312561819980218,
+      "grad_norm": 0.2952032263895049,
+      "learning_rate": 2.9058263100036642e-05,
+      "loss": 0.4521,
+      "step": 1447
+    },
+    {
+      "epoch": 1.4322453016815033,
+      "grad_norm": 0.28620576457716407,
+      "learning_rate": 2.9039941370465372e-05,
+      "loss": 0.4621,
+      "step": 1448
+    },
+    {
+      "epoch": 1.4332344213649852,
+      "grad_norm": 0.36744980164748403,
+      "learning_rate": 2.9021619640894105e-05,
+      "loss": 0.4888,
+      "step": 1449
+    },
+    {
+      "epoch": 1.434223541048467,
+      "grad_norm": 0.339812788180591,
+      "learning_rate": 2.900329791132283e-05,
+      "loss": 0.5289,
+      "step": 1450
+    },
+    {
+      "epoch": 1.4352126607319486,
+      "grad_norm": 0.31566827118837226,
+      "learning_rate": 2.898497618175156e-05,
+      "loss": 0.4492,
+      "step": 1451
+    },
+    {
+      "epoch": 1.4362017804154303,
+      "grad_norm": 0.31330393196868006,
+      "learning_rate": 2.8966654452180288e-05,
+      "loss": 0.3981,
+      "step": 1452
+    },
+    {
+      "epoch": 1.437190900098912,
+      "grad_norm": 0.3067900583902443,
+      "learning_rate": 2.8948332722609018e-05,
+      "loss": 0.4488,
+      "step": 1453
+    },
+    {
+      "epoch": 1.4381800197823937,
+      "grad_norm": 0.2968129574510213,
+      "learning_rate": 2.8930010993037744e-05,
+      "loss": 0.447,
+      "step": 1454
+    },
+    {
+      "epoch": 1.4391691394658754,
+      "grad_norm": 0.2864682872837869,
+      "learning_rate": 2.8911689263466474e-05,
+      "loss": 0.4034,
+      "step": 1455
+    },
+    {
+      "epoch": 1.4401582591493571,
+      "grad_norm": 0.6880039761239137,
+      "learning_rate": 2.88933675338952e-05,
+      "loss": 0.4305,
+      "step": 1456
+    },
+    {
+      "epoch": 1.4411473788328388,
+      "grad_norm": 0.27383935188184777,
+      "learning_rate": 2.887504580432393e-05,
+      "loss": 0.4472,
+      "step": 1457
+    },
+    {
+      "epoch": 1.4421364985163205,
+      "grad_norm": 0.29593151735190626,
+      "learning_rate": 2.8856724074752657e-05,
+      "loss": 0.4261,
+      "step": 1458
+    },
+    {
+      "epoch": 1.4431256181998022,
+      "grad_norm": 0.28273928682471794,
+      "learning_rate": 2.8838402345181386e-05,
+      "loss": 0.4034,
+      "step": 1459
+    },
+    {
+      "epoch": 1.444114737883284,
+      "grad_norm": 0.3165369665953018,
+      "learning_rate": 2.8820080615610116e-05,
+      "loss": 0.4266,
+      "step": 1460
+    },
+    {
+      "epoch": 1.4451038575667656,
+      "grad_norm": 0.3012146535540838,
+      "learning_rate": 2.8801758886038843e-05,
+      "loss": 0.4315,
+      "step": 1461
+    },
+    {
+      "epoch": 1.4460929772502473,
+      "grad_norm": 0.2711981424629738,
+      "learning_rate": 2.8783437156467576e-05,
+      "loss": 0.4184,
+      "step": 1462
+    },
+    {
+      "epoch": 1.447082096933729,
+      "grad_norm": 0.31002008293801026,
+      "learning_rate": 2.87651154268963e-05,
+      "loss": 0.4222,
+      "step": 1463
+    },
+    {
+      "epoch": 1.4480712166172107,
+      "grad_norm": 0.3539283902685812,
+      "learning_rate": 2.8746793697325032e-05,
+      "loss": 0.4322,
+      "step": 1464
+    },
+    {
+      "epoch": 1.4490603363006924,
+      "grad_norm": 0.3197987270904071,
+      "learning_rate": 2.8728471967753755e-05,
+      "loss": 0.4626,
+      "step": 1465
+    },
+    {
+      "epoch": 1.4500494559841741,
+      "grad_norm": 0.294396776618464,
+      "learning_rate": 2.8710150238182488e-05,
+      "loss": 0.4294,
+      "step": 1466
+    },
+    {
+      "epoch": 1.4510385756676558,
+      "grad_norm": 0.31250638103474127,
+      "learning_rate": 2.869182850861121e-05,
+      "loss": 0.491,
+      "step": 1467
+    },
+    {
+      "epoch": 1.4520276953511375,
+      "grad_norm": 0.29465729172509825,
+      "learning_rate": 2.8673506779039945e-05,
+      "loss": 0.4182,
+      "step": 1468
+    },
+    {
+      "epoch": 1.4530168150346192,
+      "grad_norm": 0.27683736051729835,
+      "learning_rate": 2.8655185049468668e-05,
+      "loss": 0.4176,
+      "step": 1469
+    },
+    {
+      "epoch": 1.454005934718101,
+      "grad_norm": 0.3763143498743748,
+      "learning_rate": 2.86368633198974e-05,
+      "loss": 0.4864,
+      "step": 1470
+    },
+    {
+      "epoch": 1.4549950544015826,
+      "grad_norm": 0.2951254080913707,
+      "learning_rate": 2.8618541590326127e-05,
+      "loss": 0.4084,
+      "step": 1471
+    },
+    {
+      "epoch": 1.4559841740850643,
+      "grad_norm": 0.2874094634392024,
+      "learning_rate": 2.8600219860754857e-05,
+      "loss": 0.4525,
+      "step": 1472
+    },
+    {
+      "epoch": 1.456973293768546,
+      "grad_norm": 0.34781596007553534,
+      "learning_rate": 2.8581898131183587e-05,
+      "loss": 0.4446,
+      "step": 1473
+    },
+    {
+      "epoch": 1.4579624134520277,
+      "grad_norm": 0.3118284461430052,
+      "learning_rate": 2.8563576401612313e-05,
+      "loss": 0.4084,
+      "step": 1474
+    },
+    {
+      "epoch": 1.4589515331355094,
+      "grad_norm": 0.3162361256418991,
+      "learning_rate": 2.8545254672041043e-05,
+      "loss": 0.4418,
+      "step": 1475
+    },
+    {
+      "epoch": 1.459940652818991,
+      "grad_norm": 0.38053633547795723,
+      "learning_rate": 2.852693294246977e-05,
+      "loss": 0.3583,
+      "step": 1476
+    },
+    {
+      "epoch": 1.4609297725024728,
+      "grad_norm": 0.30505081841976306,
+      "learning_rate": 2.85086112128985e-05,
+      "loss": 0.4295,
+      "step": 1477
+    },
+    {
+      "epoch": 1.4619188921859545,
+      "grad_norm": 0.3700760611954873,
+      "learning_rate": 2.8490289483327226e-05,
+      "loss": 0.4091,
+      "step": 1478
+    },
+    {
+      "epoch": 1.4629080118694362,
+      "grad_norm": 0.29239279326719575,
+      "learning_rate": 2.8471967753755955e-05,
+      "loss": 0.4535,
+      "step": 1479
+    },
+    {
+      "epoch": 1.463897131552918,
+      "grad_norm": 0.3096224600613784,
+      "learning_rate": 2.8453646024184682e-05,
+      "loss": 0.4517,
+      "step": 1480
+    },
+    {
+      "epoch": 1.4648862512363996,
+      "grad_norm": 0.29982859930862565,
+      "learning_rate": 2.8435324294613415e-05,
+      "loss": 0.3935,
+      "step": 1481
+    },
+    {
+      "epoch": 1.4658753709198813,
+      "grad_norm": 0.2972910923176461,
+      "learning_rate": 2.8417002565042138e-05,
+      "loss": 0.4544,
+      "step": 1482
+    },
+    {
+      "epoch": 1.466864490603363,
+      "grad_norm": 0.2926067116253536,
+      "learning_rate": 2.839868083547087e-05,
+      "loss": 0.415,
+      "step": 1483
+    },
+    {
+      "epoch": 1.4678536102868447,
+      "grad_norm": 0.3328585626789144,
+      "learning_rate": 2.83803591058996e-05,
+      "loss": 0.4579,
+      "step": 1484
+    },
+    {
+      "epoch": 1.4688427299703264,
+      "grad_norm": 0.3481307578459911,
+      "learning_rate": 2.8362037376328327e-05,
+      "loss": 0.4605,
+      "step": 1485
+    },
+    {
+      "epoch": 1.469831849653808,
+      "grad_norm": 0.31652707929635426,
+      "learning_rate": 2.8343715646757057e-05,
+      "loss": 0.4656,
+      "step": 1486
+    },
+    {
+      "epoch": 1.4708209693372898,
+      "grad_norm": 0.29953398282105115,
+      "learning_rate": 2.8325393917185784e-05,
+      "loss": 0.3825,
+      "step": 1487
+    },
+    {
+      "epoch": 1.4718100890207715,
+      "grad_norm": 0.3097931361873477,
+      "learning_rate": 2.8307072187614514e-05,
+      "loss": 0.3972,
+      "step": 1488
+    },
+    {
+      "epoch": 1.4727992087042532,
+      "grad_norm": 0.33071430642317845,
+      "learning_rate": 2.828875045804324e-05,
+      "loss": 0.4285,
+      "step": 1489
+    },
+    {
+      "epoch": 1.4737883283877349,
+      "grad_norm": 0.2859188800768853,
+      "learning_rate": 2.827042872847197e-05,
+      "loss": 0.418,
+      "step": 1490
+    },
+    {
+      "epoch": 1.4747774480712166,
+      "grad_norm": 0.2881058247271708,
+      "learning_rate": 2.8252106998900696e-05,
+      "loss": 0.4643,
+      "step": 1491
+    },
+    {
+      "epoch": 1.4757665677546983,
+      "grad_norm": 0.38974799158353357,
+      "learning_rate": 2.8233785269329426e-05,
+      "loss": 0.4169,
+      "step": 1492
+    },
+    {
+      "epoch": 1.47675568743818,
+      "grad_norm": 0.3275131922732535,
+      "learning_rate": 2.8215463539758152e-05,
+      "loss": 0.446,
+      "step": 1493
+    },
+    {
+      "epoch": 1.4777448071216617,
+      "grad_norm": 0.3299066979177711,
+      "learning_rate": 2.8197141810186882e-05,
+      "loss": 0.4617,
+      "step": 1494
+    },
+    {
+      "epoch": 1.4787339268051434,
+      "grad_norm": 0.364073130150921,
+      "learning_rate": 2.817882008061561e-05,
+      "loss": 0.5335,
+      "step": 1495
+    },
+    {
+      "epoch": 1.479723046488625,
+      "grad_norm": 0.3277284148931501,
+      "learning_rate": 2.816049835104434e-05,
+      "loss": 0.4456,
+      "step": 1496
+    },
+    {
+      "epoch": 1.4807121661721068,
+      "grad_norm": 0.43802618542564276,
+      "learning_rate": 2.814217662147307e-05,
+      "loss": 0.4724,
+      "step": 1497
+    },
+    {
+      "epoch": 1.4817012858555885,
+      "grad_norm": 0.2765090527927701,
+      "learning_rate": 2.8123854891901798e-05,
+      "loss": 0.3961,
+      "step": 1498
+    },
+    {
+      "epoch": 1.4826904055390702,
+      "grad_norm": 0.3416485689233916,
+      "learning_rate": 2.8105533162330528e-05,
+      "loss": 0.4166,
+      "step": 1499
+    },
+    {
+      "epoch": 1.4836795252225519,
+      "grad_norm": 0.37839623878952566,
+      "learning_rate": 2.8087211432759254e-05,
+      "loss": 0.4906,
+      "step": 1500
+    },
+    {
+      "epoch": 1.4846686449060336,
+      "grad_norm": 0.3518638957400736,
+      "learning_rate": 2.8068889703187984e-05,
+      "loss": 0.4069,
+      "step": 1501
+    },
+    {
+      "epoch": 1.4856577645895153,
+      "grad_norm": 0.33111446442715076,
+      "learning_rate": 2.805056797361671e-05,
+      "loss": 0.4293,
+      "step": 1502
+    },
+    {
+      "epoch": 1.486646884272997,
+      "grad_norm": 0.295453610129648,
+      "learning_rate": 2.803224624404544e-05,
+      "loss": 0.4292,
+      "step": 1503
+    },
+    {
+      "epoch": 1.4876360039564787,
+      "grad_norm": 0.2756874683521067,
+      "learning_rate": 2.8013924514474167e-05,
+      "loss": 0.4104,
+      "step": 1504
+    },
+    {
+      "epoch": 1.4886251236399604,
+      "grad_norm": 0.3798280905192116,
+      "learning_rate": 2.7995602784902896e-05,
+      "loss": 0.4483,
+      "step": 1505
+    },
+    {
+      "epoch": 1.489614243323442,
+      "grad_norm": 0.3414752636312388,
+      "learning_rate": 2.7977281055331623e-05,
+      "loss": 0.4637,
+      "step": 1506
+    },
+    {
+      "epoch": 1.4906033630069238,
+      "grad_norm": 0.35252371421106593,
+      "learning_rate": 2.7958959325760353e-05,
+      "loss": 0.4697,
+      "step": 1507
+    },
+    {
+      "epoch": 1.4915924826904057,
+      "grad_norm": 0.35553900870734595,
+      "learning_rate": 2.7940637596189086e-05,
+      "loss": 0.4427,
+      "step": 1508
+    },
+    {
+      "epoch": 1.4925816023738872,
+      "grad_norm": 0.27479809910211267,
+      "learning_rate": 2.792231586661781e-05,
+      "loss": 0.4329,
+      "step": 1509
+    },
+    {
+      "epoch": 1.493570722057369,
+      "grad_norm": 0.3169137202038055,
+      "learning_rate": 2.7903994137046542e-05,
+      "loss": 0.4504,
+      "step": 1510
+    },
+    {
+      "epoch": 1.4945598417408505,
+      "grad_norm": 0.3611773054393912,
+      "learning_rate": 2.7885672407475265e-05,
+      "loss": 0.3944,
+      "step": 1511
+    },
+    {
+      "epoch": 1.4955489614243325,
+      "grad_norm": 0.27848717086943314,
+      "learning_rate": 2.7867350677904e-05,
+      "loss": 0.4297,
+      "step": 1512
+    },
+    {
+      "epoch": 1.496538081107814,
+      "grad_norm": 0.2890433618635751,
+      "learning_rate": 2.784902894833272e-05,
+      "loss": 0.4256,
+      "step": 1513
+    },
+    {
+      "epoch": 1.4975272007912959,
+      "grad_norm": 0.29233252685415784,
+      "learning_rate": 2.7830707218761455e-05,
+      "loss": 0.5019,
+      "step": 1514
+    },
+    {
+      "epoch": 1.4985163204747773,
+      "grad_norm": 0.2808355518258454,
+      "learning_rate": 2.7812385489190178e-05,
+      "loss": 0.4533,
+      "step": 1515
+    },
+    {
+      "epoch": 1.4995054401582593,
+      "grad_norm": 0.26776681113551326,
+      "learning_rate": 2.779406375961891e-05,
+      "loss": 0.4154,
+      "step": 1516
+    },
+    {
+      "epoch": 1.5004945598417407,
+      "grad_norm": 0.318650708286605,
+      "learning_rate": 2.7775742030047637e-05,
+      "loss": 0.4203,
+      "step": 1517
+    },
+    {
+      "epoch": 1.5014836795252227,
+      "grad_norm": 0.26460015417559507,
+      "learning_rate": 2.7757420300476367e-05,
+      "loss": 0.4147,
+      "step": 1518
+    },
+    {
+      "epoch": 1.5024727992087041,
+      "grad_norm": 0.25425666410240005,
+      "learning_rate": 2.7739098570905093e-05,
+      "loss": 0.3944,
+      "step": 1519
+    },
+    {
+      "epoch": 1.503461918892186,
+      "grad_norm": 0.35103080270975756,
+      "learning_rate": 2.7720776841333823e-05,
+      "loss": 0.417,
+      "step": 1520
+    },
+    {
+      "epoch": 1.5044510385756675,
+      "grad_norm": 0.31288047385282897,
+      "learning_rate": 2.7702455111762553e-05,
+      "loss": 0.4376,
+      "step": 1521
+    },
+    {
+      "epoch": 1.5054401582591495,
+      "grad_norm": 0.3026304301870927,
+      "learning_rate": 2.768413338219128e-05,
+      "loss": 0.4141,
+      "step": 1522
+    },
+    {
+      "epoch": 1.506429277942631,
+      "grad_norm": 0.29749849928337113,
+      "learning_rate": 2.766581165262001e-05,
+      "loss": 0.4098,
+      "step": 1523
+    },
+    {
+      "epoch": 1.5074183976261128,
+      "grad_norm": 0.29038320359051056,
+      "learning_rate": 2.7647489923048736e-05,
+      "loss": 0.4431,
+      "step": 1524
+    },
+    {
+      "epoch": 1.5084075173095943,
+      "grad_norm": 0.2907976228534865,
+      "learning_rate": 2.7629168193477466e-05,
+      "loss": 0.4574,
+      "step": 1525
+    },
+    {
+      "epoch": 1.5093966369930762,
+      "grad_norm": 0.25183312695543336,
+      "learning_rate": 2.7610846463906192e-05,
+      "loss": 0.3744,
+      "step": 1526
+    },
+    {
+      "epoch": 1.5103857566765577,
+      "grad_norm": 0.2855912969143611,
+      "learning_rate": 2.7592524734334925e-05,
+      "loss": 0.4661,
+      "step": 1527
+    },
+    {
+      "epoch": 1.5113748763600396,
+      "grad_norm": 0.2732293198137018,
+      "learning_rate": 2.7574203004763648e-05,
+      "loss": 0.4355,
+      "step": 1528
+    },
+    {
+      "epoch": 1.5123639960435211,
+      "grad_norm": 0.3213776545316619,
+      "learning_rate": 2.755588127519238e-05,
+      "loss": 0.455,
+      "step": 1529
+    },
+    {
+      "epoch": 1.513353115727003,
+      "grad_norm": 0.29145234783713037,
+      "learning_rate": 2.7537559545621104e-05,
+      "loss": 0.4878,
+      "step": 1530
+    },
+    {
+      "epoch": 1.5143422354104845,
+      "grad_norm": 0.266071594136441,
+      "learning_rate": 2.7519237816049838e-05,
+      "loss": 0.4167,
+      "step": 1531
+    },
+    {
+      "epoch": 1.5153313550939664,
+      "grad_norm": 0.2731003398932023,
+      "learning_rate": 2.7500916086478567e-05,
+      "loss": 0.4768,
+      "step": 1532
+    },
+    {
+      "epoch": 1.516320474777448,
+      "grad_norm": 0.3097341831479785,
+      "learning_rate": 2.7482594356907294e-05,
+      "loss": 0.4474,
+      "step": 1533
+    },
+    {
+      "epoch": 1.5173095944609298,
+      "grad_norm": 0.34027538016329345,
+      "learning_rate": 2.7464272627336024e-05,
+      "loss": 0.4393,
+      "step": 1534
+    },
+    {
+      "epoch": 1.5182987141444113,
+      "grad_norm": 1.2356940989603606,
+      "learning_rate": 2.744595089776475e-05,
+      "loss": 0.4578,
+      "step": 1535
+    },
+    {
+      "epoch": 1.5192878338278932,
+      "grad_norm": 0.5459791028027153,
+      "learning_rate": 2.742762916819348e-05,
+      "loss": 0.4724,
+      "step": 1536
+    },
+    {
+      "epoch": 1.520276953511375,
+      "grad_norm": 0.31844571348235073,
+      "learning_rate": 2.7409307438622206e-05,
+      "loss": 0.4125,
+      "step": 1537
+    },
+    {
+      "epoch": 1.5212660731948566,
+      "grad_norm": 0.31015078316170824,
+      "learning_rate": 2.7390985709050936e-05,
+      "loss": 0.433,
+      "step": 1538
+    },
+    {
+      "epoch": 1.5222551928783383,
+      "grad_norm": 0.29396592937434984,
+      "learning_rate": 2.7372663979479662e-05,
+      "loss": 0.4376,
+      "step": 1539
+    },
+    {
+      "epoch": 1.52324431256182,
+      "grad_norm": 0.2911564550284436,
+      "learning_rate": 2.7354342249908392e-05,
+      "loss": 0.4112,
+      "step": 1540
+    },
+    {
+      "epoch": 1.5242334322453017,
+      "grad_norm": 0.3141108762736842,
+      "learning_rate": 2.733602052033712e-05,
+      "loss": 0.4333,
+      "step": 1541
+    },
+    {
+      "epoch": 1.5252225519287834,
+      "grad_norm": 0.28602791405930644,
+      "learning_rate": 2.731769879076585e-05,
+      "loss": 0.4535,
+      "step": 1542
+    },
+    {
+      "epoch": 1.5262116716122651,
+      "grad_norm": 0.2559604367618264,
+      "learning_rate": 2.7299377061194575e-05,
+      "loss": 0.4126,
+      "step": 1543
+    },
+    {
+      "epoch": 1.5272007912957468,
+      "grad_norm": 0.27275667409056814,
+      "learning_rate": 2.7281055331623305e-05,
+      "loss": 0.3398,
+      "step": 1544
+    },
+    {
+      "epoch": 1.5281899109792285,
+      "grad_norm": 0.27888080177130276,
+      "learning_rate": 2.7262733602052038e-05,
+      "loss": 0.396,
+      "step": 1545
+    },
+    {
+      "epoch": 1.5291790306627102,
+      "grad_norm": 0.29457676257202875,
+      "learning_rate": 2.7244411872480764e-05,
+      "loss": 0.4142,
+      "step": 1546
+    },
+    {
+      "epoch": 1.530168150346192,
+      "grad_norm": 0.2657669136258467,
+      "learning_rate": 2.7226090142909494e-05,
+      "loss": 0.3811,
+      "step": 1547
+    },
+    {
+      "epoch": 1.5311572700296736,
+      "grad_norm": 0.2787061006660649,
+      "learning_rate": 2.720776841333822e-05,
+      "loss": 0.3936,
+      "step": 1548
+    },
+    {
+      "epoch": 1.5321463897131553,
+      "grad_norm": 0.37452021826356185,
+      "learning_rate": 2.718944668376695e-05,
+      "loss": 0.4526,
+      "step": 1549
+    },
+    {
+      "epoch": 1.533135509396637,
+      "grad_norm": 0.24566974329827732,
+      "learning_rate": 2.7171124954195677e-05,
+      "loss": 0.3751,
+      "step": 1550
+    },
+    {
+      "epoch": 1.5341246290801187,
+      "grad_norm": 0.2757225886209204,
+      "learning_rate": 2.7152803224624407e-05,
+      "loss": 0.466,
+      "step": 1551
+    },
+    {
+      "epoch": 1.5351137487636004,
+      "grad_norm": 0.360202782237012,
+      "learning_rate": 2.7134481495053133e-05,
+      "loss": 0.5028,
+      "step": 1552
+    },
+    {
+      "epoch": 1.536102868447082,
+      "grad_norm": 0.28217047010151747,
+      "learning_rate": 2.7116159765481863e-05,
+      "loss": 0.4216,
+      "step": 1553
+    },
+    {
+      "epoch": 1.5370919881305638,
+      "grad_norm": 0.3369657062476821,
+      "learning_rate": 2.709783803591059e-05,
+      "loss": 0.4474,
+      "step": 1554
+    },
+    {
+      "epoch": 1.5380811078140455,
+      "grad_norm": 0.32643487155724593,
+      "learning_rate": 2.707951630633932e-05,
+      "loss": 0.4275,
+      "step": 1555
+    },
+    {
+      "epoch": 1.5390702274975272,
+      "grad_norm": 0.2817779215324222,
+      "learning_rate": 2.7061194576768052e-05,
+      "loss": 0.4371,
+      "step": 1556
+    },
+    {
+      "epoch": 1.540059347181009,
+      "grad_norm": 0.3063323029729112,
+      "learning_rate": 2.7042872847196775e-05,
+      "loss": 0.4084,
+      "step": 1557
+    },
+    {
+      "epoch": 1.5410484668644906,
+      "grad_norm": 0.26570368053936894,
+      "learning_rate": 2.702455111762551e-05,
+      "loss": 0.4377,
+      "step": 1558
+    },
+    {
+      "epoch": 1.5420375865479723,
+      "grad_norm": 0.28478093201712384,
+      "learning_rate": 2.700622938805423e-05,
+      "loss": 0.3711,
+      "step": 1559
+    },
+    {
+      "epoch": 1.543026706231454,
+      "grad_norm": 0.3527584738458635,
+      "learning_rate": 2.6987907658482965e-05,
+      "loss": 0.4467,
+      "step": 1560
+    },
+    {
+      "epoch": 1.5440158259149357,
+      "grad_norm": 0.2998161804452798,
+      "learning_rate": 2.6969585928911688e-05,
+      "loss": 0.4022,
+      "step": 1561
+    },
+    {
+      "epoch": 1.5450049455984174,
+      "grad_norm": 0.30287586677925865,
+      "learning_rate": 2.695126419934042e-05,
+      "loss": 0.4276,
+      "step": 1562
+    },
+    {
+      "epoch": 1.545994065281899,
+      "grad_norm": 0.29066288076793795,
+      "learning_rate": 2.6932942469769147e-05,
+      "loss": 0.4298,
+      "step": 1563
+    },
+    {
+      "epoch": 1.5469831849653808,
+      "grad_norm": 0.36634590156409147,
+      "learning_rate": 2.6914620740197877e-05,
+      "loss": 0.4333,
+      "step": 1564
+    },
+    {
+      "epoch": 1.5479723046488625,
+      "grad_norm": 0.3516684553628714,
+      "learning_rate": 2.6896299010626604e-05,
+      "loss": 0.474,
+      "step": 1565
+    },
+    {
+      "epoch": 1.5489614243323442,
+      "grad_norm": 0.3315473412552291,
+      "learning_rate": 2.6877977281055333e-05,
+      "loss": 0.4199,
+      "step": 1566
+    },
+    {
+      "epoch": 1.5499505440158259,
+      "grad_norm": 0.304443458898562,
+      "learning_rate": 2.685965555148406e-05,
+      "loss": 0.4685,
+      "step": 1567
+    },
+    {
+      "epoch": 1.5509396636993076,
+      "grad_norm": 0.34939766241844106,
+      "learning_rate": 2.684133382191279e-05,
+      "loss": 0.3867,
+      "step": 1568
+    },
+    {
+      "epoch": 1.5519287833827893,
+      "grad_norm": 0.3364698970448001,
+      "learning_rate": 2.682301209234152e-05,
+      "loss": 0.4314,
+      "step": 1569
+    },
+    {
+      "epoch": 1.552917903066271,
+      "grad_norm": 0.34050583019646397,
+      "learning_rate": 2.6804690362770246e-05,
+      "loss": 0.4741,
+      "step": 1570
+    },
+    {
+      "epoch": 1.5539070227497527,
+      "grad_norm": 0.26237037768227367,
+      "learning_rate": 2.6786368633198976e-05,
+      "loss": 0.4107,
+      "step": 1571
+    },
+    {
+      "epoch": 1.5548961424332344,
+      "grad_norm": 0.3461301127225192,
+      "learning_rate": 2.6768046903627702e-05,
+      "loss": 0.4541,
+      "step": 1572
+    },
+    {
+      "epoch": 1.555885262116716,
+      "grad_norm": 0.2511422428233448,
+      "learning_rate": 2.6749725174056435e-05,
+      "loss": 0.4286,
+      "step": 1573
+    },
+    {
+      "epoch": 1.5568743818001978,
+      "grad_norm": 0.6854382606786477,
+      "learning_rate": 2.6731403444485158e-05,
+      "loss": 0.5168,
+      "step": 1574
+    },
+    {
+      "epoch": 1.5578635014836797,
+      "grad_norm": 0.29587534482934763,
+      "learning_rate": 2.671308171491389e-05,
+      "loss": 0.4379,
+      "step": 1575
+    },
+    {
+      "epoch": 1.5588526211671612,
+      "grad_norm": 0.30965230188237924,
+      "learning_rate": 2.6694759985342614e-05,
+      "loss": 0.4372,
+      "step": 1576
+    },
+    {
+      "epoch": 1.559841740850643,
+      "grad_norm": 0.29630073934251444,
+      "learning_rate": 2.6676438255771348e-05,
+      "loss": 0.4082,
+      "step": 1577
+    },
+    {
+      "epoch": 1.5608308605341246,
+      "grad_norm": 0.3254718398772897,
+      "learning_rate": 2.665811652620007e-05,
+      "loss": 0.446,
+      "step": 1578
+    },
+    {
+      "epoch": 1.5618199802176065,
+      "grad_norm": 0.35742974851957743,
+      "learning_rate": 2.6639794796628804e-05,
+      "loss": 0.4064,
+      "step": 1579
+    },
+    {
+      "epoch": 1.562809099901088,
+      "grad_norm": 0.297566684364922,
+      "learning_rate": 2.6621473067057534e-05,
+      "loss": 0.3941,
+      "step": 1580
+    },
+    {
+      "epoch": 1.5637982195845699,
+      "grad_norm": 0.34824595226140137,
+      "learning_rate": 2.660315133748626e-05,
+      "loss": 0.4817,
+      "step": 1581
+    },
+    {
+      "epoch": 1.5647873392680514,
+      "grad_norm": 0.37702204793009536,
+      "learning_rate": 2.658482960791499e-05,
+      "loss": 0.4767,
+      "step": 1582
+    },
+    {
+      "epoch": 1.5657764589515333,
+      "grad_norm": 0.35461074860740005,
+      "learning_rate": 2.6566507878343716e-05,
+      "loss": 0.4296,
+      "step": 1583
+    },
+    {
+      "epoch": 1.5667655786350148,
+      "grad_norm": 0.3200818420727633,
+      "learning_rate": 2.6548186148772446e-05,
+      "loss": 0.381,
+      "step": 1584
+    },
+    {
+      "epoch": 1.5677546983184967,
+      "grad_norm": 0.425899730481969,
+      "learning_rate": 2.6529864419201173e-05,
+      "loss": 0.4158,
+      "step": 1585
+    },
+    {
+      "epoch": 1.5687438180019782,
+      "grad_norm": 1.1299623445949691,
+      "learning_rate": 2.6511542689629902e-05,
+      "loss": 0.3845,
+      "step": 1586
+    },
+    {
+      "epoch": 1.56973293768546,
+      "grad_norm": 0.2782420434679356,
+      "learning_rate": 2.649322096005863e-05,
+      "loss": 0.4361,
+      "step": 1587
+    },
+    {
+      "epoch": 1.5707220573689415,
+      "grad_norm": 0.3363493977754832,
+      "learning_rate": 2.647489923048736e-05,
+      "loss": 0.3991,
+      "step": 1588
+    },
+    {
+      "epoch": 1.5717111770524235,
+      "grad_norm": 0.3204275823714799,
+      "learning_rate": 2.6456577500916085e-05,
+      "loss": 0.5167,
+      "step": 1589
+    },
+    {
+      "epoch": 1.572700296735905,
+      "grad_norm": 0.3366195485123245,
+      "learning_rate": 2.6438255771344815e-05,
+      "loss": 0.4971,
+      "step": 1590
+    },
+    {
+      "epoch": 1.5736894164193869,
+      "grad_norm": 0.3067388676397528,
+      "learning_rate": 2.641993404177354e-05,
+      "loss": 0.4899,
+      "step": 1591
+    },
+    {
+      "epoch": 1.5746785361028683,
+      "grad_norm": 0.27590309427573345,
+      "learning_rate": 2.6401612312202274e-05,
+      "loss": 0.4055,
+      "step": 1592
+    },
+    {
+      "epoch": 1.5756676557863503,
+      "grad_norm": 0.3184897304138686,
+      "learning_rate": 2.6383290582631004e-05,
+      "loss": 0.4245,
+      "step": 1593
+    },
+    {
+      "epoch": 1.5766567754698317,
+      "grad_norm": 0.2753393949803328,
+      "learning_rate": 2.636496885305973e-05,
+      "loss": 0.4415,
+      "step": 1594
+    },
+    {
+      "epoch": 1.5776458951533137,
+      "grad_norm": 0.2696948403053459,
+      "learning_rate": 2.634664712348846e-05,
+      "loss": 0.4367,
+      "step": 1595
+    },
+    {
+      "epoch": 1.5786350148367951,
+      "grad_norm": 0.291084848882316,
+      "learning_rate": 2.6328325393917187e-05,
+      "loss": 0.4221,
+      "step": 1596
+    },
+    {
+      "epoch": 1.579624134520277,
+      "grad_norm": 0.29012800052048404,
+      "learning_rate": 2.6310003664345917e-05,
+      "loss": 0.407,
+      "step": 1597
+    },
+    {
+      "epoch": 1.5806132542037585,
+      "grad_norm": 0.32837973620046834,
+      "learning_rate": 2.6291681934774643e-05,
+      "loss": 0.4483,
+      "step": 1598
+    },
+    {
+      "epoch": 1.5816023738872405,
+      "grad_norm": 0.28011551139081314,
+      "learning_rate": 2.6273360205203373e-05,
+      "loss": 0.4711,
+      "step": 1599
+    },
+    {
+      "epoch": 1.582591493570722,
+      "grad_norm": 0.3282890674772606,
+      "learning_rate": 2.62550384756321e-05,
+      "loss": 0.3873,
+      "step": 1600
+    },
+    {
+      "epoch": 1.5835806132542039,
+      "grad_norm": 0.33875347479904816,
+      "learning_rate": 2.623671674606083e-05,
+      "loss": 0.4552,
+      "step": 1601
+    },
+    {
+      "epoch": 1.5845697329376853,
+      "grad_norm": 0.29670107000989415,
+      "learning_rate": 2.6218395016489556e-05,
+      "loss": 0.4795,
+      "step": 1602
+    },
+    {
+      "epoch": 1.5855588526211672,
+      "grad_norm": 0.265294490890646,
+      "learning_rate": 2.6200073286918285e-05,
+      "loss": 0.3973,
+      "step": 1603
+    },
+    {
+      "epoch": 1.5865479723046487,
+      "grad_norm": 0.33359772528312437,
+      "learning_rate": 2.618175155734702e-05,
+      "loss": 0.4843,
+      "step": 1604
+    },
+    {
+      "epoch": 1.5875370919881306,
+      "grad_norm": 0.4026752678559618,
+      "learning_rate": 2.616342982777574e-05,
+      "loss": 0.4425,
+      "step": 1605
+    },
+    {
+      "epoch": 1.5885262116716121,
+      "grad_norm": 0.3058367116238238,
+      "learning_rate": 2.6145108098204475e-05,
+      "loss": 0.4316,
+      "step": 1606
+    },
+    {
+      "epoch": 1.589515331355094,
+      "grad_norm": 0.30426608314843634,
+      "learning_rate": 2.6126786368633198e-05,
+      "loss": 0.4927,
+      "step": 1607
+    },
+    {
+      "epoch": 1.5905044510385755,
+      "grad_norm": 0.37917329704685165,
+      "learning_rate": 2.610846463906193e-05,
+      "loss": 0.5052,
+      "step": 1608
+    },
+    {
+      "epoch": 1.5914935707220574,
+      "grad_norm": 0.33134630981635044,
+      "learning_rate": 2.6090142909490657e-05,
+      "loss": 0.4995,
+      "step": 1609
+    },
+    {
+      "epoch": 1.592482690405539,
+      "grad_norm": 0.2991963652688359,
+      "learning_rate": 2.6071821179919387e-05,
+      "loss": 0.3927,
+      "step": 1610
+    },
+    {
+      "epoch": 1.5934718100890208,
+      "grad_norm": 0.3570948011063814,
+      "learning_rate": 2.6053499450348114e-05,
+      "loss": 0.4624,
+      "step": 1611
+    },
+    {
+      "epoch": 1.5944609297725023,
+      "grad_norm": 0.3199568594829708,
+      "learning_rate": 2.6035177720776843e-05,
+      "loss": 0.4605,
+      "step": 1612
+    },
+    {
+      "epoch": 1.5954500494559842,
+      "grad_norm": 0.33280196155936287,
+      "learning_rate": 2.601685599120557e-05,
+      "loss": 0.4746,
+      "step": 1613
+    },
+    {
+      "epoch": 1.596439169139466,
+      "grad_norm": 0.39417035896603214,
+      "learning_rate": 2.59985342616343e-05,
+      "loss": 0.4208,
+      "step": 1614
+    },
+    {
+      "epoch": 1.5974282888229476,
+      "grad_norm": 0.33066363581898495,
+      "learning_rate": 2.598021253206303e-05,
+      "loss": 0.504,
+      "step": 1615
+    },
+    {
+      "epoch": 1.5984174085064293,
+      "grad_norm": 0.3461460215044843,
+      "learning_rate": 2.5961890802491756e-05,
+      "loss": 0.4331,
+      "step": 1616
+    },
+    {
+      "epoch": 1.599406528189911,
+      "grad_norm": 0.3652275383465214,
+      "learning_rate": 2.5943569072920486e-05,
+      "loss": 0.4299,
+      "step": 1617
+    },
+    {
+      "epoch": 1.6003956478733927,
+      "grad_norm": 0.30120482055467496,
+      "learning_rate": 2.5925247343349212e-05,
+      "loss": 0.4453,
+      "step": 1618
+    },
+    {
+      "epoch": 1.6013847675568744,
+      "grad_norm": 0.30615378277352595,
+      "learning_rate": 2.5906925613777945e-05,
+      "loss": 0.4507,
+      "step": 1619
+    },
+    {
+      "epoch": 1.6023738872403561,
+      "grad_norm": 0.32283104348568503,
+      "learning_rate": 2.5888603884206668e-05,
+      "loss": 0.3887,
+      "step": 1620
+    },
+    {
+      "epoch": 1.6033630069238378,
+      "grad_norm": 0.38745300511964836,
+      "learning_rate": 2.58702821546354e-05,
+      "loss": 0.4475,
+      "step": 1621
+    },
+    {
+      "epoch": 1.6043521266073195,
+      "grad_norm": 0.29046077388027175,
+      "learning_rate": 2.5851960425064125e-05,
+      "loss": 0.4147,
+      "step": 1622
+    },
+    {
+      "epoch": 1.6053412462908012,
+      "grad_norm": 0.440454936747926,
+      "learning_rate": 2.5833638695492858e-05,
+      "loss": 0.4652,
+      "step": 1623
+    },
+    {
+      "epoch": 1.606330365974283,
+      "grad_norm": 0.6507104848117019,
+      "learning_rate": 2.581531696592158e-05,
+      "loss": 0.4032,
+      "step": 1624
+    },
+    {
+      "epoch": 1.6073194856577646,
+      "grad_norm": 0.3264866453501732,
+      "learning_rate": 2.5796995236350314e-05,
+      "loss": 0.4031,
+      "step": 1625
+    },
+    {
+      "epoch": 1.6083086053412463,
+      "grad_norm": 0.3023992104795454,
+      "learning_rate": 2.5778673506779037e-05,
+      "loss": 0.4203,
+      "step": 1626
+    },
+    {
+      "epoch": 1.609297725024728,
+      "grad_norm": 0.359657753419691,
+      "learning_rate": 2.576035177720777e-05,
+      "loss": 0.4624,
+      "step": 1627
+    },
+    {
+      "epoch": 1.6102868447082097,
+      "grad_norm": 0.3372351206140866,
+      "learning_rate": 2.57420300476365e-05,
+      "loss": 0.4056,
+      "step": 1628
+    },
+    {
+      "epoch": 1.6112759643916914,
+      "grad_norm": 0.30039862972108095,
+      "learning_rate": 2.5723708318065226e-05,
+      "loss": 0.4399,
+      "step": 1629
+    },
+    {
+      "epoch": 1.612265084075173,
+      "grad_norm": 0.30065834303511957,
+      "learning_rate": 2.5705386588493956e-05,
+      "loss": 0.4663,
+      "step": 1630
+    },
+    {
+      "epoch": 1.6132542037586548,
+      "grad_norm": 0.3410392017721668,
+      "learning_rate": 2.5687064858922683e-05,
+      "loss": 0.4458,
+      "step": 1631
+    },
+    {
+      "epoch": 1.6142433234421365,
+      "grad_norm": 0.31764577949936024,
+      "learning_rate": 2.5668743129351412e-05,
+      "loss": 0.4946,
+      "step": 1632
+    },
+    {
+      "epoch": 1.6152324431256182,
+      "grad_norm": 0.3139161488552942,
+      "learning_rate": 2.565042139978014e-05,
+      "loss": 0.45,
+      "step": 1633
+    },
+    {
+      "epoch": 1.6162215628091,
+      "grad_norm": 0.2838563703036641,
+      "learning_rate": 2.563209967020887e-05,
+      "loss": 0.3615,
+      "step": 1634
+    },
+    {
+      "epoch": 1.6172106824925816,
+      "grad_norm": 0.280910691075243,
+      "learning_rate": 2.5613777940637595e-05,
+      "loss": 0.3664,
+      "step": 1635
+    },
+    {
+      "epoch": 1.6181998021760633,
+      "grad_norm": 0.3442161215592867,
+      "learning_rate": 2.5595456211066325e-05,
+      "loss": 0.45,
+      "step": 1636
+    },
+    {
+      "epoch": 1.619188921859545,
+      "grad_norm": 0.2953185716419831,
+      "learning_rate": 2.557713448149505e-05,
+      "loss": 0.4172,
+      "step": 1637
+    },
+    {
+      "epoch": 1.6201780415430267,
+      "grad_norm": 0.2654726745970755,
+      "learning_rate": 2.5558812751923784e-05,
+      "loss": 0.3943,
+      "step": 1638
+    },
+    {
+      "epoch": 1.6211671612265084,
+      "grad_norm": 0.3200576472766354,
+      "learning_rate": 2.5540491022352514e-05,
+      "loss": 0.4265,
+      "step": 1639
+    },
+    {
+      "epoch": 1.62215628090999,
+      "grad_norm": 0.29628265645606083,
+      "learning_rate": 2.552216929278124e-05,
+      "loss": 0.4116,
+      "step": 1640
+    },
+    {
+      "epoch": 1.6231454005934718,
+      "grad_norm": 0.2841844484198089,
+      "learning_rate": 2.550384756320997e-05,
+      "loss": 0.44,
+      "step": 1641
+    },
+    {
+      "epoch": 1.6241345202769535,
+      "grad_norm": 0.2557201167123379,
+      "learning_rate": 2.5485525833638697e-05,
+      "loss": 0.3717,
+      "step": 1642
+    },
+    {
+      "epoch": 1.6251236399604352,
+      "grad_norm": 0.33250075592922157,
+      "learning_rate": 2.5467204104067427e-05,
+      "loss": 0.4857,
+      "step": 1643
+    },
+    {
+      "epoch": 1.6261127596439169,
+      "grad_norm": 0.2800845720099231,
+      "learning_rate": 2.5448882374496153e-05,
+      "loss": 0.3958,
+      "step": 1644
+    },
+    {
+      "epoch": 1.6271018793273986,
+      "grad_norm": 0.31456766982731915,
+      "learning_rate": 2.5430560644924883e-05,
+      "loss": 0.4571,
+      "step": 1645
+    },
+    {
+      "epoch": 1.6280909990108803,
+      "grad_norm": 0.277187648256086,
+      "learning_rate": 2.541223891535361e-05,
+      "loss": 0.3837,
+      "step": 1646
+    },
+    {
+      "epoch": 1.629080118694362,
+      "grad_norm": 0.26160671112794515,
+      "learning_rate": 2.539391718578234e-05,
+      "loss": 0.4197,
+      "step": 1647
+    },
+    {
+      "epoch": 1.6300692383778437,
+      "grad_norm": 0.3224446440259182,
+      "learning_rate": 2.5375595456211066e-05,
+      "loss": 0.44,
+      "step": 1648
+    },
+    {
+      "epoch": 1.6310583580613254,
+      "grad_norm": 0.2627299707116854,
+      "learning_rate": 2.5357273726639795e-05,
+      "loss": 0.4301,
+      "step": 1649
+    },
+    {
+      "epoch": 1.632047477744807,
+      "grad_norm": 0.2959085159607189,
+      "learning_rate": 2.5338951997068522e-05,
+      "loss": 0.4882,
+      "step": 1650
+    },
+    {
+      "epoch": 1.6330365974282888,
+      "grad_norm": 0.28662674449197795,
+      "learning_rate": 2.532063026749725e-05,
+      "loss": 0.4183,
+      "step": 1651
+    },
+    {
+      "epoch": 1.6340257171117705,
+      "grad_norm": 0.30546067103773544,
+      "learning_rate": 2.5302308537925985e-05,
+      "loss": 0.413,
+      "step": 1652
+    },
+    {
+      "epoch": 1.6350148367952522,
+      "grad_norm": 0.276370031240087,
+      "learning_rate": 2.5283986808354708e-05,
+      "loss": 0.4232,
+      "step": 1653
+    },
+    {
+      "epoch": 1.636003956478734,
+      "grad_norm": 0.2706115264434037,
+      "learning_rate": 2.526566507878344e-05,
+      "loss": 0.4293,
+      "step": 1654
+    },
+    {
+      "epoch": 1.6369930761622156,
+      "grad_norm": 0.30532651247215364,
+      "learning_rate": 2.5247343349212167e-05,
+      "loss": 0.4428,
+      "step": 1655
+    },
+    {
+      "epoch": 1.6379821958456975,
+      "grad_norm": 0.3187299875860798,
+      "learning_rate": 2.5229021619640897e-05,
+      "loss": 0.4933,
+      "step": 1656
+    },
+    {
+      "epoch": 1.638971315529179,
+      "grad_norm": 0.2719115674229351,
+      "learning_rate": 2.5210699890069624e-05,
+      "loss": 0.3899,
+      "step": 1657
+    },
+    {
+      "epoch": 1.6399604352126609,
+      "grad_norm": 0.27197612915344427,
+      "learning_rate": 2.5192378160498353e-05,
+      "loss": 0.3803,
+      "step": 1658
+    },
+    {
+      "epoch": 1.6409495548961424,
+      "grad_norm": 0.30000048728325507,
+      "learning_rate": 2.517405643092708e-05,
+      "loss": 0.4025,
+      "step": 1659
+    },
+    {
+      "epoch": 1.6419386745796243,
+      "grad_norm": 0.2644120267282455,
+      "learning_rate": 2.515573470135581e-05,
+      "loss": 0.3988,
+      "step": 1660
+    },
+    {
+      "epoch": 1.6429277942631058,
+      "grad_norm": 0.2741960980354914,
+      "learning_rate": 2.5137412971784536e-05,
+      "loss": 0.4587,
+      "step": 1661
+    },
+    {
+      "epoch": 1.6439169139465877,
+      "grad_norm": 0.2930096159187744,
+      "learning_rate": 2.5119091242213266e-05,
+      "loss": 0.4697,
+      "step": 1662
+    },
+    {
+      "epoch": 1.6449060336300692,
+      "grad_norm": 0.35317847881770087,
+      "learning_rate": 2.5100769512641996e-05,
+      "loss": 0.4459,
+      "step": 1663
+    },
+    {
+      "epoch": 1.645895153313551,
+      "grad_norm": 0.2870305099438775,
+      "learning_rate": 2.5082447783070722e-05,
+      "loss": 0.4329,
+      "step": 1664
+    },
+    {
+      "epoch": 1.6468842729970326,
+      "grad_norm": 0.31469807580481884,
+      "learning_rate": 2.5064126053499455e-05,
+      "loss": 0.4459,
+      "step": 1665
+    },
+    {
+      "epoch": 1.6478733926805145,
+      "grad_norm": 0.32969026971793497,
+      "learning_rate": 2.504580432392818e-05,
+      "loss": 0.4279,
+      "step": 1666
+    },
+    {
+      "epoch": 1.648862512363996,
+      "grad_norm": 0.331096333283008,
+      "learning_rate": 2.502748259435691e-05,
+      "loss": 0.4519,
+      "step": 1667
+    },
+    {
+      "epoch": 1.6498516320474779,
+      "grad_norm": 0.3441582936309551,
+      "learning_rate": 2.5009160864785635e-05,
+      "loss": 0.4669,
+      "step": 1668
+    },
+    {
+      "epoch": 1.6508407517309593,
+      "grad_norm": 0.3160027414594968,
+      "learning_rate": 2.4990839135214368e-05,
+      "loss": 0.4374,
+      "step": 1669
+    },
+    {
+      "epoch": 1.6518298714144413,
+      "grad_norm": 0.2995831571732668,
+      "learning_rate": 2.4972517405643094e-05,
+      "loss": 0.4105,
+      "step": 1670
+    },
+    {
+      "epoch": 1.6528189910979227,
+      "grad_norm": 0.4442058450146274,
+      "learning_rate": 2.4954195676071824e-05,
+      "loss": 0.4731,
+      "step": 1671
+    },
+    {
+      "epoch": 1.6538081107814047,
+      "grad_norm": 0.33416266456418664,
+      "learning_rate": 2.493587394650055e-05,
+      "loss": 0.4784,
+      "step": 1672
+    },
+    {
+      "epoch": 1.6547972304648861,
+      "grad_norm": 0.3146001816140936,
+      "learning_rate": 2.491755221692928e-05,
+      "loss": 0.4262,
+      "step": 1673
+    },
+    {
+      "epoch": 1.655786350148368,
+      "grad_norm": 0.3743726426973562,
+      "learning_rate": 2.4899230487358007e-05,
+      "loss": 0.522,
+      "step": 1674
+    },
+    {
+      "epoch": 1.6567754698318495,
+      "grad_norm": 0.37599411126000926,
+      "learning_rate": 2.4880908757786736e-05,
+      "loss": 0.4539,
+      "step": 1675
+    },
+    {
+      "epoch": 1.6577645895153315,
+      "grad_norm": 0.2971196338181042,
+      "learning_rate": 2.4862587028215463e-05,
+      "loss": 0.4425,
+      "step": 1676
+    },
+    {
+      "epoch": 1.658753709198813,
+      "grad_norm": 0.3185181610361161,
+      "learning_rate": 2.4844265298644193e-05,
+      "loss": 0.4499,
+      "step": 1677
+    },
+    {
+      "epoch": 1.6597428288822949,
+      "grad_norm": 0.34756786268548084,
+      "learning_rate": 2.4825943569072922e-05,
+      "loss": 0.4892,
+      "step": 1678
+    },
+    {
+      "epoch": 1.6607319485657763,
+      "grad_norm": 0.30828529971670104,
+      "learning_rate": 2.480762183950165e-05,
+      "loss": 0.408,
+      "step": 1679
+    },
+    {
+      "epoch": 1.6617210682492582,
+      "grad_norm": 0.3146532160887173,
+      "learning_rate": 2.478930010993038e-05,
+      "loss": 0.4639,
+      "step": 1680
+    },
+    {
+      "epoch": 1.6627101879327397,
+      "grad_norm": 0.30842421088723865,
+      "learning_rate": 2.477097838035911e-05,
+      "loss": 0.4362,
+      "step": 1681
+    },
+    {
+      "epoch": 1.6636993076162216,
+      "grad_norm": 0.3318964750018238,
+      "learning_rate": 2.4752656650787835e-05,
+      "loss": 0.4272,
+      "step": 1682
+    },
+    {
+      "epoch": 1.6646884272997031,
+      "grad_norm": 0.2719805738190358,
+      "learning_rate": 2.4734334921216565e-05,
+      "loss": 0.3888,
+      "step": 1683
+    },
+    {
+      "epoch": 1.665677546983185,
+      "grad_norm": 0.310777229938151,
+      "learning_rate": 2.4716013191645295e-05,
+      "loss": 0.4384,
+      "step": 1684
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.3443093523477439,
+      "learning_rate": 2.469769146207402e-05,
+      "loss": 0.4107,
+      "step": 1685
+    },
+    {
+      "epoch": 1.6676557863501484,
+      "grad_norm": 0.27732248345294597,
+      "learning_rate": 2.467936973250275e-05,
+      "loss": 0.4154,
+      "step": 1686
+    },
+    {
+      "epoch": 1.66864490603363,
+      "grad_norm": 0.3394675785807699,
+      "learning_rate": 2.4661048002931477e-05,
+      "loss": 0.5017,
+      "step": 1687
+    },
+    {
+      "epoch": 1.6696340257171118,
+      "grad_norm": 0.33530005890339165,
+      "learning_rate": 2.4642726273360207e-05,
+      "loss": 0.4788,
+      "step": 1688
+    },
+    {
+      "epoch": 1.6706231454005933,
+      "grad_norm": 0.3242349395154102,
+      "learning_rate": 2.4624404543788933e-05,
+      "loss": 0.4241,
+      "step": 1689
+    },
+    {
+      "epoch": 1.6716122650840752,
+      "grad_norm": 0.30760783104340705,
+      "learning_rate": 2.4606082814217663e-05,
+      "loss": 0.4403,
+      "step": 1690
+    },
+    {
+      "epoch": 1.6726013847675567,
+      "grad_norm": 0.33533199396862406,
+      "learning_rate": 2.458776108464639e-05,
+      "loss": 0.4568,
+      "step": 1691
+    },
+    {
+      "epoch": 1.6735905044510386,
+      "grad_norm": 0.26868937017240524,
+      "learning_rate": 2.456943935507512e-05,
+      "loss": 0.4603,
+      "step": 1692
+    },
+    {
+      "epoch": 1.6745796241345203,
+      "grad_norm": 0.27887094020722614,
+      "learning_rate": 2.455111762550385e-05,
+      "loss": 0.3977,
+      "step": 1693
+    },
+    {
+      "epoch": 1.675568743818002,
+      "grad_norm": 0.2952611325359872,
+      "learning_rate": 2.453279589593258e-05,
+      "loss": 0.4352,
+      "step": 1694
+    },
+    {
+      "epoch": 1.6765578635014837,
+      "grad_norm": 0.26972392989431493,
+      "learning_rate": 2.4514474166361305e-05,
+      "loss": 0.4404,
+      "step": 1695
+    },
+    {
+      "epoch": 1.6775469831849654,
+      "grad_norm": 0.26276099105184353,
+      "learning_rate": 2.4496152436790035e-05,
+      "loss": 0.4099,
+      "step": 1696
+    },
+    {
+      "epoch": 1.6785361028684471,
+      "grad_norm": 0.2834003933123841,
+      "learning_rate": 2.4477830707218762e-05,
+      "loss": 0.4044,
+      "step": 1697
+    },
+    {
+      "epoch": 1.6795252225519288,
+      "grad_norm": 0.28788506496878463,
+      "learning_rate": 2.445950897764749e-05,
+      "loss": 0.3827,
+      "step": 1698
+    },
+    {
+      "epoch": 1.6805143422354105,
+      "grad_norm": 0.31665468599783186,
+      "learning_rate": 2.4441187248076218e-05,
+      "loss": 0.4364,
+      "step": 1699
+    },
+    {
+      "epoch": 1.6815034619188922,
+      "grad_norm": 0.2578234089735481,
+      "learning_rate": 2.4422865518504948e-05,
+      "loss": 0.3777,
+      "step": 1700
+    },
+    {
+      "epoch": 1.682492581602374,
+      "grad_norm": 0.2979943738077502,
+      "learning_rate": 2.4404543788933678e-05,
+      "loss": 0.3914,
+      "step": 1701
+    },
+    {
+      "epoch": 1.6834817012858556,
+      "grad_norm": 3.020828917743508,
+      "learning_rate": 2.4386222059362404e-05,
+      "loss": 0.3896,
+      "step": 1702
+    },
+    {
+      "epoch": 1.6844708209693373,
+      "grad_norm": 0.3359718058064896,
+      "learning_rate": 2.4367900329791134e-05,
+      "loss": 0.4185,
+      "step": 1703
+    },
+    {
+      "epoch": 1.685459940652819,
+      "grad_norm": 0.2994177996182077,
+      "learning_rate": 2.434957860021986e-05,
+      "loss": 0.4086,
+      "step": 1704
+    },
+    {
+      "epoch": 1.6864490603363007,
+      "grad_norm": 0.2991038976312438,
+      "learning_rate": 2.433125687064859e-05,
+      "loss": 0.435,
+      "step": 1705
+    },
+    {
+      "epoch": 1.6874381800197824,
+      "grad_norm": 0.28661212810149267,
+      "learning_rate": 2.431293514107732e-05,
+      "loss": 0.4143,
+      "step": 1706
+    },
+    {
+      "epoch": 1.688427299703264,
+      "grad_norm": 0.37509122239648596,
+      "learning_rate": 2.429461341150605e-05,
+      "loss": 0.4484,
+      "step": 1707
+    },
+    {
+      "epoch": 1.6894164193867458,
+      "grad_norm": 0.27836601910888614,
+      "learning_rate": 2.4276291681934776e-05,
+      "loss": 0.4142,
+      "step": 1708
+    },
+    {
+      "epoch": 1.6904055390702275,
+      "grad_norm": 0.2782146839258225,
+      "learning_rate": 2.4257969952363506e-05,
+      "loss": 0.4015,
+      "step": 1709
+    },
+    {
+      "epoch": 1.6913946587537092,
+      "grad_norm": 0.313916326384413,
+      "learning_rate": 2.4239648222792232e-05,
+      "loss": 0.4149,
+      "step": 1710
+    },
+    {
+      "epoch": 1.692383778437191,
+      "grad_norm": 0.29787858936299944,
+      "learning_rate": 2.4221326493220962e-05,
+      "loss": 0.4177,
+      "step": 1711
+    },
+    {
+      "epoch": 1.6933728981206726,
+      "grad_norm": 0.2902230168937217,
+      "learning_rate": 2.420300476364969e-05,
+      "loss": 0.4459,
+      "step": 1712
+    },
+    {
+      "epoch": 1.6943620178041543,
+      "grad_norm": 0.2764799202205296,
+      "learning_rate": 2.4184683034078418e-05,
+      "loss": 0.4803,
+      "step": 1713
+    },
+    {
+      "epoch": 1.695351137487636,
+      "grad_norm": 0.3045330992531217,
+      "learning_rate": 2.4166361304507145e-05,
+      "loss": 0.4842,
+      "step": 1714
+    },
+    {
+      "epoch": 1.6963402571711177,
+      "grad_norm": 0.3029379637137786,
+      "learning_rate": 2.4148039574935874e-05,
+      "loss": 0.4554,
+      "step": 1715
+    },
+    {
+      "epoch": 1.6973293768545994,
+      "grad_norm": 0.28227347456024454,
+      "learning_rate": 2.41297178453646e-05,
+      "loss": 0.4703,
+      "step": 1716
+    },
+    {
+      "epoch": 1.698318496538081,
+      "grad_norm": 0.2943760092371024,
+      "learning_rate": 2.4111396115793334e-05,
+      "loss": 0.4384,
+      "step": 1717
+    },
+    {
+      "epoch": 1.6993076162215628,
+      "grad_norm": 0.2985609138611823,
+      "learning_rate": 2.409307438622206e-05,
+      "loss": 0.4287,
+      "step": 1718
+    },
+    {
+      "epoch": 1.7002967359050445,
+      "grad_norm": 0.26327465676428885,
+      "learning_rate": 2.407475265665079e-05,
+      "loss": 0.4177,
+      "step": 1719
+    },
+    {
+      "epoch": 1.7012858555885262,
+      "grad_norm": 0.3178230796077746,
+      "learning_rate": 2.4056430927079517e-05,
+      "loss": 0.4423,
+      "step": 1720
+    },
+    {
+      "epoch": 1.7022749752720079,
+      "grad_norm": 0.2814544147392274,
+      "learning_rate": 2.4038109197508247e-05,
+      "loss": 0.4657,
+      "step": 1721
+    },
+    {
+      "epoch": 1.7032640949554896,
+      "grad_norm": 0.3002239090911129,
+      "learning_rate": 2.4019787467936973e-05,
+      "loss": 0.4673,
+      "step": 1722
+    },
+    {
+      "epoch": 1.7042532146389713,
+      "grad_norm": 0.31781217494252273,
+      "learning_rate": 2.4001465738365703e-05,
+      "loss": 0.527,
+      "step": 1723
+    },
+    {
+      "epoch": 1.705242334322453,
+      "grad_norm": 0.3498801258975987,
+      "learning_rate": 2.3983144008794433e-05,
+      "loss": 0.4487,
+      "step": 1724
+    },
+    {
+      "epoch": 1.7062314540059347,
+      "grad_norm": 0.9087648133583388,
+      "learning_rate": 2.396482227922316e-05,
+      "loss": 0.3576,
+      "step": 1725
+    },
+    {
+      "epoch": 1.7072205736894164,
+      "grad_norm": 0.3719846637041946,
+      "learning_rate": 2.394650054965189e-05,
+      "loss": 0.4582,
+      "step": 1726
+    },
+    {
+      "epoch": 1.708209693372898,
+      "grad_norm": 0.35660510630355013,
+      "learning_rate": 2.3928178820080615e-05,
+      "loss": 0.4612,
+      "step": 1727
+    },
+    {
+      "epoch": 1.7091988130563798,
+      "grad_norm": 0.33981496021373514,
+      "learning_rate": 2.3909857090509345e-05,
+      "loss": 0.4273,
+      "step": 1728
+    },
+    {
+      "epoch": 1.7101879327398615,
+      "grad_norm": 0.3742138810240101,
+      "learning_rate": 2.3891535360938075e-05,
+      "loss": 0.4248,
+      "step": 1729
+    },
+    {
+      "epoch": 1.7111770524233432,
+      "grad_norm": 0.36187927647193935,
+      "learning_rate": 2.3873213631366805e-05,
+      "loss": 0.4379,
+      "step": 1730
+    },
+    {
+      "epoch": 1.712166172106825,
+      "grad_norm": 0.3017112790609432,
+      "learning_rate": 2.385489190179553e-05,
+      "loss": 0.4641,
+      "step": 1731
+    },
+    {
+      "epoch": 1.7131552917903066,
+      "grad_norm": 0.3798548828676941,
+      "learning_rate": 2.383657017222426e-05,
+      "loss": 0.4185,
+      "step": 1732
+    },
+    {
+      "epoch": 1.7141444114737885,
+      "grad_norm": 0.289413116025503,
+      "learning_rate": 2.3818248442652987e-05,
+      "loss": 0.3906,
+      "step": 1733
+    },
+    {
+      "epoch": 1.71513353115727,
+      "grad_norm": 0.32233281303153377,
+      "learning_rate": 2.3799926713081717e-05,
+      "loss": 0.4263,
+      "step": 1734
+    },
+    {
+      "epoch": 1.7161226508407519,
+      "grad_norm": 0.3006262895335519,
+      "learning_rate": 2.3781604983510443e-05,
+      "loss": 0.4834,
+      "step": 1735
+    },
+    {
+      "epoch": 1.7171117705242334,
+      "grad_norm": 0.2618376951089462,
+      "learning_rate": 2.3763283253939173e-05,
+      "loss": 0.3999,
+      "step": 1736
+    },
+    {
+      "epoch": 1.7181008902077153,
+      "grad_norm": 0.29424619057293433,
+      "learning_rate": 2.37449615243679e-05,
+      "loss": 0.4073,
+      "step": 1737
+    },
+    {
+      "epoch": 1.7190900098911968,
+      "grad_norm": 0.3246152750212638,
+      "learning_rate": 2.372663979479663e-05,
+      "loss": 0.4299,
+      "step": 1738
+    },
+    {
+      "epoch": 1.7200791295746787,
+      "grad_norm": 0.31160852845216713,
+      "learning_rate": 2.3708318065225356e-05,
+      "loss": 0.3979,
+      "step": 1739
+    },
+    {
+      "epoch": 1.7210682492581602,
+      "grad_norm": 0.2755114326711097,
+      "learning_rate": 2.3689996335654086e-05,
+      "loss": 0.4641,
+      "step": 1740
+    },
+    {
+      "epoch": 1.722057368941642,
+      "grad_norm": 0.30346825870958033,
+      "learning_rate": 2.3671674606082816e-05,
+      "loss": 0.4408,
+      "step": 1741
+    },
+    {
+      "epoch": 1.7230464886251236,
+      "grad_norm": 0.30129194966583844,
+      "learning_rate": 2.3653352876511545e-05,
+      "loss": 0.3837,
+      "step": 1742
+    },
+    {
+      "epoch": 1.7240356083086055,
+      "grad_norm": 0.2775790052611303,
+      "learning_rate": 2.3635031146940272e-05,
+      "loss": 0.4221,
+      "step": 1743
+    },
+    {
+      "epoch": 1.725024727992087,
+      "grad_norm": 0.265693373028082,
+      "learning_rate": 2.3616709417369e-05,
+      "loss": 0.4168,
+      "step": 1744
+    },
+    {
+      "epoch": 1.7260138476755689,
+      "grad_norm": 0.3022427971745998,
+      "learning_rate": 2.3598387687797728e-05,
+      "loss": 0.4021,
+      "step": 1745
+    },
+    {
+      "epoch": 1.7270029673590503,
+      "grad_norm": 0.308247068877542,
+      "learning_rate": 2.3580065958226458e-05,
+      "loss": 0.4473,
+      "step": 1746
+    },
+    {
+      "epoch": 1.7279920870425323,
+      "grad_norm": 0.29663390988995886,
+      "learning_rate": 2.3561744228655188e-05,
+      "loss": 0.4039,
+      "step": 1747
+    },
+    {
+      "epoch": 1.7289812067260137,
+      "grad_norm": 0.2736150704186472,
+      "learning_rate": 2.3543422499083914e-05,
+      "loss": 0.4025,
+      "step": 1748
+    },
+    {
+      "epoch": 1.7299703264094957,
+      "grad_norm": 2.823911700520109,
+      "learning_rate": 2.3525100769512644e-05,
+      "loss": 0.475,
+      "step": 1749
+    },
+    {
+      "epoch": 1.7309594460929771,
+      "grad_norm": 0.34298388253711626,
+      "learning_rate": 2.350677903994137e-05,
+      "loss": 0.4595,
+      "step": 1750
+    },
+    {
+      "epoch": 1.731948565776459,
+      "grad_norm": 0.30530109475341805,
+      "learning_rate": 2.34884573103701e-05,
+      "loss": 0.4368,
+      "step": 1751
+    },
+    {
+      "epoch": 1.7329376854599405,
+      "grad_norm": 0.32890647032684833,
+      "learning_rate": 2.347013558079883e-05,
+      "loss": 0.4635,
+      "step": 1752
+    },
+    {
+      "epoch": 1.7339268051434225,
+      "grad_norm": 0.2888669372975163,
+      "learning_rate": 2.345181385122756e-05,
+      "loss": 0.3912,
+      "step": 1753
+    },
+    {
+      "epoch": 1.734915924826904,
+      "grad_norm": 0.29633911242582683,
+      "learning_rate": 2.3433492121656286e-05,
+      "loss": 0.3674,
+      "step": 1754
+    },
+    {
+      "epoch": 1.7359050445103859,
+      "grad_norm": 0.26951749995769536,
+      "learning_rate": 2.3415170392085016e-05,
+      "loss": 0.4199,
+      "step": 1755
+    },
+    {
+      "epoch": 1.7368941641938673,
+      "grad_norm": 0.3749803881291908,
+      "learning_rate": 2.3396848662513742e-05,
+      "loss": 0.483,
+      "step": 1756
+    },
+    {
+      "epoch": 1.7378832838773492,
+      "grad_norm": 0.32853671133944534,
+      "learning_rate": 2.3378526932942472e-05,
+      "loss": 0.4501,
+      "step": 1757
+    },
+    {
+      "epoch": 1.7388724035608307,
+      "grad_norm": 0.3443696483010131,
+      "learning_rate": 2.33602052033712e-05,
+      "loss": 0.499,
+      "step": 1758
+    },
+    {
+      "epoch": 1.7398615232443126,
+      "grad_norm": 0.3145778345363139,
+      "learning_rate": 2.334188347379993e-05,
+      "loss": 0.3949,
+      "step": 1759
+    },
+    {
+      "epoch": 1.7408506429277941,
+      "grad_norm": 0.319931054528546,
+      "learning_rate": 2.3323561744228655e-05,
+      "loss": 0.3834,
+      "step": 1760
+    },
+    {
+      "epoch": 1.741839762611276,
+      "grad_norm": 0.2862423758324553,
+      "learning_rate": 2.3305240014657385e-05,
+      "loss": 0.4465,
+      "step": 1761
+    },
+    {
+      "epoch": 1.7428288822947575,
+      "grad_norm": 0.30485977879966464,
+      "learning_rate": 2.328691828508611e-05,
+      "loss": 0.401,
+      "step": 1762
+    },
+    {
+      "epoch": 1.7438180019782394,
+      "grad_norm": 0.3165930256651749,
+      "learning_rate": 2.326859655551484e-05,
+      "loss": 0.4678,
+      "step": 1763
+    },
+    {
+      "epoch": 1.744807121661721,
+      "grad_norm": 0.29246729099227736,
+      "learning_rate": 2.325027482594357e-05,
+      "loss": 0.3993,
+      "step": 1764
+    },
+    {
+      "epoch": 1.7457962413452028,
+      "grad_norm": 0.31561117841951253,
+      "learning_rate": 2.32319530963723e-05,
+      "loss": 0.4925,
+      "step": 1765
+    },
+    {
+      "epoch": 1.7467853610286843,
+      "grad_norm": 0.2888481565602988,
+      "learning_rate": 2.3213631366801027e-05,
+      "loss": 0.4381,
+      "step": 1766
+    },
+    {
+      "epoch": 1.7477744807121662,
+      "grad_norm": 0.2868705156659709,
+      "learning_rate": 2.3195309637229757e-05,
+      "loss": 0.3683,
+      "step": 1767
+    },
+    {
+      "epoch": 1.7487636003956477,
+      "grad_norm": 0.3782601564035021,
+      "learning_rate": 2.3176987907658483e-05,
+      "loss": 0.4532,
+      "step": 1768
+    },
+    {
+      "epoch": 1.7497527200791296,
+      "grad_norm": 0.2704871280008665,
+      "learning_rate": 2.3158666178087213e-05,
+      "loss": 0.4201,
+      "step": 1769
+    },
+    {
+      "epoch": 1.7507418397626113,
+      "grad_norm": 0.3875138252219907,
+      "learning_rate": 2.3140344448515943e-05,
+      "loss": 0.4805,
+      "step": 1770
+    },
+    {
+      "epoch": 1.751730959446093,
+      "grad_norm": 0.5492745343372892,
+      "learning_rate": 2.312202271894467e-05,
+      "loss": 0.4156,
+      "step": 1771
+    },
+    {
+      "epoch": 1.7527200791295747,
+      "grad_norm": 0.3204756187656851,
+      "learning_rate": 2.31037009893734e-05,
+      "loss": 0.4988,
+      "step": 1772
+    },
+    {
+      "epoch": 1.7537091988130564,
+      "grad_norm": 0.35278429265769184,
+      "learning_rate": 2.3085379259802125e-05,
+      "loss": 0.4668,
+      "step": 1773
+    },
+    {
+      "epoch": 1.7546983184965381,
+      "grad_norm": 0.2818358328142832,
+      "learning_rate": 2.3067057530230855e-05,
+      "loss": 0.4493,
+      "step": 1774
+    },
+    {
+      "epoch": 1.7556874381800198,
+      "grad_norm": 0.3770205606309304,
+      "learning_rate": 2.304873580065958e-05,
+      "loss": 0.4706,
+      "step": 1775
+    },
+    {
+      "epoch": 1.7566765578635015,
+      "grad_norm": 0.2887530974485894,
+      "learning_rate": 2.3030414071088315e-05,
+      "loss": 0.4208,
+      "step": 1776
+    },
+    {
+      "epoch": 1.7576656775469832,
+      "grad_norm": 0.38165004798947394,
+      "learning_rate": 2.301209234151704e-05,
+      "loss": 0.4299,
+      "step": 1777
+    },
+    {
+      "epoch": 1.758654797230465,
+      "grad_norm": 0.30821868136973407,
+      "learning_rate": 2.299377061194577e-05,
+      "loss": 0.4569,
+      "step": 1778
+    },
+    {
+      "epoch": 1.7596439169139466,
+      "grad_norm": 0.24167543739660413,
+      "learning_rate": 2.2975448882374497e-05,
+      "loss": 0.3777,
+      "step": 1779
+    },
+    {
+      "epoch": 1.7606330365974283,
+      "grad_norm": 0.40409075639658476,
+      "learning_rate": 2.2957127152803227e-05,
+      "loss": 0.4594,
+      "step": 1780
+    },
+    {
+      "epoch": 1.76162215628091,
+      "grad_norm": 0.3254914242691898,
+      "learning_rate": 2.2938805423231954e-05,
+      "loss": 0.4206,
+      "step": 1781
+    },
+    {
+      "epoch": 1.7626112759643917,
+      "grad_norm": 2.509631711790463,
+      "learning_rate": 2.2920483693660683e-05,
+      "loss": 0.4511,
+      "step": 1782
+    },
+    {
+      "epoch": 1.7636003956478734,
+      "grad_norm": 0.3808689413851784,
+      "learning_rate": 2.290216196408941e-05,
+      "loss": 0.4229,
+      "step": 1783
+    },
+    {
+      "epoch": 1.764589515331355,
+      "grad_norm": 0.318175279111379,
+      "learning_rate": 2.288384023451814e-05,
+      "loss": 0.4416,
+      "step": 1784
+    },
+    {
+      "epoch": 1.7655786350148368,
+      "grad_norm": 0.26964885093490887,
+      "learning_rate": 2.2865518504946866e-05,
+      "loss": 0.4192,
+      "step": 1785
+    },
+    {
+      "epoch": 1.7665677546983185,
+      "grad_norm": 0.2632332792155176,
+      "learning_rate": 2.2847196775375596e-05,
+      "loss": 0.4172,
+      "step": 1786
+    },
+    {
+      "epoch": 1.7675568743818002,
+      "grad_norm": 0.4047127793846293,
+      "learning_rate": 2.2828875045804322e-05,
+      "loss": 0.4712,
+      "step": 1787
+    },
+    {
+      "epoch": 1.768545994065282,
+      "grad_norm": 0.27723739287491106,
+      "learning_rate": 2.2810553316233055e-05,
+      "loss": 0.4164,
+      "step": 1788
+    },
+    {
+      "epoch": 1.7695351137487636,
+      "grad_norm": 0.3030859442151421,
+      "learning_rate": 2.2792231586661782e-05,
+      "loss": 0.3961,
+      "step": 1789
+    },
+    {
+      "epoch": 1.7705242334322453,
+      "grad_norm": 0.8577754720604038,
+      "learning_rate": 2.277390985709051e-05,
+      "loss": 0.504,
+      "step": 1790
+    },
+    {
+      "epoch": 1.771513353115727,
+      "grad_norm": 0.3242162051396364,
+      "learning_rate": 2.2755588127519238e-05,
+      "loss": 0.4342,
+      "step": 1791
+    },
+    {
+      "epoch": 1.7725024727992087,
+      "grad_norm": 0.3408702815957067,
+      "learning_rate": 2.2737266397947968e-05,
+      "loss": 0.5057,
+      "step": 1792
+    },
+    {
+      "epoch": 1.7734915924826904,
+      "grad_norm": 0.3542183072173877,
+      "learning_rate": 2.2718944668376698e-05,
+      "loss": 0.4232,
+      "step": 1793
+    },
+    {
+      "epoch": 1.774480712166172,
+      "grad_norm": 0.4235624797543405,
+      "learning_rate": 2.2700622938805424e-05,
+      "loss": 0.437,
+      "step": 1794
+    },
+    {
+      "epoch": 1.7754698318496538,
+      "grad_norm": 0.2789629789323769,
+      "learning_rate": 2.2682301209234154e-05,
+      "loss": 0.4422,
+      "step": 1795
+    },
+    {
+      "epoch": 1.7764589515331355,
+      "grad_norm": 0.31066306585084447,
+      "learning_rate": 2.266397947966288e-05,
+      "loss": 0.4424,
+      "step": 1796
+    },
+    {
+      "epoch": 1.7774480712166172,
+      "grad_norm": 0.33933834935357965,
+      "learning_rate": 2.264565775009161e-05,
+      "loss": 0.4203,
+      "step": 1797
+    },
+    {
+      "epoch": 1.7784371909000989,
+      "grad_norm": 0.32712238177562997,
+      "learning_rate": 2.2627336020520337e-05,
+      "loss": 0.4319,
+      "step": 1798
+    },
+    {
+      "epoch": 1.7794263105835806,
+      "grad_norm": 0.3040834782849549,
+      "learning_rate": 2.2609014290949066e-05,
+      "loss": 0.4453,
+      "step": 1799
+    },
+    {
+      "epoch": 1.7804154302670623,
+      "grad_norm": 0.30022108803178027,
+      "learning_rate": 2.2590692561377796e-05,
+      "loss": 0.4445,
+      "step": 1800
+    },
+    {
+      "epoch": 1.781404549950544,
+      "grad_norm": 0.3088237766977971,
+      "learning_rate": 2.2572370831806526e-05,
+      "loss": 0.425,
+      "step": 1801
+    },
+    {
+      "epoch": 1.7823936696340257,
+      "grad_norm": 0.25065884893036056,
+      "learning_rate": 2.2554049102235252e-05,
+      "loss": 0.4144,
+      "step": 1802
+    },
+    {
+      "epoch": 1.7833827893175074,
+      "grad_norm": 0.2975123258434565,
+      "learning_rate": 2.2535727372663982e-05,
+      "loss": 0.4606,
+      "step": 1803
+    },
+    {
+      "epoch": 1.784371909000989,
+      "grad_norm": 3.0916522188183118,
+      "learning_rate": 2.251740564309271e-05,
+      "loss": 0.5796,
+      "step": 1804
+    },
+    {
+      "epoch": 1.7853610286844708,
+      "grad_norm": 0.4161541547275752,
+      "learning_rate": 2.249908391352144e-05,
+      "loss": 0.4224,
+      "step": 1805
+    },
+    {
+      "epoch": 1.7863501483679525,
+      "grad_norm": 0.4190616672823314,
+      "learning_rate": 2.2480762183950165e-05,
+      "loss": 0.4956,
+      "step": 1806
+    },
+    {
+      "epoch": 1.7873392680514342,
+      "grad_norm": 0.7468045943261583,
+      "learning_rate": 2.2462440454378895e-05,
+      "loss": 0.506,
+      "step": 1807
+    },
+    {
+      "epoch": 1.7883283877349159,
+      "grad_norm": 0.39878637991681054,
+      "learning_rate": 2.244411872480762e-05,
+      "loss": 0.4582,
+      "step": 1808
+    },
+    {
+      "epoch": 1.7893175074183976,
+      "grad_norm": 0.3939410722134704,
+      "learning_rate": 2.242579699523635e-05,
+      "loss": 0.4213,
+      "step": 1809
+    },
+    {
+      "epoch": 1.7903066271018795,
+      "grad_norm": 0.35893540238509397,
+      "learning_rate": 2.2407475265665077e-05,
+      "loss": 0.4452,
+      "step": 1810
+    },
+    {
+      "epoch": 1.791295746785361,
+      "grad_norm": 0.3179290020697894,
+      "learning_rate": 2.2389153536093807e-05,
+      "loss": 0.4307,
+      "step": 1811
+    },
+    {
+      "epoch": 1.7922848664688429,
+      "grad_norm": 0.37222861438373844,
+      "learning_rate": 2.2370831806522537e-05,
+      "loss": 0.4481,
+      "step": 1812
+    },
+    {
+      "epoch": 1.7932739861523244,
+      "grad_norm": 0.4758358116814864,
+      "learning_rate": 2.2352510076951267e-05,
+      "loss": 0.4565,
+      "step": 1813
+    },
+    {
+      "epoch": 1.7942631058358063,
+      "grad_norm": 0.30087252373022794,
+      "learning_rate": 2.2334188347379993e-05,
+      "loss": 0.4506,
+      "step": 1814
+    },
+    {
+      "epoch": 1.7952522255192878,
+      "grad_norm": 0.38966070557634785,
+      "learning_rate": 2.2315866617808723e-05,
+      "loss": 0.4129,
+      "step": 1815
+    },
+    {
+      "epoch": 1.7962413452027697,
+      "grad_norm": 0.4977116302980233,
+      "learning_rate": 2.229754488823745e-05,
+      "loss": 0.4218,
+      "step": 1816
+    },
+    {
+      "epoch": 1.7972304648862512,
+      "grad_norm": 0.3265049192175431,
+      "learning_rate": 2.227922315866618e-05,
+      "loss": 0.418,
+      "step": 1817
+    },
+    {
+      "epoch": 1.798219584569733,
+      "grad_norm": 0.3832944235691656,
+      "learning_rate": 2.226090142909491e-05,
+      "loss": 0.4684,
+      "step": 1818
+    },
+    {
+      "epoch": 1.7992087042532146,
+      "grad_norm": 0.2958689650940704,
+      "learning_rate": 2.2242579699523635e-05,
+      "loss": 0.4074,
+      "step": 1819
+    },
+    {
+      "epoch": 1.8001978239366965,
+      "grad_norm": 0.37038788365942676,
+      "learning_rate": 2.2224257969952365e-05,
+      "loss": 0.4259,
+      "step": 1820
+    },
+    {
+      "epoch": 1.801186943620178,
+      "grad_norm": 0.3553791418334394,
+      "learning_rate": 2.220593624038109e-05,
+      "loss": 0.4669,
+      "step": 1821
+    },
+    {
+      "epoch": 1.8021760633036599,
+      "grad_norm": 0.29718199661262223,
+      "learning_rate": 2.218761451080982e-05,
+      "loss": 0.4221,
+      "step": 1822
+    },
+    {
+      "epoch": 1.8031651829871413,
+      "grad_norm": 0.3359429382037479,
+      "learning_rate": 2.2169292781238548e-05,
+      "loss": 0.397,
+      "step": 1823
+    },
+    {
+      "epoch": 1.8041543026706233,
+      "grad_norm": 0.3048041433497428,
+      "learning_rate": 2.215097105166728e-05,
+      "loss": 0.4191,
+      "step": 1824
+    },
+    {
+      "epoch": 1.8051434223541047,
+      "grad_norm": 3.3517379572451342,
+      "learning_rate": 2.2132649322096007e-05,
+      "loss": 0.4819,
+      "step": 1825
+    },
+    {
+      "epoch": 1.8061325420375867,
+      "grad_norm": 0.2867123953220445,
+      "learning_rate": 2.2114327592524737e-05,
+      "loss": 0.4172,
+      "step": 1826
+    },
+    {
+      "epoch": 1.8071216617210681,
+      "grad_norm": 0.2871478819593749,
+      "learning_rate": 2.2096005862953464e-05,
+      "loss": 0.3971,
+      "step": 1827
+    },
+    {
+      "epoch": 1.80811078140455,
+      "grad_norm": 0.3245949025237984,
+      "learning_rate": 2.2077684133382193e-05,
+      "loss": 0.4968,
+      "step": 1828
+    },
+    {
+      "epoch": 1.8090999010880315,
+      "grad_norm": 25.78091323351991,
+      "learning_rate": 2.205936240381092e-05,
+      "loss": 0.693,
+      "step": 1829
+    },
+    {
+      "epoch": 1.8100890207715135,
+      "grad_norm": 0.49724123565643946,
+      "learning_rate": 2.204104067423965e-05,
+      "loss": 0.4492,
+      "step": 1830
+    },
+    {
+      "epoch": 1.811078140454995,
+      "grad_norm": 0.3303838283801352,
+      "learning_rate": 2.2022718944668376e-05,
+      "loss": 0.4216,
+      "step": 1831
+    },
+    {
+      "epoch": 1.8120672601384769,
+      "grad_norm": 0.3510098031005022,
+      "learning_rate": 2.2004397215097106e-05,
+      "loss": 0.4758,
+      "step": 1832
+    },
+    {
+      "epoch": 1.8130563798219583,
+      "grad_norm": 0.2829388974092525,
+      "learning_rate": 2.1986075485525832e-05,
+      "loss": 0.4332,
+      "step": 1833
+    },
+    {
+      "epoch": 1.8140454995054403,
+      "grad_norm": 0.29997263346647157,
+      "learning_rate": 2.1967753755954562e-05,
+      "loss": 0.4289,
+      "step": 1834
+    },
+    {
+      "epoch": 1.8150346191889217,
+      "grad_norm": 0.31669510053999383,
+      "learning_rate": 2.1949432026383292e-05,
+      "loss": 0.4411,
+      "step": 1835
+    },
+    {
+      "epoch": 1.8160237388724036,
+      "grad_norm": 0.29684511655355245,
+      "learning_rate": 2.1931110296812022e-05,
+      "loss": 0.407,
+      "step": 1836
+    },
+    {
+      "epoch": 1.8170128585558851,
+      "grad_norm": 0.6958017772881732,
+      "learning_rate": 2.1912788567240748e-05,
+      "loss": 0.4468,
+      "step": 1837
+    },
+    {
+      "epoch": 1.818001978239367,
+      "grad_norm": 0.336206373063671,
+      "learning_rate": 2.1894466837669478e-05,
+      "loss": 0.4296,
+      "step": 1838
+    },
+    {
+      "epoch": 1.8189910979228485,
+      "grad_norm": 0.38504542640419415,
+      "learning_rate": 2.1876145108098204e-05,
+      "loss": 0.4059,
+      "step": 1839
+    },
+    {
+      "epoch": 1.8199802176063304,
+      "grad_norm": 0.32789250753473015,
+      "learning_rate": 2.1857823378526934e-05,
+      "loss": 0.4773,
+      "step": 1840
+    },
+    {
+      "epoch": 1.820969337289812,
+      "grad_norm": 0.3200012365592021,
+      "learning_rate": 2.1839501648955664e-05,
+      "loss": 0.3923,
+      "step": 1841
+    },
+    {
+      "epoch": 1.8219584569732938,
+      "grad_norm": 0.45827677015168833,
+      "learning_rate": 2.182117991938439e-05,
+      "loss": 0.4699,
+      "step": 1842
+    },
+    {
+      "epoch": 1.8229475766567753,
+      "grad_norm": 0.30103049669909526,
+      "learning_rate": 2.180285818981312e-05,
+      "loss": 0.4359,
+      "step": 1843
+    },
+    {
+      "epoch": 1.8239366963402572,
+      "grad_norm": 0.297495772208957,
+      "learning_rate": 2.1784536460241847e-05,
+      "loss": 0.4252,
+      "step": 1844
+    },
+    {
+      "epoch": 1.8249258160237387,
+      "grad_norm": 0.3680654701793383,
+      "learning_rate": 2.1766214730670576e-05,
+      "loss": 0.458,
+      "step": 1845
+    },
+    {
+      "epoch": 1.8259149357072206,
+      "grad_norm": 0.3403708954708426,
+      "learning_rate": 2.1747893001099303e-05,
+      "loss": 0.444,
+      "step": 1846
+    },
+    {
+      "epoch": 1.826904055390702,
+      "grad_norm": 0.2803891604745382,
+      "learning_rate": 2.1729571271528033e-05,
+      "loss": 0.4229,
+      "step": 1847
+    },
+    {
+      "epoch": 1.827893175074184,
+      "grad_norm": 0.2997125783852745,
+      "learning_rate": 2.1711249541956762e-05,
+      "loss": 0.4057,
+      "step": 1848
+    },
+    {
+      "epoch": 1.8288822947576657,
+      "grad_norm": 0.3371508195438753,
+      "learning_rate": 2.1692927812385492e-05,
+      "loss": 0.4581,
+      "step": 1849
+    },
+    {
+      "epoch": 1.8298714144411474,
+      "grad_norm": 0.2822053601448072,
+      "learning_rate": 2.167460608281422e-05,
+      "loss": 0.4262,
+      "step": 1850
+    },
+    {
+      "epoch": 1.8308605341246291,
+      "grad_norm": 0.33119023555410904,
+      "learning_rate": 2.165628435324295e-05,
+      "loss": 0.4673,
+      "step": 1851
+    },
+    {
+      "epoch": 1.8318496538081108,
+      "grad_norm": 0.2982548956831065,
+      "learning_rate": 2.1637962623671675e-05,
+      "loss": 0.4265,
+      "step": 1852
+    },
+    {
+      "epoch": 1.8328387734915925,
+      "grad_norm": 0.2861624181648438,
+      "learning_rate": 2.1619640894100405e-05,
+      "loss": 0.4637,
+      "step": 1853
+    },
+    {
+      "epoch": 1.8338278931750742,
+      "grad_norm": 0.27624749764651596,
+      "learning_rate": 2.160131916452913e-05,
+      "loss": 0.378,
+      "step": 1854
+    },
+    {
+      "epoch": 1.834817012858556,
+      "grad_norm": 0.30172044665288217,
+      "learning_rate": 2.158299743495786e-05,
+      "loss": 0.442,
+      "step": 1855
+    },
+    {
+      "epoch": 1.8358061325420376,
+      "grad_norm": 0.8175953310300249,
+      "learning_rate": 2.1564675705386587e-05,
+      "loss": 0.5055,
+      "step": 1856
+    },
+    {
+      "epoch": 1.8367952522255193,
+      "grad_norm": 0.30746248179532876,
+      "learning_rate": 2.1546353975815317e-05,
+      "loss": 0.45,
+      "step": 1857
+    },
+    {
+      "epoch": 1.837784371909001,
+      "grad_norm": 0.3408014710812103,
+      "learning_rate": 2.1528032246244047e-05,
+      "loss": 0.4448,
+      "step": 1858
+    },
+    {
+      "epoch": 1.8387734915924827,
+      "grad_norm": 0.6207727087629815,
+      "learning_rate": 2.1509710516672773e-05,
+      "loss": 0.4727,
+      "step": 1859
+    },
+    {
+      "epoch": 1.8397626112759644,
+      "grad_norm": 0.4225278421377986,
+      "learning_rate": 2.1491388787101503e-05,
+      "loss": 0.4452,
+      "step": 1860
+    },
+    {
+      "epoch": 1.840751730959446,
+      "grad_norm": 0.4685555138093697,
+      "learning_rate": 2.1473067057530233e-05,
+      "loss": 0.4202,
+      "step": 1861
+    },
+    {
+      "epoch": 1.8417408506429278,
+      "grad_norm": 0.2891269631180874,
+      "learning_rate": 2.145474532795896e-05,
+      "loss": 0.4393,
+      "step": 1862
+    },
+    {
+      "epoch": 1.8427299703264095,
+      "grad_norm": 0.37223238502196465,
+      "learning_rate": 2.143642359838769e-05,
+      "loss": 0.3935,
+      "step": 1863
+    },
+    {
+      "epoch": 1.8437190900098912,
+      "grad_norm": 0.3478084518953676,
+      "learning_rate": 2.141810186881642e-05,
+      "loss": 0.4112,
+      "step": 1864
+    },
+    {
+      "epoch": 1.844708209693373,
+      "grad_norm": 0.2905645567287115,
+      "learning_rate": 2.1399780139245145e-05,
+      "loss": 0.4286,
+      "step": 1865
+    },
+    {
+      "epoch": 1.8456973293768546,
+      "grad_norm": 0.37515837566952076,
+      "learning_rate": 2.1381458409673875e-05,
+      "loss": 0.5235,
+      "step": 1866
+    },
+    {
+      "epoch": 1.8466864490603363,
+      "grad_norm": 0.2951545143923494,
+      "learning_rate": 2.13631366801026e-05,
+      "loss": 0.5067,
+      "step": 1867
+    },
+    {
+      "epoch": 1.847675568743818,
+      "grad_norm": 0.2948346426391745,
+      "learning_rate": 2.134481495053133e-05,
+      "loss": 0.3966,
+      "step": 1868
+    },
+    {
+      "epoch": 1.8486646884272997,
+      "grad_norm": 0.41530734276356635,
+      "learning_rate": 2.1326493220960058e-05,
+      "loss": 0.4632,
+      "step": 1869
+    },
+    {
+      "epoch": 1.8496538081107814,
+      "grad_norm": 0.3147529275156093,
+      "learning_rate": 2.1308171491388788e-05,
+      "loss": 0.4547,
+      "step": 1870
+    },
+    {
+      "epoch": 1.850642927794263,
+      "grad_norm": 0.2927562274139951,
+      "learning_rate": 2.1289849761817514e-05,
+      "loss": 0.3962,
+      "step": 1871
+    },
+    {
+      "epoch": 1.8516320474777448,
+      "grad_norm": 0.3657012517786323,
+      "learning_rate": 2.1271528032246247e-05,
+      "loss": 0.4178,
+      "step": 1872
+    },
+    {
+      "epoch": 1.8526211671612265,
+      "grad_norm": 0.2740740734494427,
+      "learning_rate": 2.1253206302674974e-05,
+      "loss": 0.4467,
+      "step": 1873
+    },
+    {
+      "epoch": 1.8536102868447082,
+      "grad_norm": 0.3075673256037017,
+      "learning_rate": 2.1234884573103704e-05,
+      "loss": 0.4186,
+      "step": 1874
+    },
+    {
+      "epoch": 1.8545994065281899,
+      "grad_norm": 0.2920268194596183,
+      "learning_rate": 2.121656284353243e-05,
+      "loss": 0.3896,
+      "step": 1875
+    },
+    {
+      "epoch": 1.8555885262116716,
+      "grad_norm": 0.27201039572059627,
+      "learning_rate": 2.119824111396116e-05,
+      "loss": 0.4085,
+      "step": 1876
+    },
+    {
+      "epoch": 1.8565776458951533,
+      "grad_norm": 0.353148381804933,
+      "learning_rate": 2.1179919384389886e-05,
+      "loss": 0.4426,
+      "step": 1877
+    },
+    {
+      "epoch": 1.857566765578635,
+      "grad_norm": 8.40808607244907,
+      "learning_rate": 2.1161597654818616e-05,
+      "loss": 0.4798,
+      "step": 1878
+    },
+    {
+      "epoch": 1.8585558852621167,
+      "grad_norm": 0.267897511367695,
+      "learning_rate": 2.1143275925247342e-05,
+      "loss": 0.4339,
+      "step": 1879
+    },
+    {
+      "epoch": 1.8595450049455984,
+      "grad_norm": 0.31654957685044094,
+      "learning_rate": 2.1124954195676072e-05,
+      "loss": 0.4733,
+      "step": 1880
+    },
+    {
+      "epoch": 1.86053412462908,
+      "grad_norm": 0.3027075365016889,
+      "learning_rate": 2.1106632466104802e-05,
+      "loss": 0.4363,
+      "step": 1881
+    },
+    {
+      "epoch": 1.8615232443125618,
+      "grad_norm": 0.31790990073853925,
+      "learning_rate": 2.108831073653353e-05,
+      "loss": 0.4105,
+      "step": 1882
+    },
+    {
+      "epoch": 1.8625123639960435,
+      "grad_norm": 0.2843270321617449,
+      "learning_rate": 2.1069989006962258e-05,
+      "loss": 0.4514,
+      "step": 1883
+    },
+    {
+      "epoch": 1.8635014836795252,
+      "grad_norm": 0.3044166271116939,
+      "learning_rate": 2.1051667277390988e-05,
+      "loss": 0.4842,
+      "step": 1884
+    },
+    {
+      "epoch": 1.8644906033630069,
+      "grad_norm": 0.3358497677903399,
+      "learning_rate": 2.1033345547819714e-05,
+      "loss": 0.4455,
+      "step": 1885
+    },
+    {
+      "epoch": 1.8654797230464886,
+      "grad_norm": 0.30731251689927197,
+      "learning_rate": 2.1015023818248444e-05,
+      "loss": 0.4451,
+      "step": 1886
+    },
+    {
+      "epoch": 1.8664688427299705,
+      "grad_norm": 0.3143692937530264,
+      "learning_rate": 2.0996702088677174e-05,
+      "loss": 0.4269,
+      "step": 1887
+    },
+    {
+      "epoch": 1.867457962413452,
+      "grad_norm": 0.2834940816390246,
+      "learning_rate": 2.09783803591059e-05,
+      "loss": 0.412,
+      "step": 1888
+    },
+    {
+      "epoch": 1.8684470820969339,
+      "grad_norm": 0.2788270677440903,
+      "learning_rate": 2.096005862953463e-05,
+      "loss": 0.3973,
+      "step": 1889
+    },
+    {
+      "epoch": 1.8694362017804154,
+      "grad_norm": 0.3078569868447692,
+      "learning_rate": 2.0941736899963357e-05,
+      "loss": 0.4802,
+      "step": 1890
+    },
+    {
+      "epoch": 1.8704253214638973,
+      "grad_norm": 0.2882271462085658,
+      "learning_rate": 2.0923415170392087e-05,
+      "loss": 0.4506,
+      "step": 1891
+    },
+    {
+      "epoch": 1.8714144411473788,
+      "grad_norm": 0.2768064460294965,
+      "learning_rate": 2.0905093440820813e-05,
+      "loss": 0.4525,
+      "step": 1892
+    },
+    {
+      "epoch": 1.8724035608308607,
+      "grad_norm": 0.30177959635692414,
+      "learning_rate": 2.0886771711249543e-05,
+      "loss": 0.4446,
+      "step": 1893
+    },
+    {
+      "epoch": 1.8733926805143422,
+      "grad_norm": 0.30727441186455273,
+      "learning_rate": 2.086844998167827e-05,
+      "loss": 0.4138,
+      "step": 1894
+    },
+    {
+      "epoch": 1.874381800197824,
+      "grad_norm": 0.3029843395912454,
+      "learning_rate": 2.0850128252107002e-05,
+      "loss": 0.467,
+      "step": 1895
+    },
+    {
+      "epoch": 1.8753709198813056,
+      "grad_norm": 0.28610816219099716,
+      "learning_rate": 2.083180652253573e-05,
+      "loss": 0.4103,
+      "step": 1896
+    },
+    {
+      "epoch": 1.8763600395647875,
+      "grad_norm": 0.26564180151467914,
+      "learning_rate": 2.081348479296446e-05,
+      "loss": 0.413,
+      "step": 1897
+    },
+    {
+      "epoch": 1.877349159248269,
+      "grad_norm": 0.32891431091901296,
+      "learning_rate": 2.0795163063393185e-05,
+      "loss": 0.4131,
+      "step": 1898
+    },
+    {
+      "epoch": 1.8783382789317509,
+      "grad_norm": 0.2655039622432033,
+      "learning_rate": 2.0776841333821915e-05,
+      "loss": 0.421,
+      "step": 1899
+    },
+    {
+      "epoch": 1.8793273986152323,
+      "grad_norm": 0.3009633138938526,
+      "learning_rate": 2.075851960425064e-05,
+      "loss": 0.4188,
+      "step": 1900
+    },
+    {
+      "epoch": 1.8803165182987143,
+      "grad_norm": 0.29124656986978736,
+      "learning_rate": 2.074019787467937e-05,
+      "loss": 0.4162,
+      "step": 1901
+    },
+    {
+      "epoch": 1.8813056379821957,
+      "grad_norm": 0.30884019091654086,
+      "learning_rate": 2.0721876145108097e-05,
+      "loss": 0.4565,
+      "step": 1902
+    },
+    {
+      "epoch": 1.8822947576656777,
+      "grad_norm": 0.30522532353130594,
+      "learning_rate": 2.0703554415536827e-05,
+      "loss": 0.5036,
+      "step": 1903
+    },
+    {
+      "epoch": 1.8832838773491591,
+      "grad_norm": 0.3076309070429283,
+      "learning_rate": 2.0685232685965557e-05,
+      "loss": 0.4343,
+      "step": 1904
+    },
+    {
+      "epoch": 1.884272997032641,
+      "grad_norm": 0.3451728446476686,
+      "learning_rate": 2.0666910956394283e-05,
+      "loss": 0.4339,
+      "step": 1905
+    },
+    {
+      "epoch": 1.8852621167161225,
+      "grad_norm": 0.28959111272819593,
+      "learning_rate": 2.0648589226823013e-05,
+      "loss": 0.4249,
+      "step": 1906
+    },
+    {
+      "epoch": 1.8862512363996045,
+      "grad_norm": 0.30354422997001196,
+      "learning_rate": 2.0630267497251743e-05,
+      "loss": 0.4156,
+      "step": 1907
+    },
+    {
+      "epoch": 1.887240356083086,
+      "grad_norm": 0.3085053722428896,
+      "learning_rate": 2.061194576768047e-05,
+      "loss": 0.4465,
+      "step": 1908
+    },
+    {
+      "epoch": 1.8882294757665679,
+      "grad_norm": 0.3437045799968937,
+      "learning_rate": 2.05936240381092e-05,
+      "loss": 0.4523,
+      "step": 1909
+    },
+    {
+      "epoch": 1.8892185954500493,
+      "grad_norm": 0.28581058747896737,
+      "learning_rate": 2.057530230853793e-05,
+      "loss": 0.427,
+      "step": 1910
+    },
+    {
+      "epoch": 1.8902077151335313,
+      "grad_norm": 0.271508864393249,
+      "learning_rate": 2.0556980578966656e-05,
+      "loss": 0.3641,
+      "step": 1911
+    },
+    {
+      "epoch": 1.8911968348170127,
+      "grad_norm": 0.35199224588959227,
+      "learning_rate": 2.0538658849395385e-05,
+      "loss": 0.4209,
+      "step": 1912
+    },
+    {
+      "epoch": 1.8921859545004946,
+      "grad_norm": 0.2953225794878225,
+      "learning_rate": 2.0520337119824112e-05,
+      "loss": 0.3718,
+      "step": 1913
+    },
+    {
+      "epoch": 1.8931750741839761,
+      "grad_norm": 0.2738162952704808,
+      "learning_rate": 2.050201539025284e-05,
+      "loss": 0.3926,
+      "step": 1914
+    },
+    {
+      "epoch": 1.894164193867458,
+      "grad_norm": 0.26534375180654185,
+      "learning_rate": 2.0483693660681568e-05,
+      "loss": 0.3923,
+      "step": 1915
+    },
+    {
+      "epoch": 1.8951533135509395,
+      "grad_norm": 0.2631897845252683,
+      "learning_rate": 2.0465371931110298e-05,
+      "loss": 0.3955,
+      "step": 1916
+    },
+    {
+      "epoch": 1.8961424332344214,
+      "grad_norm": 0.2942843582306155,
+      "learning_rate": 2.0447050201539024e-05,
+      "loss": 0.419,
+      "step": 1917
+    },
+    {
+      "epoch": 1.897131552917903,
+      "grad_norm": 0.30867929917338793,
+      "learning_rate": 2.0428728471967754e-05,
+      "loss": 0.4338,
+      "step": 1918
+    },
+    {
+      "epoch": 1.8981206726013848,
+      "grad_norm": 0.2990621787095512,
+      "learning_rate": 2.0410406742396484e-05,
+      "loss": 0.4383,
+      "step": 1919
+    },
+    {
+      "epoch": 1.8991097922848663,
+      "grad_norm": 0.27810473804617325,
+      "learning_rate": 2.0392085012825214e-05,
+      "loss": 0.4491,
+      "step": 1920
+    },
+    {
+      "epoch": 1.9000989119683482,
+      "grad_norm": 0.33382076803749,
+      "learning_rate": 2.037376328325394e-05,
+      "loss": 0.4381,
+      "step": 1921
+    },
+    {
+      "epoch": 1.9010880316518297,
+      "grad_norm": 0.3304112930354128,
+      "learning_rate": 2.035544155368267e-05,
+      "loss": 0.5064,
+      "step": 1922
+    },
+    {
+      "epoch": 1.9020771513353116,
+      "grad_norm": 0.25783866601539174,
+      "learning_rate": 2.0337119824111396e-05,
+      "loss": 0.3897,
+      "step": 1923
+    },
+    {
+      "epoch": 1.903066271018793,
+      "grad_norm": 0.27328280868082716,
+      "learning_rate": 2.0318798094540126e-05,
+      "loss": 0.4383,
+      "step": 1924
+    },
+    {
+      "epoch": 1.904055390702275,
+      "grad_norm": 0.33204613407624645,
+      "learning_rate": 2.0300476364968852e-05,
+      "loss": 0.5073,
+      "step": 1925
+    },
+    {
+      "epoch": 1.9050445103857567,
+      "grad_norm": 0.2766857216599759,
+      "learning_rate": 2.0282154635397582e-05,
+      "loss": 0.4149,
+      "step": 1926
+    },
+    {
+      "epoch": 1.9060336300692384,
+      "grad_norm": 0.266461589601163,
+      "learning_rate": 2.0263832905826312e-05,
+      "loss": 0.4314,
+      "step": 1927
+    },
+    {
+      "epoch": 1.9070227497527201,
+      "grad_norm": 0.2512857349198211,
+      "learning_rate": 2.024551117625504e-05,
+      "loss": 0.4336,
+      "step": 1928
+    },
+    {
+      "epoch": 1.9080118694362018,
+      "grad_norm": 0.33637530648765923,
+      "learning_rate": 2.022718944668377e-05,
+      "loss": 0.4225,
+      "step": 1929
+    },
+    {
+      "epoch": 1.9090009891196835,
+      "grad_norm": 0.28208217764150173,
+      "learning_rate": 2.0208867717112495e-05,
+      "loss": 0.4725,
+      "step": 1930
+    },
+    {
+      "epoch": 1.9099901088031652,
+      "grad_norm": 0.26065398592400135,
+      "learning_rate": 2.0190545987541225e-05,
+      "loss": 0.4567,
+      "step": 1931
+    },
+    {
+      "epoch": 1.910979228486647,
+      "grad_norm": 0.28479843394920357,
+      "learning_rate": 2.0172224257969954e-05,
+      "loss": 0.4287,
+      "step": 1932
+    },
+    {
+      "epoch": 1.9119683481701286,
+      "grad_norm": 3.294579825793158,
+      "learning_rate": 2.0153902528398684e-05,
+      "loss": 0.5882,
+      "step": 1933
+    },
+    {
+      "epoch": 1.9129574678536103,
+      "grad_norm": 0.28936302581206536,
+      "learning_rate": 2.013558079882741e-05,
+      "loss": 0.5019,
+      "step": 1934
+    },
+    {
+      "epoch": 1.913946587537092,
+      "grad_norm": 0.28047234749246863,
+      "learning_rate": 2.011725906925614e-05,
+      "loss": 0.4326,
+      "step": 1935
+    },
+    {
+      "epoch": 1.9149357072205737,
+      "grad_norm": 0.3100419020654506,
+      "learning_rate": 2.0098937339684867e-05,
+      "loss": 0.4336,
+      "step": 1936
+    },
+    {
+      "epoch": 1.9159248269040554,
+      "grad_norm": 0.30316063562527656,
+      "learning_rate": 2.0080615610113597e-05,
+      "loss": 0.4482,
+      "step": 1937
+    },
+    {
+      "epoch": 1.916913946587537,
+      "grad_norm": 0.3203458184728088,
+      "learning_rate": 2.0062293880542323e-05,
+      "loss": 0.4761,
+      "step": 1938
+    },
+    {
+      "epoch": 1.9179030662710188,
+      "grad_norm": 0.319115811593246,
+      "learning_rate": 2.0043972150971053e-05,
+      "loss": 0.4387,
+      "step": 1939
+    },
+    {
+      "epoch": 1.9188921859545005,
+      "grad_norm": 0.29266033140236186,
+      "learning_rate": 2.002565042139978e-05,
+      "loss": 0.4128,
+      "step": 1940
+    },
+    {
+      "epoch": 1.9198813056379822,
+      "grad_norm": 0.2934869021796227,
+      "learning_rate": 2.000732869182851e-05,
+      "loss": 0.4266,
+      "step": 1941
+    },
+    {
+      "epoch": 1.920870425321464,
+      "grad_norm": 0.2954310231819758,
+      "learning_rate": 1.9989006962257235e-05,
+      "loss": 0.4862,
+      "step": 1942
+    },
+    {
+      "epoch": 1.9218595450049456,
+      "grad_norm": 0.2989192690361351,
+      "learning_rate": 1.997068523268597e-05,
+      "loss": 0.4284,
+      "step": 1943
+    },
+    {
+      "epoch": 1.9228486646884273,
+      "grad_norm": 0.32532736480103575,
+      "learning_rate": 1.9952363503114695e-05,
+      "loss": 0.5123,
+      "step": 1944
+    },
+    {
+      "epoch": 1.923837784371909,
+      "grad_norm": 0.279809147331516,
+      "learning_rate": 1.9934041773543425e-05,
+      "loss": 0.4593,
+      "step": 1945
+    },
+    {
+      "epoch": 1.9248269040553907,
+      "grad_norm": 0.3043919655682651,
+      "learning_rate": 1.991572004397215e-05,
+      "loss": 0.4417,
+      "step": 1946
+    },
+    {
+      "epoch": 1.9258160237388724,
+      "grad_norm": 0.2795337887573538,
+      "learning_rate": 1.989739831440088e-05,
+      "loss": 0.4287,
+      "step": 1947
+    },
+    {
+      "epoch": 1.926805143422354,
+      "grad_norm": 0.26009593666483405,
+      "learning_rate": 1.9879076584829608e-05,
+      "loss": 0.4479,
+      "step": 1948
+    },
+    {
+      "epoch": 1.9277942631058358,
+      "grad_norm": 0.25306492114507956,
+      "learning_rate": 1.9860754855258337e-05,
+      "loss": 0.4074,
+      "step": 1949
+    },
+    {
+      "epoch": 1.9287833827893175,
+      "grad_norm": 0.2706763584509426,
+      "learning_rate": 1.9842433125687067e-05,
+      "loss": 0.4687,
+      "step": 1950
+    },
+    {
+      "epoch": 1.9297725024727992,
+      "grad_norm": 0.2546219655402372,
+      "learning_rate": 1.9824111396115794e-05,
+      "loss": 0.3809,
+      "step": 1951
+    },
+    {
+      "epoch": 1.9307616221562809,
+      "grad_norm": 0.2917574146301412,
+      "learning_rate": 1.9805789666544523e-05,
+      "loss": 0.4002,
+      "step": 1952
+    },
+    {
+      "epoch": 1.9317507418397626,
+      "grad_norm": 0.914811674910379,
+      "learning_rate": 1.978746793697325e-05,
+      "loss": 0.4656,
+      "step": 1953
+    },
+    {
+      "epoch": 1.9327398615232443,
+      "grad_norm": 0.2783013811561288,
+      "learning_rate": 1.976914620740198e-05,
+      "loss": 0.4792,
+      "step": 1954
+    },
+    {
+      "epoch": 1.933728981206726,
+      "grad_norm": 0.28602656481136324,
+      "learning_rate": 1.975082447783071e-05,
+      "loss": 0.4541,
+      "step": 1955
+    },
+    {
+      "epoch": 1.9347181008902077,
+      "grad_norm": 0.2592620006394712,
+      "learning_rate": 1.973250274825944e-05,
+      "loss": 0.4348,
+      "step": 1956
+    },
+    {
+      "epoch": 1.9357072205736894,
+      "grad_norm": 0.2773762643584451,
+      "learning_rate": 1.9714181018688166e-05,
+      "loss": 0.4063,
+      "step": 1957
+    },
+    {
+      "epoch": 1.936696340257171,
+      "grad_norm": 0.28599821633612305,
+      "learning_rate": 1.9695859289116895e-05,
+      "loss": 0.4824,
+      "step": 1958
+    },
+    {
+      "epoch": 1.9376854599406528,
+      "grad_norm": 0.26672204401496125,
+      "learning_rate": 1.9677537559545622e-05,
+      "loss": 0.4232,
+      "step": 1959
+    },
+    {
+      "epoch": 1.9386745796241345,
+      "grad_norm": 0.2892625352799071,
+      "learning_rate": 1.965921582997435e-05,
+      "loss": 0.4796,
+      "step": 1960
+    },
+    {
+      "epoch": 1.9396636993076162,
+      "grad_norm": 0.2817788199429448,
+      "learning_rate": 1.9640894100403078e-05,
+      "loss": 0.402,
+      "step": 1961
+    },
+    {
+      "epoch": 1.9406528189910979,
+      "grad_norm": 0.2653710321950093,
+      "learning_rate": 1.9622572370831808e-05,
+      "loss": 0.4031,
+      "step": 1962
+    },
+    {
+      "epoch": 1.9416419386745796,
+      "grad_norm": 0.313468529227229,
+      "learning_rate": 1.9604250641260534e-05,
+      "loss": 0.469,
+      "step": 1963
+    },
+    {
+      "epoch": 1.9426310583580615,
+      "grad_norm": 0.30116180498202555,
+      "learning_rate": 1.9585928911689264e-05,
+      "loss": 0.4781,
+      "step": 1964
+    },
+    {
+      "epoch": 1.943620178041543,
+      "grad_norm": 0.36835947832919974,
+      "learning_rate": 1.956760718211799e-05,
+      "loss": 0.3505,
+      "step": 1965
+    },
+    {
+      "epoch": 1.9446092977250249,
+      "grad_norm": 0.2916022025393715,
+      "learning_rate": 1.954928545254672e-05,
+      "loss": 0.4331,
+      "step": 1966
+    },
+    {
+      "epoch": 1.9455984174085064,
+      "grad_norm": 0.24576757883657926,
+      "learning_rate": 1.953096372297545e-05,
+      "loss": 0.3918,
+      "step": 1967
+    },
+    {
+      "epoch": 1.9465875370919883,
+      "grad_norm": 0.30131909213793245,
+      "learning_rate": 1.951264199340418e-05,
+      "loss": 0.461,
+      "step": 1968
+    },
+    {
+      "epoch": 1.9475766567754698,
+      "grad_norm": 0.3339260654475112,
+      "learning_rate": 1.9494320263832906e-05,
+      "loss": 0.436,
+      "step": 1969
+    },
+    {
+      "epoch": 1.9485657764589517,
+      "grad_norm": 0.27524157139256206,
+      "learning_rate": 1.9475998534261636e-05,
+      "loss": 0.4545,
+      "step": 1970
+    },
+    {
+      "epoch": 1.9495548961424332,
+      "grad_norm": 0.2874593210645949,
+      "learning_rate": 1.9457676804690363e-05,
+      "loss": 0.4127,
+      "step": 1971
+    },
+    {
+      "epoch": 1.950544015825915,
+      "grad_norm": 0.31215326151701583,
+      "learning_rate": 1.9439355075119092e-05,
+      "loss": 0.3706,
+      "step": 1972
+    },
+    {
+      "epoch": 1.9515331355093966,
+      "grad_norm": 0.2780337765875761,
+      "learning_rate": 1.9421033345547822e-05,
+      "loss": 0.3934,
+      "step": 1973
+    },
+    {
+      "epoch": 1.9525222551928785,
+      "grad_norm": 0.3453992054984105,
+      "learning_rate": 1.940271161597655e-05,
+      "loss": 0.4671,
+      "step": 1974
+    },
+    {
+      "epoch": 1.95351137487636,
+      "grad_norm": 0.3321376938331568,
+      "learning_rate": 1.938438988640528e-05,
+      "loss": 0.4228,
+      "step": 1975
+    },
+    {
+      "epoch": 1.9545004945598419,
+      "grad_norm": 0.2928057230254324,
+      "learning_rate": 1.9366068156834005e-05,
+      "loss": 0.469,
+      "step": 1976
+    },
+    {
+      "epoch": 1.9554896142433233,
+      "grad_norm": 0.30394150855107643,
+      "learning_rate": 1.9347746427262735e-05,
+      "loss": 0.4653,
+      "step": 1977
+    },
+    {
+      "epoch": 1.9564787339268053,
+      "grad_norm": 0.2894909046020456,
+      "learning_rate": 1.932942469769146e-05,
+      "loss": 0.3932,
+      "step": 1978
+    },
+    {
+      "epoch": 1.9574678536102867,
+      "grad_norm": 0.2711875527085813,
+      "learning_rate": 1.9311102968120194e-05,
+      "loss": 0.3798,
+      "step": 1979
+    },
+    {
+      "epoch": 1.9584569732937687,
+      "grad_norm": 0.2725315165989554,
+      "learning_rate": 1.929278123854892e-05,
+      "loss": 0.3549,
+      "step": 1980
+    },
+    {
+      "epoch": 1.9594460929772501,
+      "grad_norm": 0.26597443326290904,
+      "learning_rate": 1.927445950897765e-05,
+      "loss": 0.3765,
+      "step": 1981
+    },
+    {
+      "epoch": 1.960435212660732,
+      "grad_norm": 0.27613823878296645,
+      "learning_rate": 1.9256137779406377e-05,
+      "loss": 0.3882,
+      "step": 1982
+    },
+    {
+      "epoch": 1.9614243323442135,
+      "grad_norm": 0.2851650579144139,
+      "learning_rate": 1.9237816049835107e-05,
+      "loss": 0.4073,
+      "step": 1983
+    },
+    {
+      "epoch": 1.9624134520276955,
+      "grad_norm": 0.2855817930760619,
+      "learning_rate": 1.9219494320263833e-05,
+      "loss": 0.4373,
+      "step": 1984
+    },
+    {
+      "epoch": 1.963402571711177,
+      "grad_norm": 0.2727240403962694,
+      "learning_rate": 1.9201172590692563e-05,
+      "loss": 0.4231,
+      "step": 1985
+    },
+    {
+      "epoch": 1.9643916913946589,
+      "grad_norm": 0.26875071309544885,
+      "learning_rate": 1.918285086112129e-05,
+      "loss": 0.4155,
+      "step": 1986
+    },
+    {
+      "epoch": 1.9653808110781403,
+      "grad_norm": 0.24915569601701976,
+      "learning_rate": 1.916452913155002e-05,
+      "loss": 0.3926,
+      "step": 1987
+    },
+    {
+      "epoch": 1.9663699307616223,
+      "grad_norm": 1.6073228027788682,
+      "learning_rate": 1.9146207401978746e-05,
+      "loss": 0.4279,
+      "step": 1988
+    },
+    {
+      "epoch": 1.9673590504451037,
+      "grad_norm": 0.2792233401337717,
+      "learning_rate": 1.9127885672407475e-05,
+      "loss": 0.4021,
+      "step": 1989
+    },
+    {
+      "epoch": 1.9683481701285857,
+      "grad_norm": 0.25494389574541926,
+      "learning_rate": 1.9109563942836202e-05,
+      "loss": 0.4267,
+      "step": 1990
+    },
+    {
+      "epoch": 1.9693372898120671,
+      "grad_norm": 0.2898106866186205,
+      "learning_rate": 1.9091242213264935e-05,
+      "loss": 0.4323,
+      "step": 1991
+    },
+    {
+      "epoch": 1.970326409495549,
+      "grad_norm": 0.2728511736986319,
+      "learning_rate": 1.907292048369366e-05,
+      "loss": 0.3848,
+      "step": 1992
+    },
+    {
+      "epoch": 1.9713155291790305,
+      "grad_norm": 0.24292208041411942,
+      "learning_rate": 1.905459875412239e-05,
+      "loss": 0.4054,
+      "step": 1993
+    },
+    {
+      "epoch": 1.9723046488625124,
+      "grad_norm": 0.25921473245660914,
+      "learning_rate": 1.9036277024551118e-05,
+      "loss": 0.393,
+      "step": 1994
+    },
+    {
+      "epoch": 1.973293768545994,
+      "grad_norm": 0.27928092261140336,
+      "learning_rate": 1.9017955294979847e-05,
+      "loss": 0.4445,
+      "step": 1995
+    },
+    {
+      "epoch": 1.9742828882294758,
+      "grad_norm": 0.27406269020220453,
+      "learning_rate": 1.8999633565408577e-05,
+      "loss": 0.454,
+      "step": 1996
+    },
+    {
+      "epoch": 1.9752720079129573,
+      "grad_norm": 0.2614651016454685,
+      "learning_rate": 1.8981311835837304e-05,
+      "loss": 0.3824,
+      "step": 1997
+    },
+    {
+      "epoch": 1.9762611275964392,
+      "grad_norm": 0.27737401784240373,
+      "learning_rate": 1.8962990106266033e-05,
+      "loss": 0.4434,
+      "step": 1998
+    },
+    {
+      "epoch": 1.9772502472799207,
+      "grad_norm": 0.2604415643950939,
+      "learning_rate": 1.894466837669476e-05,
+      "loss": 0.4093,
+      "step": 1999
+    },
+    {
+      "epoch": 1.9782393669634026,
+      "grad_norm": 0.24222620275410053,
+      "learning_rate": 1.892634664712349e-05,
+      "loss": 0.4325,
+      "step": 2000
+    },
+    {
+      "epoch": 1.979228486646884,
+      "grad_norm": 0.23224541289636652,
+      "learning_rate": 1.8908024917552216e-05,
+      "loss": 0.3769,
+      "step": 2001
+    },
+    {
+      "epoch": 1.980217606330366,
+      "grad_norm": 0.2723348778036034,
+      "learning_rate": 1.8889703187980946e-05,
+      "loss": 0.3764,
+      "step": 2002
+    },
+    {
+      "epoch": 1.9812067260138477,
+      "grad_norm": 0.31673784091219626,
+      "learning_rate": 1.8871381458409676e-05,
+      "loss": 0.4378,
+      "step": 2003
+    },
+    {
+      "epoch": 1.9821958456973294,
+      "grad_norm": 0.2928301587474028,
+      "learning_rate": 1.8853059728838405e-05,
+      "loss": 0.4119,
+      "step": 2004
+    },
+    {
+      "epoch": 1.9831849653808111,
+      "grad_norm": 0.24968201113913796,
+      "learning_rate": 1.8834737999267132e-05,
+      "loss": 0.4273,
+      "step": 2005
+    },
+    {
+      "epoch": 1.9841740850642928,
+      "grad_norm": 0.3322683975213439,
+      "learning_rate": 1.8816416269695862e-05,
+      "loss": 0.4857,
+      "step": 2006
+    },
+    {
+      "epoch": 1.9851632047477745,
+      "grad_norm": 0.2951583445750594,
+      "learning_rate": 1.8798094540124588e-05,
+      "loss": 0.4307,
+      "step": 2007
+    },
+    {
+      "epoch": 1.9861523244312562,
+      "grad_norm": 0.33937998935052427,
+      "learning_rate": 1.8779772810553318e-05,
+      "loss": 0.4664,
+      "step": 2008
+    },
+    {
+      "epoch": 1.987141444114738,
+      "grad_norm": 0.2923057755122759,
+      "learning_rate": 1.8761451080982044e-05,
+      "loss": 0.5091,
+      "step": 2009
+    },
+    {
+      "epoch": 1.9881305637982196,
+      "grad_norm": 0.2788462746112801,
+      "learning_rate": 1.8743129351410774e-05,
+      "loss": 0.386,
+      "step": 2010
+    },
+    {
+      "epoch": 1.9891196834817013,
+      "grad_norm": 0.2516597878121542,
+      "learning_rate": 1.87248076218395e-05,
+      "loss": 0.3403,
+      "step": 2011
+    },
+    {
+      "epoch": 1.990108803165183,
+      "grad_norm": 0.30376516700631484,
+      "learning_rate": 1.870648589226823e-05,
+      "loss": 0.4669,
+      "step": 2012
+    },
+    {
+      "epoch": 1.9910979228486647,
+      "grad_norm": 0.28097309920042796,
+      "learning_rate": 1.8688164162696957e-05,
+      "loss": 0.4018,
+      "step": 2013
+    },
+    {
+      "epoch": 1.9920870425321464,
+      "grad_norm": 0.27334460203651567,
+      "learning_rate": 1.8669842433125687e-05,
+      "loss": 0.4525,
+      "step": 2014
+    },
+    {
+      "epoch": 1.993076162215628,
+      "grad_norm": 0.29672124438553915,
+      "learning_rate": 1.8651520703554416e-05,
+      "loss": 0.454,
+      "step": 2015
+    },
+    {
+      "epoch": 1.9940652818991098,
+      "grad_norm": 0.2888799406442104,
+      "learning_rate": 1.8633198973983146e-05,
+      "loss": 0.4324,
+      "step": 2016
+    },
+    {
+      "epoch": 1.9950544015825915,
+      "grad_norm": 0.25475149445910006,
+      "learning_rate": 1.8614877244411873e-05,
+      "loss": 0.4482,
+      "step": 2017
+    },
+    {
+      "epoch": 1.9960435212660732,
+      "grad_norm": 0.29167348715567726,
+      "learning_rate": 1.8596555514840602e-05,
+      "loss": 0.399,
+      "step": 2018
+    },
+    {
+      "epoch": 1.997032640949555,
+      "grad_norm": 0.2692693993748144,
+      "learning_rate": 1.8578233785269332e-05,
+      "loss": 0.4011,
+      "step": 2019
+    },
+    {
+      "epoch": 1.9980217606330366,
+      "grad_norm": 0.2589196204328078,
+      "learning_rate": 1.855991205569806e-05,
+      "loss": 0.3859,
+      "step": 2020
+    },
+    {
+      "epoch": 1.9990108803165183,
+      "grad_norm": 0.6296008364535154,
+      "learning_rate": 1.854159032612679e-05,
+      "loss": 0.4505,
+      "step": 2021
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.3007094377136532,
+      "learning_rate": 1.8523268596555515e-05,
+      "loss": 0.3938,
+      "step": 2022
+    },
+    {
+      "epoch": 2.000989119683482,
+      "grad_norm": 0.33376807762869504,
+      "learning_rate": 1.8504946866984245e-05,
+      "loss": 0.362,
+      "step": 2023
+    },
+    {
+      "epoch": 2.0019782393669634,
+      "grad_norm": 0.2601375793910521,
+      "learning_rate": 1.848662513741297e-05,
+      "loss": 0.3527,
+      "step": 2024
+    },
+    {
+      "epoch": 2.0029673590504453,
+      "grad_norm": 0.2808677816345753,
+      "learning_rate": 1.84683034078417e-05,
+      "loss": 0.3309,
+      "step": 2025
+    },
+    {
+      "epoch": 2.003956478733927,
+      "grad_norm": 0.3021748972184023,
+      "learning_rate": 1.8449981678270427e-05,
+      "loss": 0.3377,
+      "step": 2026
+    },
+    {
+      "epoch": 2.0049455984174087,
+      "grad_norm": 0.25396738481241404,
+      "learning_rate": 1.843165994869916e-05,
+      "loss": 0.3214,
+      "step": 2027
+    },
+    {
+      "epoch": 2.00593471810089,
+      "grad_norm": 0.2522846151099154,
+      "learning_rate": 1.8413338219127887e-05,
+      "loss": 0.33,
+      "step": 2028
+    },
+    {
+      "epoch": 2.006923837784372,
+      "grad_norm": 0.2888004385011141,
+      "learning_rate": 1.8395016489556617e-05,
+      "loss": 0.3479,
+      "step": 2029
+    },
+    {
+      "epoch": 2.0079129574678536,
+      "grad_norm": 0.2587138992354896,
+      "learning_rate": 1.8376694759985343e-05,
+      "loss": 0.2811,
+      "step": 2030
+    },
+    {
+      "epoch": 2.0089020771513355,
+      "grad_norm": 0.3155016932703736,
+      "learning_rate": 1.8358373030414073e-05,
+      "loss": 0.3416,
+      "step": 2031
+    },
+    {
+      "epoch": 2.009891196834817,
+      "grad_norm": 0.2857913643432993,
+      "learning_rate": 1.83400513008428e-05,
+      "loss": 0.3571,
+      "step": 2032
+    },
+    {
+      "epoch": 2.010880316518299,
+      "grad_norm": 0.2678200050894992,
+      "learning_rate": 1.832172957127153e-05,
+      "loss": 0.3581,
+      "step": 2033
+    },
+    {
+      "epoch": 2.0118694362017804,
+      "grad_norm": 0.25644421751373253,
+      "learning_rate": 1.8303407841700256e-05,
+      "loss": 0.313,
+      "step": 2034
+    },
+    {
+      "epoch": 2.0128585558852623,
+      "grad_norm": 0.4465486558136206,
+      "learning_rate": 1.8285086112128985e-05,
+      "loss": 0.4098,
+      "step": 2035
+    },
+    {
+      "epoch": 2.013847675568744,
+      "grad_norm": 0.2833668117731138,
+      "learning_rate": 1.8266764382557712e-05,
+      "loss": 0.3211,
+      "step": 2036
+    },
+    {
+      "epoch": 2.0148367952522257,
+      "grad_norm": 0.23426073423687663,
+      "learning_rate": 1.824844265298644e-05,
+      "loss": 0.3165,
+      "step": 2037
+    },
+    {
+      "epoch": 2.015825914935707,
+      "grad_norm": 0.32693965911004064,
+      "learning_rate": 1.823012092341517e-05,
+      "loss": 0.4052,
+      "step": 2038
+    },
+    {
+      "epoch": 2.016815034619189,
+      "grad_norm": 0.3380208112798089,
+      "learning_rate": 1.82117991938439e-05,
+      "loss": 0.3321,
+      "step": 2039
+    },
+    {
+      "epoch": 2.0178041543026706,
+      "grad_norm": 0.24941143626683626,
+      "learning_rate": 1.8193477464272628e-05,
+      "loss": 0.3336,
+      "step": 2040
+    },
+    {
+      "epoch": 2.0187932739861525,
+      "grad_norm": 0.277234094183247,
+      "learning_rate": 1.8175155734701357e-05,
+      "loss": 0.353,
+      "step": 2041
+    },
+    {
+      "epoch": 2.019782393669634,
+      "grad_norm": 0.2726781199448609,
+      "learning_rate": 1.8156834005130087e-05,
+      "loss": 0.331,
+      "step": 2042
+    },
+    {
+      "epoch": 2.020771513353116,
+      "grad_norm": 0.2617540569925556,
+      "learning_rate": 1.8138512275558814e-05,
+      "loss": 0.3276,
+      "step": 2043
+    },
+    {
+      "epoch": 2.0217606330365974,
+      "grad_norm": 0.23504401210710754,
+      "learning_rate": 1.8120190545987544e-05,
+      "loss": 0.3109,
+      "step": 2044
+    },
+    {
+      "epoch": 2.0227497527200793,
+      "grad_norm": 0.2407046086285793,
+      "learning_rate": 1.810186881641627e-05,
+      "loss": 0.3374,
+      "step": 2045
+    },
+    {
+      "epoch": 2.0237388724035608,
+      "grad_norm": 0.2641964761821478,
+      "learning_rate": 1.8083547086845e-05,
+      "loss": 0.376,
+      "step": 2046
+    },
+    {
+      "epoch": 2.0247279920870427,
+      "grad_norm": 0.2882718123838396,
+      "learning_rate": 1.8065225357273726e-05,
+      "loss": 0.3328,
+      "step": 2047
+    },
+    {
+      "epoch": 2.025717111770524,
+      "grad_norm": 0.2786807025896711,
+      "learning_rate": 1.8046903627702456e-05,
+      "loss": 0.3459,
+      "step": 2048
+    },
+    {
+      "epoch": 2.026706231454006,
+      "grad_norm": 0.2539432049765403,
+      "learning_rate": 1.8028581898131182e-05,
+      "loss": 0.3647,
+      "step": 2049
+    },
+    {
+      "epoch": 2.0276953511374876,
+      "grad_norm": 0.2556731150500688,
+      "learning_rate": 1.8010260168559916e-05,
+      "loss": 0.3777,
+      "step": 2050
+    },
+    {
+      "epoch": 2.0286844708209695,
+      "grad_norm": 0.25486739832116906,
+      "learning_rate": 1.7991938438988642e-05,
+      "loss": 0.3138,
+      "step": 2051
+    },
+    {
+      "epoch": 2.029673590504451,
+      "grad_norm": 0.2774415544605617,
+      "learning_rate": 1.7973616709417372e-05,
+      "loss": 0.3343,
+      "step": 2052
+    },
+    {
+      "epoch": 2.030662710187933,
+      "grad_norm": 0.25646240486523647,
+      "learning_rate": 1.7955294979846098e-05,
+      "loss": 0.3676,
+      "step": 2053
+    },
+    {
+      "epoch": 2.0316518298714143,
+      "grad_norm": 0.23627554159821249,
+      "learning_rate": 1.7936973250274828e-05,
+      "loss": 0.297,
+      "step": 2054
+    },
+    {
+      "epoch": 2.0326409495548963,
+      "grad_norm": 0.9750055399356007,
+      "learning_rate": 1.7918651520703554e-05,
+      "loss": 0.3385,
+      "step": 2055
+    },
+    {
+      "epoch": 2.0336300692383777,
+      "grad_norm": 0.3259195566861495,
+      "learning_rate": 1.7900329791132284e-05,
+      "loss": 0.3283,
+      "step": 2056
+    },
+    {
+      "epoch": 2.0346191889218597,
+      "grad_norm": 0.34994469632467384,
+      "learning_rate": 1.788200806156101e-05,
+      "loss": 0.3335,
+      "step": 2057
+    },
+    {
+      "epoch": 2.035608308605341,
+      "grad_norm": 0.3874809670228957,
+      "learning_rate": 1.786368633198974e-05,
+      "loss": 0.3423,
+      "step": 2058
+    },
+    {
+      "epoch": 2.036597428288823,
+      "grad_norm": 0.4397999002338133,
+      "learning_rate": 1.7845364602418467e-05,
+      "loss": 0.3911,
+      "step": 2059
+    },
+    {
+      "epoch": 2.0375865479723045,
+      "grad_norm": 0.2965068166961372,
+      "learning_rate": 1.7827042872847197e-05,
+      "loss": 0.368,
+      "step": 2060
+    },
+    {
+      "epoch": 2.0385756676557865,
+      "grad_norm": 0.33913465022866834,
+      "learning_rate": 1.7808721143275926e-05,
+      "loss": 0.3485,
+      "step": 2061
+    },
+    {
+      "epoch": 2.039564787339268,
+      "grad_norm": 0.41953455881087176,
+      "learning_rate": 1.7790399413704656e-05,
+      "loss": 0.3337,
+      "step": 2062
+    },
+    {
+      "epoch": 2.04055390702275,
+      "grad_norm": 0.24963906693283444,
+      "learning_rate": 1.7772077684133383e-05,
+      "loss": 0.3388,
+      "step": 2063
+    },
+    {
+      "epoch": 2.0415430267062313,
+      "grad_norm": 0.24427178402898375,
+      "learning_rate": 1.7753755954562113e-05,
+      "loss": 0.3758,
+      "step": 2064
+    },
+    {
+      "epoch": 2.0425321463897133,
+      "grad_norm": 0.3391522202251561,
+      "learning_rate": 1.773543422499084e-05,
+      "loss": 0.3954,
+      "step": 2065
+    },
+    {
+      "epoch": 2.0435212660731947,
+      "grad_norm": 0.305105117952119,
+      "learning_rate": 1.771711249541957e-05,
+      "loss": 0.3189,
+      "step": 2066
+    },
+    {
+      "epoch": 2.0445103857566767,
+      "grad_norm": 0.2872800239846866,
+      "learning_rate": 1.76987907658483e-05,
+      "loss": 0.3776,
+      "step": 2067
+    },
+    {
+      "epoch": 2.045499505440158,
+      "grad_norm": 0.2504019307251769,
+      "learning_rate": 1.7680469036277025e-05,
+      "loss": 0.356,
+      "step": 2068
+    },
+    {
+      "epoch": 2.04648862512364,
+      "grad_norm": 0.25901454727456,
+      "learning_rate": 1.7662147306705755e-05,
+      "loss": 0.3776,
+      "step": 2069
+    },
+    {
+      "epoch": 2.0474777448071215,
+      "grad_norm": 0.29939939557269285,
+      "learning_rate": 1.764382557713448e-05,
+      "loss": 0.3126,
+      "step": 2070
+    },
+    {
+      "epoch": 2.0484668644906034,
+      "grad_norm": 0.3231004145272702,
+      "learning_rate": 1.762550384756321e-05,
+      "loss": 0.3393,
+      "step": 2071
+    },
+    {
+      "epoch": 2.049455984174085,
+      "grad_norm": 0.2476325861585833,
+      "learning_rate": 1.7607182117991937e-05,
+      "loss": 0.3463,
+      "step": 2072
+    },
+    {
+      "epoch": 2.050445103857567,
+      "grad_norm": 0.2979710367960843,
+      "learning_rate": 1.7588860388420667e-05,
+      "loss": 0.3726,
+      "step": 2073
+    },
+    {
+      "epoch": 2.0514342235410483,
+      "grad_norm": 0.2762787234379862,
+      "learning_rate": 1.7570538658849397e-05,
+      "loss": 0.3372,
+      "step": 2074
+    },
+    {
+      "epoch": 2.0524233432245302,
+      "grad_norm": 0.2525936147732162,
+      "learning_rate": 1.7552216929278127e-05,
+      "loss": 0.3224,
+      "step": 2075
+    },
+    {
+      "epoch": 2.0534124629080117,
+      "grad_norm": 0.2584635923287862,
+      "learning_rate": 1.7533895199706853e-05,
+      "loss": 0.3592,
+      "step": 2076
+    },
+    {
+      "epoch": 2.0544015825914936,
+      "grad_norm": 0.24758511782034934,
+      "learning_rate": 1.7515573470135583e-05,
+      "loss": 0.3228,
+      "step": 2077
+    },
+    {
+      "epoch": 2.055390702274975,
+      "grad_norm": 0.2883921683823937,
+      "learning_rate": 1.749725174056431e-05,
+      "loss": 0.3278,
+      "step": 2078
+    },
+    {
+      "epoch": 2.056379821958457,
+      "grad_norm": 0.3758824821850759,
+      "learning_rate": 1.747893001099304e-05,
+      "loss": 0.3069,
+      "step": 2079
+    },
+    {
+      "epoch": 2.0573689416419385,
+      "grad_norm": 0.26898488904546836,
+      "learning_rate": 1.7460608281421766e-05,
+      "loss": 0.3374,
+      "step": 2080
+    },
+    {
+      "epoch": 2.0583580613254204,
+      "grad_norm": 0.22348922505676247,
+      "learning_rate": 1.7442286551850495e-05,
+      "loss": 0.297,
+      "step": 2081
+    },
+    {
+      "epoch": 2.059347181008902,
+      "grad_norm": 0.2431620389642761,
+      "learning_rate": 1.7423964822279222e-05,
+      "loss": 0.3452,
+      "step": 2082
+    },
+    {
+      "epoch": 2.060336300692384,
+      "grad_norm": 0.2715649731140133,
+      "learning_rate": 1.7405643092707952e-05,
+      "loss": 0.3588,
+      "step": 2083
+    },
+    {
+      "epoch": 2.0613254203758653,
+      "grad_norm": 0.24583847764861538,
+      "learning_rate": 1.738732136313668e-05,
+      "loss": 0.3515,
+      "step": 2084
+    },
+    {
+      "epoch": 2.0623145400593472,
+      "grad_norm": 6.147132786755756,
+      "learning_rate": 1.7368999633565408e-05,
+      "loss": 0.6556,
+      "step": 2085
+    },
+    {
+      "epoch": 2.0633036597428287,
+      "grad_norm": 0.2739117127003737,
+      "learning_rate": 1.7350677903994138e-05,
+      "loss": 0.4016,
+      "step": 2086
+    },
+    {
+      "epoch": 2.0642927794263106,
+      "grad_norm": 0.2735119068979336,
+      "learning_rate": 1.7332356174422868e-05,
+      "loss": 0.3623,
+      "step": 2087
+    },
+    {
+      "epoch": 2.065281899109792,
+      "grad_norm": 0.2384520866225091,
+      "learning_rate": 1.7314034444851594e-05,
+      "loss": 0.3101,
+      "step": 2088
+    },
+    {
+      "epoch": 2.066271018793274,
+      "grad_norm": 0.2396852378862046,
+      "learning_rate": 1.7295712715280324e-05,
+      "loss": 0.3292,
+      "step": 2089
+    },
+    {
+      "epoch": 2.0672601384767555,
+      "grad_norm": 0.9802498145057038,
+      "learning_rate": 1.7277390985709054e-05,
+      "loss": 0.4025,
+      "step": 2090
+    },
+    {
+      "epoch": 2.0682492581602374,
+      "grad_norm": 0.26601631784836316,
+      "learning_rate": 1.725906925613778e-05,
+      "loss": 0.3529,
+      "step": 2091
+    },
+    {
+      "epoch": 2.069238377843719,
+      "grad_norm": 0.2914785211803027,
+      "learning_rate": 1.724074752656651e-05,
+      "loss": 0.3318,
+      "step": 2092
+    },
+    {
+      "epoch": 2.070227497527201,
+      "grad_norm": 0.27346245434038796,
+      "learning_rate": 1.7222425796995236e-05,
+      "loss": 0.3645,
+      "step": 2093
+    },
+    {
+      "epoch": 2.0712166172106823,
+      "grad_norm": 0.24874170019953046,
+      "learning_rate": 1.7204104067423966e-05,
+      "loss": 0.3408,
+      "step": 2094
+    },
+    {
+      "epoch": 2.072205736894164,
+      "grad_norm": 0.297906315031933,
+      "learning_rate": 1.7185782337852692e-05,
+      "loss": 0.3444,
+      "step": 2095
+    },
+    {
+      "epoch": 2.0731948565776457,
+      "grad_norm": 0.2744309054019997,
+      "learning_rate": 1.7167460608281422e-05,
+      "loss": 0.3133,
+      "step": 2096
+    },
+    {
+      "epoch": 2.0741839762611276,
+      "grad_norm": 0.24571474785996675,
+      "learning_rate": 1.714913887871015e-05,
+      "loss": 0.3276,
+      "step": 2097
+    },
+    {
+      "epoch": 2.075173095944609,
+      "grad_norm": 0.26936390193781384,
+      "learning_rate": 1.7130817149138882e-05,
+      "loss": 0.3773,
+      "step": 2098
+    },
+    {
+      "epoch": 2.076162215628091,
+      "grad_norm": 0.27132705449446876,
+      "learning_rate": 1.7112495419567608e-05,
+      "loss": 0.3354,
+      "step": 2099
+    },
+    {
+      "epoch": 2.077151335311573,
+      "grad_norm": 0.26429280776805036,
+      "learning_rate": 1.7094173689996338e-05,
+      "loss": 0.3582,
+      "step": 2100
+    },
+    {
+      "epoch": 2.0781404549950544,
+      "grad_norm": 0.2312406462125731,
+      "learning_rate": 1.7075851960425065e-05,
+      "loss": 0.3097,
+      "step": 2101
+    },
+    {
+      "epoch": 2.079129574678536,
+      "grad_norm": 0.24469123822007616,
+      "learning_rate": 1.7057530230853794e-05,
+      "loss": 0.3485,
+      "step": 2102
+    },
+    {
+      "epoch": 2.080118694362018,
+      "grad_norm": 0.2319289399183783,
+      "learning_rate": 1.703920850128252e-05,
+      "loss": 0.3115,
+      "step": 2103
+    },
+    {
+      "epoch": 2.0811078140454997,
+      "grad_norm": 0.2528130883133997,
+      "learning_rate": 1.702088677171125e-05,
+      "loss": 0.3154,
+      "step": 2104
+    },
+    {
+      "epoch": 2.082096933728981,
+      "grad_norm": 0.2530788330356479,
+      "learning_rate": 1.7002565042139977e-05,
+      "loss": 0.3571,
+      "step": 2105
+    },
+    {
+      "epoch": 2.083086053412463,
+      "grad_norm": 0.25288858939964987,
+      "learning_rate": 1.6984243312568707e-05,
+      "loss": 0.3261,
+      "step": 2106
+    },
+    {
+      "epoch": 2.0840751730959446,
+      "grad_norm": 0.2573111175053986,
+      "learning_rate": 1.6965921582997437e-05,
+      "loss": 0.3312,
+      "step": 2107
+    },
+    {
+      "epoch": 2.0850642927794265,
+      "grad_norm": 0.24283051150823262,
+      "learning_rate": 1.6947599853426163e-05,
+      "loss": 0.331,
+      "step": 2108
+    },
+    {
+      "epoch": 2.086053412462908,
+      "grad_norm": 0.7752505829926737,
+      "learning_rate": 1.6929278123854893e-05,
+      "loss": 0.343,
+      "step": 2109
+    },
+    {
+      "epoch": 2.08704253214639,
+      "grad_norm": 0.29407470466934255,
+      "learning_rate": 1.6910956394283623e-05,
+      "loss": 0.3481,
+      "step": 2110
+    },
+    {
+      "epoch": 2.0880316518298714,
+      "grad_norm": 0.24568189642774338,
+      "learning_rate": 1.689263466471235e-05,
+      "loss": 0.3132,
+      "step": 2111
+    },
+    {
+      "epoch": 2.0890207715133533,
+      "grad_norm": 0.23470405276109677,
+      "learning_rate": 1.687431293514108e-05,
+      "loss": 0.305,
+      "step": 2112
+    },
+    {
+      "epoch": 2.090009891196835,
+      "grad_norm": 0.26268923309496633,
+      "learning_rate": 1.685599120556981e-05,
+      "loss": 0.2962,
+      "step": 2113
+    },
+    {
+      "epoch": 2.0909990108803167,
+      "grad_norm": 0.26191345194398796,
+      "learning_rate": 1.6837669475998535e-05,
+      "loss": 0.3463,
+      "step": 2114
+    },
+    {
+      "epoch": 2.091988130563798,
+      "grad_norm": 1.0669555775783601,
+      "learning_rate": 1.6819347746427265e-05,
+      "loss": 0.3839,
+      "step": 2115
+    },
+    {
+      "epoch": 2.09297725024728,
+      "grad_norm": 0.2443401607557005,
+      "learning_rate": 1.680102601685599e-05,
+      "loss": 0.3496,
+      "step": 2116
+    },
+    {
+      "epoch": 2.0939663699307616,
+      "grad_norm": 0.2788783097347,
+      "learning_rate": 1.678270428728472e-05,
+      "loss": 0.3646,
+      "step": 2117
+    },
+    {
+      "epoch": 2.0949554896142435,
+      "grad_norm": 0.2808482996134958,
+      "learning_rate": 1.6764382557713447e-05,
+      "loss": 0.3222,
+      "step": 2118
+    },
+    {
+      "epoch": 2.095944609297725,
+      "grad_norm": 0.27568916240187386,
+      "learning_rate": 1.6746060828142177e-05,
+      "loss": 0.3844,
+      "step": 2119
+    },
+    {
+      "epoch": 2.096933728981207,
+      "grad_norm": 0.2581105035106014,
+      "learning_rate": 1.6727739098570904e-05,
+      "loss": 0.3115,
+      "step": 2120
+    },
+    {
+      "epoch": 2.0979228486646884,
+      "grad_norm": 0.2541721070612834,
+      "learning_rate": 1.6709417368999634e-05,
+      "loss": 0.3393,
+      "step": 2121
+    },
+    {
+      "epoch": 2.0989119683481703,
+      "grad_norm": 0.23819664141370792,
+      "learning_rate": 1.6691095639428363e-05,
+      "loss": 0.3133,
+      "step": 2122
+    },
+    {
+      "epoch": 2.0999010880316518,
+      "grad_norm": 0.2557320947503553,
+      "learning_rate": 1.6672773909857093e-05,
+      "loss": 0.3477,
+      "step": 2123
+    },
+    {
+      "epoch": 2.1008902077151337,
+      "grad_norm": 0.2271503017475219,
+      "learning_rate": 1.665445218028582e-05,
+      "loss": 0.324,
+      "step": 2124
+    },
+    {
+      "epoch": 2.101879327398615,
+      "grad_norm": 0.2863216334585617,
+      "learning_rate": 1.663613045071455e-05,
+      "loss": 0.3486,
+      "step": 2125
+    },
+    {
+      "epoch": 2.102868447082097,
+      "grad_norm": 0.2855347267238924,
+      "learning_rate": 1.6617808721143276e-05,
+      "loss": 0.3575,
+      "step": 2126
+    },
+    {
+      "epoch": 2.1038575667655786,
+      "grad_norm": 0.23052832309475305,
+      "learning_rate": 1.6599486991572006e-05,
+      "loss": 0.3211,
+      "step": 2127
+    },
+    {
+      "epoch": 2.1048466864490605,
+      "grad_norm": 0.24044171148061522,
+      "learning_rate": 1.6581165262000732e-05,
+      "loss": 0.3517,
+      "step": 2128
+    },
+    {
+      "epoch": 2.105835806132542,
+      "grad_norm": 0.26424268988397426,
+      "learning_rate": 1.6562843532429462e-05,
+      "loss": 0.3354,
+      "step": 2129
+    },
+    {
+      "epoch": 2.106824925816024,
+      "grad_norm": 0.32985801683730653,
+      "learning_rate": 1.654452180285819e-05,
+      "loss": 0.3937,
+      "step": 2130
+    },
+    {
+      "epoch": 2.1078140454995054,
+      "grad_norm": 0.2851148291510856,
+      "learning_rate": 1.6526200073286918e-05,
+      "loss": 0.369,
+      "step": 2131
+    },
+    {
+      "epoch": 2.1088031651829873,
+      "grad_norm": 0.2631999833247007,
+      "learning_rate": 1.6507878343715648e-05,
+      "loss": 0.3141,
+      "step": 2132
+    },
+    {
+      "epoch": 2.1097922848664687,
+      "grad_norm": 0.27216770136236385,
+      "learning_rate": 1.6489556614144374e-05,
+      "loss": 0.3707,
+      "step": 2133
+    },
+    {
+      "epoch": 2.1107814045499507,
+      "grad_norm": 0.26372284319992984,
+      "learning_rate": 1.6471234884573104e-05,
+      "loss": 0.3432,
+      "step": 2134
+    },
+    {
+      "epoch": 2.111770524233432,
+      "grad_norm": 2.7801144799732995,
+      "learning_rate": 1.6452913155001834e-05,
+      "loss": 0.4697,
+      "step": 2135
+    },
+    {
+      "epoch": 2.112759643916914,
+      "grad_norm": 0.3911000168082124,
+      "learning_rate": 1.6434591425430564e-05,
+      "loss": 0.3456,
+      "step": 2136
+    },
+    {
+      "epoch": 2.1137487636003955,
+      "grad_norm": 0.39596832117969594,
+      "learning_rate": 1.641626969585929e-05,
+      "loss": 0.2952,
+      "step": 2137
+    },
+    {
+      "epoch": 2.1147378832838775,
+      "grad_norm": 0.29291151380810526,
+      "learning_rate": 1.639794796628802e-05,
+      "loss": 0.3477,
+      "step": 2138
+    },
+    {
+      "epoch": 2.115727002967359,
+      "grad_norm": 0.2601124431784706,
+      "learning_rate": 1.6379626236716746e-05,
+      "loss": 0.4013,
+      "step": 2139
+    },
+    {
+      "epoch": 2.116716122650841,
+      "grad_norm": 0.3076184020471208,
+      "learning_rate": 1.6361304507145476e-05,
+      "loss": 0.3481,
+      "step": 2140
+    },
+    {
+      "epoch": 2.1177052423343223,
+      "grad_norm": 0.31115479146294667,
+      "learning_rate": 1.6342982777574203e-05,
+      "loss": 0.3502,
+      "step": 2141
+    },
+    {
+      "epoch": 2.1186943620178043,
+      "grad_norm": 0.2729157234349751,
+      "learning_rate": 1.6324661048002932e-05,
+      "loss": 0.3569,
+      "step": 2142
+    },
+    {
+      "epoch": 2.1196834817012857,
+      "grad_norm": 0.2522458263760746,
+      "learning_rate": 1.630633931843166e-05,
+      "loss": 0.3964,
+      "step": 2143
+    },
+    {
+      "epoch": 2.1206726013847677,
+      "grad_norm": 0.26441364119229976,
+      "learning_rate": 1.628801758886039e-05,
+      "loss": 0.3243,
+      "step": 2144
+    },
+    {
+      "epoch": 2.121661721068249,
+      "grad_norm": 0.26478323771121154,
+      "learning_rate": 1.6269695859289115e-05,
+      "loss": 0.3417,
+      "step": 2145
+    },
+    {
+      "epoch": 2.122650840751731,
+      "grad_norm": 5.200418924621773,
+      "learning_rate": 1.6251374129717848e-05,
+      "loss": 0.9442,
+      "step": 2146
+    },
+    {
+      "epoch": 2.1236399604352125,
+      "grad_norm": 0.26631129417116767,
+      "learning_rate": 1.6233052400146575e-05,
+      "loss": 0.3734,
+      "step": 2147
+    },
+    {
+      "epoch": 2.1246290801186944,
+      "grad_norm": 0.2397373984722219,
+      "learning_rate": 1.6214730670575304e-05,
+      "loss": 0.3142,
+      "step": 2148
+    },
+    {
+      "epoch": 2.125618199802176,
+      "grad_norm": 0.24875529064629065,
+      "learning_rate": 1.619640894100403e-05,
+      "loss": 0.3487,
+      "step": 2149
+    },
+    {
+      "epoch": 2.126607319485658,
+      "grad_norm": 0.265519222625967,
+      "learning_rate": 1.617808721143276e-05,
+      "loss": 0.319,
+      "step": 2150
+    },
+    {
+      "epoch": 2.1275964391691393,
+      "grad_norm": 0.2842533650962704,
+      "learning_rate": 1.6159765481861487e-05,
+      "loss": 0.3824,
+      "step": 2151
+    },
+    {
+      "epoch": 2.1285855588526212,
+      "grad_norm": 0.22298739828635666,
+      "learning_rate": 1.6141443752290217e-05,
+      "loss": 0.2756,
+      "step": 2152
+    },
+    {
+      "epoch": 2.1295746785361027,
+      "grad_norm": 0.25735506163209704,
+      "learning_rate": 1.6123122022718947e-05,
+      "loss": 0.3611,
+      "step": 2153
+    },
+    {
+      "epoch": 2.1305637982195846,
+      "grad_norm": 0.2818421075725635,
+      "learning_rate": 1.6104800293147673e-05,
+      "loss": 0.3923,
+      "step": 2154
+    },
+    {
+      "epoch": 2.131552917903066,
+      "grad_norm": 0.2754159452458974,
+      "learning_rate": 1.6086478563576403e-05,
+      "loss": 0.3704,
+      "step": 2155
+    },
+    {
+      "epoch": 2.132542037586548,
+      "grad_norm": 0.23952536288599155,
+      "learning_rate": 1.606815683400513e-05,
+      "loss": 0.318,
+      "step": 2156
+    },
+    {
+      "epoch": 2.1335311572700295,
+      "grad_norm": 0.25008759238683226,
+      "learning_rate": 1.604983510443386e-05,
+      "loss": 0.3357,
+      "step": 2157
+    },
+    {
+      "epoch": 2.1345202769535114,
+      "grad_norm": 0.27461726260777686,
+      "learning_rate": 1.603151337486259e-05,
+      "loss": 0.3535,
+      "step": 2158
+    },
+    {
+      "epoch": 2.135509396636993,
+      "grad_norm": 0.27788201287671715,
+      "learning_rate": 1.601319164529132e-05,
+      "loss": 0.3793,
+      "step": 2159
+    },
+    {
+      "epoch": 2.136498516320475,
+      "grad_norm": 0.26150100655181496,
+      "learning_rate": 1.5994869915720045e-05,
+      "loss": 0.3318,
+      "step": 2160
+    },
+    {
+      "epoch": 2.1374876360039563,
+      "grad_norm": 0.277747766506713,
+      "learning_rate": 1.5976548186148775e-05,
+      "loss": 0.356,
+      "step": 2161
+    },
+    {
+      "epoch": 2.1384767556874382,
+      "grad_norm": 0.25722497139475575,
+      "learning_rate": 1.59582264565775e-05,
+      "loss": 0.3627,
+      "step": 2162
+    },
+    {
+      "epoch": 2.1394658753709197,
+      "grad_norm": 0.23571960846153608,
+      "learning_rate": 1.593990472700623e-05,
+      "loss": 0.3348,
+      "step": 2163
+    },
+    {
+      "epoch": 2.1404549950544016,
+      "grad_norm": 0.2968932627364013,
+      "learning_rate": 1.5921582997434958e-05,
+      "loss": 0.3604,
+      "step": 2164
+    },
+    {
+      "epoch": 2.141444114737883,
+      "grad_norm": 0.2755392915657586,
+      "learning_rate": 1.5903261267863687e-05,
+      "loss": 0.3541,
+      "step": 2165
+    },
+    {
+      "epoch": 2.142433234421365,
+      "grad_norm": 0.23899316297498144,
+      "learning_rate": 1.5884939538292414e-05,
+      "loss": 0.3257,
+      "step": 2166
+    },
+    {
+      "epoch": 2.1434223541048465,
+      "grad_norm": 0.25067453063367884,
+      "learning_rate": 1.5866617808721144e-05,
+      "loss": 0.3296,
+      "step": 2167
+    },
+    {
+      "epoch": 2.1444114737883284,
+      "grad_norm": 0.24598510268818005,
+      "learning_rate": 1.584829607914987e-05,
+      "loss": 0.3175,
+      "step": 2168
+    },
+    {
+      "epoch": 2.14540059347181,
+      "grad_norm": 0.24842567923116413,
+      "learning_rate": 1.58299743495786e-05,
+      "loss": 0.3386,
+      "step": 2169
+    },
+    {
+      "epoch": 2.146389713155292,
+      "grad_norm": 0.279977258093171,
+      "learning_rate": 1.581165262000733e-05,
+      "loss": 0.3838,
+      "step": 2170
+    },
+    {
+      "epoch": 2.1473788328387733,
+      "grad_norm": 0.30234421674111595,
+      "learning_rate": 1.579333089043606e-05,
+      "loss": 0.3861,
+      "step": 2171
+    },
+    {
+      "epoch": 2.148367952522255,
+      "grad_norm": 0.23537629559910506,
+      "learning_rate": 1.5775009160864786e-05,
+      "loss": 0.329,
+      "step": 2172
+    },
+    {
+      "epoch": 2.1493570722057367,
+      "grad_norm": 0.27012138733219354,
+      "learning_rate": 1.5756687431293516e-05,
+      "loss": 0.3325,
+      "step": 2173
+    },
+    {
+      "epoch": 2.1503461918892186,
+      "grad_norm": 0.24908855134556376,
+      "learning_rate": 1.5738365701722242e-05,
+      "loss": 0.3055,
+      "step": 2174
+    },
+    {
+      "epoch": 2.1513353115727005,
+      "grad_norm": 0.22465760736037585,
+      "learning_rate": 1.5720043972150972e-05,
+      "loss": 0.3314,
+      "step": 2175
+    },
+    {
+      "epoch": 2.152324431256182,
+      "grad_norm": 0.24248671493308285,
+      "learning_rate": 1.57017222425797e-05,
+      "loss": 0.3247,
+      "step": 2176
+    },
+    {
+      "epoch": 2.1533135509396635,
+      "grad_norm": 0.22953998434558445,
+      "learning_rate": 1.5683400513008428e-05,
+      "loss": 0.2973,
+      "step": 2177
+    },
+    {
+      "epoch": 2.1543026706231454,
+      "grad_norm": 0.22595938467456078,
+      "learning_rate": 1.5665078783437158e-05,
+      "loss": 0.3005,
+      "step": 2178
+    },
+    {
+      "epoch": 2.1552917903066273,
+      "grad_norm": 0.2506828155398757,
+      "learning_rate": 1.5646757053865884e-05,
+      "loss": 0.3772,
+      "step": 2179
+    },
+    {
+      "epoch": 2.156280909990109,
+      "grad_norm": 0.25791925829552964,
+      "learning_rate": 1.5628435324294614e-05,
+      "loss": 0.3562,
+      "step": 2180
+    },
+    {
+      "epoch": 2.1572700296735903,
+      "grad_norm": 0.24209818620770746,
+      "learning_rate": 1.5610113594723344e-05,
+      "loss": 0.2983,
+      "step": 2181
+    },
+    {
+      "epoch": 2.158259149357072,
+      "grad_norm": 0.23728500789942555,
+      "learning_rate": 1.5591791865152074e-05,
+      "loss": 0.3153,
+      "step": 2182
+    },
+    {
+      "epoch": 2.159248269040554,
+      "grad_norm": 0.26976125221293046,
+      "learning_rate": 1.55734701355808e-05,
+      "loss": 0.3283,
+      "step": 2183
+    },
+    {
+      "epoch": 2.1602373887240356,
+      "grad_norm": 0.2401119717618975,
+      "learning_rate": 1.555514840600953e-05,
+      "loss": 0.3377,
+      "step": 2184
+    },
+    {
+      "epoch": 2.1612265084075175,
+      "grad_norm": 0.2321954749389243,
+      "learning_rate": 1.5536826676438256e-05,
+      "loss": 0.3271,
+      "step": 2185
+    },
+    {
+      "epoch": 2.162215628090999,
+      "grad_norm": 0.23834291539778332,
+      "learning_rate": 1.5518504946866986e-05,
+      "loss": 0.3227,
+      "step": 2186
+    },
+    {
+      "epoch": 2.163204747774481,
+      "grad_norm": 0.3017365420246128,
+      "learning_rate": 1.5500183217295713e-05,
+      "loss": 0.3912,
+      "step": 2187
+    },
+    {
+      "epoch": 2.1641938674579624,
+      "grad_norm": 0.25939756442962536,
+      "learning_rate": 1.5481861487724442e-05,
+      "loss": 0.3645,
+      "step": 2188
+    },
+    {
+      "epoch": 2.1651829871414443,
+      "grad_norm": 0.2671540694432964,
+      "learning_rate": 1.546353975815317e-05,
+      "loss": 0.3464,
+      "step": 2189
+    },
+    {
+      "epoch": 2.166172106824926,
+      "grad_norm": 0.24114920876599572,
+      "learning_rate": 1.54452180285819e-05,
+      "loss": 0.3318,
+      "step": 2190
+    },
+    {
+      "epoch": 2.1671612265084077,
+      "grad_norm": 0.23250896713005623,
+      "learning_rate": 1.5426896299010625e-05,
+      "loss": 0.3018,
+      "step": 2191
+    },
+    {
+      "epoch": 2.168150346191889,
+      "grad_norm": 0.25900429796077135,
+      "learning_rate": 1.5408574569439355e-05,
+      "loss": 0.3794,
+      "step": 2192
+    },
+    {
+      "epoch": 2.169139465875371,
+      "grad_norm": 0.2534934876573256,
+      "learning_rate": 1.5390252839868085e-05,
+      "loss": 0.3412,
+      "step": 2193
+    },
+    {
+      "epoch": 2.1701285855588526,
+      "grad_norm": 0.3237099924662489,
+      "learning_rate": 1.5371931110296814e-05,
+      "loss": 0.3876,
+      "step": 2194
+    },
+    {
+      "epoch": 2.1711177052423345,
+      "grad_norm": 0.2263325734121825,
+      "learning_rate": 1.535360938072554e-05,
+      "loss": 0.3421,
+      "step": 2195
+    },
+    {
+      "epoch": 2.172106824925816,
+      "grad_norm": 0.27015210617952656,
+      "learning_rate": 1.533528765115427e-05,
+      "loss": 0.325,
+      "step": 2196
+    },
+    {
+      "epoch": 2.173095944609298,
+      "grad_norm": 0.24041857827848195,
+      "learning_rate": 1.5316965921582997e-05,
+      "loss": 0.3352,
+      "step": 2197
+    },
+    {
+      "epoch": 2.1740850642927794,
+      "grad_norm": 0.2303276894642009,
+      "learning_rate": 1.5298644192011727e-05,
+      "loss": 0.3082,
+      "step": 2198
+    },
+    {
+      "epoch": 2.1750741839762613,
+      "grad_norm": 0.22582439744525307,
+      "learning_rate": 1.5280322462440457e-05,
+      "loss": 0.3294,
+      "step": 2199
+    },
+    {
+      "epoch": 2.1760633036597428,
+      "grad_norm": 0.24683399509790616,
+      "learning_rate": 1.5262000732869183e-05,
+      "loss": 0.3577,
+      "step": 2200
+    },
+    {
+      "epoch": 2.1770524233432247,
+      "grad_norm": 0.2370261131813352,
+      "learning_rate": 1.5243679003297911e-05,
+      "loss": 0.3411,
+      "step": 2201
+    },
+    {
+      "epoch": 2.178041543026706,
+      "grad_norm": 0.24781478654496875,
+      "learning_rate": 1.522535727372664e-05,
+      "loss": 0.3316,
+      "step": 2202
+    },
+    {
+      "epoch": 2.179030662710188,
+      "grad_norm": 0.23239061025797889,
+      "learning_rate": 1.5207035544155367e-05,
+      "loss": 0.3263,
+      "step": 2203
+    },
+    {
+      "epoch": 2.1800197823936696,
+      "grad_norm": 0.2232722530248141,
+      "learning_rate": 1.5188713814584096e-05,
+      "loss": 0.3081,
+      "step": 2204
+    },
+    {
+      "epoch": 2.1810089020771515,
+      "grad_norm": 0.2594820299682084,
+      "learning_rate": 1.5170392085012827e-05,
+      "loss": 0.3653,
+      "step": 2205
+    },
+    {
+      "epoch": 2.181998021760633,
+      "grad_norm": 0.24820776329147887,
+      "learning_rate": 1.5152070355441555e-05,
+      "loss": 0.3115,
+      "step": 2206
+    },
+    {
+      "epoch": 2.182987141444115,
+      "grad_norm": 0.4308536837119438,
+      "learning_rate": 1.5133748625870283e-05,
+      "loss": 0.3583,
+      "step": 2207
+    },
+    {
+      "epoch": 2.1839762611275964,
+      "grad_norm": 0.2447733062324265,
+      "learning_rate": 1.5115426896299011e-05,
+      "loss": 0.3242,
+      "step": 2208
+    },
+    {
+      "epoch": 2.1849653808110783,
+      "grad_norm": 0.255212859520126,
+      "learning_rate": 1.509710516672774e-05,
+      "loss": 0.3407,
+      "step": 2209
+    },
+    {
+      "epoch": 2.1859545004945597,
+      "grad_norm": 0.24780445254507258,
+      "learning_rate": 1.5078783437156468e-05,
+      "loss": 0.3777,
+      "step": 2210
+    },
+    {
+      "epoch": 2.1869436201780417,
+      "grad_norm": 0.22667950198277595,
+      "learning_rate": 1.5060461707585197e-05,
+      "loss": 0.322,
+      "step": 2211
+    },
+    {
+      "epoch": 2.187932739861523,
+      "grad_norm": 0.2736208075026148,
+      "learning_rate": 1.5042139978013926e-05,
+      "loss": 0.3435,
+      "step": 2212
+    },
+    {
+      "epoch": 2.188921859545005,
+      "grad_norm": 0.25076062434160956,
+      "learning_rate": 1.5023818248442654e-05,
+      "loss": 0.3341,
+      "step": 2213
+    },
+    {
+      "epoch": 2.1899109792284865,
+      "grad_norm": 0.2516337328032655,
+      "learning_rate": 1.5005496518871382e-05,
+      "loss": 0.3767,
+      "step": 2214
+    },
+    {
+      "epoch": 2.1909000989119685,
+      "grad_norm": 0.24007131228024153,
+      "learning_rate": 1.498717478930011e-05,
+      "loss": 0.3304,
+      "step": 2215
+    },
+    {
+      "epoch": 2.19188921859545,
+      "grad_norm": 0.2653627206351916,
+      "learning_rate": 1.4968853059728838e-05,
+      "loss": 0.3581,
+      "step": 2216
+    },
+    {
+      "epoch": 2.192878338278932,
+      "grad_norm": 0.23740704954394445,
+      "learning_rate": 1.495053133015757e-05,
+      "loss": 0.3322,
+      "step": 2217
+    },
+    {
+      "epoch": 2.1938674579624133,
+      "grad_norm": 0.26183283644070443,
+      "learning_rate": 1.4932209600586298e-05,
+      "loss": 0.3058,
+      "step": 2218
+    },
+    {
+      "epoch": 2.1948565776458953,
+      "grad_norm": 0.25535363753862483,
+      "learning_rate": 1.4913887871015026e-05,
+      "loss": 0.3867,
+      "step": 2219
+    },
+    {
+      "epoch": 2.1958456973293767,
+      "grad_norm": 0.2410531896962311,
+      "learning_rate": 1.4895566141443754e-05,
+      "loss": 0.3287,
+      "step": 2220
+    },
+    {
+      "epoch": 2.1968348170128587,
+      "grad_norm": 0.23864241175081596,
+      "learning_rate": 1.4877244411872482e-05,
+      "loss": 0.2957,
+      "step": 2221
+    },
+    {
+      "epoch": 2.19782393669634,
+      "grad_norm": 0.23152858294826056,
+      "learning_rate": 1.485892268230121e-05,
+      "loss": 0.3267,
+      "step": 2222
+    },
+    {
+      "epoch": 2.198813056379822,
+      "grad_norm": 0.25038544912526106,
+      "learning_rate": 1.4840600952729938e-05,
+      "loss": 0.3727,
+      "step": 2223
+    },
+    {
+      "epoch": 2.1998021760633035,
+      "grad_norm": 0.24575360880677702,
+      "learning_rate": 1.4822279223158666e-05,
+      "loss": 0.3204,
+      "step": 2224
+    },
+    {
+      "epoch": 2.2007912957467854,
+      "grad_norm": 0.2617185419544845,
+      "learning_rate": 1.4803957493587394e-05,
+      "loss": 0.3344,
+      "step": 2225
+    },
+    {
+      "epoch": 2.201780415430267,
+      "grad_norm": 0.23621554765095076,
+      "learning_rate": 1.4785635764016123e-05,
+      "loss": 0.3194,
+      "step": 2226
+    },
+    {
+      "epoch": 2.202769535113749,
+      "grad_norm": 0.2656714021948276,
+      "learning_rate": 1.476731403444485e-05,
+      "loss": 0.3251,
+      "step": 2227
+    },
+    {
+      "epoch": 2.2037586547972303,
+      "grad_norm": 0.2352678555408159,
+      "learning_rate": 1.4748992304873579e-05,
+      "loss": 0.3474,
+      "step": 2228
+    },
+    {
+      "epoch": 2.2047477744807122,
+      "grad_norm": 0.23365121088195945,
+      "learning_rate": 1.473067057530231e-05,
+      "loss": 0.3214,
+      "step": 2229
+    },
+    {
+      "epoch": 2.2057368941641937,
+      "grad_norm": 0.2455996352228685,
+      "learning_rate": 1.4712348845731038e-05,
+      "loss": 0.4093,
+      "step": 2230
+    },
+    {
+      "epoch": 2.2067260138476756,
+      "grad_norm": 0.2740975848282929,
+      "learning_rate": 1.4694027116159766e-05,
+      "loss": 0.3399,
+      "step": 2231
+    },
+    {
+      "epoch": 2.207715133531157,
+      "grad_norm": 0.2598627296070607,
+      "learning_rate": 1.4675705386588495e-05,
+      "loss": 0.3083,
+      "step": 2232
+    },
+    {
+      "epoch": 2.208704253214639,
+      "grad_norm": 0.23897597759120234,
+      "learning_rate": 1.4657383657017223e-05,
+      "loss": 0.3551,
+      "step": 2233
+    },
+    {
+      "epoch": 2.2096933728981205,
+      "grad_norm": 0.28717030628471557,
+      "learning_rate": 1.4639061927445952e-05,
+      "loss": 0.36,
+      "step": 2234
+    },
+    {
+      "epoch": 2.2106824925816024,
+      "grad_norm": 0.2551734546245449,
+      "learning_rate": 1.462074019787468e-05,
+      "loss": 0.2981,
+      "step": 2235
+    },
+    {
+      "epoch": 2.211671612265084,
+      "grad_norm": 0.22660711986812593,
+      "learning_rate": 1.4602418468303409e-05,
+      "loss": 0.3241,
+      "step": 2236
+    },
+    {
+      "epoch": 2.212660731948566,
+      "grad_norm": 0.22072364142005796,
+      "learning_rate": 1.4584096738732137e-05,
+      "loss": 0.3397,
+      "step": 2237
+    },
+    {
+      "epoch": 2.2136498516320473,
+      "grad_norm": 0.25077102419811953,
+      "learning_rate": 1.4565775009160865e-05,
+      "loss": 0.3007,
+      "step": 2238
+    },
+    {
+      "epoch": 2.2146389713155292,
+      "grad_norm": 0.23275953912915226,
+      "learning_rate": 1.4547453279589593e-05,
+      "loss": 0.3301,
+      "step": 2239
+    },
+    {
+      "epoch": 2.2156280909990107,
+      "grad_norm": 0.2552844764105821,
+      "learning_rate": 1.4529131550018321e-05,
+      "loss": 0.3359,
+      "step": 2240
+    },
+    {
+      "epoch": 2.2166172106824926,
+      "grad_norm": 0.2209011298450527,
+      "learning_rate": 1.4510809820447053e-05,
+      "loss": 0.3598,
+      "step": 2241
+    },
+    {
+      "epoch": 2.217606330365974,
+      "grad_norm": 1.480125764071416,
+      "learning_rate": 1.449248809087578e-05,
+      "loss": 0.3688,
+      "step": 2242
+    },
+    {
+      "epoch": 2.218595450049456,
+      "grad_norm": 0.30985712503457974,
+      "learning_rate": 1.4474166361304509e-05,
+      "loss": 0.3408,
+      "step": 2243
+    },
+    {
+      "epoch": 2.2195845697329375,
+      "grad_norm": 0.24892275955536927,
+      "learning_rate": 1.4455844631733237e-05,
+      "loss": 0.3085,
+      "step": 2244
+    },
+    {
+      "epoch": 2.2205736894164194,
+      "grad_norm": 0.24101322361416325,
+      "learning_rate": 1.4437522902161965e-05,
+      "loss": 0.3437,
+      "step": 2245
+    },
+    {
+      "epoch": 2.221562809099901,
+      "grad_norm": 0.2910916695987308,
+      "learning_rate": 1.4419201172590693e-05,
+      "loss": 0.342,
+      "step": 2246
+    },
+    {
+      "epoch": 2.222551928783383,
+      "grad_norm": 0.2761723063225733,
+      "learning_rate": 1.4400879443019421e-05,
+      "loss": 0.3252,
+      "step": 2247
+    },
+    {
+      "epoch": 2.2235410484668643,
+      "grad_norm": 0.23723078949172213,
+      "learning_rate": 1.438255771344815e-05,
+      "loss": 0.3535,
+      "step": 2248
+    },
+    {
+      "epoch": 2.224530168150346,
+      "grad_norm": 0.23083729839007824,
+      "learning_rate": 1.4364235983876878e-05,
+      "loss": 0.3462,
+      "step": 2249
+    },
+    {
+      "epoch": 2.2255192878338277,
+      "grad_norm": 0.24923526336688007,
+      "learning_rate": 1.4345914254305606e-05,
+      "loss": 0.3479,
+      "step": 2250
+    },
+    {
+      "epoch": 2.2265084075173096,
+      "grad_norm": 0.2365620936428639,
+      "learning_rate": 1.4327592524734334e-05,
+      "loss": 0.3341,
+      "step": 2251
+    },
+    {
+      "epoch": 2.227497527200791,
+      "grad_norm": 0.26201081319476877,
+      "learning_rate": 1.4309270795163064e-05,
+      "loss": 0.3605,
+      "step": 2252
+    },
+    {
+      "epoch": 2.228486646884273,
+      "grad_norm": 0.23329839147411013,
+      "learning_rate": 1.4290949065591793e-05,
+      "loss": 0.3275,
+      "step": 2253
+    },
+    {
+      "epoch": 2.229475766567755,
+      "grad_norm": 0.23449570191396432,
+      "learning_rate": 1.4272627336020521e-05,
+      "loss": 0.3344,
+      "step": 2254
+    },
+    {
+      "epoch": 2.2304648862512364,
+      "grad_norm": 0.25077592463879617,
+      "learning_rate": 1.425430560644925e-05,
+      "loss": 0.3158,
+      "step": 2255
+    },
+    {
+      "epoch": 2.231454005934718,
+      "grad_norm": 0.2638212095862189,
+      "learning_rate": 1.4235983876877978e-05,
+      "loss": 0.3905,
+      "step": 2256
+    },
+    {
+      "epoch": 2.2324431256182,
+      "grad_norm": 0.2281528691605368,
+      "learning_rate": 1.4217662147306708e-05,
+      "loss": 0.3068,
+      "step": 2257
+    },
+    {
+      "epoch": 2.2334322453016817,
+      "grad_norm": 0.24868299339572644,
+      "learning_rate": 1.4199340417735436e-05,
+      "loss": 0.3352,
+      "step": 2258
+    },
+    {
+      "epoch": 2.234421364985163,
+      "grad_norm": 0.25226182191275837,
+      "learning_rate": 1.4181018688164164e-05,
+      "loss": 0.3799,
+      "step": 2259
+    },
+    {
+      "epoch": 2.2354104846686447,
+      "grad_norm": 0.23719055969275504,
+      "learning_rate": 1.4162696958592892e-05,
+      "loss": 0.285,
+      "step": 2260
+    },
+    {
+      "epoch": 2.2363996043521266,
+      "grad_norm": 0.22645086719819674,
+      "learning_rate": 1.414437522902162e-05,
+      "loss": 0.3535,
+      "step": 2261
+    },
+    {
+      "epoch": 2.2373887240356085,
+      "grad_norm": 0.2378652475705354,
+      "learning_rate": 1.4126053499450348e-05,
+      "loss": 0.322,
+      "step": 2262
+    },
+    {
+      "epoch": 2.23837784371909,
+      "grad_norm": 0.2446078244591778,
+      "learning_rate": 1.4107731769879076e-05,
+      "loss": 0.3008,
+      "step": 2263
+    },
+    {
+      "epoch": 2.239366963402572,
+      "grad_norm": 0.23473892675122612,
+      "learning_rate": 1.4089410040307804e-05,
+      "loss": 0.3174,
+      "step": 2264
+    },
+    {
+      "epoch": 2.2403560830860534,
+      "grad_norm": 0.2740082445935662,
+      "learning_rate": 1.4071088310736536e-05,
+      "loss": 0.3673,
+      "step": 2265
+    },
+    {
+      "epoch": 2.2413452027695353,
+      "grad_norm": 0.3047335384705569,
+      "learning_rate": 1.4052766581165264e-05,
+      "loss": 0.3862,
+      "step": 2266
+    },
+    {
+      "epoch": 2.242334322453017,
+      "grad_norm": 0.22953115502891203,
+      "learning_rate": 1.4034444851593992e-05,
+      "loss": 0.3544,
+      "step": 2267
+    },
+    {
+      "epoch": 2.2433234421364987,
+      "grad_norm": 0.2381048121745753,
+      "learning_rate": 1.401612312202272e-05,
+      "loss": 0.385,
+      "step": 2268
+    },
+    {
+      "epoch": 2.24431256181998,
+      "grad_norm": 0.22695002295754346,
+      "learning_rate": 1.3997801392451448e-05,
+      "loss": 0.3113,
+      "step": 2269
+    },
+    {
+      "epoch": 2.245301681503462,
+      "grad_norm": 0.23494139839113953,
+      "learning_rate": 1.3979479662880176e-05,
+      "loss": 0.3006,
+      "step": 2270
+    },
+    {
+      "epoch": 2.2462908011869436,
+      "grad_norm": 0.2508107946864829,
+      "learning_rate": 1.3961157933308904e-05,
+      "loss": 0.3602,
+      "step": 2271
+    },
+    {
+      "epoch": 2.2472799208704255,
+      "grad_norm": 0.22515857517930243,
+      "learning_rate": 1.3942836203737633e-05,
+      "loss": 0.3296,
+      "step": 2272
+    },
+    {
+      "epoch": 2.248269040553907,
+      "grad_norm": 0.25467938360409764,
+      "learning_rate": 1.392451447416636e-05,
+      "loss": 0.3448,
+      "step": 2273
+    },
+    {
+      "epoch": 2.249258160237389,
+      "grad_norm": 0.30210334820145374,
+      "learning_rate": 1.3906192744595089e-05,
+      "loss": 0.3662,
+      "step": 2274
+    },
+    {
+      "epoch": 2.2502472799208704,
+      "grad_norm": 0.22040906536404206,
+      "learning_rate": 1.3887871015023819e-05,
+      "loss": 0.3222,
+      "step": 2275
+    },
+    {
+      "epoch": 2.2512363996043523,
+      "grad_norm": 0.24078053098136895,
+      "learning_rate": 1.3869549285452547e-05,
+      "loss": 0.3578,
+      "step": 2276
+    },
+    {
+      "epoch": 2.2522255192878338,
+      "grad_norm": 0.23682978022586784,
+      "learning_rate": 1.3851227555881277e-05,
+      "loss": 0.3418,
+      "step": 2277
+    },
+    {
+      "epoch": 2.2532146389713157,
+      "grad_norm": 0.25235210494112936,
+      "learning_rate": 1.3832905826310005e-05,
+      "loss": 0.342,
+      "step": 2278
+    },
+    {
+      "epoch": 2.254203758654797,
+      "grad_norm": 0.21464531728228006,
+      "learning_rate": 1.3814584096738733e-05,
+      "loss": 0.2858,
+      "step": 2279
+    },
+    {
+      "epoch": 2.255192878338279,
+      "grad_norm": 0.24947023802823698,
+      "learning_rate": 1.3796262367167463e-05,
+      "loss": 0.3581,
+      "step": 2280
+    },
+    {
+      "epoch": 2.2561819980217606,
+      "grad_norm": 0.24126311074885234,
+      "learning_rate": 1.377794063759619e-05,
+      "loss": 0.3372,
+      "step": 2281
+    },
+    {
+      "epoch": 2.2571711177052425,
+      "grad_norm": 0.24381172903295012,
+      "learning_rate": 1.3759618908024919e-05,
+      "loss": 0.3309,
+      "step": 2282
+    },
+    {
+      "epoch": 2.258160237388724,
+      "grad_norm": 0.2216901316551094,
+      "learning_rate": 1.3741297178453647e-05,
+      "loss": 0.2906,
+      "step": 2283
+    },
+    {
+      "epoch": 2.259149357072206,
+      "grad_norm": 0.22751523042939115,
+      "learning_rate": 1.3722975448882375e-05,
+      "loss": 0.327,
+      "step": 2284
+    },
+    {
+      "epoch": 2.2601384767556874,
+      "grad_norm": 0.2253518058336741,
+      "learning_rate": 1.3704653719311103e-05,
+      "loss": 0.3119,
+      "step": 2285
+    },
+    {
+      "epoch": 2.2611275964391693,
+      "grad_norm": 0.2479817150971885,
+      "learning_rate": 1.3686331989739831e-05,
+      "loss": 0.3448,
+      "step": 2286
+    },
+    {
+      "epoch": 2.2621167161226508,
+      "grad_norm": 0.2201794942968985,
+      "learning_rate": 1.366801026016856e-05,
+      "loss": 0.3606,
+      "step": 2287
+    },
+    {
+      "epoch": 2.2631058358061327,
+      "grad_norm": 0.21364590272690012,
+      "learning_rate": 1.3649688530597287e-05,
+      "loss": 0.3305,
+      "step": 2288
+    },
+    {
+      "epoch": 2.264094955489614,
+      "grad_norm": 0.24392483676889015,
+      "learning_rate": 1.3631366801026019e-05,
+      "loss": 0.3393,
+      "step": 2289
+    },
+    {
+      "epoch": 2.265084075173096,
+      "grad_norm": 0.24221769435099127,
+      "learning_rate": 1.3613045071454747e-05,
+      "loss": 0.3369,
+      "step": 2290
+    },
+    {
+      "epoch": 2.2660731948565775,
+      "grad_norm": 0.24861588082223832,
+      "learning_rate": 1.3594723341883475e-05,
+      "loss": 0.3626,
+      "step": 2291
+    },
+    {
+      "epoch": 2.2670623145400595,
+      "grad_norm": 0.23052081299132418,
+      "learning_rate": 1.3576401612312203e-05,
+      "loss": 0.3063,
+      "step": 2292
+    },
+    {
+      "epoch": 2.268051434223541,
+      "grad_norm": 0.24275686147196318,
+      "learning_rate": 1.3558079882740931e-05,
+      "loss": 0.3038,
+      "step": 2293
+    },
+    {
+      "epoch": 2.269040553907023,
+      "grad_norm": 0.22272342553004223,
+      "learning_rate": 1.353975815316966e-05,
+      "loss": 0.3094,
+      "step": 2294
+    },
+    {
+      "epoch": 2.2700296735905043,
+      "grad_norm": 0.23299197318354145,
+      "learning_rate": 1.3521436423598388e-05,
+      "loss": 0.3396,
+      "step": 2295
+    },
+    {
+      "epoch": 2.2710187932739863,
+      "grad_norm": 0.22615622555168388,
+      "learning_rate": 1.3503114694027116e-05,
+      "loss": 0.3336,
+      "step": 2296
+    },
+    {
+      "epoch": 2.2720079129574677,
+      "grad_norm": 0.2252293872395291,
+      "learning_rate": 1.3484792964455844e-05,
+      "loss": 0.2977,
+      "step": 2297
+    },
+    {
+      "epoch": 2.2729970326409497,
+      "grad_norm": 0.23457296789984822,
+      "learning_rate": 1.3466471234884574e-05,
+      "loss": 0.3399,
+      "step": 2298
+    },
+    {
+      "epoch": 2.273986152324431,
+      "grad_norm": 0.24129016983914284,
+      "learning_rate": 1.3448149505313302e-05,
+      "loss": 0.3219,
+      "step": 2299
+    },
+    {
+      "epoch": 2.274975272007913,
+      "grad_norm": 0.22141667622956007,
+      "learning_rate": 1.342982777574203e-05,
+      "loss": 0.3082,
+      "step": 2300
+    },
+    {
+      "epoch": 2.2759643916913945,
+      "grad_norm": 0.2413245232232439,
+      "learning_rate": 1.341150604617076e-05,
+      "loss": 0.3426,
+      "step": 2301
+    },
+    {
+      "epoch": 2.2769535113748764,
+      "grad_norm": 0.23696458764674602,
+      "learning_rate": 1.3393184316599488e-05,
+      "loss": 0.3759,
+      "step": 2302
+    },
+    {
+      "epoch": 2.277942631058358,
+      "grad_norm": 0.2518570901503138,
+      "learning_rate": 1.3374862587028218e-05,
+      "loss": 0.3695,
+      "step": 2303
+    },
+    {
+      "epoch": 2.27893175074184,
+      "grad_norm": 0.23086745437243067,
+      "learning_rate": 1.3356540857456946e-05,
+      "loss": 0.3369,
+      "step": 2304
+    },
+    {
+      "epoch": 2.2799208704253213,
+      "grad_norm": 0.26169884977451946,
+      "learning_rate": 1.3338219127885674e-05,
+      "loss": 0.3373,
+      "step": 2305
+    },
+    {
+      "epoch": 2.2809099901088032,
+      "grad_norm": 0.2587088888423314,
+      "learning_rate": 1.3319897398314402e-05,
+      "loss": 0.3574,
+      "step": 2306
+    },
+    {
+      "epoch": 2.2818991097922847,
+      "grad_norm": 0.22977702824971338,
+      "learning_rate": 1.330157566874313e-05,
+      "loss": 0.3457,
+      "step": 2307
+    },
+    {
+      "epoch": 2.2828882294757666,
+      "grad_norm": 0.22721408632878823,
+      "learning_rate": 1.3283253939171858e-05,
+      "loss": 0.308,
+      "step": 2308
+    },
+    {
+      "epoch": 2.283877349159248,
+      "grad_norm": 0.2519093156052609,
+      "learning_rate": 1.3264932209600586e-05,
+      "loss": 0.3589,
+      "step": 2309
+    },
+    {
+      "epoch": 2.28486646884273,
+      "grad_norm": 0.26712020676598086,
+      "learning_rate": 1.3246610480029314e-05,
+      "loss": 0.3624,
+      "step": 2310
+    },
+    {
+      "epoch": 2.2858555885262115,
+      "grad_norm": 0.2326596051044427,
+      "learning_rate": 1.3228288750458042e-05,
+      "loss": 0.3335,
+      "step": 2311
+    },
+    {
+      "epoch": 2.2868447082096934,
+      "grad_norm": 0.2765369772968104,
+      "learning_rate": 1.320996702088677e-05,
+      "loss": 0.3757,
+      "step": 2312
+    },
+    {
+      "epoch": 2.287833827893175,
+      "grad_norm": 0.25058824305887195,
+      "learning_rate": 1.3191645291315502e-05,
+      "loss": 0.3808,
+      "step": 2313
+    },
+    {
+      "epoch": 2.288822947576657,
+      "grad_norm": 0.2248644826213857,
+      "learning_rate": 1.317332356174423e-05,
+      "loss": 0.3213,
+      "step": 2314
+    },
+    {
+      "epoch": 2.2898120672601383,
+      "grad_norm": 0.23779476179693096,
+      "learning_rate": 1.3155001832172958e-05,
+      "loss": 0.3293,
+      "step": 2315
+    },
+    {
+      "epoch": 2.2908011869436202,
+      "grad_norm": 0.2723138257723175,
+      "learning_rate": 1.3136680102601686e-05,
+      "loss": 0.3428,
+      "step": 2316
+    },
+    {
+      "epoch": 2.2917903066271017,
+      "grad_norm": 0.2616699975935992,
+      "learning_rate": 1.3118358373030415e-05,
+      "loss": 0.3454,
+      "step": 2317
+    },
+    {
+      "epoch": 2.2927794263105836,
+      "grad_norm": 0.24102049258430722,
+      "learning_rate": 1.3100036643459143e-05,
+      "loss": 0.3333,
+      "step": 2318
+    },
+    {
+      "epoch": 2.293768545994065,
+      "grad_norm": 0.23650754741019725,
+      "learning_rate": 1.308171491388787e-05,
+      "loss": 0.3123,
+      "step": 2319
+    },
+    {
+      "epoch": 2.294757665677547,
+      "grad_norm": 0.37538745041783184,
+      "learning_rate": 1.3063393184316599e-05,
+      "loss": 0.369,
+      "step": 2320
+    },
+    {
+      "epoch": 2.2957467853610285,
+      "grad_norm": 0.23666732747554203,
+      "learning_rate": 1.3045071454745329e-05,
+      "loss": 0.316,
+      "step": 2321
+    },
+    {
+      "epoch": 2.2967359050445104,
+      "grad_norm": 0.22954722561453486,
+      "learning_rate": 1.3026749725174057e-05,
+      "loss": 0.3216,
+      "step": 2322
+    },
+    {
+      "epoch": 2.297725024727992,
+      "grad_norm": 0.2490149729269961,
+      "learning_rate": 1.3008427995602785e-05,
+      "loss": 0.3144,
+      "step": 2323
+    },
+    {
+      "epoch": 2.298714144411474,
+      "grad_norm": 0.25850900249259257,
+      "learning_rate": 1.2990106266031515e-05,
+      "loss": 0.3166,
+      "step": 2324
+    },
+    {
+      "epoch": 2.2997032640949557,
+      "grad_norm": 0.26057122713059194,
+      "learning_rate": 1.2971784536460243e-05,
+      "loss": 0.3526,
+      "step": 2325
+    },
+    {
+      "epoch": 2.300692383778437,
+      "grad_norm": 0.24491367081125778,
+      "learning_rate": 1.2953462806888973e-05,
+      "loss": 0.3525,
+      "step": 2326
+    },
+    {
+      "epoch": 2.3016815034619187,
+      "grad_norm": 0.2760339712557652,
+      "learning_rate": 1.29351410773177e-05,
+      "loss": 0.3596,
+      "step": 2327
+    },
+    {
+      "epoch": 2.3026706231454006,
+      "grad_norm": 0.24073219741729474,
+      "learning_rate": 1.2916819347746429e-05,
+      "loss": 0.3269,
+      "step": 2328
+    },
+    {
+      "epoch": 2.3036597428288825,
+      "grad_norm": 0.2703548932443427,
+      "learning_rate": 1.2898497618175157e-05,
+      "loss": 0.3219,
+      "step": 2329
+    },
+    {
+      "epoch": 2.304648862512364,
+      "grad_norm": 0.2687127704366253,
+      "learning_rate": 1.2880175888603885e-05,
+      "loss": 0.3417,
+      "step": 2330
+    },
+    {
+      "epoch": 2.3056379821958455,
+      "grad_norm": 0.25270161397798035,
+      "learning_rate": 1.2861854159032613e-05,
+      "loss": 0.3714,
+      "step": 2331
+    },
+    {
+      "epoch": 2.3066271018793274,
+      "grad_norm": 0.22146824391487552,
+      "learning_rate": 1.2843532429461341e-05,
+      "loss": 0.3217,
+      "step": 2332
+    },
+    {
+      "epoch": 2.3076162215628093,
+      "grad_norm": 0.2558572987685534,
+      "learning_rate": 1.282521069989007e-05,
+      "loss": 0.3753,
+      "step": 2333
+    },
+    {
+      "epoch": 2.308605341246291,
+      "grad_norm": 0.25523094828283194,
+      "learning_rate": 1.2806888970318798e-05,
+      "loss": 0.3419,
+      "step": 2334
+    },
+    {
+      "epoch": 2.3095944609297723,
+      "grad_norm": 0.24281022901050803,
+      "learning_rate": 1.2788567240747526e-05,
+      "loss": 0.3487,
+      "step": 2335
+    },
+    {
+      "epoch": 2.310583580613254,
+      "grad_norm": 0.2748946608527433,
+      "learning_rate": 1.2770245511176257e-05,
+      "loss": 0.3927,
+      "step": 2336
+    },
+    {
+      "epoch": 2.311572700296736,
+      "grad_norm": 0.24688576678949595,
+      "learning_rate": 1.2751923781604985e-05,
+      "loss": 0.3877,
+      "step": 2337
+    },
+    {
+      "epoch": 2.3125618199802176,
+      "grad_norm": 0.27181775950372444,
+      "learning_rate": 1.2733602052033713e-05,
+      "loss": 0.3822,
+      "step": 2338
+    },
+    {
+      "epoch": 2.313550939663699,
+      "grad_norm": 0.22921656595976417,
+      "learning_rate": 1.2715280322462441e-05,
+      "loss": 0.2951,
+      "step": 2339
+    },
+    {
+      "epoch": 2.314540059347181,
+      "grad_norm": 0.2575726813551104,
+      "learning_rate": 1.269695859289117e-05,
+      "loss": 0.3406,
+      "step": 2340
+    },
+    {
+      "epoch": 2.315529179030663,
+      "grad_norm": 0.2555589633497033,
+      "learning_rate": 1.2678636863319898e-05,
+      "loss": 0.3396,
+      "step": 2341
+    },
+    {
+      "epoch": 2.3165182987141444,
+      "grad_norm": 0.2364187403064698,
+      "learning_rate": 1.2660315133748626e-05,
+      "loss": 0.377,
+      "step": 2342
+    },
+    {
+      "epoch": 2.3175074183976263,
+      "grad_norm": 0.2302096798238319,
+      "learning_rate": 1.2641993404177354e-05,
+      "loss": 0.3319,
+      "step": 2343
+    },
+    {
+      "epoch": 2.318496538081108,
+      "grad_norm": 0.24888650385331362,
+      "learning_rate": 1.2623671674606084e-05,
+      "loss": 0.3066,
+      "step": 2344
+    },
+    {
+      "epoch": 2.3194856577645897,
+      "grad_norm": 0.2773610707597904,
+      "learning_rate": 1.2605349945034812e-05,
+      "loss": 0.3414,
+      "step": 2345
+    },
+    {
+      "epoch": 2.320474777448071,
+      "grad_norm": 0.2154430712098479,
+      "learning_rate": 1.258702821546354e-05,
+      "loss": 0.2968,
+      "step": 2346
+    },
+    {
+      "epoch": 2.321463897131553,
+      "grad_norm": 0.2338043867905404,
+      "learning_rate": 1.2568706485892268e-05,
+      "loss": 0.3118,
+      "step": 2347
+    },
+    {
+      "epoch": 2.3224530168150346,
+      "grad_norm": 0.24602167457573554,
+      "learning_rate": 1.2550384756320998e-05,
+      "loss": 0.3702,
+      "step": 2348
+    },
+    {
+      "epoch": 2.3234421364985165,
+      "grad_norm": 0.2327990964511457,
+      "learning_rate": 1.2532063026749728e-05,
+      "loss": 0.3523,
+      "step": 2349
+    },
+    {
+      "epoch": 2.324431256181998,
+      "grad_norm": 0.2589528675798955,
+      "learning_rate": 1.2513741297178456e-05,
+      "loss": 0.3864,
+      "step": 2350
+    },
+    {
+      "epoch": 2.32542037586548,
+      "grad_norm": 0.26176208543972257,
+      "learning_rate": 1.2495419567607184e-05,
+      "loss": 0.3336,
+      "step": 2351
+    },
+    {
+      "epoch": 2.3264094955489614,
+      "grad_norm": 0.2844298379253604,
+      "learning_rate": 1.2477097838035912e-05,
+      "loss": 0.3492,
+      "step": 2352
+    },
+    {
+      "epoch": 2.3273986152324433,
+      "grad_norm": 0.2546167667799685,
+      "learning_rate": 1.245877610846464e-05,
+      "loss": 0.3555,
+      "step": 2353
+    },
+    {
+      "epoch": 2.3283877349159248,
+      "grad_norm": 0.2696964327271954,
+      "learning_rate": 1.2440454378893368e-05,
+      "loss": 0.3706,
+      "step": 2354
+    },
+    {
+      "epoch": 2.3293768545994067,
+      "grad_norm": 0.31827984227427614,
+      "learning_rate": 1.2422132649322096e-05,
+      "loss": 0.332,
+      "step": 2355
+    },
+    {
+      "epoch": 2.330365974282888,
+      "grad_norm": 0.2240056160701879,
+      "learning_rate": 1.2403810919750824e-05,
+      "loss": 0.3328,
+      "step": 2356
+    },
+    {
+      "epoch": 2.33135509396637,
+      "grad_norm": 0.2517740200145697,
+      "learning_rate": 1.2385489190179554e-05,
+      "loss": 0.372,
+      "step": 2357
+    },
+    {
+      "epoch": 2.3323442136498516,
+      "grad_norm": 0.24940675998881365,
+      "learning_rate": 1.2367167460608282e-05,
+      "loss": 0.3959,
+      "step": 2358
+    },
+    {
+      "epoch": 2.3333333333333335,
+      "grad_norm": 0.2459642937733617,
+      "learning_rate": 1.234884573103701e-05,
+      "loss": 0.3492,
+      "step": 2359
+    },
+    {
+      "epoch": 2.334322453016815,
+      "grad_norm": 0.22768174832698604,
+      "learning_rate": 1.2330524001465739e-05,
+      "loss": 0.3404,
+      "step": 2360
+    },
+    {
+      "epoch": 2.335311572700297,
+      "grad_norm": 0.22086915383890057,
+      "learning_rate": 1.2312202271894467e-05,
+      "loss": 0.3099,
+      "step": 2361
+    },
+    {
+      "epoch": 2.3363006923837784,
+      "grad_norm": 0.22113520177366824,
+      "learning_rate": 1.2293880542323195e-05,
+      "loss": 0.3393,
+      "step": 2362
+    },
+    {
+      "epoch": 2.3372898120672603,
+      "grad_norm": 2.991113459889468,
+      "learning_rate": 1.2275558812751925e-05,
+      "loss": 0.3309,
+      "step": 2363
+    },
+    {
+      "epoch": 2.3382789317507418,
+      "grad_norm": 0.29054610148791177,
+      "learning_rate": 1.2257237083180653e-05,
+      "loss": 0.3655,
+      "step": 2364
+    },
+    {
+      "epoch": 2.3392680514342237,
+      "grad_norm": 0.24093202591978707,
+      "learning_rate": 1.2238915353609381e-05,
+      "loss": 0.3428,
+      "step": 2365
+    },
+    {
+      "epoch": 2.340257171117705,
+      "grad_norm": 0.23736300980535247,
+      "learning_rate": 1.2220593624038109e-05,
+      "loss": 0.3229,
+      "step": 2366
+    },
+    {
+      "epoch": 2.341246290801187,
+      "grad_norm": 0.35258591516934035,
+      "learning_rate": 1.2202271894466839e-05,
+      "loss": 0.3363,
+      "step": 2367
+    },
+    {
+      "epoch": 2.3422354104846685,
+      "grad_norm": 0.26676374709095013,
+      "learning_rate": 1.2183950164895567e-05,
+      "loss": 0.3211,
+      "step": 2368
+    },
+    {
+      "epoch": 2.3432245301681505,
+      "grad_norm": 0.2351851020565668,
+      "learning_rate": 1.2165628435324295e-05,
+      "loss": 0.35,
+      "step": 2369
+    },
+    {
+      "epoch": 2.344213649851632,
+      "grad_norm": 0.22232852533104622,
+      "learning_rate": 1.2147306705753025e-05,
+      "loss": 0.31,
+      "step": 2370
+    },
+    {
+      "epoch": 2.345202769535114,
+      "grad_norm": 0.23680359247744562,
+      "learning_rate": 1.2128984976181753e-05,
+      "loss": 0.3478,
+      "step": 2371
+    },
+    {
+      "epoch": 2.3461918892185953,
+      "grad_norm": 0.24988403986844843,
+      "learning_rate": 1.2110663246610481e-05,
+      "loss": 0.3635,
+      "step": 2372
+    },
+    {
+      "epoch": 2.3471810089020773,
+      "grad_norm": 0.24767501874015097,
+      "learning_rate": 1.2092341517039209e-05,
+      "loss": 0.3883,
+      "step": 2373
+    },
+    {
+      "epoch": 2.3481701285855587,
+      "grad_norm": 0.2521283236697681,
+      "learning_rate": 1.2074019787467937e-05,
+      "loss": 0.4225,
+      "step": 2374
+    },
+    {
+      "epoch": 2.3491592482690407,
+      "grad_norm": 0.24675041358192287,
+      "learning_rate": 1.2055698057896667e-05,
+      "loss": 0.3663,
+      "step": 2375
+    },
+    {
+      "epoch": 2.350148367952522,
+      "grad_norm": 0.24311636349663465,
+      "learning_rate": 1.2037376328325395e-05,
+      "loss": 0.3759,
+      "step": 2376
+    },
+    {
+      "epoch": 2.351137487636004,
+      "grad_norm": 0.24263994864791805,
+      "learning_rate": 1.2019054598754123e-05,
+      "loss": 0.38,
+      "step": 2377
+    },
+    {
+      "epoch": 2.3521266073194855,
+      "grad_norm": 0.25402272877139787,
+      "learning_rate": 1.2000732869182851e-05,
+      "loss": 0.31,
+      "step": 2378
+    },
+    {
+      "epoch": 2.3531157270029674,
+      "grad_norm": 0.2495073709635195,
+      "learning_rate": 1.198241113961158e-05,
+      "loss": 0.3659,
+      "step": 2379
+    },
+    {
+      "epoch": 2.354104846686449,
+      "grad_norm": 0.2298654377866694,
+      "learning_rate": 1.1964089410040308e-05,
+      "loss": 0.3209,
+      "step": 2380
+    },
+    {
+      "epoch": 2.355093966369931,
+      "grad_norm": 0.28863897336682676,
+      "learning_rate": 1.1945767680469037e-05,
+      "loss": 0.3417,
+      "step": 2381
+    },
+    {
+      "epoch": 2.3560830860534123,
+      "grad_norm": 0.41489670024603986,
+      "learning_rate": 1.1927445950897766e-05,
+      "loss": 0.3503,
+      "step": 2382
+    },
+    {
+      "epoch": 2.3570722057368942,
+      "grad_norm": 0.2205422492533083,
+      "learning_rate": 1.1909124221326494e-05,
+      "loss": 0.3529,
+      "step": 2383
+    },
+    {
+      "epoch": 2.3580613254203757,
+      "grad_norm": 0.2364704954092965,
+      "learning_rate": 1.1890802491755222e-05,
+      "loss": 0.3625,
+      "step": 2384
+    },
+    {
+      "epoch": 2.3590504451038576,
+      "grad_norm": 0.2476419419748873,
+      "learning_rate": 1.187248076218395e-05,
+      "loss": 0.3181,
+      "step": 2385
+    },
+    {
+      "epoch": 2.360039564787339,
+      "grad_norm": 0.24541438121836218,
+      "learning_rate": 1.1854159032612678e-05,
+      "loss": 0.3931,
+      "step": 2386
+    },
+    {
+      "epoch": 2.361028684470821,
+      "grad_norm": 0.23259123379943467,
+      "learning_rate": 1.1835837303041408e-05,
+      "loss": 0.3599,
+      "step": 2387
+    },
+    {
+      "epoch": 2.3620178041543025,
+      "grad_norm": 0.24998847686378414,
+      "learning_rate": 1.1817515573470136e-05,
+      "loss": 0.3454,
+      "step": 2388
+    },
+    {
+      "epoch": 2.3630069238377844,
+      "grad_norm": 0.2488071258698388,
+      "learning_rate": 1.1799193843898864e-05,
+      "loss": 0.3633,
+      "step": 2389
+    },
+    {
+      "epoch": 2.363996043521266,
+      "grad_norm": 0.22105819528108095,
+      "learning_rate": 1.1780872114327594e-05,
+      "loss": 0.332,
+      "step": 2390
+    },
+    {
+      "epoch": 2.364985163204748,
+      "grad_norm": 0.2334569251699502,
+      "learning_rate": 1.1762550384756322e-05,
+      "loss": 0.3415,
+      "step": 2391
+    },
+    {
+      "epoch": 2.3659742828882293,
+      "grad_norm": 0.27195122783585196,
+      "learning_rate": 1.174422865518505e-05,
+      "loss": 0.3633,
+      "step": 2392
+    },
+    {
+      "epoch": 2.3669634025717112,
+      "grad_norm": 0.2687234363535735,
+      "learning_rate": 1.172590692561378e-05,
+      "loss": 0.3223,
+      "step": 2393
+    },
+    {
+      "epoch": 2.3679525222551927,
+      "grad_norm": 0.27418695750053923,
+      "learning_rate": 1.1707585196042508e-05,
+      "loss": 0.348,
+      "step": 2394
+    },
+    {
+      "epoch": 2.3689416419386746,
+      "grad_norm": 0.2250392951845017,
+      "learning_rate": 1.1689263466471236e-05,
+      "loss": 0.3367,
+      "step": 2395
+    },
+    {
+      "epoch": 2.369930761622156,
+      "grad_norm": 0.2428040281312013,
+      "learning_rate": 1.1670941736899964e-05,
+      "loss": 0.3316,
+      "step": 2396
+    },
+    {
+      "epoch": 2.370919881305638,
+      "grad_norm": 0.257913044877425,
+      "learning_rate": 1.1652620007328692e-05,
+      "loss": 0.3541,
+      "step": 2397
+    },
+    {
+      "epoch": 2.3719090009891195,
+      "grad_norm": 0.24888673843156847,
+      "learning_rate": 1.163429827775742e-05,
+      "loss": 0.2872,
+      "step": 2398
+    },
+    {
+      "epoch": 2.3728981206726014,
+      "grad_norm": 0.22992256914847725,
+      "learning_rate": 1.161597654818615e-05,
+      "loss": 0.3668,
+      "step": 2399
+    },
+    {
+      "epoch": 2.373887240356083,
+      "grad_norm": 0.23105699693670317,
+      "learning_rate": 1.1597654818614878e-05,
+      "loss": 0.3354,
+      "step": 2400
+    },
+    {
+      "epoch": 2.374876360039565,
+      "grad_norm": 0.23878238019277057,
+      "learning_rate": 1.1579333089043606e-05,
+      "loss": 0.3339,
+      "step": 2401
+    },
+    {
+      "epoch": 2.3758654797230463,
+      "grad_norm": 0.23787914735351112,
+      "learning_rate": 1.1561011359472335e-05,
+      "loss": 0.3409,
+      "step": 2402
+    },
+    {
+      "epoch": 2.376854599406528,
+      "grad_norm": 0.21777157204428282,
+      "learning_rate": 1.1542689629901063e-05,
+      "loss": 0.3152,
+      "step": 2403
+    },
+    {
+      "epoch": 2.37784371909001,
+      "grad_norm": 0.23084255588965574,
+      "learning_rate": 1.152436790032979e-05,
+      "loss": 0.3406,
+      "step": 2404
+    },
+    {
+      "epoch": 2.3788328387734916,
+      "grad_norm": 0.23000546672119904,
+      "learning_rate": 1.150604617075852e-05,
+      "loss": 0.3362,
+      "step": 2405
+    },
+    {
+      "epoch": 2.379821958456973,
+      "grad_norm": 0.24929393096074584,
+      "learning_rate": 1.1487724441187249e-05,
+      "loss": 0.3395,
+      "step": 2406
+    },
+    {
+      "epoch": 2.380811078140455,
+      "grad_norm": 0.2253847100064449,
+      "learning_rate": 1.1469402711615977e-05,
+      "loss": 0.2843,
+      "step": 2407
+    },
+    {
+      "epoch": 2.381800197823937,
+      "grad_norm": 0.21736855996444068,
+      "learning_rate": 1.1451080982044705e-05,
+      "loss": 0.3224,
+      "step": 2408
+    },
+    {
+      "epoch": 2.3827893175074184,
+      "grad_norm": 0.26077964199405224,
+      "learning_rate": 1.1432759252473433e-05,
+      "loss": 0.3754,
+      "step": 2409
+    },
+    {
+      "epoch": 2.3837784371909,
+      "grad_norm": 0.23441329142827347,
+      "learning_rate": 1.1414437522902161e-05,
+      "loss": 0.3193,
+      "step": 2410
+    },
+    {
+      "epoch": 2.384767556874382,
+      "grad_norm": 0.22460852238892978,
+      "learning_rate": 1.1396115793330891e-05,
+      "loss": 0.3019,
+      "step": 2411
+    },
+    {
+      "epoch": 2.3857566765578637,
+      "grad_norm": 0.22061710031425286,
+      "learning_rate": 1.1377794063759619e-05,
+      "loss": 0.329,
+      "step": 2412
+    },
+    {
+      "epoch": 2.386745796241345,
+      "grad_norm": 0.23283333531597325,
+      "learning_rate": 1.1359472334188349e-05,
+      "loss": 0.3475,
+      "step": 2413
+    },
+    {
+      "epoch": 2.3877349159248267,
+      "grad_norm": 0.23549258278998214,
+      "learning_rate": 1.1341150604617077e-05,
+      "loss": 0.3681,
+      "step": 2414
+    },
+    {
+      "epoch": 2.3887240356083086,
+      "grad_norm": 0.23746707960246552,
+      "learning_rate": 1.1322828875045805e-05,
+      "loss": 0.3218,
+      "step": 2415
+    },
+    {
+      "epoch": 2.3897131552917905,
+      "grad_norm": 0.21232937070558,
+      "learning_rate": 1.1304507145474533e-05,
+      "loss": 0.3096,
+      "step": 2416
+    },
+    {
+      "epoch": 2.390702274975272,
+      "grad_norm": 0.22158582934961538,
+      "learning_rate": 1.1286185415903263e-05,
+      "loss": 0.3158,
+      "step": 2417
+    },
+    {
+      "epoch": 2.3916913946587535,
+      "grad_norm": 0.2481025761788115,
+      "learning_rate": 1.1267863686331991e-05,
+      "loss": 0.3196,
+      "step": 2418
+    },
+    {
+      "epoch": 2.3926805143422354,
+      "grad_norm": 0.24337539698694738,
+      "learning_rate": 1.124954195676072e-05,
+      "loss": 0.3259,
+      "step": 2419
+    },
+    {
+      "epoch": 2.3936696340257173,
+      "grad_norm": 0.228969243738579,
+      "learning_rate": 1.1231220227189447e-05,
+      "loss": 0.3427,
+      "step": 2420
+    },
+    {
+      "epoch": 2.394658753709199,
+      "grad_norm": 0.22025974633504555,
+      "learning_rate": 1.1212898497618175e-05,
+      "loss": 0.3483,
+      "step": 2421
+    },
+    {
+      "epoch": 2.3956478733926807,
+      "grad_norm": 0.26184165101054785,
+      "learning_rate": 1.1194576768046904e-05,
+      "loss": 0.3828,
+      "step": 2422
+    },
+    {
+      "epoch": 2.396636993076162,
+      "grad_norm": 0.24478098303647366,
+      "learning_rate": 1.1176255038475633e-05,
+      "loss": 0.3599,
+      "step": 2423
+    },
+    {
+      "epoch": 2.397626112759644,
+      "grad_norm": 0.24261159243488178,
+      "learning_rate": 1.1157933308904361e-05,
+      "loss": 0.3112,
+      "step": 2424
+    },
+    {
+      "epoch": 2.3986152324431256,
+      "grad_norm": 0.22301944660446946,
+      "learning_rate": 1.113961157933309e-05,
+      "loss": 0.3214,
+      "step": 2425
+    },
+    {
+      "epoch": 2.3996043521266075,
+      "grad_norm": 0.24202065872251996,
+      "learning_rate": 1.1121289849761818e-05,
+      "loss": 0.3373,
+      "step": 2426
+    },
+    {
+      "epoch": 2.400593471810089,
+      "grad_norm": 0.26782140193309023,
+      "learning_rate": 1.1102968120190546e-05,
+      "loss": 0.3926,
+      "step": 2427
+    },
+    {
+      "epoch": 2.401582591493571,
+      "grad_norm": 0.24029910736845295,
+      "learning_rate": 1.1084646390619274e-05,
+      "loss": 0.368,
+      "step": 2428
+    },
+    {
+      "epoch": 2.4025717111770524,
+      "grad_norm": 0.2407671258225506,
+      "learning_rate": 1.1066324661048004e-05,
+      "loss": 0.3423,
+      "step": 2429
+    },
+    {
+      "epoch": 2.4035608308605343,
+      "grad_norm": 0.24572982261117407,
+      "learning_rate": 1.1048002931476732e-05,
+      "loss": 0.3482,
+      "step": 2430
+    },
+    {
+      "epoch": 2.4045499505440158,
+      "grad_norm": 0.2788166813071352,
+      "learning_rate": 1.102968120190546e-05,
+      "loss": 0.3574,
+      "step": 2431
+    },
+    {
+      "epoch": 2.4055390702274977,
+      "grad_norm": 0.23296254668053415,
+      "learning_rate": 1.1011359472334188e-05,
+      "loss": 0.3137,
+      "step": 2432
+    },
+    {
+      "epoch": 2.406528189910979,
+      "grad_norm": 0.24801668753216083,
+      "learning_rate": 1.0993037742762916e-05,
+      "loss": 0.3509,
+      "step": 2433
+    },
+    {
+      "epoch": 2.407517309594461,
+      "grad_norm": 0.25891621196791104,
+      "learning_rate": 1.0974716013191646e-05,
+      "loss": 0.3266,
+      "step": 2434
+    },
+    {
+      "epoch": 2.4085064292779426,
+      "grad_norm": 0.25721058330749236,
+      "learning_rate": 1.0956394283620374e-05,
+      "loss": 0.3299,
+      "step": 2435
+    },
+    {
+      "epoch": 2.4094955489614245,
+      "grad_norm": 0.23910805211819738,
+      "learning_rate": 1.0938072554049102e-05,
+      "loss": 0.3222,
+      "step": 2436
+    },
+    {
+      "epoch": 2.410484668644906,
+      "grad_norm": 0.7585722434930996,
+      "learning_rate": 1.0919750824477832e-05,
+      "loss": 0.3307,
+      "step": 2437
+    },
+    {
+      "epoch": 2.411473788328388,
+      "grad_norm": 0.2622094628177202,
+      "learning_rate": 1.090142909490656e-05,
+      "loss": 0.3657,
+      "step": 2438
+    },
+    {
+      "epoch": 2.4124629080118694,
+      "grad_norm": 0.24659250738864202,
+      "learning_rate": 1.0883107365335288e-05,
+      "loss": 0.3411,
+      "step": 2439
+    },
+    {
+      "epoch": 2.4134520276953513,
+      "grad_norm": 0.24754083987784184,
+      "learning_rate": 1.0864785635764016e-05,
+      "loss": 0.3949,
+      "step": 2440
+    },
+    {
+      "epoch": 2.4144411473788328,
+      "grad_norm": 0.2300523326797196,
+      "learning_rate": 1.0846463906192746e-05,
+      "loss": 0.3217,
+      "step": 2441
+    },
+    {
+      "epoch": 2.4154302670623147,
+      "grad_norm": 0.268740945166187,
+      "learning_rate": 1.0828142176621474e-05,
+      "loss": 0.3416,
+      "step": 2442
+    },
+    {
+      "epoch": 2.416419386745796,
+      "grad_norm": 0.25252939297829813,
+      "learning_rate": 1.0809820447050202e-05,
+      "loss": 0.3397,
+      "step": 2443
+    },
+    {
+      "epoch": 2.417408506429278,
+      "grad_norm": 0.2530969379846874,
+      "learning_rate": 1.079149871747893e-05,
+      "loss": 0.3864,
+      "step": 2444
+    },
+    {
+      "epoch": 2.4183976261127595,
+      "grad_norm": 0.24562207385794602,
+      "learning_rate": 1.0773176987907659e-05,
+      "loss": 0.3548,
+      "step": 2445
+    },
+    {
+      "epoch": 2.4193867457962415,
+      "grad_norm": 0.2441336356073428,
+      "learning_rate": 1.0754855258336387e-05,
+      "loss": 0.3508,
+      "step": 2446
+    },
+    {
+      "epoch": 2.420375865479723,
+      "grad_norm": 0.33254572988738407,
+      "learning_rate": 1.0736533528765116e-05,
+      "loss": 0.3308,
+      "step": 2447
+    },
+    {
+      "epoch": 2.421364985163205,
+      "grad_norm": 0.33598359004717754,
+      "learning_rate": 1.0718211799193845e-05,
+      "loss": 0.3371,
+      "step": 2448
+    },
+    {
+      "epoch": 2.4223541048466863,
+      "grad_norm": 0.24536171632195142,
+      "learning_rate": 1.0699890069622573e-05,
+      "loss": 0.346,
+      "step": 2449
+    },
+    {
+      "epoch": 2.4233432245301683,
+      "grad_norm": 0.220072693747127,
+      "learning_rate": 1.06815683400513e-05,
+      "loss": 0.3051,
+      "step": 2450
+    },
+    {
+      "epoch": 2.4243323442136497,
+      "grad_norm": 0.26313981909407597,
+      "learning_rate": 1.0663246610480029e-05,
+      "loss": 0.3351,
+      "step": 2451
+    },
+    {
+      "epoch": 2.4253214638971317,
+      "grad_norm": 0.2710866024316405,
+      "learning_rate": 1.0644924880908757e-05,
+      "loss": 0.3164,
+      "step": 2452
+    },
+    {
+      "epoch": 2.426310583580613,
+      "grad_norm": 0.25396458294554114,
+      "learning_rate": 1.0626603151337487e-05,
+      "loss": 0.3463,
+      "step": 2453
+    },
+    {
+      "epoch": 2.427299703264095,
+      "grad_norm": 0.2252749311775426,
+      "learning_rate": 1.0608281421766215e-05,
+      "loss": 0.354,
+      "step": 2454
+    },
+    {
+      "epoch": 2.4282888229475765,
+      "grad_norm": 0.2606999859832173,
+      "learning_rate": 1.0589959692194943e-05,
+      "loss": 0.3511,
+      "step": 2455
+    },
+    {
+      "epoch": 2.4292779426310585,
+      "grad_norm": 0.37868283734687525,
+      "learning_rate": 1.0571637962623671e-05,
+      "loss": 0.361,
+      "step": 2456
+    },
+    {
+      "epoch": 2.43026706231454,
+      "grad_norm": 0.23778601344518252,
+      "learning_rate": 1.0553316233052401e-05,
+      "loss": 0.3375,
+      "step": 2457
+    },
+    {
+      "epoch": 2.431256181998022,
+      "grad_norm": 0.23759219268753948,
+      "learning_rate": 1.0534994503481129e-05,
+      "loss": 0.3371,
+      "step": 2458
+    },
+    {
+      "epoch": 2.4322453016815033,
+      "grad_norm": 0.22579121090246385,
+      "learning_rate": 1.0516672773909857e-05,
+      "loss": 0.3615,
+      "step": 2459
+    },
+    {
+      "epoch": 2.4332344213649852,
+      "grad_norm": 0.23369877437828138,
+      "learning_rate": 1.0498351044338587e-05,
+      "loss": 0.375,
+      "step": 2460
+    },
+    {
+      "epoch": 2.4342235410484667,
+      "grad_norm": 0.22961846338817127,
+      "learning_rate": 1.0480029314767315e-05,
+      "loss": 0.3484,
+      "step": 2461
+    },
+    {
+      "epoch": 2.4352126607319486,
+      "grad_norm": 0.22887264558327763,
+      "learning_rate": 1.0461707585196043e-05,
+      "loss": 0.3385,
+      "step": 2462
+    },
+    {
+      "epoch": 2.43620178041543,
+      "grad_norm": 0.25032092756268276,
+      "learning_rate": 1.0443385855624771e-05,
+      "loss": 0.3151,
+      "step": 2463
+    },
+    {
+      "epoch": 2.437190900098912,
+      "grad_norm": 0.23728928057357426,
+      "learning_rate": 1.0425064126053501e-05,
+      "loss": 0.3813,
+      "step": 2464
+    },
+    {
+      "epoch": 2.4381800197823935,
+      "grad_norm": 0.2438827034129295,
+      "learning_rate": 1.040674239648223e-05,
+      "loss": 0.3659,
+      "step": 2465
+    },
+    {
+      "epoch": 2.4391691394658754,
+      "grad_norm": 0.2778214140943854,
+      "learning_rate": 1.0388420666910957e-05,
+      "loss": 0.3513,
+      "step": 2466
+    },
+    {
+      "epoch": 2.440158259149357,
+      "grad_norm": 0.22815945841865654,
+      "learning_rate": 1.0370098937339686e-05,
+      "loss": 0.3447,
+      "step": 2467
+    },
+    {
+      "epoch": 2.441147378832839,
+      "grad_norm": 0.23528405331897878,
+      "learning_rate": 1.0351777207768414e-05,
+      "loss": 0.3352,
+      "step": 2468
+    },
+    {
+      "epoch": 2.4421364985163203,
+      "grad_norm": 0.235995080768111,
+      "learning_rate": 1.0333455478197142e-05,
+      "loss": 0.3263,
+      "step": 2469
+    },
+    {
+      "epoch": 2.4431256181998022,
+      "grad_norm": 0.21678114600038997,
+      "learning_rate": 1.0315133748625872e-05,
+      "loss": 0.3112,
+      "step": 2470
+    },
+    {
+      "epoch": 2.4441147378832837,
+      "grad_norm": 0.25806394891487444,
+      "learning_rate": 1.02968120190546e-05,
+      "loss": 0.3551,
+      "step": 2471
+    },
+    {
+      "epoch": 2.4451038575667656,
+      "grad_norm": 0.26401170188018913,
+      "learning_rate": 1.0278490289483328e-05,
+      "loss": 0.39,
+      "step": 2472
+    },
+    {
+      "epoch": 2.446092977250247,
+      "grad_norm": 0.21740446151393822,
+      "learning_rate": 1.0260168559912056e-05,
+      "loss": 0.3286,
+      "step": 2473
+    },
+    {
+      "epoch": 2.447082096933729,
+      "grad_norm": 0.23443927897272304,
+      "learning_rate": 1.0241846830340784e-05,
+      "loss": 0.3652,
+      "step": 2474
+    },
+    {
+      "epoch": 2.4480712166172105,
+      "grad_norm": 0.218480918457683,
+      "learning_rate": 1.0223525100769512e-05,
+      "loss": 0.3383,
+      "step": 2475
+    },
+    {
+      "epoch": 2.4490603363006924,
+      "grad_norm": 0.2203346851930569,
+      "learning_rate": 1.0205203371198242e-05,
+      "loss": 0.3401,
+      "step": 2476
+    },
+    {
+      "epoch": 2.450049455984174,
+      "grad_norm": 0.22869849971032385,
+      "learning_rate": 1.018688164162697e-05,
+      "loss": 0.3308,
+      "step": 2477
+    },
+    {
+      "epoch": 2.451038575667656,
+      "grad_norm": 0.2363096068659232,
+      "learning_rate": 1.0168559912055698e-05,
+      "loss": 0.3628,
+      "step": 2478
+    },
+    {
+      "epoch": 2.4520276953511377,
+      "grad_norm": 0.21724450496142547,
+      "learning_rate": 1.0150238182484426e-05,
+      "loss": 0.3435,
+      "step": 2479
+    },
+    {
+      "epoch": 2.453016815034619,
+      "grad_norm": 0.24107977424377153,
+      "learning_rate": 1.0131916452913156e-05,
+      "loss": 0.365,
+      "step": 2480
+    },
+    {
+      "epoch": 2.4540059347181007,
+      "grad_norm": 0.24828390832939212,
+      "learning_rate": 1.0113594723341884e-05,
+      "loss": 0.3366,
+      "step": 2481
+    },
+    {
+      "epoch": 2.4549950544015826,
+      "grad_norm": 0.20569807648782604,
+      "learning_rate": 1.0095272993770612e-05,
+      "loss": 0.2761,
+      "step": 2482
+    },
+    {
+      "epoch": 2.4559841740850645,
+      "grad_norm": 0.2322910616182908,
+      "learning_rate": 1.0076951264199342e-05,
+      "loss": 0.3741,
+      "step": 2483
+    },
+    {
+      "epoch": 2.456973293768546,
+      "grad_norm": 0.24079586769347716,
+      "learning_rate": 1.005862953462807e-05,
+      "loss": 0.3657,
+      "step": 2484
+    },
+    {
+      "epoch": 2.4579624134520275,
+      "grad_norm": 0.24451423706174355,
+      "learning_rate": 1.0040307805056798e-05,
+      "loss": 0.3375,
+      "step": 2485
+    },
+    {
+      "epoch": 2.4589515331355094,
+      "grad_norm": 0.21789521602688885,
+      "learning_rate": 1.0021986075485526e-05,
+      "loss": 0.3115,
+      "step": 2486
+    },
+    {
+      "epoch": 2.4599406528189913,
+      "grad_norm": 0.2367007818716319,
+      "learning_rate": 1.0003664345914255e-05,
+      "loss": 0.3434,
+      "step": 2487
+    },
+    {
+      "epoch": 2.460929772502473,
+      "grad_norm": 0.2307538273441414,
+      "learning_rate": 9.985342616342984e-06,
+      "loss": 0.3279,
+      "step": 2488
+    },
+    {
+      "epoch": 2.4619188921859543,
+      "grad_norm": 0.2366577365331995,
+      "learning_rate": 9.967020886771712e-06,
+      "loss": 0.3405,
+      "step": 2489
+    },
+    {
+      "epoch": 2.462908011869436,
+      "grad_norm": 0.21292878456865053,
+      "learning_rate": 9.94869915720044e-06,
+      "loss": 0.3271,
+      "step": 2490
+    },
+    {
+      "epoch": 2.463897131552918,
+      "grad_norm": 0.21294135453400145,
+      "learning_rate": 9.930377427629169e-06,
+      "loss": 0.312,
+      "step": 2491
+    },
+    {
+      "epoch": 2.4648862512363996,
+      "grad_norm": 0.2406136238914857,
+      "learning_rate": 9.912055698057897e-06,
+      "loss": 0.3537,
+      "step": 2492
+    },
+    {
+      "epoch": 2.465875370919881,
+      "grad_norm": 0.25484410374642147,
+      "learning_rate": 9.893733968486625e-06,
+      "loss": 0.3134,
+      "step": 2493
+    },
+    {
+      "epoch": 2.466864490603363,
+      "grad_norm": 0.2620293766154584,
+      "learning_rate": 9.875412238915355e-06,
+      "loss": 0.4176,
+      "step": 2494
+    },
+    {
+      "epoch": 2.467853610286845,
+      "grad_norm": 0.23944770550967767,
+      "learning_rate": 9.857090509344083e-06,
+      "loss": 0.356,
+      "step": 2495
+    },
+    {
+      "epoch": 2.4688427299703264,
+      "grad_norm": 0.25086364175393056,
+      "learning_rate": 9.838768779772811e-06,
+      "loss": 0.3575,
+      "step": 2496
+    },
+    {
+      "epoch": 2.469831849653808,
+      "grad_norm": 0.23501454699640065,
+      "learning_rate": 9.820447050201539e-06,
+      "loss": 0.3876,
+      "step": 2497
+    },
+    {
+      "epoch": 2.47082096933729,
+      "grad_norm": 0.2551766847463871,
+      "learning_rate": 9.802125320630267e-06,
+      "loss": 0.3367,
+      "step": 2498
+    },
+    {
+      "epoch": 2.4718100890207717,
+      "grad_norm": 0.2361580571158426,
+      "learning_rate": 9.783803591058995e-06,
+      "loss": 0.4009,
+      "step": 2499
+    },
+    {
+      "epoch": 2.472799208704253,
+      "grad_norm": 1.4900062991141605,
+      "learning_rate": 9.765481861487725e-06,
+      "loss": 0.3581,
+      "step": 2500
+    },
+    {
+      "epoch": 2.473788328387735,
+      "grad_norm": 0.23073936660230762,
+      "learning_rate": 9.747160131916453e-06,
+      "loss": 0.3565,
+      "step": 2501
+    },
+    {
+      "epoch": 2.4747774480712166,
+      "grad_norm": 0.24682663116703085,
+      "learning_rate": 9.728838402345181e-06,
+      "loss": 0.3535,
+      "step": 2502
+    },
+    {
+      "epoch": 2.4757665677546985,
+      "grad_norm": 0.27034996608934037,
+      "learning_rate": 9.710516672773911e-06,
+      "loss": 0.3916,
+      "step": 2503
+    },
+    {
+      "epoch": 2.47675568743818,
+      "grad_norm": 0.2303037854565598,
+      "learning_rate": 9.69219494320264e-06,
+      "loss": 0.3436,
+      "step": 2504
+    },
+    {
+      "epoch": 2.477744807121662,
+      "grad_norm": 0.23585596331964975,
+      "learning_rate": 9.673873213631367e-06,
+      "loss": 0.3118,
+      "step": 2505
+    },
+    {
+      "epoch": 2.4787339268051434,
+      "grad_norm": 0.23726271220235548,
+      "learning_rate": 9.655551484060097e-06,
+      "loss": 0.3394,
+      "step": 2506
+    },
+    {
+      "epoch": 2.4797230464886253,
+      "grad_norm": 0.2487286182774457,
+      "learning_rate": 9.637229754488825e-06,
+      "loss": 0.3573,
+      "step": 2507
+    },
+    {
+      "epoch": 2.4807121661721068,
+      "grad_norm": 0.2698429096752246,
+      "learning_rate": 9.618908024917553e-06,
+      "loss": 0.3876,
+      "step": 2508
+    },
+    {
+      "epoch": 2.4817012858555887,
+      "grad_norm": 0.24070333522567572,
+      "learning_rate": 9.600586295346281e-06,
+      "loss": 0.3465,
+      "step": 2509
+    },
+    {
+      "epoch": 2.48269040553907,
+      "grad_norm": 0.24206724387701353,
+      "learning_rate": 9.58226456577501e-06,
+      "loss": 0.3652,
+      "step": 2510
+    },
+    {
+      "epoch": 2.483679525222552,
+      "grad_norm": 0.22545629834132364,
+      "learning_rate": 9.563942836203738e-06,
+      "loss": 0.3275,
+      "step": 2511
+    },
+    {
+      "epoch": 2.4846686449060336,
+      "grad_norm": 0.23858808844243973,
+      "learning_rate": 9.545621106632467e-06,
+      "loss": 0.3505,
+      "step": 2512
+    },
+    {
+      "epoch": 2.4856577645895155,
+      "grad_norm": 0.2728863075026196,
+      "learning_rate": 9.527299377061196e-06,
+      "loss": 0.3541,
+      "step": 2513
+    },
+    {
+      "epoch": 2.486646884272997,
+      "grad_norm": 0.23294695154773748,
+      "learning_rate": 9.508977647489924e-06,
+      "loss": 0.3574,
+      "step": 2514
+    },
+    {
+      "epoch": 2.487636003956479,
+      "grad_norm": 0.24124966388283223,
+      "learning_rate": 9.490655917918652e-06,
+      "loss": 0.3243,
+      "step": 2515
+    },
+    {
+      "epoch": 2.4886251236399604,
+      "grad_norm": 0.2320389035173641,
+      "learning_rate": 9.47233418834738e-06,
+      "loss": 0.3245,
+      "step": 2516
+    },
+    {
+      "epoch": 2.4896142433234423,
+      "grad_norm": 0.26700657514143444,
+      "learning_rate": 9.454012458776108e-06,
+      "loss": 0.3852,
+      "step": 2517
+    },
+    {
+      "epoch": 2.4906033630069238,
+      "grad_norm": 0.2507593747565587,
+      "learning_rate": 9.435690729204838e-06,
+      "loss": 0.3476,
+      "step": 2518
+    },
+    {
+      "epoch": 2.4915924826904057,
+      "grad_norm": 0.21748192964113497,
+      "learning_rate": 9.417368999633566e-06,
+      "loss": 0.3334,
+      "step": 2519
+    },
+    {
+      "epoch": 2.492581602373887,
+      "grad_norm": 0.25264072777527896,
+      "learning_rate": 9.399047270062294e-06,
+      "loss": 0.3685,
+      "step": 2520
+    },
+    {
+      "epoch": 2.493570722057369,
+      "grad_norm": 0.24839443510243983,
+      "learning_rate": 9.380725540491022e-06,
+      "loss": 0.3523,
+      "step": 2521
+    },
+    {
+      "epoch": 2.4945598417408505,
+      "grad_norm": 0.22023654916565122,
+      "learning_rate": 9.36240381091975e-06,
+      "loss": 0.32,
+      "step": 2522
+    },
+    {
+      "epoch": 2.4955489614243325,
+      "grad_norm": 0.2441603761072515,
+      "learning_rate": 9.344082081348478e-06,
+      "loss": 0.3677,
+      "step": 2523
+    },
+    {
+      "epoch": 2.496538081107814,
+      "grad_norm": 0.23410429303428584,
+      "learning_rate": 9.325760351777208e-06,
+      "loss": 0.3282,
+      "step": 2524
+    },
+    {
+      "epoch": 2.497527200791296,
+      "grad_norm": 0.5284404116687907,
+      "learning_rate": 9.307438622205936e-06,
+      "loss": 0.3842,
+      "step": 2525
+    },
+    {
+      "epoch": 2.4985163204747773,
+      "grad_norm": 0.2597369164991658,
+      "learning_rate": 9.289116892634666e-06,
+      "loss": 0.3564,
+      "step": 2526
+    },
+    {
+      "epoch": 2.4995054401582593,
+      "grad_norm": 1.5905916143954588,
+      "learning_rate": 9.270795163063394e-06,
+      "loss": 0.4346,
+      "step": 2527
+    },
+    {
+      "epoch": 2.5004945598417407,
+      "grad_norm": 0.2406750160968987,
+      "learning_rate": 9.252473433492122e-06,
+      "loss": 0.3277,
+      "step": 2528
+    },
+    {
+      "epoch": 2.5014836795252227,
+      "grad_norm": 0.25687044415825083,
+      "learning_rate": 9.23415170392085e-06,
+      "loss": 0.3453,
+      "step": 2529
+    },
+    {
+      "epoch": 2.502472799208704,
+      "grad_norm": 0.25135048181774106,
+      "learning_rate": 9.21582997434958e-06,
+      "loss": 0.3547,
+      "step": 2530
+    },
+    {
+      "epoch": 2.503461918892186,
+      "grad_norm": 0.24560280016284752,
+      "learning_rate": 9.197508244778308e-06,
+      "loss": 0.3661,
+      "step": 2531
+    },
+    {
+      "epoch": 2.5044510385756675,
+      "grad_norm": 0.49064750366441084,
+      "learning_rate": 9.179186515207036e-06,
+      "loss": 0.3812,
+      "step": 2532
+    },
+    {
+      "epoch": 2.5054401582591495,
+      "grad_norm": 0.22993415291097227,
+      "learning_rate": 9.160864785635765e-06,
+      "loss": 0.3637,
+      "step": 2533
+    },
+    {
+      "epoch": 2.506429277942631,
+      "grad_norm": 0.21741000097273247,
+      "learning_rate": 9.142543056064493e-06,
+      "loss": 0.3319,
+      "step": 2534
+    },
+    {
+      "epoch": 2.507418397626113,
+      "grad_norm": 0.22510085134096802,
+      "learning_rate": 9.12422132649322e-06,
+      "loss": 0.3054,
+      "step": 2535
+    },
+    {
+      "epoch": 2.5084075173095943,
+      "grad_norm": 0.23033682251844267,
+      "learning_rate": 9.10589959692195e-06,
+      "loss": 0.3684,
+      "step": 2536
+    },
+    {
+      "epoch": 2.5093966369930762,
+      "grad_norm": 0.23336881930405867,
+      "learning_rate": 9.087577867350679e-06,
+      "loss": 0.336,
+      "step": 2537
+    },
+    {
+      "epoch": 2.5103857566765577,
+      "grad_norm": 0.25835519741672547,
+      "learning_rate": 9.069256137779407e-06,
+      "loss": 0.3665,
+      "step": 2538
+    },
+    {
+      "epoch": 2.5113748763600396,
+      "grad_norm": 0.21527278722575882,
+      "learning_rate": 9.050934408208135e-06,
+      "loss": 0.3271,
+      "step": 2539
+    },
+    {
+      "epoch": 2.512363996043521,
+      "grad_norm": 0.22846804776774987,
+      "learning_rate": 9.032612678636863e-06,
+      "loss": 0.3634,
+      "step": 2540
+    },
+    {
+      "epoch": 2.513353115727003,
+      "grad_norm": 0.2461420001985178,
+      "learning_rate": 9.014290949065591e-06,
+      "loss": 0.3618,
+      "step": 2541
+    },
+    {
+      "epoch": 2.5143422354104845,
+      "grad_norm": 0.24677056500039793,
+      "learning_rate": 8.995969219494321e-06,
+      "loss": 0.3474,
+      "step": 2542
+    },
+    {
+      "epoch": 2.5153313550939664,
+      "grad_norm": 0.23552827235851204,
+      "learning_rate": 8.977647489923049e-06,
+      "loss": 0.3477,
+      "step": 2543
+    },
+    {
+      "epoch": 2.516320474777448,
+      "grad_norm": 0.23544874721670853,
+      "learning_rate": 8.959325760351777e-06,
+      "loss": 0.3665,
+      "step": 2544
+    },
+    {
+      "epoch": 2.51730959446093,
+      "grad_norm": 0.25589519603994987,
+      "learning_rate": 8.941004030780505e-06,
+      "loss": 0.3676,
+      "step": 2545
+    },
+    {
+      "epoch": 2.5182987141444113,
+      "grad_norm": 0.22529131811822634,
+      "learning_rate": 8.922682301209233e-06,
+      "loss": 0.342,
+      "step": 2546
+    },
+    {
+      "epoch": 2.5192878338278932,
+      "grad_norm": 0.22830805413949087,
+      "learning_rate": 8.904360571637963e-06,
+      "loss": 0.3228,
+      "step": 2547
+    },
+    {
+      "epoch": 2.5202769535113747,
+      "grad_norm": 0.24283043796243683,
+      "learning_rate": 8.886038842066691e-06,
+      "loss": 0.3521,
+      "step": 2548
+    },
+    {
+      "epoch": 2.5212660731948566,
+      "grad_norm": 0.23539048865709836,
+      "learning_rate": 8.86771711249542e-06,
+      "loss": 0.3232,
+      "step": 2549
+    },
+    {
+      "epoch": 2.5222551928783385,
+      "grad_norm": 0.22416815087474565,
+      "learning_rate": 8.84939538292415e-06,
+      "loss": 0.3502,
+      "step": 2550
+    },
+    {
+      "epoch": 2.52324431256182,
+      "grad_norm": 0.24356963148470098,
+      "learning_rate": 8.831073653352877e-06,
+      "loss": 0.3041,
+      "step": 2551
+    },
+    {
+      "epoch": 2.5242334322453015,
+      "grad_norm": 0.2395352023996427,
+      "learning_rate": 8.812751923781605e-06,
+      "loss": 0.3787,
+      "step": 2552
+    },
+    {
+      "epoch": 2.5252225519287834,
+      "grad_norm": 0.23644859570044596,
+      "learning_rate": 8.794430194210334e-06,
+      "loss": 0.3657,
+      "step": 2553
+    },
+    {
+      "epoch": 2.5262116716122653,
+      "grad_norm": 0.253996407977228,
+      "learning_rate": 8.776108464639063e-06,
+      "loss": 0.3445,
+      "step": 2554
+    },
+    {
+      "epoch": 2.527200791295747,
+      "grad_norm": 0.251282958120267,
+      "learning_rate": 8.757786735067792e-06,
+      "loss": 0.3536,
+      "step": 2555
+    },
+    {
+      "epoch": 2.5281899109792283,
+      "grad_norm": 0.45586735138497564,
+      "learning_rate": 8.73946500549652e-06,
+      "loss": 0.3641,
+      "step": 2556
+    },
+    {
+      "epoch": 2.52917903066271,
+      "grad_norm": 0.25052516019444043,
+      "learning_rate": 8.721143275925248e-06,
+      "loss": 0.3669,
+      "step": 2557
+    },
+    {
+      "epoch": 2.530168150346192,
+      "grad_norm": 0.23437819091702222,
+      "learning_rate": 8.702821546353976e-06,
+      "loss": 0.3283,
+      "step": 2558
+    },
+    {
+      "epoch": 2.5311572700296736,
+      "grad_norm": 0.22284762353121085,
+      "learning_rate": 8.684499816782704e-06,
+      "loss": 0.3145,
+      "step": 2559
+    },
+    {
+      "epoch": 2.532146389713155,
+      "grad_norm": 0.22449486144864736,
+      "learning_rate": 8.666178087211434e-06,
+      "loss": 0.3553,
+      "step": 2560
+    },
+    {
+      "epoch": 2.533135509396637,
+      "grad_norm": 0.2431824560198144,
+      "learning_rate": 8.647856357640162e-06,
+      "loss": 0.353,
+      "step": 2561
+    },
+    {
+      "epoch": 2.534124629080119,
+      "grad_norm": 0.24220316126556765,
+      "learning_rate": 8.62953462806889e-06,
+      "loss": 0.3698,
+      "step": 2562
+    },
+    {
+      "epoch": 2.5351137487636004,
+      "grad_norm": 0.24561350691928416,
+      "learning_rate": 8.611212898497618e-06,
+      "loss": 0.3541,
+      "step": 2563
+    },
+    {
+      "epoch": 2.536102868447082,
+      "grad_norm": 0.26058782429161903,
+      "learning_rate": 8.592891168926346e-06,
+      "loss": 0.358,
+      "step": 2564
+    },
+    {
+      "epoch": 2.537091988130564,
+      "grad_norm": 0.24483285029632448,
+      "learning_rate": 8.574569439355074e-06,
+      "loss": 0.3722,
+      "step": 2565
+    },
+    {
+      "epoch": 2.5380811078140457,
+      "grad_norm": 0.22463143941644045,
+      "learning_rate": 8.556247709783804e-06,
+      "loss": 0.3344,
+      "step": 2566
+    },
+    {
+      "epoch": 2.539070227497527,
+      "grad_norm": 0.25857496371914634,
+      "learning_rate": 8.537925980212532e-06,
+      "loss": 0.3955,
+      "step": 2567
+    },
+    {
+      "epoch": 2.5400593471810087,
+      "grad_norm": 0.24794143406441915,
+      "learning_rate": 8.51960425064126e-06,
+      "loss": 0.3279,
+      "step": 2568
+    },
+    {
+      "epoch": 2.5410484668644906,
+      "grad_norm": 0.23454478908544515,
+      "learning_rate": 8.501282521069988e-06,
+      "loss": 0.3387,
+      "step": 2569
+    },
+    {
+      "epoch": 2.5420375865479725,
+      "grad_norm": 0.218775146778518,
+      "learning_rate": 8.482960791498718e-06,
+      "loss": 0.3149,
+      "step": 2570
+    },
+    {
+      "epoch": 2.543026706231454,
+      "grad_norm": 0.256130200586046,
+      "learning_rate": 8.464639061927446e-06,
+      "loss": 0.3491,
+      "step": 2571
+    },
+    {
+      "epoch": 2.5440158259149355,
+      "grad_norm": 0.21696788649308552,
+      "learning_rate": 8.446317332356175e-06,
+      "loss": 0.3194,
+      "step": 2572
+    },
+    {
+      "epoch": 2.5450049455984174,
+      "grad_norm": 0.214921895223905,
+      "learning_rate": 8.427995602784904e-06,
+      "loss": 0.3066,
+      "step": 2573
+    },
+    {
+      "epoch": 2.5459940652818993,
+      "grad_norm": 0.222786093431417,
+      "learning_rate": 8.409673873213632e-06,
+      "loss": 0.3633,
+      "step": 2574
+    },
+    {
+      "epoch": 2.546983184965381,
+      "grad_norm": 0.23510985038285132,
+      "learning_rate": 8.39135214364236e-06,
+      "loss": 0.3647,
+      "step": 2575
+    },
+    {
+      "epoch": 2.5479723046488623,
+      "grad_norm": 0.2192600372412503,
+      "learning_rate": 8.373030414071089e-06,
+      "loss": 0.3552,
+      "step": 2576
+    },
+    {
+      "epoch": 2.548961424332344,
+      "grad_norm": 0.21452070249266664,
+      "learning_rate": 8.354708684499817e-06,
+      "loss": 0.3031,
+      "step": 2577
+    },
+    {
+      "epoch": 2.549950544015826,
+      "grad_norm": 0.2286175179188525,
+      "learning_rate": 8.336386954928547e-06,
+      "loss": 0.3338,
+      "step": 2578
+    },
+    {
+      "epoch": 2.5509396636993076,
+      "grad_norm": 0.24433394067425576,
+      "learning_rate": 8.318065225357275e-06,
+      "loss": 0.3519,
+      "step": 2579
+    },
+    {
+      "epoch": 2.551928783382789,
+      "grad_norm": 0.2188927821523692,
+      "learning_rate": 8.299743495786003e-06,
+      "loss": 0.2997,
+      "step": 2580
+    },
+    {
+      "epoch": 2.552917903066271,
+      "grad_norm": 0.23903928008474262,
+      "learning_rate": 8.281421766214731e-06,
+      "loss": 0.3582,
+      "step": 2581
+    },
+    {
+      "epoch": 2.553907022749753,
+      "grad_norm": 0.2252037417774991,
+      "learning_rate": 8.263100036643459e-06,
+      "loss": 0.3158,
+      "step": 2582
+    },
+    {
+      "epoch": 2.5548961424332344,
+      "grad_norm": 0.22547563235668888,
+      "learning_rate": 8.244778307072187e-06,
+      "loss": 0.3161,
+      "step": 2583
+    },
+    {
+      "epoch": 2.555885262116716,
+      "grad_norm": 0.212708651970397,
+      "learning_rate": 8.226456577500917e-06,
+      "loss": 0.3021,
+      "step": 2584
+    },
+    {
+      "epoch": 2.5568743818001978,
+      "grad_norm": 0.2211913667645755,
+      "learning_rate": 8.208134847929645e-06,
+      "loss": 0.3355,
+      "step": 2585
+    },
+    {
+      "epoch": 2.5578635014836797,
+      "grad_norm": 0.2422032730001762,
+      "learning_rate": 8.189813118358373e-06,
+      "loss": 0.3753,
+      "step": 2586
+    },
+    {
+      "epoch": 2.558852621167161,
+      "grad_norm": 0.26729432792289365,
+      "learning_rate": 8.171491388787101e-06,
+      "loss": 0.399,
+      "step": 2587
+    },
+    {
+      "epoch": 2.559841740850643,
+      "grad_norm": 0.25427480702090216,
+      "learning_rate": 8.15316965921583e-06,
+      "loss": 0.3876,
+      "step": 2588
+    },
+    {
+      "epoch": 2.5608308605341246,
+      "grad_norm": 0.23376302569182988,
+      "learning_rate": 8.134847929644557e-06,
+      "loss": 0.3499,
+      "step": 2589
+    },
+    {
+      "epoch": 2.5618199802176065,
+      "grad_norm": 0.21819970135465813,
+      "learning_rate": 8.116526200073287e-06,
+      "loss": 0.3337,
+      "step": 2590
+    },
+    {
+      "epoch": 2.562809099901088,
+      "grad_norm": 0.249529986101218,
+      "learning_rate": 8.098204470502015e-06,
+      "loss": 0.358,
+      "step": 2591
+    },
+    {
+      "epoch": 2.56379821958457,
+      "grad_norm": 0.2365993729449896,
+      "learning_rate": 8.079882740930744e-06,
+      "loss": 0.3484,
+      "step": 2592
+    },
+    {
+      "epoch": 2.5647873392680514,
+      "grad_norm": 0.23483393877228428,
+      "learning_rate": 8.061561011359473e-06,
+      "loss": 0.3566,
+      "step": 2593
+    },
+    {
+      "epoch": 2.5657764589515333,
+      "grad_norm": 0.222182938428324,
+      "learning_rate": 8.043239281788201e-06,
+      "loss": 0.3106,
+      "step": 2594
+    },
+    {
+      "epoch": 2.5667655786350148,
+      "grad_norm": 0.2094235992786249,
+      "learning_rate": 8.02491755221693e-06,
+      "loss": 0.3308,
+      "step": 2595
+    },
+    {
+      "epoch": 2.5677546983184967,
+      "grad_norm": 0.22678718769630118,
+      "learning_rate": 8.00659582264566e-06,
+      "loss": 0.3054,
+      "step": 2596
+    },
+    {
+      "epoch": 2.568743818001978,
+      "grad_norm": 0.2309026367438794,
+      "learning_rate": 7.988274093074387e-06,
+      "loss": 0.3136,
+      "step": 2597
+    },
+    {
+      "epoch": 2.56973293768546,
+      "grad_norm": 0.22667618009371604,
+      "learning_rate": 7.969952363503116e-06,
+      "loss": 0.349,
+      "step": 2598
+    },
+    {
+      "epoch": 2.5707220573689415,
+      "grad_norm": 0.21117846126384685,
+      "learning_rate": 7.951630633931844e-06,
+      "loss": 0.327,
+      "step": 2599
+    },
+    {
+      "epoch": 2.5717111770524235,
+      "grad_norm": 0.21092844530263816,
+      "learning_rate": 7.933308904360572e-06,
+      "loss": 0.2968,
+      "step": 2600
+    },
+    {
+      "epoch": 2.572700296735905,
+      "grad_norm": 0.22503274613024515,
+      "learning_rate": 7.9149871747893e-06,
+      "loss": 0.348,
+      "step": 2601
+    },
+    {
+      "epoch": 2.573689416419387,
+      "grad_norm": 0.22080704154411201,
+      "learning_rate": 7.89666544521803e-06,
+      "loss": 0.3173,
+      "step": 2602
+    },
+    {
+      "epoch": 2.5746785361028683,
+      "grad_norm": 0.20732028945319075,
+      "learning_rate": 7.878343715646758e-06,
+      "loss": 0.3168,
+      "step": 2603
+    },
+    {
+      "epoch": 2.5756676557863503,
+      "grad_norm": 0.2380503672520683,
+      "learning_rate": 7.860021986075486e-06,
+      "loss": 0.3353,
+      "step": 2604
+    },
+    {
+      "epoch": 2.5766567754698317,
+      "grad_norm": 0.21944010685108145,
+      "learning_rate": 7.841700256504214e-06,
+      "loss": 0.3154,
+      "step": 2605
+    },
+    {
+      "epoch": 2.5776458951533137,
+      "grad_norm": 0.5291657134084721,
+      "learning_rate": 7.823378526932942e-06,
+      "loss": 0.3486,
+      "step": 2606
+    },
+    {
+      "epoch": 2.578635014836795,
+      "grad_norm": 0.2287926036524713,
+      "learning_rate": 7.805056797361672e-06,
+      "loss": 0.3367,
+      "step": 2607
+    },
+    {
+      "epoch": 2.579624134520277,
+      "grad_norm": 0.24607517669943713,
+      "learning_rate": 7.7867350677904e-06,
+      "loss": 0.3822,
+      "step": 2608
+    },
+    {
+      "epoch": 2.5806132542037585,
+      "grad_norm": 0.30753285728195096,
+      "learning_rate": 7.768413338219128e-06,
+      "loss": 0.331,
+      "step": 2609
+    },
+    {
+      "epoch": 2.5816023738872405,
+      "grad_norm": 0.23264272610772255,
+      "learning_rate": 7.750091608647856e-06,
+      "loss": 0.3177,
+      "step": 2610
+    },
+    {
+      "epoch": 2.582591493570722,
+      "grad_norm": 0.24746556520732035,
+      "learning_rate": 7.731769879076584e-06,
+      "loss": 0.3255,
+      "step": 2611
+    },
+    {
+      "epoch": 2.583580613254204,
+      "grad_norm": 0.2586316776392973,
+      "learning_rate": 7.713448149505313e-06,
+      "loss": 0.3122,
+      "step": 2612
+    },
+    {
+      "epoch": 2.5845697329376853,
+      "grad_norm": 0.22370719892578067,
+      "learning_rate": 7.695126419934042e-06,
+      "loss": 0.3386,
+      "step": 2613
+    },
+    {
+      "epoch": 2.5855588526211672,
+      "grad_norm": 0.22463072855787217,
+      "learning_rate": 7.67680469036277e-06,
+      "loss": 0.3192,
+      "step": 2614
+    },
+    {
+      "epoch": 2.5865479723046487,
+      "grad_norm": 0.23719441390072404,
+      "learning_rate": 7.658482960791499e-06,
+      "loss": 0.3573,
+      "step": 2615
+    },
+    {
+      "epoch": 2.5875370919881306,
+      "grad_norm": 0.607168367699493,
+      "learning_rate": 7.640161231220228e-06,
+      "loss": 0.4113,
+      "step": 2616
+    },
+    {
+      "epoch": 2.588526211671612,
+      "grad_norm": 0.21084308256123116,
+      "learning_rate": 7.621839501648956e-06,
+      "loss": 0.3162,
+      "step": 2617
+    },
+    {
+      "epoch": 2.589515331355094,
+      "grad_norm": 0.22956176268509806,
+      "learning_rate": 7.603517772077684e-06,
+      "loss": 0.3383,
+      "step": 2618
+    },
+    {
+      "epoch": 2.5905044510385755,
+      "grad_norm": 0.22913294281991195,
+      "learning_rate": 7.5851960425064135e-06,
+      "loss": 0.3425,
+      "step": 2619
+    },
+    {
+      "epoch": 2.5914935707220574,
+      "grad_norm": 0.23260244421957282,
+      "learning_rate": 7.566874312935142e-06,
+      "loss": 0.3409,
+      "step": 2620
+    },
+    {
+      "epoch": 2.592482690405539,
+      "grad_norm": 0.2436083162884712,
+      "learning_rate": 7.54855258336387e-06,
+      "loss": 0.3385,
+      "step": 2621
+    },
+    {
+      "epoch": 2.593471810089021,
+      "grad_norm": 0.21262219531288126,
+      "learning_rate": 7.530230853792599e-06,
+      "loss": 0.3135,
+      "step": 2622
+    },
+    {
+      "epoch": 2.5944609297725023,
+      "grad_norm": 0.2368450650064414,
+      "learning_rate": 7.511909124221327e-06,
+      "loss": 0.385,
+      "step": 2623
+    },
+    {
+      "epoch": 2.5954500494559842,
+      "grad_norm": 0.215786394243701,
+      "learning_rate": 7.493587394650055e-06,
+      "loss": 0.3208,
+      "step": 2624
+    },
+    {
+      "epoch": 2.596439169139466,
+      "grad_norm": 0.238337529875864,
+      "learning_rate": 7.475265665078785e-06,
+      "loss": 0.3352,
+      "step": 2625
+    },
+    {
+      "epoch": 2.5974282888229476,
+      "grad_norm": 0.20979045932789106,
+      "learning_rate": 7.456943935507513e-06,
+      "loss": 0.3257,
+      "step": 2626
+    },
+    {
+      "epoch": 2.598417408506429,
+      "grad_norm": 0.24130139437210824,
+      "learning_rate": 7.438622205936241e-06,
+      "loss": 0.3548,
+      "step": 2627
+    },
+    {
+      "epoch": 2.599406528189911,
+      "grad_norm": 0.2258956518218783,
+      "learning_rate": 7.420300476364969e-06,
+      "loss": 0.3752,
+      "step": 2628
+    },
+    {
+      "epoch": 2.600395647873393,
+      "grad_norm": 0.23656689766867856,
+      "learning_rate": 7.401978746793697e-06,
+      "loss": 0.3919,
+      "step": 2629
+    },
+    {
+      "epoch": 2.6013847675568744,
+      "grad_norm": 0.20797130671298258,
+      "learning_rate": 7.383657017222425e-06,
+      "loss": 0.295,
+      "step": 2630
+    },
+    {
+      "epoch": 2.602373887240356,
+      "grad_norm": 0.2123407700433041,
+      "learning_rate": 7.365335287651155e-06,
+      "loss": 0.3318,
+      "step": 2631
+    },
+    {
+      "epoch": 2.603363006923838,
+      "grad_norm": 0.25028617103387574,
+      "learning_rate": 7.347013558079883e-06,
+      "loss": 0.3695,
+      "step": 2632
+    },
+    {
+      "epoch": 2.6043521266073197,
+      "grad_norm": 0.23738737755032072,
+      "learning_rate": 7.328691828508611e-06,
+      "loss": 0.3469,
+      "step": 2633
+    },
+    {
+      "epoch": 2.605341246290801,
+      "grad_norm": 0.230389214849109,
+      "learning_rate": 7.31037009893734e-06,
+      "loss": 0.3563,
+      "step": 2634
+    },
+    {
+      "epoch": 2.6063303659742827,
+      "grad_norm": 0.21597950801778004,
+      "learning_rate": 7.292048369366068e-06,
+      "loss": 0.3339,
+      "step": 2635
+    },
+    {
+      "epoch": 2.6073194856577646,
+      "grad_norm": 0.2349603015018135,
+      "learning_rate": 7.2737266397947965e-06,
+      "loss": 0.3586,
+      "step": 2636
+    },
+    {
+      "epoch": 2.6083086053412465,
+      "grad_norm": 0.2083573369468888,
+      "learning_rate": 7.255404910223526e-06,
+      "loss": 0.3004,
+      "step": 2637
+    },
+    {
+      "epoch": 2.609297725024728,
+      "grad_norm": 0.2375417304044579,
+      "learning_rate": 7.2370831806522544e-06,
+      "loss": 0.3667,
+      "step": 2638
+    },
+    {
+      "epoch": 2.6102868447082095,
+      "grad_norm": 0.21283724617762306,
+      "learning_rate": 7.2187614510809825e-06,
+      "loss": 0.3242,
+      "step": 2639
+    },
+    {
+      "epoch": 2.6112759643916914,
+      "grad_norm": 0.24402416859387327,
+      "learning_rate": 7.200439721509711e-06,
+      "loss": 0.3715,
+      "step": 2640
+    },
+    {
+      "epoch": 2.6122650840751733,
+      "grad_norm": 0.22429238323329195,
+      "learning_rate": 7.182117991938439e-06,
+      "loss": 0.3266,
+      "step": 2641
+    },
+    {
+      "epoch": 2.613254203758655,
+      "grad_norm": 0.21784385253259106,
+      "learning_rate": 7.163796262367167e-06,
+      "loss": 0.3087,
+      "step": 2642
+    },
+    {
+      "epoch": 2.6142433234421363,
+      "grad_norm": 0.22124966929792098,
+      "learning_rate": 7.145474532795897e-06,
+      "loss": 0.3418,
+      "step": 2643
+    },
+    {
+      "epoch": 2.615232443125618,
+      "grad_norm": 0.21365105225452002,
+      "learning_rate": 7.127152803224625e-06,
+      "loss": 0.3277,
+      "step": 2644
+    },
+    {
+      "epoch": 2.6162215628091,
+      "grad_norm": 0.22073371217981072,
+      "learning_rate": 7.108831073653354e-06,
+      "loss": 0.3436,
+      "step": 2645
+    },
+    {
+      "epoch": 2.6172106824925816,
+      "grad_norm": 0.23269085111573848,
+      "learning_rate": 7.090509344082082e-06,
+      "loss": 0.3352,
+      "step": 2646
+    },
+    {
+      "epoch": 2.618199802176063,
+      "grad_norm": 0.20633980623132578,
+      "learning_rate": 7.07218761451081e-06,
+      "loss": 0.3013,
+      "step": 2647
+    },
+    {
+      "epoch": 2.619188921859545,
+      "grad_norm": 0.2241331658702146,
+      "learning_rate": 7.053865884939538e-06,
+      "loss": 0.3542,
+      "step": 2648
+    },
+    {
+      "epoch": 2.620178041543027,
+      "grad_norm": 0.2289694050769472,
+      "learning_rate": 7.035544155368268e-06,
+      "loss": 0.3261,
+      "step": 2649
+    },
+    {
+      "epoch": 2.6211671612265084,
+      "grad_norm": 0.22213256212288926,
+      "learning_rate": 7.017222425796996e-06,
+      "loss": 0.3227,
+      "step": 2650
+    },
+    {
+      "epoch": 2.62215628090999,
+      "grad_norm": 0.22863504343907765,
+      "learning_rate": 6.998900696225724e-06,
+      "loss": 0.3628,
+      "step": 2651
+    },
+    {
+      "epoch": 2.623145400593472,
+      "grad_norm": 0.2166138187893422,
+      "learning_rate": 6.980578966654452e-06,
+      "loss": 0.3213,
+      "step": 2652
+    },
+    {
+      "epoch": 2.6241345202769537,
+      "grad_norm": 0.22061018086704073,
+      "learning_rate": 6.96225723708318e-06,
+      "loss": 0.362,
+      "step": 2653
+    },
+    {
+      "epoch": 2.625123639960435,
+      "grad_norm": 0.24304965680542823,
+      "learning_rate": 6.943935507511909e-06,
+      "loss": 0.3757,
+      "step": 2654
+    },
+    {
+      "epoch": 2.6261127596439167,
+      "grad_norm": 0.2166210140761748,
+      "learning_rate": 6.925613777940638e-06,
+      "loss": 0.3167,
+      "step": 2655
+    },
+    {
+      "epoch": 2.6271018793273986,
+      "grad_norm": 0.21031218581565844,
+      "learning_rate": 6.907292048369366e-06,
+      "loss": 0.3391,
+      "step": 2656
+    },
+    {
+      "epoch": 2.6280909990108805,
+      "grad_norm": 0.20543467160833437,
+      "learning_rate": 6.888970318798095e-06,
+      "loss": 0.3058,
+      "step": 2657
+    },
+    {
+      "epoch": 2.629080118694362,
+      "grad_norm": 0.23810595397235643,
+      "learning_rate": 6.8706485892268234e-06,
+      "loss": 0.3524,
+      "step": 2658
+    },
+    {
+      "epoch": 2.6300692383778435,
+      "grad_norm": 0.21740713887368684,
+      "learning_rate": 6.8523268596555516e-06,
+      "loss": 0.3436,
+      "step": 2659
+    },
+    {
+      "epoch": 2.6310583580613254,
+      "grad_norm": 0.20995936719262198,
+      "learning_rate": 6.83400513008428e-06,
+      "loss": 0.3251,
+      "step": 2660
+    },
+    {
+      "epoch": 2.6320474777448073,
+      "grad_norm": 0.2303528115218695,
+      "learning_rate": 6.8156834005130095e-06,
+      "loss": 0.3675,
+      "step": 2661
+    },
+    {
+      "epoch": 2.6330365974282888,
+      "grad_norm": 0.22458667063645485,
+      "learning_rate": 6.797361670941738e-06,
+      "loss": 0.3362,
+      "step": 2662
+    },
+    {
+      "epoch": 2.6340257171117702,
+      "grad_norm": 0.20914465008907085,
+      "learning_rate": 6.779039941370466e-06,
+      "loss": 0.3254,
+      "step": 2663
+    },
+    {
+      "epoch": 2.635014836795252,
+      "grad_norm": 0.23181001869723658,
+      "learning_rate": 6.760718211799194e-06,
+      "loss": 0.3387,
+      "step": 2664
+    },
+    {
+      "epoch": 2.636003956478734,
+      "grad_norm": 0.2223165254541124,
+      "learning_rate": 6.742396482227922e-06,
+      "loss": 0.3498,
+      "step": 2665
+    },
+    {
+      "epoch": 2.6369930761622156,
+      "grad_norm": 0.22334071001886052,
+      "learning_rate": 6.724074752656651e-06,
+      "loss": 0.3115,
+      "step": 2666
+    },
+    {
+      "epoch": 2.6379821958456975,
+      "grad_norm": 0.23398152025847294,
+      "learning_rate": 6.70575302308538e-06,
+      "loss": 0.3507,
+      "step": 2667
+    },
+    {
+      "epoch": 2.638971315529179,
+      "grad_norm": 0.22539172159114598,
+      "learning_rate": 6.687431293514109e-06,
+      "loss": 0.3611,
+      "step": 2668
+    },
+    {
+      "epoch": 2.639960435212661,
+      "grad_norm": 0.21097633122850826,
+      "learning_rate": 6.669109563942837e-06,
+      "loss": 0.3183,
+      "step": 2669
+    },
+    {
+      "epoch": 2.6409495548961424,
+      "grad_norm": 0.2445964426483477,
+      "learning_rate": 6.650787834371565e-06,
+      "loss": 0.3399,
+      "step": 2670
+    },
+    {
+      "epoch": 2.6419386745796243,
+      "grad_norm": 0.23695631235504286,
+      "learning_rate": 6.632466104800293e-06,
+      "loss": 0.3412,
+      "step": 2671
+    },
+    {
+      "epoch": 2.6429277942631058,
+      "grad_norm": 0.2534807066261069,
+      "learning_rate": 6.614144375229021e-06,
+      "loss": 0.3811,
+      "step": 2672
+    },
+    {
+      "epoch": 2.6439169139465877,
+      "grad_norm": 0.2352492192731664,
+      "learning_rate": 6.595822645657751e-06,
+      "loss": 0.3265,
+      "step": 2673
+    },
+    {
+      "epoch": 2.644906033630069,
+      "grad_norm": 0.22325401378241108,
+      "learning_rate": 6.577500916086479e-06,
+      "loss": 0.338,
+      "step": 2674
+    },
+    {
+      "epoch": 2.645895153313551,
+      "grad_norm": 0.2355023381434876,
+      "learning_rate": 6.559179186515207e-06,
+      "loss": 0.3439,
+      "step": 2675
+    },
+    {
+      "epoch": 2.6468842729970326,
+      "grad_norm": 0.25844539764836716,
+      "learning_rate": 6.540857456943935e-06,
+      "loss": 0.3465,
+      "step": 2676
+    },
+    {
+      "epoch": 2.6478733926805145,
+      "grad_norm": 0.20204122613105294,
+      "learning_rate": 6.522535727372664e-06,
+      "loss": 0.326,
+      "step": 2677
+    },
+    {
+      "epoch": 2.648862512363996,
+      "grad_norm": 0.20091765692132063,
+      "learning_rate": 6.5042139978013925e-06,
+      "loss": 0.3146,
+      "step": 2678
+    },
+    {
+      "epoch": 2.649851632047478,
+      "grad_norm": 0.2151974965408436,
+      "learning_rate": 6.485892268230121e-06,
+      "loss": 0.3508,
+      "step": 2679
+    },
+    {
+      "epoch": 2.6508407517309593,
+      "grad_norm": 0.25164157106976864,
+      "learning_rate": 6.46757053865885e-06,
+      "loss": 0.3793,
+      "step": 2680
+    },
+    {
+      "epoch": 2.6518298714144413,
+      "grad_norm": 0.23370693030036446,
+      "learning_rate": 6.4492488090875785e-06,
+      "loss": 0.4006,
+      "step": 2681
+    },
+    {
+      "epoch": 2.6528189910979227,
+      "grad_norm": 0.23987905015984629,
+      "learning_rate": 6.430927079516307e-06,
+      "loss": 0.3613,
+      "step": 2682
+    },
+    {
+      "epoch": 2.6538081107814047,
+      "grad_norm": 0.21594181062915788,
+      "learning_rate": 6.412605349945035e-06,
+      "loss": 0.3307,
+      "step": 2683
+    },
+    {
+      "epoch": 2.654797230464886,
+      "grad_norm": 0.2290158725540692,
+      "learning_rate": 6.394283620373763e-06,
+      "loss": 0.3062,
+      "step": 2684
+    },
+    {
+      "epoch": 2.655786350148368,
+      "grad_norm": 0.22948262260424798,
+      "learning_rate": 6.375961890802493e-06,
+      "loss": 0.3722,
+      "step": 2685
+    },
+    {
+      "epoch": 2.6567754698318495,
+      "grad_norm": 0.22027398416939878,
+      "learning_rate": 6.357640161231221e-06,
+      "loss": 0.3253,
+      "step": 2686
+    },
+    {
+      "epoch": 2.6577645895153315,
+      "grad_norm": 0.22047131128891237,
+      "learning_rate": 6.339318431659949e-06,
+      "loss": 0.3447,
+      "step": 2687
+    },
+    {
+      "epoch": 2.658753709198813,
+      "grad_norm": 0.21264120578875886,
+      "learning_rate": 6.320996702088677e-06,
+      "loss": 0.3178,
+      "step": 2688
+    },
+    {
+      "epoch": 2.659742828882295,
+      "grad_norm": 0.21403466908930086,
+      "learning_rate": 6.302674972517406e-06,
+      "loss": 0.3263,
+      "step": 2689
+    },
+    {
+      "epoch": 2.6607319485657763,
+      "grad_norm": 0.2340128370458701,
+      "learning_rate": 6.284353242946134e-06,
+      "loss": 0.344,
+      "step": 2690
+    },
+    {
+      "epoch": 2.6617210682492582,
+      "grad_norm": 0.21709601415825422,
+      "learning_rate": 6.266031513374864e-06,
+      "loss": 0.332,
+      "step": 2691
+    },
+    {
+      "epoch": 2.6627101879327397,
+      "grad_norm": 0.20573493768846976,
+      "learning_rate": 6.247709783803592e-06,
+      "loss": 0.2896,
+      "step": 2692
+    },
+    {
+      "epoch": 2.6636993076162216,
+      "grad_norm": 0.2175633594597353,
+      "learning_rate": 6.22938805423232e-06,
+      "loss": 0.3255,
+      "step": 2693
+    },
+    {
+      "epoch": 2.664688427299703,
+      "grad_norm": 0.2208890382225565,
+      "learning_rate": 6.211066324661048e-06,
+      "loss": 0.3241,
+      "step": 2694
+    },
+    {
+      "epoch": 2.665677546983185,
+      "grad_norm": 0.22648814414163115,
+      "learning_rate": 6.192744595089777e-06,
+      "loss": 0.3803,
+      "step": 2695
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.2065203676472046,
+      "learning_rate": 6.174422865518505e-06,
+      "loss": 0.3179,
+      "step": 2696
+    },
+    {
+      "epoch": 2.6676557863501484,
+      "grad_norm": 0.22119342098303224,
+      "learning_rate": 6.156101135947233e-06,
+      "loss": 0.3375,
+      "step": 2697
+    },
+    {
+      "epoch": 2.66864490603363,
+      "grad_norm": 0.22093559612372538,
+      "learning_rate": 6.137779406375962e-06,
+      "loss": 0.3514,
+      "step": 2698
+    },
+    {
+      "epoch": 2.669634025717112,
+      "grad_norm": 0.4155710402624548,
+      "learning_rate": 6.1194576768046904e-06,
+      "loss": 0.4384,
+      "step": 2699
+    },
+    {
+      "epoch": 2.6706231454005933,
+      "grad_norm": 0.21197542896961055,
+      "learning_rate": 6.101135947233419e-06,
+      "loss": 0.3247,
+      "step": 2700
+    },
+    {
+      "epoch": 2.6716122650840752,
+      "grad_norm": 0.20867372426418582,
+      "learning_rate": 6.0828142176621475e-06,
+      "loss": 0.3162,
+      "step": 2701
+    },
+    {
+      "epoch": 2.6726013847675567,
+      "grad_norm": 0.21283459025003898,
+      "learning_rate": 6.0644924880908765e-06,
+      "loss": 0.3088,
+      "step": 2702
+    },
+    {
+      "epoch": 2.6735905044510386,
+      "grad_norm": 0.27512316233142525,
+      "learning_rate": 6.0461707585196046e-06,
+      "loss": 0.3543,
+      "step": 2703
+    },
+    {
+      "epoch": 2.6745796241345206,
+      "grad_norm": 0.23578251994420446,
+      "learning_rate": 6.0278490289483335e-06,
+      "loss": 0.3408,
+      "step": 2704
+    },
+    {
+      "epoch": 2.675568743818002,
+      "grad_norm": 0.23108104721689834,
+      "learning_rate": 6.009527299377062e-06,
+      "loss": 0.3547,
+      "step": 2705
+    },
+    {
+      "epoch": 2.6765578635014835,
+      "grad_norm": 0.22990919690448694,
+      "learning_rate": 5.99120556980579e-06,
+      "loss": 0.3583,
+      "step": 2706
+    },
+    {
+      "epoch": 2.6775469831849654,
+      "grad_norm": 0.22608119799904844,
+      "learning_rate": 5.972883840234519e-06,
+      "loss": 0.3468,
+      "step": 2707
+    },
+    {
+      "epoch": 2.6785361028684473,
+      "grad_norm": 0.23420424171951532,
+      "learning_rate": 5.954562110663247e-06,
+      "loss": 0.3711,
+      "step": 2708
+    },
+    {
+      "epoch": 2.679525222551929,
+      "grad_norm": 0.22283499002319432,
+      "learning_rate": 5.936240381091975e-06,
+      "loss": 0.3433,
+      "step": 2709
+    },
+    {
+      "epoch": 2.6805143422354103,
+      "grad_norm": 0.25165983826778154,
+      "learning_rate": 5.917918651520704e-06,
+      "loss": 0.3577,
+      "step": 2710
+    },
+    {
+      "epoch": 2.681503461918892,
+      "grad_norm": 0.21381831548899768,
+      "learning_rate": 5.899596921949432e-06,
+      "loss": 0.3243,
+      "step": 2711
+    },
+    {
+      "epoch": 2.682492581602374,
+      "grad_norm": 0.24859770910561865,
+      "learning_rate": 5.881275192378161e-06,
+      "loss": 0.3932,
+      "step": 2712
+    },
+    {
+      "epoch": 2.6834817012858556,
+      "grad_norm": 0.22835997974466907,
+      "learning_rate": 5.86295346280689e-06,
+      "loss": 0.358,
+      "step": 2713
+    },
+    {
+      "epoch": 2.684470820969337,
+      "grad_norm": 0.21603309757069702,
+      "learning_rate": 5.844631733235618e-06,
+      "loss": 0.3227,
+      "step": 2714
+    },
+    {
+      "epoch": 2.685459940652819,
+      "grad_norm": 0.22420474785394226,
+      "learning_rate": 5.826310003664346e-06,
+      "loss": 0.3219,
+      "step": 2715
+    },
+    {
+      "epoch": 2.686449060336301,
+      "grad_norm": 0.211435200550928,
+      "learning_rate": 5.807988274093075e-06,
+      "loss": 0.3027,
+      "step": 2716
+    },
+    {
+      "epoch": 2.6874381800197824,
+      "grad_norm": 0.2244513347827355,
+      "learning_rate": 5.789666544521803e-06,
+      "loss": 0.3397,
+      "step": 2717
+    },
+    {
+      "epoch": 2.688427299703264,
+      "grad_norm": 0.21717002471268115,
+      "learning_rate": 5.771344814950531e-06,
+      "loss": 0.3334,
+      "step": 2718
+    },
+    {
+      "epoch": 2.689416419386746,
+      "grad_norm": 0.23016598747660916,
+      "learning_rate": 5.75302308537926e-06,
+      "loss": 0.3375,
+      "step": 2719
+    },
+    {
+      "epoch": 2.6904055390702277,
+      "grad_norm": 0.21015267166960389,
+      "learning_rate": 5.734701355807988e-06,
+      "loss": 0.3159,
+      "step": 2720
+    },
+    {
+      "epoch": 2.691394658753709,
+      "grad_norm": 0.2108694801384789,
+      "learning_rate": 5.7163796262367165e-06,
+      "loss": 0.3308,
+      "step": 2721
+    },
+    {
+      "epoch": 2.6923837784371907,
+      "grad_norm": 0.23368408762172113,
+      "learning_rate": 5.6980578966654455e-06,
+      "loss": 0.3768,
+      "step": 2722
+    },
+    {
+      "epoch": 2.6933728981206726,
+      "grad_norm": 0.23439430393969224,
+      "learning_rate": 5.679736167094174e-06,
+      "loss": 0.3017,
+      "step": 2723
+    },
+    {
+      "epoch": 2.6943620178041545,
+      "grad_norm": 0.23942635675606497,
+      "learning_rate": 5.6614144375229025e-06,
+      "loss": 0.3689,
+      "step": 2724
+    },
+    {
+      "epoch": 2.695351137487636,
+      "grad_norm": 0.22177373379695609,
+      "learning_rate": 5.6430927079516315e-06,
+      "loss": 0.3521,
+      "step": 2725
+    },
+    {
+      "epoch": 2.6963402571711175,
+      "grad_norm": 0.21547233338756122,
+      "learning_rate": 5.62477097838036e-06,
+      "loss": 0.3416,
+      "step": 2726
+    },
+    {
+      "epoch": 2.6973293768545994,
+      "grad_norm": 0.21774106460684997,
+      "learning_rate": 5.606449248809088e-06,
+      "loss": 0.3218,
+      "step": 2727
+    },
+    {
+      "epoch": 2.6983184965380813,
+      "grad_norm": 0.2334321077543478,
+      "learning_rate": 5.588127519237817e-06,
+      "loss": 0.358,
+      "step": 2728
+    },
+    {
+      "epoch": 2.699307616221563,
+      "grad_norm": 0.24139566192301815,
+      "learning_rate": 5.569805789666545e-06,
+      "loss": 0.3771,
+      "step": 2729
+    },
+    {
+      "epoch": 2.7002967359050443,
+      "grad_norm": 0.2540431339935207,
+      "learning_rate": 5.551484060095273e-06,
+      "loss": 0.3657,
+      "step": 2730
+    },
+    {
+      "epoch": 2.701285855588526,
+      "grad_norm": 0.5909800839666434,
+      "learning_rate": 5.533162330524002e-06,
+      "loss": 0.3952,
+      "step": 2731
+    },
+    {
+      "epoch": 2.702274975272008,
+      "grad_norm": 0.2503607288953179,
+      "learning_rate": 5.51484060095273e-06,
+      "loss": 0.3793,
+      "step": 2732
+    },
+    {
+      "epoch": 2.7032640949554896,
+      "grad_norm": 0.2223796649689281,
+      "learning_rate": 5.496518871381458e-06,
+      "loss": 0.3031,
+      "step": 2733
+    },
+    {
+      "epoch": 2.704253214638971,
+      "grad_norm": 0.23538524556735074,
+      "learning_rate": 5.478197141810187e-06,
+      "loss": 0.3706,
+      "step": 2734
+    },
+    {
+      "epoch": 2.705242334322453,
+      "grad_norm": 0.2229912934237323,
+      "learning_rate": 5.459875412238916e-06,
+      "loss": 0.3253,
+      "step": 2735
+    },
+    {
+      "epoch": 2.706231454005935,
+      "grad_norm": 0.22679198948267829,
+      "learning_rate": 5.441553682667644e-06,
+      "loss": 0.3297,
+      "step": 2736
+    },
+    {
+      "epoch": 2.7072205736894164,
+      "grad_norm": 0.21590619745768205,
+      "learning_rate": 5.423231953096373e-06,
+      "loss": 0.3169,
+      "step": 2737
+    },
+    {
+      "epoch": 2.708209693372898,
+      "grad_norm": 0.2217628704155836,
+      "learning_rate": 5.404910223525101e-06,
+      "loss": 0.3624,
+      "step": 2738
+    },
+    {
+      "epoch": 2.7091988130563798,
+      "grad_norm": 0.2169003913914285,
+      "learning_rate": 5.386588493953829e-06,
+      "loss": 0.3112,
+      "step": 2739
+    },
+    {
+      "epoch": 2.7101879327398617,
+      "grad_norm": 0.23975530450787874,
+      "learning_rate": 5.368266764382558e-06,
+      "loss": 0.371,
+      "step": 2740
+    },
+    {
+      "epoch": 2.711177052423343,
+      "grad_norm": 0.22760491329000107,
+      "learning_rate": 5.349945034811286e-06,
+      "loss": 0.3033,
+      "step": 2741
+    },
+    {
+      "epoch": 2.712166172106825,
+      "grad_norm": 0.2126172357592932,
+      "learning_rate": 5.3316233052400145e-06,
+      "loss": 0.3459,
+      "step": 2742
+    },
+    {
+      "epoch": 2.7131552917903066,
+      "grad_norm": 0.24447754746061037,
+      "learning_rate": 5.3133015756687434e-06,
+      "loss": 0.3579,
+      "step": 2743
+    },
+    {
+      "epoch": 2.7141444114737885,
+      "grad_norm": 0.21923248450136007,
+      "learning_rate": 5.2949798460974715e-06,
+      "loss": 0.3553,
+      "step": 2744
+    },
+    {
+      "epoch": 2.71513353115727,
+      "grad_norm": 0.23324322920894322,
+      "learning_rate": 5.2766581165262005e-06,
+      "loss": 0.3876,
+      "step": 2745
+    },
+    {
+      "epoch": 2.716122650840752,
+      "grad_norm": 0.21651127289013336,
+      "learning_rate": 5.258336386954929e-06,
+      "loss": 0.3271,
+      "step": 2746
+    },
+    {
+      "epoch": 2.7171117705242334,
+      "grad_norm": 0.2278600935466184,
+      "learning_rate": 5.2400146573836576e-06,
+      "loss": 0.3767,
+      "step": 2747
+    },
+    {
+      "epoch": 2.7181008902077153,
+      "grad_norm": 0.2102279454382535,
+      "learning_rate": 5.221692927812386e-06,
+      "loss": 0.2996,
+      "step": 2748
+    },
+    {
+      "epoch": 2.7190900098911968,
+      "grad_norm": 0.19846329542181648,
+      "learning_rate": 5.203371198241115e-06,
+      "loss": 0.2969,
+      "step": 2749
+    },
+    {
+      "epoch": 2.7200791295746787,
+      "grad_norm": 0.23474124954742392,
+      "learning_rate": 5.185049468669843e-06,
+      "loss": 0.3563,
+      "step": 2750
+    },
+    {
+      "epoch": 2.72106824925816,
+      "grad_norm": 0.2133239490347135,
+      "learning_rate": 5.166727739098571e-06,
+      "loss": 0.3625,
+      "step": 2751
+    },
+    {
+      "epoch": 2.722057368941642,
+      "grad_norm": 0.2358371050129765,
+      "learning_rate": 5.1484060095273e-06,
+      "loss": 0.3468,
+      "step": 2752
+    },
+    {
+      "epoch": 2.7230464886251236,
+      "grad_norm": 0.20255887262973588,
+      "learning_rate": 5.130084279956028e-06,
+      "loss": 0.3045,
+      "step": 2753
+    },
+    {
+      "epoch": 2.7240356083086055,
+      "grad_norm": 0.22868689020977842,
+      "learning_rate": 5.111762550384756e-06,
+      "loss": 0.3525,
+      "step": 2754
+    },
+    {
+      "epoch": 2.725024727992087,
+      "grad_norm": 0.24113962478912296,
+      "learning_rate": 5.093440820813485e-06,
+      "loss": 0.3564,
+      "step": 2755
+    },
+    {
+      "epoch": 2.726013847675569,
+      "grad_norm": 0.2442213322729859,
+      "learning_rate": 5.075119091242213e-06,
+      "loss": 0.3272,
+      "step": 2756
+    },
+    {
+      "epoch": 2.7270029673590503,
+      "grad_norm": 0.22564109275269484,
+      "learning_rate": 5.056797361670942e-06,
+      "loss": 0.3294,
+      "step": 2757
+    },
+    {
+      "epoch": 2.7279920870425323,
+      "grad_norm": 0.2187439459162259,
+      "learning_rate": 5.038475632099671e-06,
+      "loss": 0.3428,
+      "step": 2758
+    },
+    {
+      "epoch": 2.7289812067260137,
+      "grad_norm": 0.22198247455708378,
+      "learning_rate": 5.020153902528399e-06,
+      "loss": 0.3523,
+      "step": 2759
+    },
+    {
+      "epoch": 2.7299703264094957,
+      "grad_norm": 0.20648216241698295,
+      "learning_rate": 5.001832172957127e-06,
+      "loss": 0.2998,
+      "step": 2760
+    },
+    {
+      "epoch": 2.730959446092977,
+      "grad_norm": 0.21591111811349886,
+      "learning_rate": 4.983510443385856e-06,
+      "loss": 0.3502,
+      "step": 2761
+    },
+    {
+      "epoch": 2.731948565776459,
+      "grad_norm": 0.20875229932327685,
+      "learning_rate": 4.965188713814584e-06,
+      "loss": 0.319,
+      "step": 2762
+    },
+    {
+      "epoch": 2.7329376854599405,
+      "grad_norm": 0.2309618563362465,
+      "learning_rate": 4.9468669842433124e-06,
+      "loss": 0.3301,
+      "step": 2763
+    },
+    {
+      "epoch": 2.7339268051434225,
+      "grad_norm": 0.21705046113121196,
+      "learning_rate": 4.928545254672041e-06,
+      "loss": 0.3328,
+      "step": 2764
+    },
+    {
+      "epoch": 2.734915924826904,
+      "grad_norm": 0.2027787922422683,
+      "learning_rate": 4.9102235251007695e-06,
+      "loss": 0.3022,
+      "step": 2765
+    },
+    {
+      "epoch": 2.735905044510386,
+      "grad_norm": 0.2133690055551515,
+      "learning_rate": 4.891901795529498e-06,
+      "loss": 0.3357,
+      "step": 2766
+    },
+    {
+      "epoch": 2.7368941641938673,
+      "grad_norm": 0.22559371081701649,
+      "learning_rate": 4.873580065958227e-06,
+      "loss": 0.3545,
+      "step": 2767
+    },
+    {
+      "epoch": 2.7378832838773492,
+      "grad_norm": 0.21203557203334927,
+      "learning_rate": 4.8552583363869555e-06,
+      "loss": 0.3328,
+      "step": 2768
+    },
+    {
+      "epoch": 2.7388724035608307,
+      "grad_norm": 0.21294747265250044,
+      "learning_rate": 4.836936606815684e-06,
+      "loss": 0.332,
+      "step": 2769
+    },
+    {
+      "epoch": 2.7398615232443126,
+      "grad_norm": 0.21123677919400954,
+      "learning_rate": 4.818614877244413e-06,
+      "loss": 0.3414,
+      "step": 2770
+    },
+    {
+      "epoch": 2.740850642927794,
+      "grad_norm": 0.21659310977778334,
+      "learning_rate": 4.800293147673141e-06,
+      "loss": 0.3264,
+      "step": 2771
+    },
+    {
+      "epoch": 2.741839762611276,
+      "grad_norm": 0.22960667507714003,
+      "learning_rate": 4.781971418101869e-06,
+      "loss": 0.374,
+      "step": 2772
+    },
+    {
+      "epoch": 2.7428288822947575,
+      "grad_norm": 0.21339735395164217,
+      "learning_rate": 4.763649688530598e-06,
+      "loss": 0.3446,
+      "step": 2773
+    },
+    {
+      "epoch": 2.7438180019782394,
+      "grad_norm": 0.23992675800975746,
+      "learning_rate": 4.745327958959326e-06,
+      "loss": 0.3627,
+      "step": 2774
+    },
+    {
+      "epoch": 2.744807121661721,
+      "grad_norm": 0.21653441504113724,
+      "learning_rate": 4.727006229388054e-06,
+      "loss": 0.3292,
+      "step": 2775
+    },
+    {
+      "epoch": 2.745796241345203,
+      "grad_norm": 0.20765781658155638,
+      "learning_rate": 4.708684499816783e-06,
+      "loss": 0.3369,
+      "step": 2776
+    },
+    {
+      "epoch": 2.7467853610286843,
+      "grad_norm": 0.2137923966682356,
+      "learning_rate": 4.690362770245511e-06,
+      "loss": 0.3316,
+      "step": 2777
+    },
+    {
+      "epoch": 2.7477744807121662,
+      "grad_norm": 0.22951511263981078,
+      "learning_rate": 4.672041040674239e-06,
+      "loss": 0.3232,
+      "step": 2778
+    },
+    {
+      "epoch": 2.7487636003956477,
+      "grad_norm": 0.21668420363059104,
+      "learning_rate": 4.653719311102968e-06,
+      "loss": 0.3133,
+      "step": 2779
+    },
+    {
+      "epoch": 2.7497527200791296,
+      "grad_norm": 0.20870298094145542,
+      "learning_rate": 4.635397581531697e-06,
+      "loss": 0.3079,
+      "step": 2780
+    },
+    {
+      "epoch": 2.750741839762611,
+      "grad_norm": 0.21695592576488162,
+      "learning_rate": 4.617075851960425e-06,
+      "loss": 0.3365,
+      "step": 2781
+    },
+    {
+      "epoch": 2.751730959446093,
+      "grad_norm": 0.23452800703138157,
+      "learning_rate": 4.598754122389154e-06,
+      "loss": 0.3605,
+      "step": 2782
+    },
+    {
+      "epoch": 2.752720079129575,
+      "grad_norm": 0.22386479424333205,
+      "learning_rate": 4.580432392817882e-06,
+      "loss": 0.3766,
+      "step": 2783
+    },
+    {
+      "epoch": 2.7537091988130564,
+      "grad_norm": 0.2500059904791562,
+      "learning_rate": 4.56211066324661e-06,
+      "loss": 0.3552,
+      "step": 2784
+    },
+    {
+      "epoch": 2.754698318496538,
+      "grad_norm": 0.2120523623817133,
+      "learning_rate": 4.543788933675339e-06,
+      "loss": 0.3178,
+      "step": 2785
+    },
+    {
+      "epoch": 2.75568743818002,
+      "grad_norm": 0.2151292399138165,
+      "learning_rate": 4.5254672041040675e-06,
+      "loss": 0.3671,
+      "step": 2786
+    },
+    {
+      "epoch": 2.7566765578635017,
+      "grad_norm": 0.21872770644925504,
+      "learning_rate": 4.507145474532796e-06,
+      "loss": 0.3641,
+      "step": 2787
+    },
+    {
+      "epoch": 2.757665677546983,
+      "grad_norm": 0.24772664908497757,
+      "learning_rate": 4.4888237449615246e-06,
+      "loss": 0.3734,
+      "step": 2788
+    },
+    {
+      "epoch": 2.7586547972304647,
+      "grad_norm": 0.2642668933133229,
+      "learning_rate": 4.470502015390253e-06,
+      "loss": 0.3684,
+      "step": 2789
+    },
+    {
+      "epoch": 2.7596439169139466,
+      "grad_norm": 0.2221777432007733,
+      "learning_rate": 4.452180285818982e-06,
+      "loss": 0.349,
+      "step": 2790
+    },
+    {
+      "epoch": 2.7606330365974285,
+      "grad_norm": 0.2196073440373851,
+      "learning_rate": 4.43385855624771e-06,
+      "loss": 0.3489,
+      "step": 2791
+    },
+    {
+      "epoch": 2.76162215628091,
+      "grad_norm": 0.23391467333346672,
+      "learning_rate": 4.415536826676439e-06,
+      "loss": 0.3598,
+      "step": 2792
+    },
+    {
+      "epoch": 2.7626112759643915,
+      "grad_norm": 0.2115533472124657,
+      "learning_rate": 4.397215097105167e-06,
+      "loss": 0.3081,
+      "step": 2793
+    },
+    {
+      "epoch": 2.7636003956478734,
+      "grad_norm": 0.2271678820385135,
+      "learning_rate": 4.378893367533896e-06,
+      "loss": 0.3716,
+      "step": 2794
+    },
+    {
+      "epoch": 2.7645895153313553,
+      "grad_norm": 0.23767597252465578,
+      "learning_rate": 4.360571637962624e-06,
+      "loss": 0.3591,
+      "step": 2795
+    },
+    {
+      "epoch": 2.765578635014837,
+      "grad_norm": 0.2066536302237579,
+      "learning_rate": 4.342249908391352e-06,
+      "loss": 0.3413,
+      "step": 2796
+    },
+    {
+      "epoch": 2.7665677546983183,
+      "grad_norm": 0.21845630072729733,
+      "learning_rate": 4.323928178820081e-06,
+      "loss": 0.3577,
+      "step": 2797
+    },
+    {
+      "epoch": 2.7675568743818,
+      "grad_norm": 0.21779485602289012,
+      "learning_rate": 4.305606449248809e-06,
+      "loss": 0.3468,
+      "step": 2798
+    },
+    {
+      "epoch": 2.768545994065282,
+      "grad_norm": 0.22232401786465894,
+      "learning_rate": 4.287284719677537e-06,
+      "loss": 0.3144,
+      "step": 2799
+    },
+    {
+      "epoch": 2.7695351137487636,
+      "grad_norm": 0.23008435815779882,
+      "learning_rate": 4.268962990106266e-06,
+      "loss": 0.3433,
+      "step": 2800
+    },
+    {
+      "epoch": 2.770524233432245,
+      "grad_norm": 0.2226756211496331,
+      "learning_rate": 4.250641260534994e-06,
+      "loss": 0.375,
+      "step": 2801
+    },
+    {
+      "epoch": 2.771513353115727,
+      "grad_norm": 0.21695887700850297,
+      "learning_rate": 4.232319530963723e-06,
+      "loss": 0.3465,
+      "step": 2802
+    },
+    {
+      "epoch": 2.772502472799209,
+      "grad_norm": 0.22223208742172174,
+      "learning_rate": 4.213997801392452e-06,
+      "loss": 0.2978,
+      "step": 2803
+    },
+    {
+      "epoch": 2.7734915924826904,
+      "grad_norm": 0.2170797804111916,
+      "learning_rate": 4.19567607182118e-06,
+      "loss": 0.3213,
+      "step": 2804
+    },
+    {
+      "epoch": 2.774480712166172,
+      "grad_norm": 0.22027515254777197,
+      "learning_rate": 4.177354342249908e-06,
+      "loss": 0.3443,
+      "step": 2805
+    },
+    {
+      "epoch": 2.775469831849654,
+      "grad_norm": 0.22207428166028975,
+      "learning_rate": 4.159032612678637e-06,
+      "loss": 0.3573,
+      "step": 2806
+    },
+    {
+      "epoch": 2.7764589515331357,
+      "grad_norm": 0.2087620444374386,
+      "learning_rate": 4.1407108831073654e-06,
+      "loss": 0.3414,
+      "step": 2807
+    },
+    {
+      "epoch": 2.777448071216617,
+      "grad_norm": 0.2078913845907797,
+      "learning_rate": 4.1223891535360936e-06,
+      "loss": 0.3213,
+      "step": 2808
+    },
+    {
+      "epoch": 2.7784371909000987,
+      "grad_norm": 0.23077801939274703,
+      "learning_rate": 4.1040674239648225e-06,
+      "loss": 0.3524,
+      "step": 2809
+    },
+    {
+      "epoch": 2.7794263105835806,
+      "grad_norm": 0.21731853730142003,
+      "learning_rate": 4.085745694393551e-06,
+      "loss": 0.3441,
+      "step": 2810
+    },
+    {
+      "epoch": 2.7804154302670625,
+      "grad_norm": 0.216573556930499,
+      "learning_rate": 4.067423964822279e-06,
+      "loss": 0.3576,
+      "step": 2811
+    },
+    {
+      "epoch": 2.781404549950544,
+      "grad_norm": 0.21120665151959334,
+      "learning_rate": 4.049102235251008e-06,
+      "loss": 0.3431,
+      "step": 2812
+    },
+    {
+      "epoch": 2.7823936696340255,
+      "grad_norm": 0.23307858510646606,
+      "learning_rate": 4.030780505679737e-06,
+      "loss": 0.3603,
+      "step": 2813
+    },
+    {
+      "epoch": 2.7833827893175074,
+      "grad_norm": 0.2107086413941475,
+      "learning_rate": 4.012458776108465e-06,
+      "loss": 0.3332,
+      "step": 2814
+    },
+    {
+      "epoch": 2.7843719090009893,
+      "grad_norm": 0.23230553012383037,
+      "learning_rate": 3.994137046537194e-06,
+      "loss": 0.3313,
+      "step": 2815
+    },
+    {
+      "epoch": 2.7853610286844708,
+      "grad_norm": 0.25073644328468253,
+      "learning_rate": 3.975815316965922e-06,
+      "loss": 0.4522,
+      "step": 2816
+    },
+    {
+      "epoch": 2.7863501483679523,
+      "grad_norm": 0.22850963515675785,
+      "learning_rate": 3.95749358739465e-06,
+      "loss": 0.3318,
+      "step": 2817
+    },
+    {
+      "epoch": 2.787339268051434,
+      "grad_norm": 0.22594036641456755,
+      "learning_rate": 3.939171857823379e-06,
+      "loss": 0.3078,
+      "step": 2818
+    },
+    {
+      "epoch": 2.788328387734916,
+      "grad_norm": 0.23726950060282387,
+      "learning_rate": 3.920850128252107e-06,
+      "loss": 0.3667,
+      "step": 2819
+    },
+    {
+      "epoch": 2.7893175074183976,
+      "grad_norm": 0.21765453645873728,
+      "learning_rate": 3.902528398680836e-06,
+      "loss": 0.3253,
+      "step": 2820
+    },
+    {
+      "epoch": 2.7903066271018795,
+      "grad_norm": 0.21731756160867208,
+      "learning_rate": 3.884206669109564e-06,
+      "loss": 0.3354,
+      "step": 2821
+    },
+    {
+      "epoch": 2.791295746785361,
+      "grad_norm": 0.20040051432164183,
+      "learning_rate": 3.865884939538292e-06,
+      "loss": 0.3006,
+      "step": 2822
+    },
+    {
+      "epoch": 2.792284866468843,
+      "grad_norm": 0.22186912403842526,
+      "learning_rate": 3.847563209967021e-06,
+      "loss": 0.3171,
+      "step": 2823
+    },
+    {
+      "epoch": 2.7932739861523244,
+      "grad_norm": 0.22337977682594518,
+      "learning_rate": 3.829241480395749e-06,
+      "loss": 0.3805,
+      "step": 2824
+    },
+    {
+      "epoch": 2.7942631058358063,
+      "grad_norm": 0.22332412301847868,
+      "learning_rate": 3.810919750824478e-06,
+      "loss": 0.3363,
+      "step": 2825
+    },
+    {
+      "epoch": 2.7952522255192878,
+      "grad_norm": 0.2184887090680327,
+      "learning_rate": 3.7925980212532068e-06,
+      "loss": 0.3253,
+      "step": 2826
+    },
+    {
+      "epoch": 2.7962413452027697,
+      "grad_norm": 0.2326535028536087,
+      "learning_rate": 3.774276291681935e-06,
+      "loss": 0.3413,
+      "step": 2827
+    },
+    {
+      "epoch": 2.797230464886251,
+      "grad_norm": 0.22537246867937816,
+      "learning_rate": 3.7559545621106634e-06,
+      "loss": 0.3787,
+      "step": 2828
+    },
+    {
+      "epoch": 2.798219584569733,
+      "grad_norm": 0.21512167988752373,
+      "learning_rate": 3.7376328325393924e-06,
+      "loss": 0.3419,
+      "step": 2829
+    },
+    {
+      "epoch": 2.7992087042532146,
+      "grad_norm": 0.21070682115918318,
+      "learning_rate": 3.7193111029681205e-06,
+      "loss": 0.3435,
+      "step": 2830
+    },
+    {
+      "epoch": 2.8001978239366965,
+      "grad_norm": 0.23029319903291864,
+      "learning_rate": 3.7009893733968486e-06,
+      "loss": 0.3486,
+      "step": 2831
+    },
+    {
+      "epoch": 2.801186943620178,
+      "grad_norm": 0.2138444420546789,
+      "learning_rate": 3.6826676438255776e-06,
+      "loss": 0.3415,
+      "step": 2832
+    },
+    {
+      "epoch": 2.80217606330366,
+      "grad_norm": 0.20357209160670728,
+      "learning_rate": 3.6643459142543057e-06,
+      "loss": 0.3293,
+      "step": 2833
+    },
+    {
+      "epoch": 2.8031651829871413,
+      "grad_norm": 0.25026208901493,
+      "learning_rate": 3.646024184683034e-06,
+      "loss": 0.3354,
+      "step": 2834
+    },
+    {
+      "epoch": 2.8041543026706233,
+      "grad_norm": 0.2011884823421506,
+      "learning_rate": 3.627702455111763e-06,
+      "loss": 0.3096,
+      "step": 2835
+    },
+    {
+      "epoch": 2.8051434223541047,
+      "grad_norm": 0.2394174392476079,
+      "learning_rate": 3.6093807255404913e-06,
+      "loss": 0.37,
+      "step": 2836
+    },
+    {
+      "epoch": 2.8061325420375867,
+      "grad_norm": 0.2255311976493184,
+      "learning_rate": 3.5910589959692194e-06,
+      "loss": 0.3478,
+      "step": 2837
+    },
+    {
+      "epoch": 2.807121661721068,
+      "grad_norm": 0.24237047406156542,
+      "learning_rate": 3.5727372663979483e-06,
+      "loss": 0.3575,
+      "step": 2838
+    },
+    {
+      "epoch": 2.80811078140455,
+      "grad_norm": 0.21059337790432422,
+      "learning_rate": 3.554415536826677e-06,
+      "loss": 0.3284,
+      "step": 2839
+    },
+    {
+      "epoch": 2.8090999010880315,
+      "grad_norm": 0.20503847170160883,
+      "learning_rate": 3.536093807255405e-06,
+      "loss": 0.3233,
+      "step": 2840
+    },
+    {
+      "epoch": 2.8100890207715135,
+      "grad_norm": 0.21119961563953477,
+      "learning_rate": 3.517772077684134e-06,
+      "loss": 0.3272,
+      "step": 2841
+    },
+    {
+      "epoch": 2.811078140454995,
+      "grad_norm": 0.19978901112842018,
+      "learning_rate": 3.499450348112862e-06,
+      "loss": 0.3026,
+      "step": 2842
+    },
+    {
+      "epoch": 2.812067260138477,
+      "grad_norm": 0.21915361223504046,
+      "learning_rate": 3.48112861854159e-06,
+      "loss": 0.3428,
+      "step": 2843
+    },
+    {
+      "epoch": 2.8130563798219583,
+      "grad_norm": 0.2235259457896506,
+      "learning_rate": 3.462806888970319e-06,
+      "loss": 0.3043,
+      "step": 2844
+    },
+    {
+      "epoch": 2.8140454995054403,
+      "grad_norm": 0.2178345130349941,
+      "learning_rate": 3.4444851593990477e-06,
+      "loss": 0.3428,
+      "step": 2845
+    },
+    {
+      "epoch": 2.8150346191889217,
+      "grad_norm": 0.21372597462198137,
+      "learning_rate": 3.4261634298277758e-06,
+      "loss": 0.2926,
+      "step": 2846
+    },
+    {
+      "epoch": 2.8160237388724036,
+      "grad_norm": 0.22720651208550413,
+      "learning_rate": 3.4078417002565047e-06,
+      "loss": 0.3482,
+      "step": 2847
+    },
+    {
+      "epoch": 2.817012858555885,
+      "grad_norm": 0.22006740899272062,
+      "learning_rate": 3.389519970685233e-06,
+      "loss": 0.3427,
+      "step": 2848
+    },
+    {
+      "epoch": 2.818001978239367,
+      "grad_norm": 0.2159529003979577,
+      "learning_rate": 3.371198241113961e-06,
+      "loss": 0.3423,
+      "step": 2849
+    },
+    {
+      "epoch": 2.8189910979228485,
+      "grad_norm": 0.20214518327063988,
+      "learning_rate": 3.35287651154269e-06,
+      "loss": 0.3143,
+      "step": 2850
+    },
+    {
+      "epoch": 2.8199802176063304,
+      "grad_norm": 0.21793790452801143,
+      "learning_rate": 3.3345547819714185e-06,
+      "loss": 0.3113,
+      "step": 2851
+    },
+    {
+      "epoch": 2.820969337289812,
+      "grad_norm": 0.20335863728522915,
+      "learning_rate": 3.3162330524001466e-06,
+      "loss": 0.3059,
+      "step": 2852
+    },
+    {
+      "epoch": 2.821958456973294,
+      "grad_norm": 0.20270595996542862,
+      "learning_rate": 3.2979113228288755e-06,
+      "loss": 0.3272,
+      "step": 2853
+    },
+    {
+      "epoch": 2.8229475766567753,
+      "grad_norm": 1.1077539805296455,
+      "learning_rate": 3.2795895932576036e-06,
+      "loss": 0.3856,
+      "step": 2854
+    },
+    {
+      "epoch": 2.8239366963402572,
+      "grad_norm": 0.24132309930564733,
+      "learning_rate": 3.261267863686332e-06,
+      "loss": 0.3957,
+      "step": 2855
+    },
+    {
+      "epoch": 2.8249258160237387,
+      "grad_norm": 0.2275466649899499,
+      "learning_rate": 3.2429461341150607e-06,
+      "loss": 0.3769,
+      "step": 2856
+    },
+    {
+      "epoch": 2.8259149357072206,
+      "grad_norm": 0.25008670282611023,
+      "learning_rate": 3.2246244045437892e-06,
+      "loss": 0.35,
+      "step": 2857
+    },
+    {
+      "epoch": 2.826904055390702,
+      "grad_norm": 0.22773541395218408,
+      "learning_rate": 3.2063026749725174e-06,
+      "loss": 0.3784,
+      "step": 2858
+    },
+    {
+      "epoch": 2.827893175074184,
+      "grad_norm": 0.20609422624498253,
+      "learning_rate": 3.1879809454012463e-06,
+      "loss": 0.3239,
+      "step": 2859
+    },
+    {
+      "epoch": 2.8288822947576655,
+      "grad_norm": 0.20460207000166866,
+      "learning_rate": 3.1696592158299744e-06,
+      "loss": 0.3215,
+      "step": 2860
+    },
+    {
+      "epoch": 2.8298714144411474,
+      "grad_norm": 0.23129338204484665,
+      "learning_rate": 3.151337486258703e-06,
+      "loss": 0.3938,
+      "step": 2861
+    },
+    {
+      "epoch": 2.8308605341246293,
+      "grad_norm": 0.22362717643248406,
+      "learning_rate": 3.133015756687432e-06,
+      "loss": 0.3617,
+      "step": 2862
+    },
+    {
+      "epoch": 2.831849653808111,
+      "grad_norm": 0.21117103548385116,
+      "learning_rate": 3.11469402711616e-06,
+      "loss": 0.3237,
+      "step": 2863
+    },
+    {
+      "epoch": 2.8328387734915923,
+      "grad_norm": 0.20671124421870285,
+      "learning_rate": 3.0963722975448886e-06,
+      "loss": 0.3112,
+      "step": 2864
+    },
+    {
+      "epoch": 2.833827893175074,
+      "grad_norm": 0.22217972694620672,
+      "learning_rate": 3.0780505679736167e-06,
+      "loss": 0.3634,
+      "step": 2865
+    },
+    {
+      "epoch": 2.834817012858556,
+      "grad_norm": 0.2123227501555754,
+      "learning_rate": 3.0597288384023452e-06,
+      "loss": 0.3181,
+      "step": 2866
+    },
+    {
+      "epoch": 2.8358061325420376,
+      "grad_norm": 0.20096780688164737,
+      "learning_rate": 3.0414071088310737e-06,
+      "loss": 0.3491,
+      "step": 2867
+    },
+    {
+      "epoch": 2.836795252225519,
+      "grad_norm": 0.21924673434018688,
+      "learning_rate": 3.0230853792598023e-06,
+      "loss": 0.3443,
+      "step": 2868
+    },
+    {
+      "epoch": 2.837784371909001,
+      "grad_norm": 0.2286827845733992,
+      "learning_rate": 3.004763649688531e-06,
+      "loss": 0.3668,
+      "step": 2869
+    },
+    {
+      "epoch": 2.838773491592483,
+      "grad_norm": 0.21092705025428599,
+      "learning_rate": 2.9864419201172594e-06,
+      "loss": 0.338,
+      "step": 2870
+    },
+    {
+      "epoch": 2.8397626112759644,
+      "grad_norm": 0.2148124154519086,
+      "learning_rate": 2.9681201905459875e-06,
+      "loss": 0.3545,
+      "step": 2871
+    },
+    {
+      "epoch": 2.840751730959446,
+      "grad_norm": 0.2133227511589604,
+      "learning_rate": 2.949798460974716e-06,
+      "loss": 0.3347,
+      "step": 2872
+    },
+    {
+      "epoch": 2.841740850642928,
+      "grad_norm": 0.2208561637151365,
+      "learning_rate": 2.931476731403445e-06,
+      "loss": 0.3511,
+      "step": 2873
+    },
+    {
+      "epoch": 2.8427299703264097,
+      "grad_norm": 0.2211175652037006,
+      "learning_rate": 2.913155001832173e-06,
+      "loss": 0.3802,
+      "step": 2874
+    },
+    {
+      "epoch": 2.843719090009891,
+      "grad_norm": 0.22087087502836644,
+      "learning_rate": 2.8948332722609016e-06,
+      "loss": 0.3668,
+      "step": 2875
+    },
+    {
+      "epoch": 2.8447082096933727,
+      "grad_norm": 0.22955939394628602,
+      "learning_rate": 2.87651154268963e-06,
+      "loss": 0.3035,
+      "step": 2876
+    },
+    {
+      "epoch": 2.8456973293768546,
+      "grad_norm": 0.20911592349861238,
+      "learning_rate": 2.8581898131183583e-06,
+      "loss": 0.3487,
+      "step": 2877
+    },
+    {
+      "epoch": 2.8466864490603365,
+      "grad_norm": 0.20740229129970458,
+      "learning_rate": 2.839868083547087e-06,
+      "loss": 0.2983,
+      "step": 2878
+    },
+    {
+      "epoch": 2.847675568743818,
+      "grad_norm": 0.1982895805702466,
+      "learning_rate": 2.8215463539758157e-06,
+      "loss": 0.29,
+      "step": 2879
+    },
+    {
+      "epoch": 2.8486646884272995,
+      "grad_norm": 0.21506799683461428,
+      "learning_rate": 2.803224624404544e-06,
+      "loss": 0.3596,
+      "step": 2880
+    },
+    {
+      "epoch": 2.8496538081107814,
+      "grad_norm": 0.21618734491043085,
+      "learning_rate": 2.7849028948332724e-06,
+      "loss": 0.3111,
+      "step": 2881
+    },
+    {
+      "epoch": 2.8506429277942633,
+      "grad_norm": 0.2123248042207484,
+      "learning_rate": 2.766581165262001e-06,
+      "loss": 0.3545,
+      "step": 2882
+    },
+    {
+      "epoch": 2.851632047477745,
+      "grad_norm": 0.22214805828542464,
+      "learning_rate": 2.748259435690729e-06,
+      "loss": 0.358,
+      "step": 2883
+    },
+    {
+      "epoch": 2.8526211671612263,
+      "grad_norm": 0.2004977051100376,
+      "learning_rate": 2.729937706119458e-06,
+      "loss": 0.3267,
+      "step": 2884
+    },
+    {
+      "epoch": 2.853610286844708,
+      "grad_norm": 0.21295550797121635,
+      "learning_rate": 2.7116159765481865e-06,
+      "loss": 0.3417,
+      "step": 2885
+    },
+    {
+      "epoch": 2.85459940652819,
+      "grad_norm": 0.2103516015542619,
+      "learning_rate": 2.6932942469769146e-06,
+      "loss": 0.3276,
+      "step": 2886
+    },
+    {
+      "epoch": 2.8555885262116716,
+      "grad_norm": 0.20954170278454848,
+      "learning_rate": 2.674972517405643e-06,
+      "loss": 0.3189,
+      "step": 2887
+    },
+    {
+      "epoch": 2.856577645895153,
+      "grad_norm": 0.20360676369415176,
+      "learning_rate": 2.6566507878343717e-06,
+      "loss": 0.319,
+      "step": 2888
+    },
+    {
+      "epoch": 2.857566765578635,
+      "grad_norm": 0.21277689756209228,
+      "learning_rate": 2.6383290582631003e-06,
+      "loss": 0.3326,
+      "step": 2889
+    },
+    {
+      "epoch": 2.858555885262117,
+      "grad_norm": 0.21752596509030658,
+      "learning_rate": 2.6200073286918288e-06,
+      "loss": 0.3731,
+      "step": 2890
+    },
+    {
+      "epoch": 2.8595450049455984,
+      "grad_norm": 0.2138648178065751,
+      "learning_rate": 2.6016855991205573e-06,
+      "loss": 0.3556,
+      "step": 2891
+    },
+    {
+      "epoch": 2.86053412462908,
+      "grad_norm": 0.21421289059170068,
+      "learning_rate": 2.5833638695492854e-06,
+      "loss": 0.3449,
+      "step": 2892
+    },
+    {
+      "epoch": 2.8615232443125618,
+      "grad_norm": 0.22802713240073696,
+      "learning_rate": 2.565042139978014e-06,
+      "loss": 0.339,
+      "step": 2893
+    },
+    {
+      "epoch": 2.8625123639960437,
+      "grad_norm": 0.20891785110011957,
+      "learning_rate": 2.5467204104067425e-06,
+      "loss": 0.3272,
+      "step": 2894
+    },
+    {
+      "epoch": 2.863501483679525,
+      "grad_norm": 0.2154505874086274,
+      "learning_rate": 2.528398680835471e-06,
+      "loss": 0.3493,
+      "step": 2895
+    },
+    {
+      "epoch": 2.8644906033630066,
+      "grad_norm": 0.21116137690722792,
+      "learning_rate": 2.5100769512641996e-06,
+      "loss": 0.3261,
+      "step": 2896
+    },
+    {
+      "epoch": 2.8654797230464886,
+      "grad_norm": 0.20872464139224442,
+      "learning_rate": 2.491755221692928e-06,
+      "loss": 0.3091,
+      "step": 2897
+    },
+    {
+      "epoch": 2.8664688427299705,
+      "grad_norm": 0.22172791728030142,
+      "learning_rate": 2.4734334921216562e-06,
+      "loss": 0.3674,
+      "step": 2898
+    },
+    {
+      "epoch": 2.867457962413452,
+      "grad_norm": 0.22760995786242147,
+      "learning_rate": 2.4551117625503848e-06,
+      "loss": 0.3728,
+      "step": 2899
+    },
+    {
+      "epoch": 2.868447082096934,
+      "grad_norm": 0.22014957904362617,
+      "learning_rate": 2.4367900329791133e-06,
+      "loss": 0.3356,
+      "step": 2900
+    },
+    {
+      "epoch": 2.8694362017804154,
+      "grad_norm": 0.21304917465749867,
+      "learning_rate": 2.418468303407842e-06,
+      "loss": 0.3579,
+      "step": 2901
+    },
+    {
+      "epoch": 2.8704253214638973,
+      "grad_norm": 0.19788363376546905,
+      "learning_rate": 2.4001465738365704e-06,
+      "loss": 0.3202,
+      "step": 2902
+    },
+    {
+      "epoch": 2.8714144411473788,
+      "grad_norm": 0.21988418201730414,
+      "learning_rate": 2.381824844265299e-06,
+      "loss": 0.3181,
+      "step": 2903
+    },
+    {
+      "epoch": 2.8724035608308607,
+      "grad_norm": 0.20738796103864413,
+      "learning_rate": 2.363503114694027e-06,
+      "loss": 0.3477,
+      "step": 2904
+    },
+    {
+      "epoch": 2.873392680514342,
+      "grad_norm": 0.21762364979569407,
+      "learning_rate": 2.3451813851227555e-06,
+      "loss": 0.3439,
+      "step": 2905
+    },
+    {
+      "epoch": 2.874381800197824,
+      "grad_norm": 0.2143578729116185,
+      "learning_rate": 2.326859655551484e-06,
+      "loss": 0.3348,
+      "step": 2906
+    },
+    {
+      "epoch": 2.8753709198813056,
+      "grad_norm": 0.232885047628153,
+      "learning_rate": 2.3085379259802126e-06,
+      "loss": 0.3748,
+      "step": 2907
+    },
+    {
+      "epoch": 2.8763600395647875,
+      "grad_norm": 0.21077048066014784,
+      "learning_rate": 2.290216196408941e-06,
+      "loss": 0.3429,
+      "step": 2908
+    },
+    {
+      "epoch": 2.877349159248269,
+      "grad_norm": 0.21001095217110033,
+      "learning_rate": 2.2718944668376697e-06,
+      "loss": 0.3032,
+      "step": 2909
+    },
+    {
+      "epoch": 2.878338278931751,
+      "grad_norm": 0.2166268520185779,
+      "learning_rate": 2.253572737266398e-06,
+      "loss": 0.3875,
+      "step": 2910
+    },
+    {
+      "epoch": 2.8793273986152323,
+      "grad_norm": 0.21898528679644236,
+      "learning_rate": 2.2352510076951263e-06,
+      "loss": 0.3256,
+      "step": 2911
+    },
+    {
+      "epoch": 2.8803165182987143,
+      "grad_norm": 0.22735390510825595,
+      "learning_rate": 2.216929278123855e-06,
+      "loss": 0.4083,
+      "step": 2912
+    },
+    {
+      "epoch": 2.8813056379821957,
+      "grad_norm": 0.23062531916202664,
+      "learning_rate": 2.1986075485525834e-06,
+      "loss": 0.372,
+      "step": 2913
+    },
+    {
+      "epoch": 2.8822947576656777,
+      "grad_norm": 0.21477846290717173,
+      "learning_rate": 2.180285818981312e-06,
+      "loss": 0.3271,
+      "step": 2914
+    },
+    {
+      "epoch": 2.883283877349159,
+      "grad_norm": 0.2117459683443643,
+      "learning_rate": 2.1619640894100405e-06,
+      "loss": 0.3577,
+      "step": 2915
+    },
+    {
+      "epoch": 2.884272997032641,
+      "grad_norm": 0.21937476541837353,
+      "learning_rate": 2.1436423598387686e-06,
+      "loss": 0.3423,
+      "step": 2916
+    },
+    {
+      "epoch": 2.8852621167161225,
+      "grad_norm": 0.1976009681830841,
+      "learning_rate": 2.125320630267497e-06,
+      "loss": 0.3253,
+      "step": 2917
+    },
+    {
+      "epoch": 2.8862512363996045,
+      "grad_norm": 0.21302952806939138,
+      "learning_rate": 2.106998900696226e-06,
+      "loss": 0.3305,
+      "step": 2918
+    },
+    {
+      "epoch": 2.887240356083086,
+      "grad_norm": 0.206815293905744,
+      "learning_rate": 2.088677171124954e-06,
+      "loss": 0.3496,
+      "step": 2919
+    },
+    {
+      "epoch": 2.888229475766568,
+      "grad_norm": 0.1995894143318209,
+      "learning_rate": 2.0703554415536827e-06,
+      "loss": 0.3112,
+      "step": 2920
+    },
+    {
+      "epoch": 2.8892185954500493,
+      "grad_norm": 0.19955543207743653,
+      "learning_rate": 2.0520337119824113e-06,
+      "loss": 0.3113,
+      "step": 2921
+    },
+    {
+      "epoch": 2.8902077151335313,
+      "grad_norm": 0.2158562754144166,
+      "learning_rate": 2.0337119824111394e-06,
+      "loss": 0.3698,
+      "step": 2922
+    },
+    {
+      "epoch": 2.8911968348170127,
+      "grad_norm": 0.20864063973766814,
+      "learning_rate": 2.0153902528398683e-06,
+      "loss": 0.3441,
+      "step": 2923
+    },
+    {
+      "epoch": 2.8921859545004946,
+      "grad_norm": 0.2119247973375731,
+      "learning_rate": 1.997068523268597e-06,
+      "loss": 0.3437,
+      "step": 2924
+    },
+    {
+      "epoch": 2.893175074183976,
+      "grad_norm": 0.2232073700071952,
+      "learning_rate": 1.978746793697325e-06,
+      "loss": 0.3929,
+      "step": 2925
+    },
+    {
+      "epoch": 2.894164193867458,
+      "grad_norm": 0.20286778897421978,
+      "learning_rate": 1.9604250641260535e-06,
+      "loss": 0.3237,
+      "step": 2926
+    },
+    {
+      "epoch": 2.8951533135509395,
+      "grad_norm": 0.2092640542037701,
+      "learning_rate": 1.942103334554782e-06,
+      "loss": 0.2943,
+      "step": 2927
+    },
+    {
+      "epoch": 2.8961424332344214,
+      "grad_norm": 0.2088218793092618,
+      "learning_rate": 1.9237816049835106e-06,
+      "loss": 0.3332,
+      "step": 2928
+    },
+    {
+      "epoch": 2.897131552917903,
+      "grad_norm": 0.20119421841154572,
+      "learning_rate": 1.905459875412239e-06,
+      "loss": 0.319,
+      "step": 2929
+    },
+    {
+      "epoch": 2.898120672601385,
+      "grad_norm": 0.20219374342024857,
+      "learning_rate": 1.8871381458409674e-06,
+      "loss": 0.3187,
+      "step": 2930
+    },
+    {
+      "epoch": 2.8991097922848663,
+      "grad_norm": 0.19777287722260406,
+      "learning_rate": 1.8688164162696962e-06,
+      "loss": 0.3202,
+      "step": 2931
+    },
+    {
+      "epoch": 2.9000989119683482,
+      "grad_norm": 0.20427420158126633,
+      "learning_rate": 1.8504946866984243e-06,
+      "loss": 0.322,
+      "step": 2932
+    },
+    {
+      "epoch": 2.9010880316518297,
+      "grad_norm": 0.21551527284059774,
+      "learning_rate": 1.8321729571271528e-06,
+      "loss": 0.3069,
+      "step": 2933
+    },
+    {
+      "epoch": 2.9020771513353116,
+      "grad_norm": 0.199753848958888,
+      "learning_rate": 1.8138512275558816e-06,
+      "loss": 0.291,
+      "step": 2934
+    },
+    {
+      "epoch": 2.903066271018793,
+      "grad_norm": 0.2313478126214845,
+      "learning_rate": 1.7955294979846097e-06,
+      "loss": 0.368,
+      "step": 2935
+    },
+    {
+      "epoch": 2.904055390702275,
+      "grad_norm": 0.20699383705522323,
+      "learning_rate": 1.7772077684133384e-06,
+      "loss": 0.3402,
+      "step": 2936
+    },
+    {
+      "epoch": 2.905044510385757,
+      "grad_norm": 0.2187505761089255,
+      "learning_rate": 1.758886038842067e-06,
+      "loss": 0.3623,
+      "step": 2937
+    },
+    {
+      "epoch": 2.9060336300692384,
+      "grad_norm": 0.19345606798141826,
+      "learning_rate": 1.740564309270795e-06,
+      "loss": 0.3104,
+      "step": 2938
+    },
+    {
+      "epoch": 2.90702274975272,
+      "grad_norm": 0.21864657860270537,
+      "learning_rate": 1.7222425796995238e-06,
+      "loss": 0.3415,
+      "step": 2939
+    },
+    {
+      "epoch": 2.908011869436202,
+      "grad_norm": 0.19683732715540603,
+      "learning_rate": 1.7039208501282524e-06,
+      "loss": 0.2874,
+      "step": 2940
+    },
+    {
+      "epoch": 2.9090009891196837,
+      "grad_norm": 0.21266519134677597,
+      "learning_rate": 1.6855991205569805e-06,
+      "loss": 0.3586,
+      "step": 2941
+    },
+    {
+      "epoch": 2.909990108803165,
+      "grad_norm": 0.21398924276198206,
+      "learning_rate": 1.6672773909857092e-06,
+      "loss": 0.3229,
+      "step": 2942
+    },
+    {
+      "epoch": 2.9109792284866467,
+      "grad_norm": 0.22495179099039847,
+      "learning_rate": 1.6489556614144378e-06,
+      "loss": 0.3834,
+      "step": 2943
+    },
+    {
+      "epoch": 2.9119683481701286,
+      "grad_norm": 0.2024056953522565,
+      "learning_rate": 1.630633931843166e-06,
+      "loss": 0.3348,
+      "step": 2944
+    },
+    {
+      "epoch": 2.9129574678536105,
+      "grad_norm": 0.21753481027936725,
+      "learning_rate": 1.6123122022718946e-06,
+      "loss": 0.3871,
+      "step": 2945
+    },
+    {
+      "epoch": 2.913946587537092,
+      "grad_norm": 0.23414035730320035,
+      "learning_rate": 1.5939904727006232e-06,
+      "loss": 0.403,
+      "step": 2946
+    },
+    {
+      "epoch": 2.9149357072205735,
+      "grad_norm": 0.20610158192122174,
+      "learning_rate": 1.5756687431293515e-06,
+      "loss": 0.3576,
+      "step": 2947
+    },
+    {
+      "epoch": 2.9159248269040554,
+      "grad_norm": 0.2264080372941922,
+      "learning_rate": 1.55734701355808e-06,
+      "loss": 0.3619,
+      "step": 2948
+    },
+    {
+      "epoch": 2.9169139465875373,
+      "grad_norm": 0.20401730729882686,
+      "learning_rate": 1.5390252839868083e-06,
+      "loss": 0.3416,
+      "step": 2949
+    },
+    {
+      "epoch": 2.917903066271019,
+      "grad_norm": 0.20630091994141,
+      "learning_rate": 1.5207035544155369e-06,
+      "loss": 0.3122,
+      "step": 2950
+    },
+    {
+      "epoch": 2.9188921859545003,
+      "grad_norm": 0.2284728589041083,
+      "learning_rate": 1.5023818248442654e-06,
+      "loss": 0.3559,
+      "step": 2951
+    },
+    {
+      "epoch": 2.919881305637982,
+      "grad_norm": 0.21523792128505556,
+      "learning_rate": 1.4840600952729937e-06,
+      "loss": 0.3322,
+      "step": 2952
+    },
+    {
+      "epoch": 2.920870425321464,
+      "grad_norm": 0.21968342654749495,
+      "learning_rate": 1.4657383657017225e-06,
+      "loss": 0.3689,
+      "step": 2953
+    },
+    {
+      "epoch": 2.9218595450049456,
+      "grad_norm": 0.22690637293425686,
+      "learning_rate": 1.4474166361304508e-06,
+      "loss": 0.3335,
+      "step": 2954
+    },
+    {
+      "epoch": 2.922848664688427,
+      "grad_norm": 0.23072015845383623,
+      "learning_rate": 1.4290949065591791e-06,
+      "loss": 0.383,
+      "step": 2955
+    },
+    {
+      "epoch": 2.923837784371909,
+      "grad_norm": 0.2227436812072427,
+      "learning_rate": 1.4107731769879079e-06,
+      "loss": 0.3514,
+      "step": 2956
+    },
+    {
+      "epoch": 2.924826904055391,
+      "grad_norm": 0.23110200374260195,
+      "learning_rate": 1.3924514474166362e-06,
+      "loss": 0.4092,
+      "step": 2957
+    },
+    {
+      "epoch": 2.9258160237388724,
+      "grad_norm": 0.20321476984964387,
+      "learning_rate": 1.3741297178453645e-06,
+      "loss": 0.3342,
+      "step": 2958
+    },
+    {
+      "epoch": 2.926805143422354,
+      "grad_norm": 0.18821044499165784,
+      "learning_rate": 1.3558079882740933e-06,
+      "loss": 0.2944,
+      "step": 2959
+    },
+    {
+      "epoch": 2.927794263105836,
+      "grad_norm": 0.2133703549279981,
+      "learning_rate": 1.3374862587028216e-06,
+      "loss": 0.3528,
+      "step": 2960
+    },
+    {
+      "epoch": 2.9287833827893177,
+      "grad_norm": 0.19949471309765202,
+      "learning_rate": 1.3191645291315501e-06,
+      "loss": 0.3012,
+      "step": 2961
+    },
+    {
+      "epoch": 2.929772502472799,
+      "grad_norm": 0.20980980030921756,
+      "learning_rate": 1.3008427995602787e-06,
+      "loss": 0.3459,
+      "step": 2962
+    },
+    {
+      "epoch": 2.9307616221562807,
+      "grad_norm": 0.20593464559520894,
+      "learning_rate": 1.282521069989007e-06,
+      "loss": 0.3328,
+      "step": 2963
+    },
+    {
+      "epoch": 2.9317507418397626,
+      "grad_norm": 0.20716599210238562,
+      "learning_rate": 1.2641993404177355e-06,
+      "loss": 0.3473,
+      "step": 2964
+    },
+    {
+      "epoch": 2.9327398615232445,
+      "grad_norm": 0.22715584225040256,
+      "learning_rate": 1.245877610846464e-06,
+      "loss": 0.3487,
+      "step": 2965
+    },
+    {
+      "epoch": 2.933728981206726,
+      "grad_norm": 0.19923436499191036,
+      "learning_rate": 1.2275558812751924e-06,
+      "loss": 0.3245,
+      "step": 2966
+    },
+    {
+      "epoch": 2.9347181008902075,
+      "grad_norm": 0.21117769437499914,
+      "learning_rate": 1.209234151703921e-06,
+      "loss": 0.3417,
+      "step": 2967
+    },
+    {
+      "epoch": 2.9357072205736894,
+      "grad_norm": 0.2198013877108107,
+      "learning_rate": 1.1909124221326494e-06,
+      "loss": 0.3786,
+      "step": 2968
+    },
+    {
+      "epoch": 2.9366963402571713,
+      "grad_norm": 0.20015001058595291,
+      "learning_rate": 1.1725906925613778e-06,
+      "loss": 0.3256,
+      "step": 2969
+    },
+    {
+      "epoch": 2.9376854599406528,
+      "grad_norm": 0.21276982029063424,
+      "learning_rate": 1.1542689629901063e-06,
+      "loss": 0.3417,
+      "step": 2970
+    },
+    {
+      "epoch": 2.9386745796241343,
+      "grad_norm": 0.22527713295979443,
+      "learning_rate": 1.1359472334188348e-06,
+      "loss": 0.3625,
+      "step": 2971
+    },
+    {
+      "epoch": 2.939663699307616,
+      "grad_norm": 0.201432544398078,
+      "learning_rate": 1.1176255038475632e-06,
+      "loss": 0.3073,
+      "step": 2972
+    },
+    {
+      "epoch": 2.940652818991098,
+      "grad_norm": 7.943101719255052,
+      "learning_rate": 1.0993037742762917e-06,
+      "loss": 0.8247,
+      "step": 2973
+    },
+    {
+      "epoch": 2.9416419386745796,
+      "grad_norm": 0.21340135227018642,
+      "learning_rate": 1.0809820447050202e-06,
+      "loss": 0.3784,
+      "step": 2974
+    },
+    {
+      "epoch": 2.9426310583580615,
+      "grad_norm": 0.2113735158072839,
+      "learning_rate": 1.0626603151337486e-06,
+      "loss": 0.3574,
+      "step": 2975
+    },
+    {
+      "epoch": 2.943620178041543,
+      "grad_norm": 0.19609645027272843,
+      "learning_rate": 1.044338585562477e-06,
+      "loss": 0.3304,
+      "step": 2976
+    },
+    {
+      "epoch": 2.944609297725025,
+      "grad_norm": 0.21099389986750994,
+      "learning_rate": 1.0260168559912056e-06,
+      "loss": 0.346,
+      "step": 2977
+    },
+    {
+      "epoch": 2.9455984174085064,
+      "grad_norm": 0.21796864077492323,
+      "learning_rate": 1.0076951264199342e-06,
+      "loss": 0.3519,
+      "step": 2978
+    },
+    {
+      "epoch": 2.9465875370919883,
+      "grad_norm": 0.2382189042915159,
+      "learning_rate": 9.893733968486625e-07,
+      "loss": 0.3673,
+      "step": 2979
+    },
+    {
+      "epoch": 2.9475766567754698,
+      "grad_norm": 0.20419821656474832,
+      "learning_rate": 9.71051667277391e-07,
+      "loss": 0.3464,
+      "step": 2980
+    },
+    {
+      "epoch": 2.9485657764589517,
+      "grad_norm": 0.22395363781798322,
+      "learning_rate": 9.527299377061195e-07,
+      "loss": 0.3279,
+      "step": 2981
+    },
+    {
+      "epoch": 2.949554896142433,
+      "grad_norm": 0.20253721100448752,
+      "learning_rate": 9.344082081348481e-07,
+      "loss": 0.3368,
+      "step": 2982
+    },
+    {
+      "epoch": 2.950544015825915,
+      "grad_norm": 0.2022713938179965,
+      "learning_rate": 9.160864785635764e-07,
+      "loss": 0.3311,
+      "step": 2983
+    },
+    {
+      "epoch": 2.9515331355093966,
+      "grad_norm": 0.20595080269509453,
+      "learning_rate": 8.977647489923048e-07,
+      "loss": 0.352,
+      "step": 2984
+    },
+    {
+      "epoch": 2.9525222551928785,
+      "grad_norm": 0.1987103967446778,
+      "learning_rate": 8.794430194210335e-07,
+      "loss": 0.3465,
+      "step": 2985
+    },
+    {
+      "epoch": 2.95351137487636,
+      "grad_norm": 0.2057786134643462,
+      "learning_rate": 8.611212898497619e-07,
+      "loss": 0.3466,
+      "step": 2986
+    },
+    {
+      "epoch": 2.954500494559842,
+      "grad_norm": 0.21967452126576467,
+      "learning_rate": 8.427995602784902e-07,
+      "loss": 0.3539,
+      "step": 2987
+    },
+    {
+      "epoch": 2.9554896142433233,
+      "grad_norm": 0.22985902892440865,
+      "learning_rate": 8.244778307072189e-07,
+      "loss": 0.3911,
+      "step": 2988
+    },
+    {
+      "epoch": 2.9564787339268053,
+      "grad_norm": 0.2201316492538097,
+      "learning_rate": 8.061561011359473e-07,
+      "loss": 0.3493,
+      "step": 2989
+    },
+    {
+      "epoch": 2.9574678536102867,
+      "grad_norm": 0.19954226377424544,
+      "learning_rate": 7.878343715646757e-07,
+      "loss": 0.3463,
+      "step": 2990
+    },
+    {
+      "epoch": 2.9584569732937687,
+      "grad_norm": 0.20222148981802693,
+      "learning_rate": 7.695126419934042e-07,
+      "loss": 0.3135,
+      "step": 2991
+    },
+    {
+      "epoch": 2.95944609297725,
+      "grad_norm": 0.21871651088863459,
+      "learning_rate": 7.511909124221327e-07,
+      "loss": 0.3676,
+      "step": 2992
+    },
+    {
+      "epoch": 2.960435212660732,
+      "grad_norm": 0.21927670774593283,
+      "learning_rate": 7.328691828508612e-07,
+      "loss": 0.3351,
+      "step": 2993
+    },
+    {
+      "epoch": 2.9614243323442135,
+      "grad_norm": 0.208780563559218,
+      "learning_rate": 7.145474532795896e-07,
+      "loss": 0.3447,
+      "step": 2994
+    },
+    {
+      "epoch": 2.9624134520276955,
+      "grad_norm": 0.2126920301550537,
+      "learning_rate": 6.962257237083181e-07,
+      "loss": 0.3608,
+      "step": 2995
+    },
+    {
+      "epoch": 2.963402571711177,
+      "grad_norm": 0.20548958069933887,
+      "learning_rate": 6.779039941370466e-07,
+      "loss": 0.3304,
+      "step": 2996
+    },
+    {
+      "epoch": 2.964391691394659,
+      "grad_norm": 0.2036679215154334,
+      "learning_rate": 6.595822645657751e-07,
+      "loss": 0.3463,
+      "step": 2997
+    },
+    {
+      "epoch": 2.9653808110781403,
+      "grad_norm": 0.21172321613940526,
+      "learning_rate": 6.412605349945035e-07,
+      "loss": 0.3599,
+      "step": 2998
+    },
+    {
+      "epoch": 2.9663699307616223,
+      "grad_norm": 0.21160961055287664,
+      "learning_rate": 6.22938805423232e-07,
+      "loss": 0.3891,
+      "step": 2999
+    },
+    {
+      "epoch": 2.9673590504451037,
+      "grad_norm": 0.21327213001738943,
+      "learning_rate": 6.046170758519605e-07,
+      "loss": 0.3691,
+      "step": 3000
+    },
+    {
+      "epoch": 2.9683481701285857,
+      "grad_norm": 0.21547058365602503,
+      "learning_rate": 5.862953462806889e-07,
+      "loss": 0.3385,
+      "step": 3001
+    },
+    {
+      "epoch": 2.969337289812067,
+      "grad_norm": 0.20224924064139008,
+      "learning_rate": 5.679736167094174e-07,
+      "loss": 0.3285,
+      "step": 3002
+    },
+    {
+      "epoch": 2.970326409495549,
+      "grad_norm": 0.219988718165209,
+      "learning_rate": 5.496518871381459e-07,
+      "loss": 0.3431,
+      "step": 3003
+    },
+    {
+      "epoch": 2.9713155291790305,
+      "grad_norm": 0.22283985943183246,
+      "learning_rate": 5.313301575668743e-07,
+      "loss": 0.3386,
+      "step": 3004
+    },
+    {
+      "epoch": 2.9723046488625124,
+      "grad_norm": 0.20042527762749443,
+      "learning_rate": 5.130084279956028e-07,
+      "loss": 0.3339,
+      "step": 3005
+    },
+    {
+      "epoch": 2.973293768545994,
+      "grad_norm": 0.9140786441671352,
+      "learning_rate": 4.946866984243312e-07,
+      "loss": 0.3563,
+      "step": 3006
+    },
+    {
+      "epoch": 2.974282888229476,
+      "grad_norm": 0.20417408391682282,
+      "learning_rate": 4.763649688530597e-07,
+      "loss": 0.3401,
+      "step": 3007
+    },
+    {
+      "epoch": 2.9752720079129573,
+      "grad_norm": 0.20602866956947444,
+      "learning_rate": 4.580432392817882e-07,
+      "loss": 0.3247,
+      "step": 3008
+    },
+    {
+      "epoch": 2.9762611275964392,
+      "grad_norm": 0.21276857507850871,
+      "learning_rate": 4.3972150971051674e-07,
+      "loss": 0.3594,
+      "step": 3009
+    },
+    {
+      "epoch": 2.9772502472799207,
+      "grad_norm": 0.19943196340758107,
+      "learning_rate": 4.213997801392451e-07,
+      "loss": 0.3296,
+      "step": 3010
+    },
+    {
+      "epoch": 2.9782393669634026,
+      "grad_norm": 0.21268206131905573,
+      "learning_rate": 4.0307805056797366e-07,
+      "loss": 0.3423,
+      "step": 3011
+    },
+    {
+      "epoch": 2.979228486646884,
+      "grad_norm": 0.2057146467485511,
+      "learning_rate": 3.847563209967021e-07,
+      "loss": 0.3186,
+      "step": 3012
+    },
+    {
+      "epoch": 2.980217606330366,
+      "grad_norm": 0.20744284251947426,
+      "learning_rate": 3.664345914254306e-07,
+      "loss": 0.3208,
+      "step": 3013
+    },
+    {
+      "epoch": 2.9812067260138475,
+      "grad_norm": 0.20532924838610409,
+      "learning_rate": 3.4811286185415905e-07,
+      "loss": 0.3296,
+      "step": 3014
+    },
+    {
+      "epoch": 2.9821958456973294,
+      "grad_norm": 0.2033415582326357,
+      "learning_rate": 3.2979113228288753e-07,
+      "loss": 0.3257,
+      "step": 3015
+    },
+    {
+      "epoch": 2.9831849653808113,
+      "grad_norm": 0.20754506694784877,
+      "learning_rate": 3.11469402711616e-07,
+      "loss": 0.3296,
+      "step": 3016
+    },
+    {
+      "epoch": 2.984174085064293,
+      "grad_norm": 0.20629067666131826,
+      "learning_rate": 2.9314767314034444e-07,
+      "loss": 0.3189,
+      "step": 3017
+    },
+    {
+      "epoch": 2.9851632047477743,
+      "grad_norm": 0.22938650262981103,
+      "learning_rate": 2.748259435690729e-07,
+      "loss": 0.3406,
+      "step": 3018
+    },
+    {
+      "epoch": 2.986152324431256,
+      "grad_norm": 0.20636895308921352,
+      "learning_rate": 2.565042139978014e-07,
+      "loss": 0.3344,
+      "step": 3019
+    },
+    {
+      "epoch": 2.987141444114738,
+      "grad_norm": 0.20357483574027163,
+      "learning_rate": 2.3818248442652986e-07,
+      "loss": 0.3484,
+      "step": 3020
+    },
+    {
+      "epoch": 2.9881305637982196,
+      "grad_norm": 0.214430972642966,
+      "learning_rate": 2.1986075485525837e-07,
+      "loss": 0.3679,
+      "step": 3021
+    },
+    {
+      "epoch": 2.989119683481701,
+      "grad_norm": 0.19589997688875407,
+      "learning_rate": 2.0153902528398683e-07,
+      "loss": 0.3265,
+      "step": 3022
+    },
+    {
+      "epoch": 2.990108803165183,
+      "grad_norm": 0.23345971886153838,
+      "learning_rate": 1.832172957127153e-07,
+      "loss": 0.3985,
+      "step": 3023
+    },
+    {
+      "epoch": 2.991097922848665,
+      "grad_norm": 0.1981503597152391,
+      "learning_rate": 1.6489556614144377e-07,
+      "loss": 0.3099,
+      "step": 3024
+    },
+    {
+      "epoch": 2.9920870425321464,
+      "grad_norm": 0.2145800019812682,
+      "learning_rate": 1.4657383657017222e-07,
+      "loss": 0.3628,
+      "step": 3025
+    },
+    {
+      "epoch": 2.993076162215628,
+      "grad_norm": 0.1993615760861862,
+      "learning_rate": 1.282521069989007e-07,
+      "loss": 0.3104,
+      "step": 3026
+    },
+    {
+      "epoch": 2.99406528189911,
+      "grad_norm": 0.22261786031016298,
+      "learning_rate": 1.0993037742762919e-07,
+      "loss": 0.373,
+      "step": 3027
+    },
+    {
+      "epoch": 2.9950544015825917,
+      "grad_norm": 0.2022644187391328,
+      "learning_rate": 9.160864785635765e-08,
+      "loss": 0.3516,
+      "step": 3028
+    },
+    {
+      "epoch": 2.996043521266073,
+      "grad_norm": 0.19548068811302774,
+      "learning_rate": 7.328691828508611e-08,
+      "loss": 0.3303,
+      "step": 3029
+    },
+    {
+      "epoch": 2.9970326409495547,
+      "grad_norm": 0.202037553129905,
+      "learning_rate": 5.496518871381459e-08,
+      "loss": 0.3579,
+      "step": 3030
+    },
+    {
+      "epoch": 2.9980217606330366,
+      "grad_norm": 0.19575234484936066,
+      "learning_rate": 3.6643459142543055e-08,
+      "loss": 0.3171,
+      "step": 3031
+    },
+    {
+      "epoch": 2.9990108803165185,
+      "grad_norm": 0.21172137281490067,
+      "learning_rate": 1.8321729571271528e-08,
+      "loss": 0.3511,
+      "step": 3032
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.20754262490189151,
+      "learning_rate": 0.0,
+      "loss": 0.3455,
+      "step": 3033
+    },
+    {
+      "epoch": 3.0,
+      "step": 3033,
+      "total_flos": 2.5830390820744724e+18,
+      "train_loss": 0.5209268033307966,
+      "train_runtime": 175746.8725,
+      "train_samples_per_second": 0.276,
+      "train_steps_per_second": 0.017
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 3033,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.5830390820744724e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}