diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5923 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9910873440285206,
+  "eval_steps": 500,
+  "global_step": 840,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0035650623885918,
+      "grad_norm": 62.25356074807307,
+      "learning_rate": 0.0,
+      "loss": 10.8811,
+      "step": 1
+    },
+    {
+      "epoch": 0.0071301247771836,
+      "grad_norm": 63.34153655032229,
+      "learning_rate": 5.952380952380952e-07,
+      "loss": 10.9628,
+      "step": 2
+    },
+    {
+      "epoch": 0.0106951871657754,
+      "grad_norm": 64.60631124464966,
+      "learning_rate": 1.1904761904761904e-06,
+      "loss": 10.8185,
+      "step": 3
+    },
+    {
+      "epoch": 0.0142602495543672,
+      "grad_norm": 64.90607827544154,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 11.0275,
+      "step": 4
+    },
+    {
+      "epoch": 0.017825311942959002,
+      "grad_norm": 64.73689735450964,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 10.8488,
+      "step": 5
+    },
+    {
+      "epoch": 0.0213903743315508,
+      "grad_norm": 73.12135364808142,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 10.6226,
+      "step": 6
+    },
+    {
+      "epoch": 0.024955436720142603,
+      "grad_norm": 110.51257804000987,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 8.9896,
+      "step": 7
+    },
+    {
+      "epoch": 0.0285204991087344,
+      "grad_norm": 127.09914155280674,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 8.3582,
+      "step": 8
+    },
+    {
+      "epoch": 0.03208556149732621,
+      "grad_norm": 56.41528000922067,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 3.1244,
+      "step": 9
+    },
+    {
+      "epoch": 0.035650623885918005,
+      "grad_norm": 45.46349768277812,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 2.6617,
+      "step": 10
+    },
+    {
+      "epoch": 0.0392156862745098,
+      "grad_norm": 29.160845033365973,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 2.0849,
+      "step": 11
+    },
+    {
+      "epoch": 0.0427807486631016,
+      "grad_norm": 6.219944229621047,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 1.253,
+      "step": 12
+    },
+    {
+      "epoch": 0.04634581105169341,
+      "grad_norm": 4.346668066966246,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 1.1396,
+      "step": 13
+    },
+    {
+      "epoch": 0.049910873440285206,
+      "grad_norm": 3.12992790659393,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 1.05,
+      "step": 14
+    },
+    {
+      "epoch": 0.053475935828877004,
+      "grad_norm": 2.311291422441743,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.9959,
+      "step": 15
+    },
+    {
+      "epoch": 0.0570409982174688,
+      "grad_norm": 1.511460028748143,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.8525,
+      "step": 16
+    },
+    {
+      "epoch": 0.06060606060606061,
+      "grad_norm": 59.26795218706408,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.8456,
+      "step": 17
+    },
+    {
+      "epoch": 0.06417112299465241,
+      "grad_norm": 10.97327574530585,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 0.8061,
+      "step": 18
+    },
+    {
+      "epoch": 0.0677361853832442,
+      "grad_norm": 1.3121248826474081,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.792,
+      "step": 19
+    },
+    {
+      "epoch": 0.07130124777183601,
+      "grad_norm": 0.9576686655782182,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 0.7412,
+      "step": 20
+    },
+    {
+      "epoch": 0.0748663101604278,
+      "grad_norm": 0.7441435239014126,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.7062,
+      "step": 21
+    },
+    {
+      "epoch": 0.0784313725490196,
+      "grad_norm": 0.7248722173091678,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7148,
+      "step": 22
+    },
+    {
+      "epoch": 0.08199643493761141,
+      "grad_norm": 0.7968291336524947,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.6616,
+      "step": 23
+    },
+    {
+      "epoch": 0.0855614973262032,
+      "grad_norm": 0.8286552983998217,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 0.6477,
+      "step": 24
+    },
+    {
+      "epoch": 0.08912655971479501,
+      "grad_norm": 0.586257217180834,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.631,
+      "step": 25
+    },
+    {
+      "epoch": 0.09269162210338681,
+      "grad_norm": 0.6224873735906401,
+      "learning_rate": 1.4880952380952381e-05,
+      "loss": 0.6233,
+      "step": 26
+    },
+    {
+      "epoch": 0.0962566844919786,
+      "grad_norm": 0.7139111744471022,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.6253,
+      "step": 27
+    },
+    {
+      "epoch": 0.09982174688057041,
+      "grad_norm": 0.5965919016200115,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.6324,
+      "step": 28
+    },
+    {
+      "epoch": 0.10338680926916222,
+      "grad_norm": 0.5508056718833139,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.6162,
+      "step": 29
+    },
+    {
+      "epoch": 0.10695187165775401,
+      "grad_norm": 0.4552042545183718,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.5746,
+      "step": 30
+    },
+    {
+      "epoch": 0.11051693404634581,
+      "grad_norm": 0.5425382552847986,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.5587,
+      "step": 31
+    },
+    {
+      "epoch": 0.1140819964349376,
+      "grad_norm": 0.449897372746649,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.5677,
+      "step": 32
+    },
+    {
+      "epoch": 0.11764705882352941,
+      "grad_norm": 0.3947664617534591,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.5439,
+      "step": 33
+    },
+    {
+      "epoch": 0.12121212121212122,
+      "grad_norm": 0.4121123285638678,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.5602,
+      "step": 34
+    },
+    {
+      "epoch": 0.12477718360071301,
+      "grad_norm": 0.43887220586326897,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.5491,
+      "step": 35
+    },
+    {
+      "epoch": 0.12834224598930483,
+      "grad_norm": 0.3873242373328585,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.5453,
+      "step": 36
+    },
+    {
+      "epoch": 0.1319073083778966,
+      "grad_norm": 0.31052871729634857,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.5195,
+      "step": 37
+    },
+    {
+      "epoch": 0.1354723707664884,
+      "grad_norm": 0.3001351627276006,
+      "learning_rate": 2.2023809523809524e-05,
+      "loss": 0.5245,
+      "step": 38
+    },
+    {
+      "epoch": 0.13903743315508021,
+      "grad_norm": 0.349423838742291,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.5224,
+      "step": 39
+    },
+    {
+      "epoch": 0.14260249554367202,
+      "grad_norm": 0.319359304088445,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.544,
+      "step": 40
+    },
+    {
+      "epoch": 0.14616755793226383,
+      "grad_norm": 0.27461777012554645,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.5149,
+      "step": 41
+    },
+    {
+      "epoch": 0.1497326203208556,
+      "grad_norm": 0.2814961230172167,
+      "learning_rate": 2.4404761904761906e-05,
+      "loss": 0.5009,
+      "step": 42
+    },
+    {
+      "epoch": 0.1532976827094474,
+      "grad_norm": 0.2842242135519138,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5079,
+      "step": 43
+    },
+    {
+      "epoch": 0.1568627450980392,
+      "grad_norm": 0.286153037622253,
+      "learning_rate": 2.5595238095238093e-05,
+      "loss": 0.5188,
+      "step": 44
+    },
+    {
+      "epoch": 0.16042780748663102,
+      "grad_norm": 0.2752904486136685,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.4963,
+      "step": 45
+    },
+    {
+      "epoch": 0.16399286987522282,
+      "grad_norm": 0.30914734868606475,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.5105,
+      "step": 46
+    },
+    {
+      "epoch": 0.16755793226381463,
+      "grad_norm": 0.2531995065894658,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.5054,
+      "step": 47
+    },
+    {
+      "epoch": 0.1711229946524064,
+      "grad_norm": 0.28821891663049387,
+      "learning_rate": 2.797619047619048e-05,
+      "loss": 0.515,
+      "step": 48
+    },
+    {
+      "epoch": 0.1746880570409982,
+      "grad_norm": 0.31652362666244127,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.4961,
+      "step": 49
+    },
+    {
+      "epoch": 0.17825311942959002,
+      "grad_norm": 0.26685098567865917,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.5029,
+      "step": 50
+    },
+    {
+      "epoch": 0.18181818181818182,
+      "grad_norm": 0.2904411460814608,
+      "learning_rate": 2.9761904761904762e-05,
+      "loss": 0.4909,
+      "step": 51
+    },
+    {
+      "epoch": 0.18538324420677363,
+      "grad_norm": 0.2771658397515614,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.4962,
+      "step": 52
+    },
+    {
+      "epoch": 0.1889483065953654,
+      "grad_norm": 0.24265179411740379,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.4995,
+      "step": 53
+    },
+    {
+      "epoch": 0.1925133689839572,
+      "grad_norm": 0.25203343737636374,
+      "learning_rate": 3.154761904761905e-05,
+      "loss": 0.4903,
+      "step": 54
+    },
+    {
+      "epoch": 0.19607843137254902,
+      "grad_norm": 0.24111536974960926,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.4926,
+      "step": 55
+    },
+    {
+      "epoch": 0.19964349376114082,
+      "grad_norm": 0.22834656564001263,
+      "learning_rate": 3.273809523809524e-05,
+      "loss": 0.4868,
+      "step": 56
+    },
+    {
+      "epoch": 0.20320855614973263,
+      "grad_norm": 0.2586635101404632,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.4835,
+      "step": 57
+    },
+    {
+      "epoch": 0.20677361853832443,
+      "grad_norm": 0.2631948305000637,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.479,
+      "step": 58
+    },
+    {
+      "epoch": 0.2103386809269162,
+      "grad_norm": 0.2426046855383363,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.4519,
+      "step": 59
+    },
+    {
+      "epoch": 0.21390374331550802,
+      "grad_norm": 0.22692280708702078,
+      "learning_rate": 3.511904761904762e-05,
+      "loss": 0.4807,
+      "step": 60
+    },
+    {
+      "epoch": 0.21746880570409982,
+      "grad_norm": 0.25328022488252383,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.469,
+      "step": 61
+    },
+    {
+      "epoch": 0.22103386809269163,
+      "grad_norm": 0.23618560071991695,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.4887,
+      "step": 62
+    },
+    {
+      "epoch": 0.22459893048128343,
+      "grad_norm": 0.24142699578051036,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.4857,
+      "step": 63
+    },
+    {
+      "epoch": 0.2281639928698752,
+      "grad_norm": 0.2434109164965224,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.4687,
+      "step": 64
+    },
+    {
+      "epoch": 0.23172905525846701,
+      "grad_norm": 0.24251289507246285,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.4424,
+      "step": 65
+    },
+    {
+      "epoch": 0.23529411764705882,
+      "grad_norm": 0.24663568533437502,
+      "learning_rate": 3.8690476190476195e-05,
+      "loss": 0.4465,
+      "step": 66
+    },
+    {
+      "epoch": 0.23885918003565063,
+      "grad_norm": 0.25161202678965316,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.4705,
+      "step": 67
+    },
+    {
+      "epoch": 0.24242424242424243,
+      "grad_norm": 0.2376282440517162,
+      "learning_rate": 3.9880952380952386e-05,
+      "loss": 0.4593,
+      "step": 68
+    },
+    {
+      "epoch": 0.24598930481283424,
+      "grad_norm": 0.25132574043293215,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.4553,
+      "step": 69
+    },
+    {
+      "epoch": 0.24955436720142601,
+      "grad_norm": 0.24573153429841577,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.4717,
+      "step": 70
+    },
+    {
+      "epoch": 0.2531194295900178,
+      "grad_norm": 0.2610886442266147,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4695,
+      "step": 71
+    },
+    {
+      "epoch": 0.25668449197860965,
+      "grad_norm": 0.2702286140750662,
+      "learning_rate": 4.226190476190476e-05,
+      "loss": 0.4434,
+      "step": 72
+    },
+    {
+      "epoch": 0.26024955436720143,
+      "grad_norm": 0.24772449699472507,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.4608,
+      "step": 73
+    },
+    {
+      "epoch": 0.2638146167557932,
+      "grad_norm": 0.24634704246937988,
+      "learning_rate": 4.345238095238096e-05,
+      "loss": 0.443,
+      "step": 74
+    },
+    {
+      "epoch": 0.26737967914438504,
+      "grad_norm": 0.3213709518904763,
+      "learning_rate": 4.404761904761905e-05,
+      "loss": 0.4608,
+      "step": 75
+    },
+    {
+      "epoch": 0.2709447415329768,
+      "grad_norm": 0.2653876420601825,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.4722,
+      "step": 76
+    },
+    {
+      "epoch": 0.27450980392156865,
+      "grad_norm": 0.3279620195307733,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.4601,
+      "step": 77
+    },
+    {
+      "epoch": 0.27807486631016043,
+      "grad_norm": 0.234887355366621,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4538,
+      "step": 78
+    },
+    {
+      "epoch": 0.2816399286987522,
+      "grad_norm": 0.265048583877912,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.463,
+      "step": 79
+    },
+    {
+      "epoch": 0.28520499108734404,
+      "grad_norm": 0.2720335404453364,
+      "learning_rate": 4.7023809523809525e-05,
+      "loss": 0.4401,
+      "step": 80
+    },
+    {
+      "epoch": 0.2887700534759358,
+      "grad_norm": 0.2531355298441409,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.4432,
+      "step": 81
+    },
+    {
+      "epoch": 0.29233511586452765,
+      "grad_norm": 0.26544674963410764,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.4673,
+      "step": 82
+    },
+    {
+      "epoch": 0.29590017825311943,
+      "grad_norm": 0.2664030790126549,
+      "learning_rate": 4.880952380952381e-05,
+      "loss": 0.4321,
+      "step": 83
+    },
+    {
+      "epoch": 0.2994652406417112,
+      "grad_norm": 0.23949544369445178,
+      "learning_rate": 4.940476190476191e-05,
+      "loss": 0.4313,
+      "step": 84
+    },
+    {
+      "epoch": 0.30303030303030304,
+      "grad_norm": 0.2385297012224782,
+      "learning_rate": 5e-05,
+      "loss": 0.4355,
+      "step": 85
+    },
+    {
+      "epoch": 0.3065953654188948,
+      "grad_norm": 0.2647946606355624,
+      "learning_rate": 4.993386243386244e-05,
+      "loss": 0.4194,
+      "step": 86
+    },
+    {
+      "epoch": 0.31016042780748665,
+      "grad_norm": 0.24687716628686734,
+      "learning_rate": 4.986772486772487e-05,
+      "loss": 0.441,
+      "step": 87
+    },
+    {
+      "epoch": 0.3137254901960784,
+      "grad_norm": 0.46551829644076037,
+      "learning_rate": 4.9801587301587306e-05,
+      "loss": 0.4391,
+      "step": 88
+    },
+    {
+      "epoch": 0.3172905525846702,
+      "grad_norm": 0.31302204392561833,
+      "learning_rate": 4.973544973544973e-05,
+      "loss": 0.4449,
+      "step": 89
+    },
+    {
+      "epoch": 0.32085561497326204,
+      "grad_norm": 0.264761862193315,
+      "learning_rate": 4.966931216931217e-05,
+      "loss": 0.4424,
+      "step": 90
+    },
+    {
+      "epoch": 0.3244206773618538,
+      "grad_norm": 0.381903500905413,
+      "learning_rate": 4.960317460317461e-05,
+      "loss": 0.4655,
+      "step": 91
+    },
+    {
+      "epoch": 0.32798573975044565,
+      "grad_norm": 0.27476297954320916,
+      "learning_rate": 4.9537037037037035e-05,
+      "loss": 0.4593,
+      "step": 92
+    },
+    {
+      "epoch": 0.3315508021390374,
+      "grad_norm": 0.31246265611714974,
+      "learning_rate": 4.9470899470899475e-05,
+      "loss": 0.4553,
+      "step": 93
+    },
+    {
+      "epoch": 0.33511586452762926,
+      "grad_norm": 0.2681791787347031,
+      "learning_rate": 4.940476190476191e-05,
+      "loss": 0.4359,
+      "step": 94
+    },
+    {
+      "epoch": 0.33868092691622104,
+      "grad_norm": 0.3532364039554794,
+      "learning_rate": 4.933862433862434e-05,
+      "loss": 0.4582,
+      "step": 95
+    },
+    {
+      "epoch": 0.3422459893048128,
+      "grad_norm": 0.536601467727922,
+      "learning_rate": 4.927248677248678e-05,
+      "loss": 0.453,
+      "step": 96
+    },
+    {
+      "epoch": 0.34581105169340465,
+      "grad_norm": 0.36735199918403505,
+      "learning_rate": 4.9206349206349204e-05,
+      "loss": 0.4499,
+      "step": 97
+    },
+    {
+      "epoch": 0.3493761140819964,
+      "grad_norm": 0.27837599560034726,
+      "learning_rate": 4.914021164021164e-05,
+      "loss": 0.4462,
+      "step": 98
+    },
+    {
+      "epoch": 0.35294117647058826,
+      "grad_norm": 0.29511697326074254,
+      "learning_rate": 4.9074074074074075e-05,
+      "loss": 0.4433,
+      "step": 99
+    },
+    {
+      "epoch": 0.35650623885918004,
+      "grad_norm": 0.2496929774220167,
+      "learning_rate": 4.900793650793651e-05,
+      "loss": 0.4372,
+      "step": 100
+    },
+    {
+      "epoch": 0.3600713012477718,
+      "grad_norm": 0.318739015686309,
+      "learning_rate": 4.894179894179895e-05,
+      "loss": 0.4364,
+      "step": 101
+    },
+    {
+      "epoch": 0.36363636363636365,
+      "grad_norm": 0.2752956866938909,
+      "learning_rate": 4.887566137566138e-05,
+      "loss": 0.4459,
+      "step": 102
+    },
+    {
+      "epoch": 0.3672014260249554,
+      "grad_norm": 0.2715543379851255,
+      "learning_rate": 4.880952380952381e-05,
+      "loss": 0.4259,
+      "step": 103
+    },
+    {
+      "epoch": 0.37076648841354726,
+      "grad_norm": 0.281979003784767,
+      "learning_rate": 4.8743386243386244e-05,
+      "loss": 0.4336,
+      "step": 104
+    },
+    {
+      "epoch": 0.37433155080213903,
+      "grad_norm": 0.30582126461131515,
+      "learning_rate": 4.8677248677248676e-05,
+      "loss": 0.4164,
+      "step": 105
+    },
+    {
+      "epoch": 0.3778966131907308,
+      "grad_norm": 0.27204216125058184,
+      "learning_rate": 4.8611111111111115e-05,
+      "loss": 0.4566,
+      "step": 106
+    },
+    {
+      "epoch": 0.38146167557932265,
+      "grad_norm": 0.3077553257782486,
+      "learning_rate": 4.854497354497355e-05,
+      "loss": 0.4436,
+      "step": 107
+    },
+    {
+      "epoch": 0.3850267379679144,
+      "grad_norm": 0.2582497704226908,
+      "learning_rate": 4.847883597883598e-05,
+      "loss": 0.4454,
+      "step": 108
+    },
+    {
+      "epoch": 0.38859180035650626,
+      "grad_norm": 0.27308361708396445,
+      "learning_rate": 4.841269841269841e-05,
+      "loss": 0.4346,
+      "step": 109
+    },
+    {
+      "epoch": 0.39215686274509803,
+      "grad_norm": 0.28612056247486717,
+      "learning_rate": 4.834656084656085e-05,
+      "loss": 0.421,
+      "step": 110
+    },
+    {
+      "epoch": 0.39572192513368987,
+      "grad_norm": 0.23314032211249056,
+      "learning_rate": 4.8280423280423284e-05,
+      "loss": 0.4435,
+      "step": 111
+    },
+    {
+      "epoch": 0.39928698752228164,
+      "grad_norm": 0.34212642191662695,
+      "learning_rate": 4.8214285714285716e-05,
+      "loss": 0.4273,
+      "step": 112
+    },
+    {
+      "epoch": 0.4028520499108734,
+      "grad_norm": 0.29337786387812703,
+      "learning_rate": 4.814814814814815e-05,
+      "loss": 0.4357,
+      "step": 113
+    },
+    {
+      "epoch": 0.40641711229946526,
+      "grad_norm": 0.27109618432529725,
+      "learning_rate": 4.808201058201058e-05,
+      "loss": 0.4363,
+      "step": 114
+    },
+    {
+      "epoch": 0.40998217468805703,
+      "grad_norm": 0.3038476673166368,
+      "learning_rate": 4.801587301587302e-05,
+      "loss": 0.4279,
+      "step": 115
+    },
+    {
+      "epoch": 0.41354723707664887,
+      "grad_norm": 0.2469736444113806,
+      "learning_rate": 4.794973544973545e-05,
+      "loss": 0.4689,
+      "step": 116
+    },
+    {
+      "epoch": 0.41711229946524064,
+      "grad_norm": 0.31628664078397917,
+      "learning_rate": 4.7883597883597884e-05,
+      "loss": 0.4584,
+      "step": 117
+    },
+    {
+      "epoch": 0.4206773618538324,
+      "grad_norm": 0.27210195175368657,
+      "learning_rate": 4.781746031746032e-05,
+      "loss": 0.4442,
+      "step": 118
+    },
+    {
+      "epoch": 0.42424242424242425,
+      "grad_norm": 0.29793682134027866,
+      "learning_rate": 4.7751322751322756e-05,
+      "loss": 0.436,
+      "step": 119
+    },
+    {
+      "epoch": 0.42780748663101603,
+      "grad_norm": 0.26088109260891573,
+      "learning_rate": 4.768518518518519e-05,
+      "loss": 0.4057,
+      "step": 120
+    },
+    {
+      "epoch": 0.43137254901960786,
+      "grad_norm": 0.24807856529355293,
+      "learning_rate": 4.761904761904762e-05,
+      "loss": 0.4381,
+      "step": 121
+    },
+    {
+      "epoch": 0.43493761140819964,
+      "grad_norm": 0.29062120195456376,
+      "learning_rate": 4.755291005291005e-05,
+      "loss": 0.4207,
+      "step": 122
+    },
+    {
+      "epoch": 0.4385026737967914,
+      "grad_norm": 0.2301732983656001,
+      "learning_rate": 4.748677248677249e-05,
+      "loss": 0.4427,
+      "step": 123
+    },
+    {
+      "epoch": 0.44206773618538325,
+      "grad_norm": 0.2809220314960326,
+      "learning_rate": 4.7420634920634924e-05,
+      "loss": 0.4236,
+      "step": 124
+    },
+    {
+      "epoch": 0.44563279857397503,
+      "grad_norm": 0.27862378467899274,
+      "learning_rate": 4.7354497354497356e-05,
+      "loss": 0.4289,
+      "step": 125
+    },
+    {
+      "epoch": 0.44919786096256686,
+      "grad_norm": 0.24697030715822402,
+      "learning_rate": 4.7288359788359796e-05,
+      "loss": 0.4269,
+      "step": 126
+    },
+    {
+      "epoch": 0.45276292335115864,
+      "grad_norm": 0.24423057839210063,
+      "learning_rate": 4.722222222222222e-05,
+      "loss": 0.4194,
+      "step": 127
+    },
+    {
+      "epoch": 0.4563279857397504,
+      "grad_norm": 0.2408884488998936,
+      "learning_rate": 4.715608465608466e-05,
+      "loss": 0.426,
+      "step": 128
+    },
+    {
+      "epoch": 0.45989304812834225,
+      "grad_norm": 0.2614135791546272,
+      "learning_rate": 4.708994708994709e-05,
+      "loss": 0.4364,
+      "step": 129
+    },
+    {
+      "epoch": 0.46345811051693403,
+      "grad_norm": 0.23844122206644586,
+      "learning_rate": 4.7023809523809525e-05,
+      "loss": 0.4377,
+      "step": 130
+    },
+    {
+      "epoch": 0.46702317290552586,
+      "grad_norm": 0.265778831668479,
+      "learning_rate": 4.6957671957671964e-05,
+      "loss": 0.4314,
+      "step": 131
+    },
+    {
+      "epoch": 0.47058823529411764,
+      "grad_norm": 0.23138874142061824,
+      "learning_rate": 4.689153439153439e-05,
+      "loss": 0.4313,
+      "step": 132
+    },
+    {
+      "epoch": 0.4741532976827095,
+      "grad_norm": 0.22179860641065724,
+      "learning_rate": 4.682539682539683e-05,
+      "loss": 0.4069,
+      "step": 133
+    },
+    {
+      "epoch": 0.47771836007130125,
+      "grad_norm": 0.2562645948187469,
+      "learning_rate": 4.675925925925926e-05,
+      "loss": 0.4292,
+      "step": 134
+    },
+    {
+      "epoch": 0.48128342245989303,
+      "grad_norm": 0.23376151529223232,
+      "learning_rate": 4.669312169312169e-05,
+      "loss": 0.4285,
+      "step": 135
+    },
+    {
+      "epoch": 0.48484848484848486,
+      "grad_norm": 0.23472720093625568,
+      "learning_rate": 4.662698412698413e-05,
+      "loss": 0.4401,
+      "step": 136
+    },
+    {
+      "epoch": 0.48841354723707664,
+      "grad_norm": 0.2532654700682945,
+      "learning_rate": 4.656084656084656e-05,
+      "loss": 0.4304,
+      "step": 137
+    },
+    {
+      "epoch": 0.4919786096256685,
+      "grad_norm": 0.24261824814971444,
+      "learning_rate": 4.6494708994709e-05,
+      "loss": 0.44,
+      "step": 138
+    },
+    {
+      "epoch": 0.49554367201426025,
+      "grad_norm": 0.2778687376633609,
+      "learning_rate": 4.642857142857143e-05,
+      "loss": 0.4278,
+      "step": 139
+    },
+    {
+      "epoch": 0.49910873440285203,
+      "grad_norm": 0.27072720802599815,
+      "learning_rate": 4.636243386243386e-05,
+      "loss": 0.4208,
+      "step": 140
+    },
+    {
+      "epoch": 0.5026737967914439,
+      "grad_norm": 0.26809024161843764,
+      "learning_rate": 4.62962962962963e-05,
+      "loss": 0.4115,
+      "step": 141
+    },
+    {
+      "epoch": 0.5062388591800356,
+      "grad_norm": 0.27301520879205193,
+      "learning_rate": 4.623015873015873e-05,
+      "loss": 0.4143,
+      "step": 142
+    },
+    {
+      "epoch": 0.5098039215686274,
+      "grad_norm": 0.21258310038027722,
+      "learning_rate": 4.6164021164021165e-05,
+      "loss": 0.4094,
+      "step": 143
+    },
+    {
+      "epoch": 0.5133689839572193,
+      "grad_norm": 0.27565959869265977,
+      "learning_rate": 4.60978835978836e-05,
+      "loss": 0.4276,
+      "step": 144
+    },
+    {
+      "epoch": 0.5169340463458111,
+      "grad_norm": 0.23979071267793753,
+      "learning_rate": 4.603174603174603e-05,
+      "loss": 0.4169,
+      "step": 145
+    },
+    {
+      "epoch": 0.5204991087344029,
+      "grad_norm": 0.22408054886193862,
+      "learning_rate": 4.596560846560847e-05,
+      "loss": 0.4157,
+      "step": 146
+    },
+    {
+      "epoch": 0.5240641711229946,
+      "grad_norm": 0.23450567179482062,
+      "learning_rate": 4.58994708994709e-05,
+      "loss": 0.4138,
+      "step": 147
+    },
+    {
+      "epoch": 0.5276292335115864,
+      "grad_norm": 0.23468120943540086,
+      "learning_rate": 4.5833333333333334e-05,
+      "loss": 0.4166,
+      "step": 148
+    },
+    {
+      "epoch": 0.5311942959001783,
+      "grad_norm": 0.2579273413286011,
+      "learning_rate": 4.576719576719577e-05,
+      "loss": 0.4179,
+      "step": 149
+    },
+    {
+      "epoch": 0.5347593582887701,
+      "grad_norm": 0.2485692321518658,
+      "learning_rate": 4.5701058201058205e-05,
+      "loss": 0.4162,
+      "step": 150
+    },
+    {
+      "epoch": 0.5383244206773619,
+      "grad_norm": 0.27766667484064855,
+      "learning_rate": 4.563492063492064e-05,
+      "loss": 0.4245,
+      "step": 151
+    },
+    {
+      "epoch": 0.5418894830659536,
+      "grad_norm": 0.24976722360513506,
+      "learning_rate": 4.556878306878307e-05,
+      "loss": 0.4258,
+      "step": 152
+    },
+    {
+      "epoch": 0.5454545454545454,
+      "grad_norm": 0.22706378628597793,
+      "learning_rate": 4.55026455026455e-05,
+      "loss": 0.3991,
+      "step": 153
+    },
+    {
+      "epoch": 0.5490196078431373,
+      "grad_norm": 0.2206903220736772,
+      "learning_rate": 4.543650793650794e-05,
+      "loss": 0.4104,
+      "step": 154
+    },
+    {
+      "epoch": 0.5525846702317291,
+      "grad_norm": 0.2619442593138655,
+      "learning_rate": 4.5370370370370374e-05,
+      "loss": 0.4369,
+      "step": 155
+    },
+    {
+      "epoch": 0.5561497326203209,
+      "grad_norm": 0.2267571470319652,
+      "learning_rate": 4.5304232804232806e-05,
+      "loss": 0.3997,
+      "step": 156
+    },
+    {
+      "epoch": 0.5597147950089126,
+      "grad_norm": 0.2776880161706508,
+      "learning_rate": 4.523809523809524e-05,
+      "loss": 0.4185,
+      "step": 157
+    },
+    {
+      "epoch": 0.5632798573975044,
+      "grad_norm": 0.24448259444463818,
+      "learning_rate": 4.517195767195768e-05,
+      "loss": 0.4087,
+      "step": 158
+    },
+    {
+      "epoch": 0.5668449197860963,
+      "grad_norm": 0.2808871378889792,
+      "learning_rate": 4.510582010582011e-05,
+      "loss": 0.4094,
+      "step": 159
+    },
+    {
+      "epoch": 0.5704099821746881,
+      "grad_norm": 0.2612547745838505,
+      "learning_rate": 4.503968253968254e-05,
+      "loss": 0.4149,
+      "step": 160
+    },
+    {
+      "epoch": 0.5739750445632799,
+      "grad_norm": 0.26145670746830635,
+      "learning_rate": 4.4973544973544974e-05,
+      "loss": 0.4041,
+      "step": 161
+    },
+    {
+      "epoch": 0.5775401069518716,
+      "grad_norm": 0.2557816140970961,
+      "learning_rate": 4.490740740740741e-05,
+      "loss": 0.4085,
+      "step": 162
+    },
+    {
+      "epoch": 0.5811051693404634,
+      "grad_norm": 0.3156030009402418,
+      "learning_rate": 4.4841269841269846e-05,
+      "loss": 0.4106,
+      "step": 163
+    },
+    {
+      "epoch": 0.5846702317290553,
+      "grad_norm": 0.2765054527757168,
+      "learning_rate": 4.477513227513228e-05,
+      "loss": 0.4372,
+      "step": 164
+    },
+    {
+      "epoch": 0.5882352941176471,
+      "grad_norm": 0.28892535349787296,
+      "learning_rate": 4.470899470899471e-05,
+      "loss": 0.4017,
+      "step": 165
+    },
+    {
+      "epoch": 0.5918003565062389,
+      "grad_norm": 0.24560106861290157,
+      "learning_rate": 4.464285714285715e-05,
+      "loss": 0.4009,
+      "step": 166
+    },
+    {
+      "epoch": 0.5953654188948306,
+      "grad_norm": 0.29644842423000184,
+      "learning_rate": 4.4576719576719575e-05,
+      "loss": 0.4167,
+      "step": 167
+    },
+    {
+      "epoch": 0.5989304812834224,
+      "grad_norm": 0.24415243860480915,
+      "learning_rate": 4.4510582010582014e-05,
+      "loss": 0.3987,
+      "step": 168
+    },
+    {
+      "epoch": 0.6024955436720143,
+      "grad_norm": 0.3167118254207159,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.4,
+      "step": 169
+    },
+    {
+      "epoch": 0.6060606060606061,
+      "grad_norm": 0.22236137381951881,
+      "learning_rate": 4.437830687830688e-05,
+      "loss": 0.4174,
+      "step": 170
+    },
+    {
+      "epoch": 0.6096256684491979,
+      "grad_norm": 0.3297518572409423,
+      "learning_rate": 4.431216931216932e-05,
+      "loss": 0.4324,
+      "step": 171
+    },
+    {
+      "epoch": 0.6131907308377896,
+      "grad_norm": 0.21983656827765505,
+      "learning_rate": 4.4246031746031744e-05,
+      "loss": 0.3969,
+      "step": 172
+    },
+    {
+      "epoch": 0.6167557932263814,
+      "grad_norm": 0.3018769063113944,
+      "learning_rate": 4.417989417989418e-05,
+      "loss": 0.4103,
+      "step": 173
+    },
+    {
+      "epoch": 0.6203208556149733,
+      "grad_norm": 0.2796078390446288,
+      "learning_rate": 4.4113756613756615e-05,
+      "loss": 0.4288,
+      "step": 174
+    },
+    {
+      "epoch": 0.6238859180035651,
+      "grad_norm": 0.32394169337639817,
+      "learning_rate": 4.404761904761905e-05,
+      "loss": 0.414,
+      "step": 175
+    },
+    {
+      "epoch": 0.6274509803921569,
+      "grad_norm": 0.29751053987105625,
+      "learning_rate": 4.3981481481481486e-05,
+      "loss": 0.4278,
+      "step": 176
+    },
+    {
+      "epoch": 0.6310160427807486,
+      "grad_norm": 0.2877197120249485,
+      "learning_rate": 4.391534391534391e-05,
+      "loss": 0.4092,
+      "step": 177
+    },
+    {
+      "epoch": 0.6345811051693404,
+      "grad_norm": 0.239853761954047,
+      "learning_rate": 4.384920634920635e-05,
+      "loss": 0.396,
+      "step": 178
+    },
+    {
+      "epoch": 0.6381461675579323,
+      "grad_norm": 0.2730616777174013,
+      "learning_rate": 4.378306878306879e-05,
+      "loss": 0.4054,
+      "step": 179
+    },
+    {
+      "epoch": 0.6417112299465241,
+      "grad_norm": 0.24263893360720792,
+      "learning_rate": 4.3716931216931216e-05,
+      "loss": 0.4185,
+      "step": 180
+    },
+    {
+      "epoch": 0.6452762923351159,
+      "grad_norm": 0.2831166441668498,
+      "learning_rate": 4.3650793650793655e-05,
+      "loss": 0.4028,
+      "step": 181
+    },
+    {
+      "epoch": 0.6488413547237076,
+      "grad_norm": 0.24750943083324203,
+      "learning_rate": 4.358465608465609e-05,
+      "loss": 0.4094,
+      "step": 182
+    },
+    {
+      "epoch": 0.6524064171122995,
+      "grad_norm": 0.26841919588025537,
+      "learning_rate": 4.351851851851852e-05,
+      "loss": 0.3927,
+      "step": 183
+    },
+    {
+      "epoch": 0.6559714795008913,
+      "grad_norm": 0.2943573088784056,
+      "learning_rate": 4.345238095238096e-05,
+      "loss": 0.4081,
+      "step": 184
+    },
+    {
+      "epoch": 0.6595365418894831,
+      "grad_norm": 0.2667939434935582,
+      "learning_rate": 4.3386243386243384e-05,
+      "loss": 0.3934,
+      "step": 185
+    },
+    {
+      "epoch": 0.6631016042780749,
+      "grad_norm": 0.26790375173659237,
+      "learning_rate": 4.332010582010582e-05,
+      "loss": 0.4019,
+      "step": 186
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.27370711973081785,
+      "learning_rate": 4.3253968253968256e-05,
+      "loss": 0.4077,
+      "step": 187
+    },
+    {
+      "epoch": 0.6702317290552585,
+      "grad_norm": 0.24058227563470871,
+      "learning_rate": 4.318783068783069e-05,
+      "loss": 0.4141,
+      "step": 188
+    },
+    {
+      "epoch": 0.6737967914438503,
+      "grad_norm": 0.23154757759323977,
+      "learning_rate": 4.312169312169313e-05,
+      "loss": 0.407,
+      "step": 189
+    },
+    {
+      "epoch": 0.6773618538324421,
+      "grad_norm": 0.22498833365843765,
+      "learning_rate": 4.305555555555556e-05,
+      "loss": 0.3933,
+      "step": 190
+    },
+    {
+      "epoch": 0.6809269162210339,
+      "grad_norm": 0.22452672437754953,
+      "learning_rate": 4.298941798941799e-05,
+      "loss": 0.3857,
+      "step": 191
+    },
+    {
+      "epoch": 0.6844919786096256,
+      "grad_norm": 0.2650901868655939,
+      "learning_rate": 4.2923280423280424e-05,
+      "loss": 0.4279,
+      "step": 192
+    },
+    {
+      "epoch": 0.6880570409982175,
+      "grad_norm": 0.22332574059189586,
+      "learning_rate": 4.2857142857142856e-05,
+      "loss": 0.4059,
+      "step": 193
+    },
+    {
+      "epoch": 0.6916221033868093,
+      "grad_norm": 0.2847586702299382,
+      "learning_rate": 4.2791005291005295e-05,
+      "loss": 0.4259,
+      "step": 194
+    },
+    {
+      "epoch": 0.6951871657754011,
+      "grad_norm": 0.22700690845900814,
+      "learning_rate": 4.272486772486773e-05,
+      "loss": 0.3962,
+      "step": 195
+    },
+    {
+      "epoch": 0.6987522281639929,
+      "grad_norm": 0.28201686556027095,
+      "learning_rate": 4.265873015873016e-05,
+      "loss": 0.4042,
+      "step": 196
+    },
+    {
+      "epoch": 0.7023172905525846,
+      "grad_norm": 0.2907900083355,
+      "learning_rate": 4.259259259259259e-05,
+      "loss": 0.4068,
+      "step": 197
+    },
+    {
+      "epoch": 0.7058823529411765,
+      "grad_norm": 0.2817684786178529,
+      "learning_rate": 4.252645502645503e-05,
+      "loss": 0.4256,
+      "step": 198
+    },
+    {
+      "epoch": 0.7094474153297683,
+      "grad_norm": 0.23658805574290273,
+      "learning_rate": 4.2460317460317464e-05,
+      "loss": 0.4134,
+      "step": 199
+    },
+    {
+      "epoch": 0.7130124777183601,
+      "grad_norm": 0.299118581601962,
+      "learning_rate": 4.2394179894179896e-05,
+      "loss": 0.4,
+      "step": 200
+    },
+    {
+      "epoch": 0.7165775401069518,
+      "grad_norm": 0.22642959006356225,
+      "learning_rate": 4.232804232804233e-05,
+      "loss": 0.3969,
+      "step": 201
+    },
+    {
+      "epoch": 0.7201426024955436,
+      "grad_norm": 0.294036556226744,
+      "learning_rate": 4.226190476190476e-05,
+      "loss": 0.4107,
+      "step": 202
+    },
+    {
+      "epoch": 0.7237076648841355,
+      "grad_norm": 0.24160245270385666,
+      "learning_rate": 4.21957671957672e-05,
+      "loss": 0.382,
+      "step": 203
+    },
+    {
+      "epoch": 0.7272727272727273,
+      "grad_norm": 0.28164728386845256,
+      "learning_rate": 4.212962962962963e-05,
+      "loss": 0.4132,
+      "step": 204
+    },
+    {
+      "epoch": 0.7308377896613191,
+      "grad_norm": 0.2450804579049508,
+      "learning_rate": 4.2063492063492065e-05,
+      "loss": 0.3983,
+      "step": 205
+    },
+    {
+      "epoch": 0.7344028520499108,
+      "grad_norm": 0.29109396787862757,
+      "learning_rate": 4.1997354497354504e-05,
+      "loss": 0.4007,
+      "step": 206
+    },
+    {
+      "epoch": 0.7379679144385026,
+      "grad_norm": 0.24958376816661956,
+      "learning_rate": 4.193121693121693e-05,
+      "loss": 0.3956,
+      "step": 207
+    },
+    {
+      "epoch": 0.7415329768270945,
+      "grad_norm": 0.23820864539293313,
+      "learning_rate": 4.186507936507937e-05,
+      "loss": 0.4221,
+      "step": 208
+    },
+    {
+      "epoch": 0.7450980392156863,
+      "grad_norm": 0.26200178980585,
+      "learning_rate": 4.17989417989418e-05,
+      "loss": 0.4257,
+      "step": 209
+    },
+    {
+      "epoch": 0.7486631016042781,
+      "grad_norm": 0.24414664760552993,
+      "learning_rate": 4.173280423280423e-05,
+      "loss": 0.4003,
+      "step": 210
+    },
+    {
+      "epoch": 0.7522281639928698,
+      "grad_norm": 0.2533921698544666,
+      "learning_rate": 4.166666666666667e-05,
+      "loss": 0.4062,
+      "step": 211
+    },
+    {
+      "epoch": 0.7557932263814616,
+      "grad_norm": 0.2504001539743044,
+      "learning_rate": 4.16005291005291e-05,
+      "loss": 0.4189,
+      "step": 212
+    },
+    {
+      "epoch": 0.7593582887700535,
+      "grad_norm": 0.23928316170865524,
+      "learning_rate": 4.153439153439154e-05,
+      "loss": 0.3944,
+      "step": 213
+    },
+    {
+      "epoch": 0.7629233511586453,
+      "grad_norm": 0.2645861063012114,
+      "learning_rate": 4.1468253968253976e-05,
+      "loss": 0.3947,
+      "step": 214
+    },
+    {
+      "epoch": 0.7664884135472371,
+      "grad_norm": 0.22707135799155675,
+      "learning_rate": 4.14021164021164e-05,
+      "loss": 0.3936,
+      "step": 215
+    },
+    {
+      "epoch": 0.7700534759358288,
+      "grad_norm": 0.29697865960516906,
+      "learning_rate": 4.133597883597884e-05,
+      "loss": 0.3904,
+      "step": 216
+    },
+    {
+      "epoch": 0.7736185383244206,
+      "grad_norm": 0.2541438054999295,
+      "learning_rate": 4.126984126984127e-05,
+      "loss": 0.4016,
+      "step": 217
+    },
+    {
+      "epoch": 0.7771836007130125,
+      "grad_norm": 0.2651520340472921,
+      "learning_rate": 4.1203703703703705e-05,
+      "loss": 0.4031,
+      "step": 218
+    },
+    {
+      "epoch": 0.7807486631016043,
+      "grad_norm": 0.2736426681369592,
+      "learning_rate": 4.1137566137566144e-05,
+      "loss": 0.4083,
+      "step": 219
+    },
+    {
+      "epoch": 0.7843137254901961,
+      "grad_norm": 0.24492131241592885,
+      "learning_rate": 4.107142857142857e-05,
+      "loss": 0.3769,
+      "step": 220
+    },
+    {
+      "epoch": 0.7878787878787878,
+      "grad_norm": 0.32076393929352504,
+      "learning_rate": 4.100529100529101e-05,
+      "loss": 0.4077,
+      "step": 221
+    },
+    {
+      "epoch": 0.7914438502673797,
+      "grad_norm": 0.2432796932049507,
+      "learning_rate": 4.093915343915344e-05,
+      "loss": 0.3944,
+      "step": 222
+    },
+    {
+      "epoch": 0.7950089126559715,
+      "grad_norm": 0.24006766733911922,
+      "learning_rate": 4.0873015873015874e-05,
+      "loss": 0.4159,
+      "step": 223
+    },
+    {
+      "epoch": 0.7985739750445633,
+      "grad_norm": 0.24351824821378842,
+      "learning_rate": 4.080687830687831e-05,
+      "loss": 0.4095,
+      "step": 224
+    },
+    {
+      "epoch": 0.8021390374331551,
+      "grad_norm": 0.254040621726078,
+      "learning_rate": 4.074074074074074e-05,
+      "loss": 0.4143,
+      "step": 225
+    },
+    {
+      "epoch": 0.8057040998217468,
+      "grad_norm": 0.24953740357738588,
+      "learning_rate": 4.067460317460318e-05,
+      "loss": 0.3934,
+      "step": 226
+    },
+    {
+      "epoch": 0.8092691622103387,
+      "grad_norm": 0.23446919982313796,
+      "learning_rate": 4.060846560846561e-05,
+      "loss": 0.397,
+      "step": 227
+    },
+    {
+      "epoch": 0.8128342245989305,
+      "grad_norm": 1.1184687580226778,
+      "learning_rate": 4.054232804232804e-05,
+      "loss": 0.4145,
+      "step": 228
+    },
+    {
+      "epoch": 0.8163992869875223,
+      "grad_norm": 0.24030583562025895,
+      "learning_rate": 4.047619047619048e-05,
+      "loss": 0.4082,
+      "step": 229
+    },
+    {
+      "epoch": 0.8199643493761141,
+      "grad_norm": 0.23074956675703204,
+      "learning_rate": 4.041005291005291e-05,
+      "loss": 0.3896,
+      "step": 230
+    },
+    {
+      "epoch": 0.8235294117647058,
+      "grad_norm": 0.23855598907716088,
+      "learning_rate": 4.0343915343915346e-05,
+      "loss": 0.4,
+      "step": 231
+    },
+    {
+      "epoch": 0.8270944741532977,
+      "grad_norm": 0.22311458939873455,
+      "learning_rate": 4.027777777777778e-05,
+      "loss": 0.3998,
+      "step": 232
+    },
+    {
+      "epoch": 0.8306595365418895,
+      "grad_norm": 0.2591834324449582,
+      "learning_rate": 4.021164021164021e-05,
+      "loss": 0.4103,
+      "step": 233
+    },
+    {
+      "epoch": 0.8342245989304813,
+      "grad_norm": 0.25039760052292054,
+      "learning_rate": 4.014550264550265e-05,
+      "loss": 0.3932,
+      "step": 234
+    },
+    {
+      "epoch": 0.8377896613190731,
+      "grad_norm": 0.23158442910111424,
+      "learning_rate": 4.007936507936508e-05,
+      "loss": 0.3851,
+      "step": 235
+    },
+    {
+      "epoch": 0.8413547237076648,
+      "grad_norm": 0.23426142880407433,
+      "learning_rate": 4.0013227513227514e-05,
+      "loss": 0.4025,
+      "step": 236
+    },
+    {
+      "epoch": 0.8449197860962567,
+      "grad_norm": 0.24173692192926743,
+      "learning_rate": 3.9947089947089946e-05,
+      "loss": 0.4019,
+      "step": 237
+    },
+    {
+      "epoch": 0.8484848484848485,
+      "grad_norm": 0.22440029996301472,
+      "learning_rate": 3.9880952380952386e-05,
+      "loss": 0.4009,
+      "step": 238
+    },
+    {
+      "epoch": 0.8520499108734403,
+      "grad_norm": 0.26041889005228147,
+      "learning_rate": 3.981481481481482e-05,
+      "loss": 0.3973,
+      "step": 239
+    },
+    {
+      "epoch": 0.8556149732620321,
+      "grad_norm": 0.2179629721584941,
+      "learning_rate": 3.974867724867725e-05,
+      "loss": 0.3871,
+      "step": 240
+    },
+    {
+      "epoch": 0.8591800356506238,
+      "grad_norm": 0.2800252858543578,
+      "learning_rate": 3.968253968253968e-05,
+      "loss": 0.3941,
+      "step": 241
+    },
+    {
+      "epoch": 0.8627450980392157,
+      "grad_norm": 0.24810834353925612,
+      "learning_rate": 3.9616402116402115e-05,
+      "loss": 0.383,
+      "step": 242
+    },
+    {
+      "epoch": 0.8663101604278075,
+      "grad_norm": 0.22945317419350042,
+      "learning_rate": 3.9550264550264554e-05,
+      "loss": 0.4121,
+      "step": 243
+    },
+    {
+      "epoch": 0.8698752228163993,
+      "grad_norm": 0.3141896822071523,
+      "learning_rate": 3.9484126984126986e-05,
+      "loss": 0.3945,
+      "step": 244
+    },
+    {
+      "epoch": 0.8734402852049911,
+      "grad_norm": 0.26909231165849823,
+      "learning_rate": 3.941798941798942e-05,
+      "loss": 0.4044,
+      "step": 245
+    },
+    {
+      "epoch": 0.8770053475935828,
+      "grad_norm": 0.23341130848645644,
+      "learning_rate": 3.935185185185186e-05,
+      "loss": 0.3922,
+      "step": 246
+    },
+    {
+      "epoch": 0.8805704099821747,
+      "grad_norm": 0.29087768840385597,
+      "learning_rate": 3.928571428571429e-05,
+      "loss": 0.404,
+      "step": 247
+    },
+    {
+      "epoch": 0.8841354723707665,
+      "grad_norm": 0.2532983239560777,
+      "learning_rate": 3.921957671957672e-05,
+      "loss": 0.3843,
+      "step": 248
+    },
+    {
+      "epoch": 0.8877005347593583,
+      "grad_norm": 0.2583121749262775,
+      "learning_rate": 3.9153439153439155e-05,
+      "loss": 0.3931,
+      "step": 249
+    },
+    {
+      "epoch": 0.8912655971479501,
+      "grad_norm": 0.2948774186680668,
+      "learning_rate": 3.908730158730159e-05,
+      "loss": 0.4124,
+      "step": 250
+    },
+    {
+      "epoch": 0.8948306595365418,
+      "grad_norm": 0.2396549716981653,
+      "learning_rate": 3.9021164021164026e-05,
+      "loss": 0.4057,
+      "step": 251
+    },
+    {
+      "epoch": 0.8983957219251337,
+      "grad_norm": 0.2915903101730862,
+      "learning_rate": 3.895502645502646e-05,
+      "loss": 0.4148,
+      "step": 252
+    },
+    {
+      "epoch": 0.9019607843137255,
+      "grad_norm": 0.2545140074294779,
+      "learning_rate": 3.888888888888889e-05,
+      "loss": 0.3771,
+      "step": 253
+    },
+    {
+      "epoch": 0.9055258467023173,
+      "grad_norm": 0.2640502870137779,
+      "learning_rate": 3.882275132275133e-05,
+      "loss": 0.3944,
+      "step": 254
+    },
+    {
+      "epoch": 0.9090909090909091,
+      "grad_norm": 0.24374886368613632,
+      "learning_rate": 3.8756613756613755e-05,
+      "loss": 0.3987,
+      "step": 255
+    },
+    {
+      "epoch": 0.9126559714795008,
+      "grad_norm": 0.2809600878706678,
+      "learning_rate": 3.8690476190476195e-05,
+      "loss": 0.4008,
+      "step": 256
+    },
+    {
+      "epoch": 0.9162210338680927,
+      "grad_norm": 0.23268059257007717,
+      "learning_rate": 3.862433862433863e-05,
+      "loss": 0.4009,
+      "step": 257
+    },
+    {
+      "epoch": 0.9197860962566845,
+      "grad_norm": 0.24048049926800477,
+      "learning_rate": 3.855820105820106e-05,
+      "loss": 0.4117,
+      "step": 258
+    },
+    {
+      "epoch": 0.9233511586452763,
+      "grad_norm": 0.2649369281049162,
+      "learning_rate": 3.84920634920635e-05,
+      "loss": 0.3884,
+      "step": 259
+    },
+    {
+      "epoch": 0.9269162210338681,
+      "grad_norm": 0.237155661723028,
+      "learning_rate": 3.8425925925925924e-05,
+      "loss": 0.3933,
+      "step": 260
+    },
+    {
+      "epoch": 0.93048128342246,
+      "grad_norm": 0.2485602109598096,
+      "learning_rate": 3.835978835978836e-05,
+      "loss": 0.4106,
+      "step": 261
+    },
+    {
+      "epoch": 0.9340463458110517,
+      "grad_norm": 0.2694399856848455,
+      "learning_rate": 3.8293650793650795e-05,
+      "loss": 0.4032,
+      "step": 262
+    },
+    {
+      "epoch": 0.9376114081996435,
+      "grad_norm": 0.2648634493737914,
+      "learning_rate": 3.822751322751323e-05,
+      "loss": 0.4025,
+      "step": 263
+    },
+    {
+      "epoch": 0.9411764705882353,
+      "grad_norm": 0.22659539775045073,
+      "learning_rate": 3.816137566137567e-05,
+      "loss": 0.3929,
+      "step": 264
+    },
+    {
+      "epoch": 0.9447415329768271,
+      "grad_norm": 0.22448119053294815,
+      "learning_rate": 3.809523809523809e-05,
+      "loss": 0.3985,
+      "step": 265
+    },
+    {
+      "epoch": 0.948306595365419,
+      "grad_norm": 0.23605897424000963,
+      "learning_rate": 3.802910052910053e-05,
+      "loss": 0.3902,
+      "step": 266
+    },
+    {
+      "epoch": 0.9518716577540107,
+      "grad_norm": 0.21567841013952768,
+      "learning_rate": 3.7962962962962964e-05,
+      "loss": 0.4031,
+      "step": 267
+    },
+    {
+      "epoch": 0.9554367201426025,
+      "grad_norm": 0.226582206009011,
+      "learning_rate": 3.7896825396825396e-05,
+      "loss": 0.4034,
+      "step": 268
+    },
+    {
+      "epoch": 0.9590017825311943,
+      "grad_norm": 0.21410813536621254,
+      "learning_rate": 3.7830687830687835e-05,
+      "loss": 0.388,
+      "step": 269
+    },
+    {
+      "epoch": 0.9625668449197861,
+      "grad_norm": 0.22635417721134862,
+      "learning_rate": 3.776455026455027e-05,
+      "loss": 0.3981,
+      "step": 270
+    },
+    {
+      "epoch": 0.966131907308378,
+      "grad_norm": 0.24549930492308253,
+      "learning_rate": 3.76984126984127e-05,
+      "loss": 0.3873,
+      "step": 271
+    },
+    {
+      "epoch": 0.9696969696969697,
+      "grad_norm": 0.23311920805361547,
+      "learning_rate": 3.763227513227514e-05,
+      "loss": 0.3914,
+      "step": 272
+    },
+    {
+      "epoch": 0.9732620320855615,
+      "grad_norm": 0.2219324754241415,
+      "learning_rate": 3.7566137566137564e-05,
+      "loss": 0.3816,
+      "step": 273
+    },
+    {
+      "epoch": 0.9768270944741533,
+      "grad_norm": 0.25439583328415083,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.3926,
+      "step": 274
+    },
+    {
+      "epoch": 0.9803921568627451,
+      "grad_norm": 0.20271930292703347,
+      "learning_rate": 3.7433862433862436e-05,
+      "loss": 0.4029,
+      "step": 275
+    },
+    {
+      "epoch": 0.983957219251337,
+      "grad_norm": 0.2940260926073234,
+      "learning_rate": 3.736772486772487e-05,
+      "loss": 0.3966,
+      "step": 276
+    },
+    {
+      "epoch": 0.9875222816399287,
+      "grad_norm": 0.24208283274203934,
+      "learning_rate": 3.730158730158731e-05,
+      "loss": 0.3888,
+      "step": 277
+    },
+    {
+      "epoch": 0.9910873440285205,
+      "grad_norm": 0.24392643811450568,
+      "learning_rate": 3.723544973544973e-05,
+      "loss": 0.4067,
+      "step": 278
+    },
+    {
+      "epoch": 0.9946524064171123,
+      "grad_norm": 0.21490414167190403,
+      "learning_rate": 3.716931216931217e-05,
+      "loss": 0.3932,
+      "step": 279
+    },
+    {
+      "epoch": 0.9982174688057041,
+      "grad_norm": 0.24374249860990013,
+      "learning_rate": 3.7103174603174604e-05,
+      "loss": 0.4113,
+      "step": 280
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.24374249860990013,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 0.3928,
+      "step": 281
+    },
+    {
+      "epoch": 1.0035650623885919,
+      "grad_norm": 0.31525908521769425,
+      "learning_rate": 3.6970899470899476e-05,
+      "loss": 0.3232,
+      "step": 282
+    },
+    {
+      "epoch": 1.0071301247771836,
+      "grad_norm": 0.2643939964306856,
+      "learning_rate": 3.690476190476191e-05,
+      "loss": 0.3263,
+      "step": 283
+    },
+    {
+      "epoch": 1.0106951871657754,
+      "grad_norm": 0.23332086398141738,
+      "learning_rate": 3.683862433862434e-05,
+      "loss": 0.3111,
+      "step": 284
+    },
+    {
+      "epoch": 1.014260249554367,
+      "grad_norm": 0.2421961703766304,
+      "learning_rate": 3.677248677248677e-05,
+      "loss": 0.328,
+      "step": 285
+    },
+    {
+      "epoch": 1.017825311942959,
+      "grad_norm": 0.280155384464068,
+      "learning_rate": 3.6706349206349205e-05,
+      "loss": 0.3226,
+      "step": 286
+    },
+    {
+      "epoch": 1.0213903743315509,
+      "grad_norm": 0.25066624899200685,
+      "learning_rate": 3.6640211640211644e-05,
+      "loss": 0.3158,
+      "step": 287
+    },
+    {
+      "epoch": 1.0249554367201426,
+      "grad_norm": 0.2417536178673186,
+      "learning_rate": 3.6574074074074076e-05,
+      "loss": 0.3046,
+      "step": 288
+    },
+    {
+      "epoch": 1.0285204991087344,
+      "grad_norm": 0.30826071577139075,
+      "learning_rate": 3.650793650793651e-05,
+      "loss": 0.3194,
+      "step": 289
+    },
+    {
+      "epoch": 1.032085561497326,
+      "grad_norm": 0.2309068352595092,
+      "learning_rate": 3.644179894179894e-05,
+      "loss": 0.3361,
+      "step": 290
+    },
+    {
+      "epoch": 1.035650623885918,
+      "grad_norm": 0.31008229547524313,
+      "learning_rate": 3.637566137566138e-05,
+      "loss": 0.3177,
+      "step": 291
+    },
+    {
+      "epoch": 1.0392156862745099,
+      "grad_norm": 0.25545372484549134,
+      "learning_rate": 3.630952380952381e-05,
+      "loss": 0.3101,
+      "step": 292
+    },
+    {
+      "epoch": 1.0427807486631016,
+      "grad_norm": 0.23184466757198224,
+      "learning_rate": 3.6243386243386245e-05,
+      "loss": 0.3168,
+      "step": 293
+    },
+    {
+      "epoch": 1.0463458110516934,
+      "grad_norm": 0.24811631331610906,
+      "learning_rate": 3.617724867724868e-05,
+      "loss": 0.3129,
+      "step": 294
+    },
+    {
+      "epoch": 1.049910873440285,
+      "grad_norm": 0.2709627739010445,
+      "learning_rate": 3.611111111111111e-05,
+      "loss": 0.3218,
+      "step": 295
+    },
+    {
+      "epoch": 1.053475935828877,
+      "grad_norm": 0.24141072938856623,
+      "learning_rate": 3.604497354497355e-05,
+      "loss": 0.3572,
+      "step": 296
+    },
+    {
+      "epoch": 1.0570409982174689,
+      "grad_norm": 0.24005354065055498,
+      "learning_rate": 3.597883597883598e-05,
+      "loss": 0.3094,
+      "step": 297
+    },
+    {
+      "epoch": 1.0606060606060606,
+      "grad_norm": 0.23461155846998907,
+      "learning_rate": 3.591269841269841e-05,
+      "loss": 0.3435,
+      "step": 298
+    },
+    {
+      "epoch": 1.0641711229946524,
+      "grad_norm": 0.24835238198211415,
+      "learning_rate": 3.584656084656085e-05,
+      "loss": 0.3177,
+      "step": 299
+    },
+    {
+      "epoch": 1.067736185383244,
+      "grad_norm": 0.2300499445112996,
+      "learning_rate": 3.578042328042328e-05,
+      "loss": 0.3141,
+      "step": 300
+    },
+    {
+      "epoch": 1.071301247771836,
+      "grad_norm": 0.26029542032508707,
+      "learning_rate": 3.571428571428572e-05,
+      "loss": 0.3113,
+      "step": 301
+    },
+    {
+      "epoch": 1.0748663101604279,
+      "grad_norm": 0.23241089227926529,
+      "learning_rate": 3.564814814814815e-05,
+      "loss": 0.3378,
+      "step": 302
+    },
+    {
+      "epoch": 1.0784313725490196,
+      "grad_norm": 0.23749395321909145,
+      "learning_rate": 3.558201058201058e-05,
+      "loss": 0.3242,
+      "step": 303
+    },
+    {
+      "epoch": 1.0819964349376114,
+      "grad_norm": 0.24313241085963547,
+      "learning_rate": 3.551587301587302e-05,
+      "loss": 0.3349,
+      "step": 304
+    },
+    {
+      "epoch": 1.085561497326203,
+      "grad_norm": 0.2340991342429292,
+      "learning_rate": 3.5449735449735446e-05,
+      "loss": 0.3246,
+      "step": 305
+    },
+    {
+      "epoch": 1.089126559714795,
+      "grad_norm": 0.21496476570019948,
+      "learning_rate": 3.5383597883597885e-05,
+      "loss": 0.3147,
+      "step": 306
+    },
+    {
+      "epoch": 1.0926916221033869,
+      "grad_norm": 0.21939962942096003,
+      "learning_rate": 3.5317460317460324e-05,
+      "loss": 0.307,
+      "step": 307
+    },
+    {
+      "epoch": 1.0962566844919786,
+      "grad_norm": 0.22099659624474047,
+      "learning_rate": 3.525132275132275e-05,
+      "loss": 0.329,
+      "step": 308
+    },
+    {
+      "epoch": 1.0998217468805704,
+      "grad_norm": 0.2228425219747887,
+      "learning_rate": 3.518518518518519e-05,
+      "loss": 0.3498,
+      "step": 309
+    },
+    {
+      "epoch": 1.1033868092691623,
+      "grad_norm": 0.22052635486023414,
+      "learning_rate": 3.511904761904762e-05,
+      "loss": 0.3222,
+      "step": 310
+    },
+    {
+      "epoch": 1.106951871657754,
+      "grad_norm": 0.23938237725827405,
+      "learning_rate": 3.5052910052910054e-05,
+      "loss": 0.3166,
+      "step": 311
+    },
+    {
+      "epoch": 1.1105169340463459,
+      "grad_norm": 0.18838380661352314,
+      "learning_rate": 3.498677248677249e-05,
+      "loss": 0.3108,
+      "step": 312
+    },
+    {
+      "epoch": 1.1140819964349375,
+      "grad_norm": 0.21278577476667734,
+      "learning_rate": 3.492063492063492e-05,
+      "loss": 0.3013,
+      "step": 313
+    },
+    {
+      "epoch": 1.1176470588235294,
+      "grad_norm": 0.23972432756615197,
+      "learning_rate": 3.485449735449736e-05,
+      "loss": 0.3209,
+      "step": 314
+    },
+    {
+      "epoch": 1.121212121212121,
+      "grad_norm": 0.2045475199546501,
+      "learning_rate": 3.478835978835979e-05,
+      "loss": 0.3431,
+      "step": 315
+    },
+    {
+      "epoch": 1.124777183600713,
+      "grad_norm": 0.239737979677932,
+      "learning_rate": 3.472222222222222e-05,
+      "loss": 0.3078,
+      "step": 316
+    },
+    {
+      "epoch": 1.1283422459893049,
+      "grad_norm": 0.2265813527306897,
+      "learning_rate": 3.465608465608466e-05,
+      "loss": 0.307,
+      "step": 317
+    },
+    {
+      "epoch": 1.1319073083778965,
+      "grad_norm": 0.23322266902477815,
+      "learning_rate": 3.458994708994709e-05,
+      "loss": 0.3321,
+      "step": 318
+    },
+    {
+      "epoch": 1.1354723707664884,
+      "grad_norm": 0.22011164204847558,
+      "learning_rate": 3.4523809523809526e-05,
+      "loss": 0.306,
+      "step": 319
+    },
+    {
+      "epoch": 1.1390374331550803,
+      "grad_norm": 0.2098345409866113,
+      "learning_rate": 3.445767195767196e-05,
+      "loss": 0.3049,
+      "step": 320
+    },
+    {
+      "epoch": 1.142602495543672,
+      "grad_norm": 0.24133953330582658,
+      "learning_rate": 3.439153439153439e-05,
+      "loss": 0.3114,
+      "step": 321
+    },
+    {
+      "epoch": 1.1461675579322639,
+      "grad_norm": 0.22005478195742428,
+      "learning_rate": 3.432539682539683e-05,
+      "loss": 0.3182,
+      "step": 322
+    },
+    {
+      "epoch": 1.1497326203208555,
+      "grad_norm": 0.20786230459043104,
+      "learning_rate": 3.425925925925926e-05,
+      "loss": 0.3537,
+      "step": 323
+    },
+    {
+      "epoch": 1.1532976827094474,
+      "grad_norm": 0.2366867581735452,
+      "learning_rate": 3.4193121693121694e-05,
+      "loss": 0.3224,
+      "step": 324
+    },
+    {
+      "epoch": 1.156862745098039,
+      "grad_norm": 0.24427952405073164,
+      "learning_rate": 3.412698412698413e-05,
+      "loss": 0.3073,
+      "step": 325
+    },
+    {
+      "epoch": 1.160427807486631,
+      "grad_norm": 0.2148177713455186,
+      "learning_rate": 3.406084656084656e-05,
+      "loss": 0.3157,
+      "step": 326
+    },
+    {
+      "epoch": 1.1639928698752229,
+      "grad_norm": 0.2154428562433842,
+      "learning_rate": 3.3994708994709e-05,
+      "loss": 0.311,
+      "step": 327
+    },
+    {
+      "epoch": 1.1675579322638145,
+      "grad_norm": 0.21365271236501948,
+      "learning_rate": 3.392857142857143e-05,
+      "loss": 0.3243,
+      "step": 328
+    },
+    {
+      "epoch": 1.1711229946524064,
+      "grad_norm": 0.23106780723739542,
+      "learning_rate": 3.386243386243386e-05,
+      "loss": 0.3199,
+      "step": 329
+    },
+    {
+      "epoch": 1.1746880570409983,
+      "grad_norm": 0.20378310493243906,
+      "learning_rate": 3.3796296296296295e-05,
+      "loss": 0.3083,
+      "step": 330
+    },
+    {
+      "epoch": 1.17825311942959,
+      "grad_norm": 0.20121967286938067,
+      "learning_rate": 3.3730158730158734e-05,
+      "loss": 0.3128,
+      "step": 331
+    },
+    {
+      "epoch": 1.1818181818181819,
+      "grad_norm": 0.20294246609153418,
+      "learning_rate": 3.3664021164021167e-05,
+      "loss": 0.3324,
+      "step": 332
+    },
+    {
+      "epoch": 1.1853832442067735,
+      "grad_norm": 0.2248678683673094,
+      "learning_rate": 3.35978835978836e-05,
+      "loss": 0.3346,
+      "step": 333
+    },
+    {
+      "epoch": 1.1889483065953654,
+      "grad_norm": 0.21814903914505945,
+      "learning_rate": 3.353174603174603e-05,
+      "loss": 0.329,
+      "step": 334
+    },
+    {
+      "epoch": 1.192513368983957,
+      "grad_norm": 0.23109119938194214,
+      "learning_rate": 3.3465608465608464e-05,
+      "loss": 0.3277,
+      "step": 335
+    },
+    {
+      "epoch": 1.196078431372549,
+      "grad_norm": 0.2102214004221497,
+      "learning_rate": 3.33994708994709e-05,
+      "loss": 0.3183,
+      "step": 336
+    },
+    {
+      "epoch": 1.1996434937611409,
+      "grad_norm": 0.22577045157632822,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.3148,
+      "step": 337
+    },
+    {
+      "epoch": 1.2032085561497325,
+      "grad_norm": 0.22203965301536938,
+      "learning_rate": 3.326719576719577e-05,
+      "loss": 0.3199,
+      "step": 338
+    },
+    {
+      "epoch": 1.2067736185383244,
+      "grad_norm": 0.20423159452569417,
+      "learning_rate": 3.3201058201058206e-05,
+      "loss": 0.3416,
+      "step": 339
+    },
+    {
+      "epoch": 1.2103386809269163,
+      "grad_norm": 0.22529813188482867,
+      "learning_rate": 3.313492063492064e-05,
+      "loss": 0.3259,
+      "step": 340
+    },
+    {
+      "epoch": 1.213903743315508,
+      "grad_norm": 0.20744922802895027,
+      "learning_rate": 3.306878306878307e-05,
+      "loss": 0.3423,
+      "step": 341
+    },
+    {
+      "epoch": 1.2174688057040999,
+      "grad_norm": 0.2014799372147007,
+      "learning_rate": 3.30026455026455e-05,
+      "loss": 0.3237,
+      "step": 342
+    },
+    {
+      "epoch": 1.2210338680926915,
+      "grad_norm": 0.22958550701571454,
+      "learning_rate": 3.2936507936507936e-05,
+      "loss": 0.2992,
+      "step": 343
+    },
+    {
+      "epoch": 1.2245989304812834,
+      "grad_norm": 0.20428869907816727,
+      "learning_rate": 3.2870370370370375e-05,
+      "loss": 0.3177,
+      "step": 344
+    },
+    {
+      "epoch": 1.228163992869875,
+      "grad_norm": 0.2101497453066906,
+      "learning_rate": 3.280423280423281e-05,
+      "loss": 0.3142,
+      "step": 345
+    },
+    {
+      "epoch": 1.231729055258467,
+      "grad_norm": 0.21451339833045976,
+      "learning_rate": 3.273809523809524e-05,
+      "loss": 0.3149,
+      "step": 346
+    },
+    {
+      "epoch": 1.2352941176470589,
+      "grad_norm": 0.23260946277162806,
+      "learning_rate": 3.267195767195768e-05,
+      "loss": 0.3331,
+      "step": 347
+    },
+    {
+      "epoch": 1.2388591800356505,
+      "grad_norm": 0.23823291796387266,
+      "learning_rate": 3.2605820105820104e-05,
+      "loss": 0.3208,
+      "step": 348
+    },
+    {
+      "epoch": 1.2424242424242424,
+      "grad_norm": 0.223429828748214,
+      "learning_rate": 3.253968253968254e-05,
+      "loss": 0.3209,
+      "step": 349
+    },
+    {
+      "epoch": 1.2459893048128343,
+      "grad_norm": 0.22417249378396703,
+      "learning_rate": 3.2473544973544976e-05,
+      "loss": 0.3208,
+      "step": 350
+    },
+    {
+      "epoch": 1.249554367201426,
+      "grad_norm": 0.21454657917012931,
+      "learning_rate": 3.240740740740741e-05,
+      "loss": 0.3189,
+      "step": 351
+    },
+    {
+      "epoch": 1.2531194295900179,
+      "grad_norm": 0.21817658653352878,
+      "learning_rate": 3.234126984126985e-05,
+      "loss": 0.3084,
+      "step": 352
+    },
+    {
+      "epoch": 1.2566844919786098,
+      "grad_norm": 0.1884735660283043,
+      "learning_rate": 3.227513227513227e-05,
+      "loss": 0.3134,
+      "step": 353
+    },
+    {
+      "epoch": 1.2602495543672014,
+      "grad_norm": 0.2606760022111759,
+      "learning_rate": 3.220899470899471e-05,
+      "loss": 0.3085,
+      "step": 354
+    },
+    {
+      "epoch": 1.263814616755793,
+      "grad_norm": 0.23061214698334195,
+      "learning_rate": 3.2142857142857144e-05,
+      "loss": 0.3239,
+      "step": 355
+    },
+    {
+      "epoch": 1.267379679144385,
+      "grad_norm": 0.21811541377841318,
+      "learning_rate": 3.2076719576719576e-05,
+      "loss": 0.3137,
+      "step": 356
+    },
+    {
+      "epoch": 1.2709447415329769,
+      "grad_norm": 0.25088605692404314,
+      "learning_rate": 3.2010582010582015e-05,
+      "loss": 0.3392,
+      "step": 357
+    },
+    {
+      "epoch": 1.2745098039215685,
+      "grad_norm": 0.23016827074966836,
+      "learning_rate": 3.194444444444444e-05,
+      "loss": 0.3299,
+      "step": 358
+    },
+    {
+      "epoch": 1.2780748663101604,
+      "grad_norm": 0.2421974975026712,
+      "learning_rate": 3.187830687830688e-05,
+      "loss": 0.3237,
+      "step": 359
+    },
+    {
+      "epoch": 1.2816399286987523,
+      "grad_norm": 0.2543131477092282,
+      "learning_rate": 3.181216931216931e-05,
+      "loss": 0.3126,
+      "step": 360
+    },
+    {
+      "epoch": 1.285204991087344,
+      "grad_norm": 0.22904682830909853,
+      "learning_rate": 3.1746031746031745e-05,
+      "loss": 0.3211,
+      "step": 361
+    },
+    {
+      "epoch": 1.2887700534759359,
+      "grad_norm": 0.22398104295979468,
+      "learning_rate": 3.1679894179894184e-05,
+      "loss": 0.3232,
+      "step": 362
+    },
+    {
+      "epoch": 1.2923351158645278,
+      "grad_norm": 0.24975834285084594,
+      "learning_rate": 3.1613756613756616e-05,
+      "loss": 0.3153,
+      "step": 363
+    },
+    {
+      "epoch": 1.2959001782531194,
+      "grad_norm": 0.22166086727233478,
+      "learning_rate": 3.154761904761905e-05,
+      "loss": 0.3067,
+      "step": 364
+    },
+    {
+      "epoch": 1.299465240641711,
+      "grad_norm": 0.22728639132886716,
+      "learning_rate": 3.148148148148148e-05,
+      "loss": 0.3463,
+      "step": 365
+    },
+    {
+      "epoch": 1.303030303030303,
+      "grad_norm": 0.2667820946041398,
+      "learning_rate": 3.141534391534391e-05,
+      "loss": 0.3472,
+      "step": 366
+    },
+    {
+      "epoch": 1.3065953654188949,
+      "grad_norm": 0.2145208746622264,
+      "learning_rate": 3.134920634920635e-05,
+      "loss": 0.317,
+      "step": 367
+    },
+    {
+      "epoch": 1.3101604278074865,
+      "grad_norm": 0.24071269924062685,
+      "learning_rate": 3.1283068783068784e-05,
+      "loss": 0.304,
+      "step": 368
+    },
+    {
+      "epoch": 1.3137254901960784,
+      "grad_norm": 0.2424499639825663,
+      "learning_rate": 3.121693121693122e-05,
+      "loss": 0.32,
+      "step": 369
+    },
+    {
+      "epoch": 1.3172905525846703,
+      "grad_norm": 0.20473032753551487,
+      "learning_rate": 3.1150793650793656e-05,
+      "loss": 0.292,
+      "step": 370
+    },
+    {
+      "epoch": 1.320855614973262,
+      "grad_norm": 0.1953875986255421,
+      "learning_rate": 3.108465608465609e-05,
+      "loss": 0.3084,
+      "step": 371
+    },
+    {
+      "epoch": 1.3244206773618539,
+      "grad_norm": 0.22922922374079685,
+      "learning_rate": 3.101851851851852e-05,
+      "loss": 0.3159,
+      "step": 372
+    },
+    {
+      "epoch": 1.3279857397504458,
+      "grad_norm": 0.20405010410911523,
+      "learning_rate": 3.095238095238095e-05,
+      "loss": 0.3133,
+      "step": 373
+    },
+    {
+      "epoch": 1.3315508021390374,
+      "grad_norm": 0.20777966401029974,
+      "learning_rate": 3.0886243386243385e-05,
+      "loss": 0.3157,
+      "step": 374
+    },
+    {
+      "epoch": 1.3351158645276293,
+      "grad_norm": 0.19928997526305234,
+      "learning_rate": 3.0820105820105824e-05,
+      "loss": 0.3362,
+      "step": 375
+    },
+    {
+      "epoch": 1.338680926916221,
+      "grad_norm": 0.19646067759416652,
+      "learning_rate": 3.075396825396826e-05,
+      "loss": 0.31,
+      "step": 376
+    },
+    {
+      "epoch": 1.3422459893048129,
+      "grad_norm": 0.21947779820244392,
+      "learning_rate": 3.068783068783069e-05,
+      "loss": 0.3072,
+      "step": 377
+    },
+    {
+      "epoch": 1.3458110516934045,
+      "grad_norm": 0.19872860796933098,
+      "learning_rate": 3.062169312169312e-05,
+      "loss": 0.3175,
+      "step": 378
+    },
+    {
+      "epoch": 1.3493761140819964,
+      "grad_norm": 0.19946649624117538,
+      "learning_rate": 3.055555555555556e-05,
+      "loss": 0.3239,
+      "step": 379
+    },
+    {
+      "epoch": 1.3529411764705883,
+      "grad_norm": 0.2078138628282042,
+      "learning_rate": 3.048941798941799e-05,
+      "loss": 0.2922,
+      "step": 380
+    },
+    {
+      "epoch": 1.35650623885918,
+      "grad_norm": 0.1955787152157713,
+      "learning_rate": 3.0423280423280425e-05,
+      "loss": 0.3182,
+      "step": 381
+    },
+    {
+      "epoch": 1.3600713012477719,
+      "grad_norm": 0.23271966402434482,
+      "learning_rate": 3.0357142857142857e-05,
+      "loss": 0.3036,
+      "step": 382
+    },
+    {
+      "epoch": 1.3636363636363638,
+      "grad_norm": 0.22857850997488346,
+      "learning_rate": 3.0291005291005293e-05,
+      "loss": 0.3159,
+      "step": 383
+    },
+    {
+      "epoch": 1.3672014260249554,
+      "grad_norm": 0.2267874097792744,
+      "learning_rate": 3.022486772486773e-05,
+      "loss": 0.3149,
+      "step": 384
+    },
+    {
+      "epoch": 1.3707664884135473,
+      "grad_norm": 0.22725414410059314,
+      "learning_rate": 3.0158730158730158e-05,
+      "loss": 0.3172,
+      "step": 385
+    },
+    {
+      "epoch": 1.374331550802139,
+      "grad_norm": 0.2186908817676311,
+      "learning_rate": 3.0092592592592593e-05,
+      "loss": 0.3124,
+      "step": 386
+    },
+    {
+      "epoch": 1.3778966131907309,
+      "grad_norm": 0.235405338128003,
+      "learning_rate": 3.002645502645503e-05,
+      "loss": 0.3084,
+      "step": 387
+    },
+    {
+      "epoch": 1.3814616755793225,
+      "grad_norm": 0.22225336010296162,
+      "learning_rate": 2.996031746031746e-05,
+      "loss": 0.3336,
+      "step": 388
+    },
+    {
+      "epoch": 1.3850267379679144,
+      "grad_norm": 0.23442197746715424,
+      "learning_rate": 2.9894179894179897e-05,
+      "loss": 0.3118,
+      "step": 389
+    },
+    {
+      "epoch": 1.3885918003565063,
+      "grad_norm": 0.2234369540100837,
+      "learning_rate": 2.9828042328042326e-05,
+      "loss": 0.3207,
+      "step": 390
+    },
+    {
+      "epoch": 1.392156862745098,
+      "grad_norm": 0.2185234046206301,
+      "learning_rate": 2.9761904761904762e-05,
+      "loss": 0.3085,
+      "step": 391
+    },
+    {
+      "epoch": 1.3957219251336899,
+      "grad_norm": 0.2134003673102712,
+      "learning_rate": 2.9695767195767198e-05,
+      "loss": 0.3214,
+      "step": 392
+    },
+    {
+      "epoch": 1.3992869875222818,
+      "grad_norm": 0.20171738258937869,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.3143,
+      "step": 393
+    },
+    {
+      "epoch": 1.4028520499108734,
+      "grad_norm": 0.20129735323023007,
+      "learning_rate": 2.9563492063492066e-05,
+      "loss": 0.3321,
+      "step": 394
+    },
+    {
+      "epoch": 1.4064171122994653,
+      "grad_norm": 0.24207239646252252,
+      "learning_rate": 2.94973544973545e-05,
+      "loss": 0.3174,
+      "step": 395
+    },
+    {
+      "epoch": 1.409982174688057,
+      "grad_norm": 0.21684155716428563,
+      "learning_rate": 2.943121693121693e-05,
+      "loss": 0.3215,
+      "step": 396
+    },
+    {
+      "epoch": 1.4135472370766489,
+      "grad_norm": 0.18802246374467288,
+      "learning_rate": 2.9365079365079366e-05,
+      "loss": 0.3267,
+      "step": 397
+    },
+    {
+      "epoch": 1.4171122994652405,
+      "grad_norm": 0.25304556950939483,
+      "learning_rate": 2.92989417989418e-05,
+      "loss": 0.3132,
+      "step": 398
+    },
+    {
+      "epoch": 1.4206773618538324,
+      "grad_norm": 0.195088342368623,
+      "learning_rate": 2.9232804232804234e-05,
+      "loss": 0.3134,
+      "step": 399
+    },
+    {
+      "epoch": 1.4242424242424243,
+      "grad_norm": 0.2182881644927651,
+      "learning_rate": 2.916666666666667e-05,
+      "loss": 0.3023,
+      "step": 400
+    },
+    {
+      "epoch": 1.427807486631016,
+      "grad_norm": 0.21802121501801272,
+      "learning_rate": 2.91005291005291e-05,
+      "loss": 0.3255,
+      "step": 401
+    },
+    {
+      "epoch": 1.4313725490196079,
+      "grad_norm": 0.22312274126829879,
+      "learning_rate": 2.9034391534391538e-05,
+      "loss": 0.3415,
+      "step": 402
+    },
+    {
+      "epoch": 1.4349376114081998,
+      "grad_norm": 0.2137822003172708,
+      "learning_rate": 2.8968253968253974e-05,
+      "loss": 0.3273,
+      "step": 403
+    },
+    {
+      "epoch": 1.4385026737967914,
+      "grad_norm": 0.20495235889031405,
+      "learning_rate": 2.8902116402116402e-05,
+      "loss": 0.3032,
+      "step": 404
+    },
+    {
+      "epoch": 1.4420677361853833,
+      "grad_norm": 0.21655780326309493,
+      "learning_rate": 2.8835978835978838e-05,
+      "loss": 0.2982,
+      "step": 405
+    },
+    {
+      "epoch": 1.445632798573975,
+      "grad_norm": 0.18728577234073096,
+      "learning_rate": 2.876984126984127e-05,
+      "loss": 0.3148,
+      "step": 406
+    },
+    {
+      "epoch": 1.4491978609625669,
+      "grad_norm": 0.27778033796942686,
+      "learning_rate": 2.8703703703703706e-05,
+      "loss": 0.3508,
+      "step": 407
+    },
+    {
+      "epoch": 1.4527629233511585,
+      "grad_norm": 0.21086920971978912,
+      "learning_rate": 2.8637566137566142e-05,
+      "loss": 0.3342,
+      "step": 408
+    },
+    {
+      "epoch": 1.4563279857397504,
+      "grad_norm": 0.24682677159593425,
+      "learning_rate": 2.857142857142857e-05,
+      "loss": 0.291,
+      "step": 409
+    },
+    {
+      "epoch": 1.4598930481283423,
+      "grad_norm": 0.23663541260303997,
+      "learning_rate": 2.8505291005291007e-05,
+      "loss": 0.3191,
+      "step": 410
+    },
+    {
+      "epoch": 1.463458110516934,
+      "grad_norm": 0.20703760427505377,
+      "learning_rate": 2.8439153439153442e-05,
+      "loss": 0.3156,
+      "step": 411
+    },
+    {
+      "epoch": 1.4670231729055259,
+      "grad_norm": 0.23157149797254634,
+      "learning_rate": 2.8373015873015875e-05,
+      "loss": 0.3182,
+      "step": 412
+    },
+    {
+      "epoch": 1.4705882352941178,
+      "grad_norm": 0.2555845619255346,
+      "learning_rate": 2.830687830687831e-05,
+      "loss": 0.3564,
+      "step": 413
+    },
+    {
+      "epoch": 1.4741532976827094,
+      "grad_norm": 0.23490581994262758,
+      "learning_rate": 2.824074074074074e-05,
+      "loss": 0.3115,
+      "step": 414
+    },
+    {
+      "epoch": 1.4777183600713013,
+      "grad_norm": 0.22821518514441477,
+      "learning_rate": 2.8174603174603175e-05,
+      "loss": 0.3235,
+      "step": 415
+    },
+    {
+      "epoch": 1.481283422459893,
+      "grad_norm": 0.2820139045137352,
+      "learning_rate": 2.810846560846561e-05,
+      "loss": 0.3135,
+      "step": 416
+    },
+    {
+      "epoch": 1.4848484848484849,
+      "grad_norm": 0.23511851918723767,
+      "learning_rate": 2.8042328042328043e-05,
+      "loss": 0.3265,
+      "step": 417
+    },
+    {
+      "epoch": 1.4884135472370765,
+      "grad_norm": 0.18941098466588235,
+      "learning_rate": 2.797619047619048e-05,
+      "loss": 0.3069,
+      "step": 418
+    },
+    {
+      "epoch": 1.4919786096256684,
+      "grad_norm": 0.2698400080197567,
+      "learning_rate": 2.7910052910052914e-05,
+      "loss": 0.3181,
+      "step": 419
+    },
+    {
+      "epoch": 1.4955436720142603,
+      "grad_norm": 0.19750385207292742,
+      "learning_rate": 2.7843915343915343e-05,
+      "loss": 0.3133,
+      "step": 420
+    },
+    {
+      "epoch": 1.499108734402852,
+      "grad_norm": 0.2457376269880891,
+      "learning_rate": 2.777777777777778e-05,
+      "loss": 0.3002,
+      "step": 421
+    },
+    {
+      "epoch": 1.5026737967914439,
+      "grad_norm": 0.23329609907620164,
+      "learning_rate": 2.771164021164021e-05,
+      "loss": 0.3156,
+      "step": 422
+    },
+    {
+      "epoch": 1.5062388591800357,
+      "grad_norm": 0.23430129532705443,
+      "learning_rate": 2.7645502645502647e-05,
+      "loss": 0.3176,
+      "step": 423
+    },
+    {
+      "epoch": 1.5098039215686274,
+      "grad_norm": 0.25234952298801444,
+      "learning_rate": 2.7579365079365083e-05,
+      "loss": 0.3119,
+      "step": 424
+    },
+    {
+      "epoch": 1.5133689839572193,
+      "grad_norm": 0.23759599694113478,
+      "learning_rate": 2.7513227513227512e-05,
+      "loss": 0.3331,
+      "step": 425
+    },
+    {
+      "epoch": 1.5169340463458112,
+      "grad_norm": 0.19856205142922717,
+      "learning_rate": 2.7447089947089948e-05,
+      "loss": 0.3407,
+      "step": 426
+    },
+    {
+      "epoch": 1.5204991087344029,
+      "grad_norm": 0.22875622301603105,
+      "learning_rate": 2.7380952380952383e-05,
+      "loss": 0.3131,
+      "step": 427
+    },
+    {
+      "epoch": 1.5240641711229945,
+      "grad_norm": 0.211348806364257,
+      "learning_rate": 2.7314814814814816e-05,
+      "loss": 0.3347,
+      "step": 428
+    },
+    {
+      "epoch": 1.5276292335115864,
+      "grad_norm": 0.207907332665488,
+      "learning_rate": 2.724867724867725e-05,
+      "loss": 0.3158,
+      "step": 429
+    },
+    {
+      "epoch": 1.5311942959001783,
+      "grad_norm": 0.19899412036573233,
+      "learning_rate": 2.718253968253968e-05,
+      "loss": 0.312,
+      "step": 430
+    },
+    {
+      "epoch": 1.53475935828877,
+      "grad_norm": 0.21311604856517985,
+      "learning_rate": 2.7116402116402116e-05,
+      "loss": 0.3124,
+      "step": 431
+    },
+    {
+      "epoch": 1.5383244206773619,
+      "grad_norm": 0.2014625972162956,
+      "learning_rate": 2.7050264550264555e-05,
+      "loss": 0.3162,
+      "step": 432
+    },
+    {
+      "epoch": 1.5418894830659537,
+      "grad_norm": 0.19373860855670413,
+      "learning_rate": 2.6984126984126984e-05,
+      "loss": 0.3112,
+      "step": 433
+    },
+    {
+      "epoch": 1.5454545454545454,
+      "grad_norm": 0.19713483071582547,
+      "learning_rate": 2.691798941798942e-05,
+      "loss": 0.3171,
+      "step": 434
+    },
+    {
+      "epoch": 1.5490196078431373,
+      "grad_norm": 0.19257566048437327,
+      "learning_rate": 2.6851851851851855e-05,
+      "loss": 0.3226,
+      "step": 435
+    },
+    {
+      "epoch": 1.5525846702317292,
+      "grad_norm": 0.21404202539070205,
+      "learning_rate": 2.6785714285714288e-05,
+      "loss": 0.3371,
+      "step": 436
+    },
+    {
+      "epoch": 1.5561497326203209,
+      "grad_norm": 0.19710075946165687,
+      "learning_rate": 2.6719576719576723e-05,
+      "loss": 0.3448,
+      "step": 437
+    },
+    {
+      "epoch": 1.5597147950089125,
+      "grad_norm": 0.21792173119539968,
+      "learning_rate": 2.6653439153439152e-05,
+      "loss": 0.3031,
+      "step": 438
+    },
+    {
+      "epoch": 1.5632798573975044,
+      "grad_norm": 0.21255538393677179,
+      "learning_rate": 2.6587301587301588e-05,
+      "loss": 0.31,
+      "step": 439
+    },
+    {
+      "epoch": 1.5668449197860963,
+      "grad_norm": 0.18813863894262572,
+      "learning_rate": 2.6521164021164024e-05,
+      "loss": 0.2907,
+      "step": 440
+    },
+    {
+      "epoch": 1.570409982174688,
+      "grad_norm": 0.21602785826591622,
+      "learning_rate": 2.6455026455026456e-05,
+      "loss": 0.3057,
+      "step": 441
+    },
+    {
+      "epoch": 1.5739750445632799,
+      "grad_norm": 0.1942519255792355,
+      "learning_rate": 2.6388888888888892e-05,
+      "loss": 0.3102,
+      "step": 442
+    },
+    {
+      "epoch": 1.5775401069518717,
+      "grad_norm": 0.2221582968637408,
+      "learning_rate": 2.6322751322751328e-05,
+      "loss": 0.3328,
+      "step": 443
+    },
+    {
+      "epoch": 1.5811051693404634,
+      "grad_norm": 0.21145901562944847,
+      "learning_rate": 2.6256613756613757e-05,
+      "loss": 0.3254,
+      "step": 444
+    },
+    {
+      "epoch": 1.5846702317290553,
+      "grad_norm": 0.22508556227477491,
+      "learning_rate": 2.6190476190476192e-05,
+      "loss": 0.3175,
+      "step": 445
+    },
+    {
+      "epoch": 1.5882352941176472,
+      "grad_norm": 0.20314366613181892,
+      "learning_rate": 2.6124338624338625e-05,
+      "loss": 0.3148,
+      "step": 446
+    },
+    {
+      "epoch": 1.5918003565062389,
+      "grad_norm": 0.21827587275771088,
+      "learning_rate": 2.605820105820106e-05,
+      "loss": 0.3126,
+      "step": 447
+    },
+    {
+      "epoch": 1.5953654188948305,
+      "grad_norm": 0.2254164069621931,
+      "learning_rate": 2.5992063492063496e-05,
+      "loss": 0.3144,
+      "step": 448
+    },
+    {
+      "epoch": 1.5989304812834224,
+      "grad_norm": 0.19932246187679203,
+      "learning_rate": 2.5925925925925925e-05,
+      "loss": 0.3041,
+      "step": 449
+    },
+    {
+      "epoch": 1.6024955436720143,
+      "grad_norm": 0.2158641937943349,
+      "learning_rate": 2.585978835978836e-05,
+      "loss": 0.3085,
+      "step": 450
+    },
+    {
+      "epoch": 1.606060606060606,
+      "grad_norm": 0.1905542158673529,
+      "learning_rate": 2.5793650793650796e-05,
+      "loss": 0.3167,
+      "step": 451
+    },
+    {
+      "epoch": 1.6096256684491979,
+      "grad_norm": 0.20315358302779118,
+      "learning_rate": 2.572751322751323e-05,
+      "loss": 0.3154,
+      "step": 452
+    },
+    {
+      "epoch": 1.6131907308377897,
+      "grad_norm": 0.193452367891943,
+      "learning_rate": 2.5661375661375664e-05,
+      "loss": 0.3223,
+      "step": 453
+    },
+    {
+      "epoch": 1.6167557932263814,
+      "grad_norm": 0.22721465187398188,
+      "learning_rate": 2.5595238095238093e-05,
+      "loss": 0.3394,
+      "step": 454
+    },
+    {
+      "epoch": 1.6203208556149733,
+      "grad_norm": 0.20086768014805884,
+      "learning_rate": 2.552910052910053e-05,
+      "loss": 0.3179,
+      "step": 455
+    },
+    {
+      "epoch": 1.6238859180035652,
+      "grad_norm": 0.24530653751967776,
+      "learning_rate": 2.5462962962962965e-05,
+      "loss": 0.3025,
+      "step": 456
+    },
+    {
+      "epoch": 1.6274509803921569,
+      "grad_norm": 0.1858810547245731,
+      "learning_rate": 2.5396825396825397e-05,
+      "loss": 0.3138,
+      "step": 457
+    },
+    {
+      "epoch": 1.6310160427807485,
+      "grad_norm": 0.20313601202486925,
+      "learning_rate": 2.5330687830687833e-05,
+      "loss": 0.3116,
+      "step": 458
+    },
+    {
+      "epoch": 1.6345811051693404,
+      "grad_norm": 0.21827188635216663,
+      "learning_rate": 2.526455026455027e-05,
+      "loss": 0.3137,
+      "step": 459
+    },
+    {
+      "epoch": 1.6381461675579323,
+      "grad_norm": 0.19048557451431505,
+      "learning_rate": 2.5198412698412697e-05,
+      "loss": 0.3213,
+      "step": 460
+    },
+    {
+      "epoch": 1.641711229946524,
+      "grad_norm": 0.20925899471456255,
+      "learning_rate": 2.5132275132275137e-05,
+      "loss": 0.3164,
+      "step": 461
+    },
+    {
+      "epoch": 1.6452762923351159,
+      "grad_norm": 0.187818912758532,
+      "learning_rate": 2.5066137566137565e-05,
+      "loss": 0.3086,
+      "step": 462
+    },
+    {
+      "epoch": 1.6488413547237077,
+      "grad_norm": 0.2108500113583357,
+      "learning_rate": 2.5e-05,
+      "loss": 0.2946,
+      "step": 463
+    },
+    {
+      "epoch": 1.6524064171122994,
+      "grad_norm": 0.22006164729951747,
+      "learning_rate": 2.4933862433862434e-05,
+      "loss": 0.3132,
+      "step": 464
+    },
+    {
+      "epoch": 1.6559714795008913,
+      "grad_norm": 0.1718321072626545,
+      "learning_rate": 2.4867724867724866e-05,
+      "loss": 0.2994,
+      "step": 465
+    },
+    {
+      "epoch": 1.6595365418894832,
+      "grad_norm": 0.19692734876215362,
+      "learning_rate": 2.4801587301587305e-05,
+      "loss": 0.3154,
+      "step": 466
+    },
+    {
+      "epoch": 1.6631016042780749,
+      "grad_norm": 0.19177460327043558,
+      "learning_rate": 2.4735449735449737e-05,
+      "loss": 0.2989,
+      "step": 467
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.19446284476682252,
+      "learning_rate": 2.466931216931217e-05,
+      "loss": 0.3178,
+      "step": 468
+    },
+    {
+      "epoch": 1.6702317290552586,
+      "grad_norm": 0.206334913085213,
+      "learning_rate": 2.4603174603174602e-05,
+      "loss": 0.3209,
+      "step": 469
+    },
+    {
+      "epoch": 1.6737967914438503,
+      "grad_norm": 0.18459079448974214,
+      "learning_rate": 2.4537037037037038e-05,
+      "loss": 0.3051,
+      "step": 470
+    },
+    {
+      "epoch": 1.677361853832442,
+      "grad_norm": 0.22131143477388124,
+      "learning_rate": 2.4470899470899473e-05,
+      "loss": 0.3224,
+      "step": 471
+    },
+    {
+      "epoch": 1.6809269162210339,
+      "grad_norm": 0.2018833561138814,
+      "learning_rate": 2.4404761904761906e-05,
+      "loss": 0.3063,
+      "step": 472
+    },
+    {
+      "epoch": 1.6844919786096257,
+      "grad_norm": 0.19529747692485203,
+      "learning_rate": 2.4338624338624338e-05,
+      "loss": 0.3029,
+      "step": 473
+    },
+    {
+      "epoch": 1.6880570409982174,
+      "grad_norm": 0.1985249170117421,
+      "learning_rate": 2.4272486772486774e-05,
+      "loss": 0.315,
+      "step": 474
+    },
+    {
+      "epoch": 1.6916221033868093,
+      "grad_norm": 0.2122147883680695,
+      "learning_rate": 2.4206349206349206e-05,
+      "loss": 0.3231,
+      "step": 475
+    },
+    {
+      "epoch": 1.6951871657754012,
+      "grad_norm": 0.19799282746927058,
+      "learning_rate": 2.4140211640211642e-05,
+      "loss": 0.3246,
+      "step": 476
+    },
+    {
+      "epoch": 1.6987522281639929,
+      "grad_norm": 0.18208717336322142,
+      "learning_rate": 2.4074074074074074e-05,
+      "loss": 0.3072,
+      "step": 477
+    },
+    {
+      "epoch": 1.7023172905525845,
+      "grad_norm": 0.18859955414014584,
+      "learning_rate": 2.400793650793651e-05,
+      "loss": 0.31,
+      "step": 478
+    },
+    {
+      "epoch": 1.7058823529411766,
+      "grad_norm": 0.20921657858349363,
+      "learning_rate": 2.3941798941798942e-05,
+      "loss": 0.3019,
+      "step": 479
+    },
+    {
+      "epoch": 1.7094474153297683,
+      "grad_norm": 0.1982707597041992,
+      "learning_rate": 2.3875661375661378e-05,
+      "loss": 0.3236,
+      "step": 480
+    },
+    {
+      "epoch": 1.71301247771836,
+      "grad_norm": 0.6201765879281012,
+      "learning_rate": 2.380952380952381e-05,
+      "loss": 0.3375,
+      "step": 481
+    },
+    {
+      "epoch": 1.7165775401069518,
+      "grad_norm": 0.2138071118729244,
+      "learning_rate": 2.3743386243386246e-05,
+      "loss": 0.322,
+      "step": 482
+    },
+    {
+      "epoch": 1.7201426024955437,
+      "grad_norm": 0.19203048123376498,
+      "learning_rate": 2.3677248677248678e-05,
+      "loss": 0.3377,
+      "step": 483
+    },
+    {
+      "epoch": 1.7237076648841354,
+      "grad_norm": 0.20670478397464165,
+      "learning_rate": 2.361111111111111e-05,
+      "loss": 0.2984,
+      "step": 484
+    },
+    {
+      "epoch": 1.7272727272727273,
+      "grad_norm": 0.19575126167850404,
+      "learning_rate": 2.3544973544973546e-05,
+      "loss": 0.2983,
+      "step": 485
+    },
+    {
+      "epoch": 1.7308377896613192,
+      "grad_norm": 0.20376687236376323,
+      "learning_rate": 2.3478835978835982e-05,
+      "loss": 0.3259,
+      "step": 486
+    },
+    {
+      "epoch": 1.7344028520499108,
+      "grad_norm": 0.21982992315657082,
+      "learning_rate": 2.3412698412698414e-05,
+      "loss": 0.322,
+      "step": 487
+    },
+    {
+      "epoch": 1.7379679144385025,
+      "grad_norm": 0.22001788508011383,
+      "learning_rate": 2.3346560846560847e-05,
+      "loss": 0.3091,
+      "step": 488
+    },
+    {
+      "epoch": 1.7415329768270946,
+      "grad_norm": 0.20523105925579282,
+      "learning_rate": 2.328042328042328e-05,
+      "loss": 0.3192,
+      "step": 489
+    },
+    {
+      "epoch": 1.7450980392156863,
+      "grad_norm": 0.1920781074195229,
+      "learning_rate": 2.3214285714285715e-05,
+      "loss": 0.319,
+      "step": 490
+    },
+    {
+      "epoch": 1.748663101604278,
+      "grad_norm": 0.19352944079340595,
+      "learning_rate": 2.314814814814815e-05,
+      "loss": 0.3105,
+      "step": 491
+    },
+    {
+      "epoch": 1.7522281639928698,
+      "grad_norm": 0.19727799586219957,
+      "learning_rate": 2.3082010582010583e-05,
+      "loss": 0.3205,
+      "step": 492
+    },
+    {
+      "epoch": 1.7557932263814617,
+      "grad_norm": 0.20042603100895678,
+      "learning_rate": 2.3015873015873015e-05,
+      "loss": 0.3268,
+      "step": 493
+    },
+    {
+      "epoch": 1.7593582887700534,
+      "grad_norm": 0.19361171882220973,
+      "learning_rate": 2.294973544973545e-05,
+      "loss": 0.3071,
+      "step": 494
+    },
+    {
+      "epoch": 1.7629233511586453,
+      "grad_norm": 0.19629470643534186,
+      "learning_rate": 2.2883597883597886e-05,
+      "loss": 0.3101,
+      "step": 495
+    },
+    {
+      "epoch": 1.7664884135472372,
+      "grad_norm": 0.18396617832822676,
+      "learning_rate": 2.281746031746032e-05,
+      "loss": 0.3129,
+      "step": 496
+    },
+    {
+      "epoch": 1.7700534759358288,
+      "grad_norm": 0.18636190007697212,
+      "learning_rate": 2.275132275132275e-05,
+      "loss": 0.3094,
+      "step": 497
+    },
+    {
+      "epoch": 1.7736185383244205,
+      "grad_norm": 0.1964579757416564,
+      "learning_rate": 2.2685185185185187e-05,
+      "loss": 0.3363,
+      "step": 498
+    },
+    {
+      "epoch": 1.7771836007130126,
+      "grad_norm": 0.19011894291228393,
+      "learning_rate": 2.261904761904762e-05,
+      "loss": 0.3124,
+      "step": 499
+    },
+    {
+      "epoch": 1.7807486631016043,
+      "grad_norm": 0.19100253216681698,
+      "learning_rate": 2.2552910052910055e-05,
+      "loss": 0.3073,
+      "step": 500
+    },
+    {
+      "epoch": 1.784313725490196,
+      "grad_norm": 0.19155681109591394,
+      "learning_rate": 2.2486772486772487e-05,
+      "loss": 0.3191,
+      "step": 501
+    },
+    {
+      "epoch": 1.7878787878787878,
+      "grad_norm": 0.19511419962319346,
+      "learning_rate": 2.2420634920634923e-05,
+      "loss": 0.3214,
+      "step": 502
+    },
+    {
+      "epoch": 1.7914438502673797,
+      "grad_norm": 0.2026017731839353,
+      "learning_rate": 2.2354497354497355e-05,
+      "loss": 0.3135,
+      "step": 503
+    },
+    {
+      "epoch": 1.7950089126559714,
+      "grad_norm": 0.18532543183794659,
+      "learning_rate": 2.2288359788359788e-05,
+      "loss": 0.3106,
+      "step": 504
+    },
+    {
+      "epoch": 1.7985739750445633,
+      "grad_norm": 0.19506570228396244,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 0.316,
+      "step": 505
+    },
+    {
+      "epoch": 1.8021390374331552,
+      "grad_norm": 0.1802530807977306,
+      "learning_rate": 2.215608465608466e-05,
+      "loss": 0.314,
+      "step": 506
+    },
+    {
+      "epoch": 1.8057040998217468,
+      "grad_norm": 0.18281275029331645,
+      "learning_rate": 2.208994708994709e-05,
+      "loss": 0.298,
+      "step": 507
+    },
+    {
+      "epoch": 1.8092691622103387,
+      "grad_norm": 0.2033769544690045,
+      "learning_rate": 2.2023809523809524e-05,
+      "loss": 0.3134,
+      "step": 508
+    },
+    {
+      "epoch": 1.8128342245989306,
+      "grad_norm": 0.20470062724847538,
+      "learning_rate": 2.1957671957671956e-05,
+      "loss": 0.3335,
+      "step": 509
+    },
+    {
+      "epoch": 1.8163992869875223,
+      "grad_norm": 0.20744080793585934,
+      "learning_rate": 2.1891534391534395e-05,
+      "loss": 0.3121,
+      "step": 510
+    },
+    {
+      "epoch": 1.819964349376114,
+      "grad_norm": 0.19960817053723556,
+      "learning_rate": 2.1825396825396827e-05,
+      "loss": 0.3255,
+      "step": 511
+    },
+    {
+      "epoch": 1.8235294117647058,
+      "grad_norm": 0.21301446901563517,
+      "learning_rate": 2.175925925925926e-05,
+      "loss": 0.3034,
+      "step": 512
+    },
+    {
+      "epoch": 1.8270944741532977,
+      "grad_norm": 0.1720975540476051,
+      "learning_rate": 2.1693121693121692e-05,
+      "loss": 0.3036,
+      "step": 513
+    },
+    {
+      "epoch": 1.8306595365418894,
+      "grad_norm": 0.1826003900042692,
+      "learning_rate": 2.1626984126984128e-05,
+      "loss": 0.3018,
+      "step": 514
+    },
+    {
+      "epoch": 1.8342245989304813,
+      "grad_norm": 0.20147130173757077,
+      "learning_rate": 2.1560846560846563e-05,
+      "loss": 0.3318,
+      "step": 515
+    },
+    {
+      "epoch": 1.8377896613190732,
+      "grad_norm": 0.18404204428041754,
+      "learning_rate": 2.1494708994708996e-05,
+      "loss": 0.3038,
+      "step": 516
+    },
+    {
+      "epoch": 1.8413547237076648,
+      "grad_norm": 0.19111299538038287,
+      "learning_rate": 2.1428571428571428e-05,
+      "loss": 0.3128,
+      "step": 517
+    },
+    {
+      "epoch": 1.8449197860962567,
+      "grad_norm": 0.20337474833215646,
+      "learning_rate": 2.1362433862433864e-05,
+      "loss": 0.3197,
+      "step": 518
+    },
+    {
+      "epoch": 1.8484848484848486,
+      "grad_norm": 0.19879999181430622,
+      "learning_rate": 2.1296296296296296e-05,
+      "loss": 0.3266,
+      "step": 519
+    },
+    {
+      "epoch": 1.8520499108734403,
+      "grad_norm": 0.20268350702396395,
+      "learning_rate": 2.1230158730158732e-05,
+      "loss": 0.3261,
+      "step": 520
+    },
+    {
+      "epoch": 1.855614973262032,
+      "grad_norm": 0.1978630832424863,
+      "learning_rate": 2.1164021164021164e-05,
+      "loss": 0.3174,
+      "step": 521
+    },
+    {
+      "epoch": 1.8591800356506238,
+      "grad_norm": 0.19066987311728278,
+      "learning_rate": 2.10978835978836e-05,
+      "loss": 0.3134,
+      "step": 522
+    },
+    {
+      "epoch": 1.8627450980392157,
+      "grad_norm": 0.19136715218429953,
+      "learning_rate": 2.1031746031746032e-05,
+      "loss": 0.3192,
+      "step": 523
+    },
+    {
+      "epoch": 1.8663101604278074,
+      "grad_norm": 0.17741415888344655,
+      "learning_rate": 2.0965608465608465e-05,
+      "loss": 0.3292,
+      "step": 524
+    },
+    {
+      "epoch": 1.8698752228163993,
+      "grad_norm": 0.20622345716547077,
+      "learning_rate": 2.08994708994709e-05,
+      "loss": 0.3207,
+      "step": 525
+    },
+    {
+      "epoch": 1.8734402852049912,
+      "grad_norm": 0.19206052200932816,
+      "learning_rate": 2.0833333333333336e-05,
+      "loss": 0.3073,
+      "step": 526
+    },
+    {
+      "epoch": 1.8770053475935828,
+      "grad_norm": 0.18773393374361333,
+      "learning_rate": 2.076719576719577e-05,
+      "loss": 0.32,
+      "step": 527
+    },
+    {
+      "epoch": 1.8805704099821747,
+      "grad_norm": 0.18476804251865922,
+      "learning_rate": 2.07010582010582e-05,
+      "loss": 0.3178,
+      "step": 528
+    },
+    {
+      "epoch": 1.8841354723707666,
+      "grad_norm": 0.18986790460862826,
+      "learning_rate": 2.0634920634920636e-05,
+      "loss": 0.2883,
+      "step": 529
+    },
+    {
+      "epoch": 1.8877005347593583,
+      "grad_norm": 0.18203573374121396,
+      "learning_rate": 2.0568783068783072e-05,
+      "loss": 0.3158,
+      "step": 530
+    },
+    {
+      "epoch": 1.89126559714795,
+      "grad_norm": 0.2006719351083745,
+      "learning_rate": 2.0502645502645504e-05,
+      "loss": 0.3138,
+      "step": 531
+    },
+    {
+      "epoch": 1.8948306595365418,
+      "grad_norm": 0.1997495050446407,
+      "learning_rate": 2.0436507936507937e-05,
+      "loss": 0.3235,
+      "step": 532
+    },
+    {
+      "epoch": 1.8983957219251337,
+      "grad_norm": 0.2054907426815646,
+      "learning_rate": 2.037037037037037e-05,
+      "loss": 0.3126,
+      "step": 533
+    },
+    {
+      "epoch": 1.9019607843137254,
+      "grad_norm": 0.1779392814638193,
+      "learning_rate": 2.0304232804232805e-05,
+      "loss": 0.3319,
+      "step": 534
+    },
+    {
+      "epoch": 1.9055258467023173,
+      "grad_norm": 0.2167109468946905,
+      "learning_rate": 2.023809523809524e-05,
+      "loss": 0.3128,
+      "step": 535
+    },
+    {
+      "epoch": 1.9090909090909092,
+      "grad_norm": 0.20007305255831664,
+      "learning_rate": 2.0171957671957673e-05,
+      "loss": 0.3062,
+      "step": 536
+    },
+    {
+      "epoch": 1.9126559714795008,
+      "grad_norm": 0.20850480971141566,
+      "learning_rate": 2.0105820105820105e-05,
+      "loss": 0.3039,
+      "step": 537
+    },
+    {
+      "epoch": 1.9162210338680927,
+      "grad_norm": 0.21014985365028963,
+      "learning_rate": 2.003968253968254e-05,
+      "loss": 0.3102,
+      "step": 538
+    },
+    {
+      "epoch": 1.9197860962566846,
+      "grad_norm": 0.20117803870157286,
+      "learning_rate": 1.9973544973544973e-05,
+      "loss": 0.3283,
+      "step": 539
+    },
+    {
+      "epoch": 1.9233511586452763,
+      "grad_norm": 0.19908458402971088,
+      "learning_rate": 1.990740740740741e-05,
+      "loss": 0.2971,
+      "step": 540
+    },
+    {
+      "epoch": 1.926916221033868,
+      "grad_norm": 0.19568998947570748,
+      "learning_rate": 1.984126984126984e-05,
+      "loss": 0.3032,
+      "step": 541
+    },
+    {
+      "epoch": 1.93048128342246,
+      "grad_norm": 0.2007046132747606,
+      "learning_rate": 1.9775132275132277e-05,
+      "loss": 0.3266,
+      "step": 542
+    },
+    {
+      "epoch": 1.9340463458110517,
+      "grad_norm": 0.21266855942935473,
+      "learning_rate": 1.970899470899471e-05,
+      "loss": 0.307,
+      "step": 543
+    },
+    {
+      "epoch": 1.9376114081996434,
+      "grad_norm": 0.18835217135991375,
+      "learning_rate": 1.9642857142857145e-05,
+      "loss": 0.3103,
+      "step": 544
+    },
+    {
+      "epoch": 1.9411764705882353,
+      "grad_norm": 0.2041891346519106,
+      "learning_rate": 1.9576719576719577e-05,
+      "loss": 0.313,
+      "step": 545
+    },
+    {
+      "epoch": 1.9447415329768272,
+      "grad_norm": 0.21439827355853566,
+      "learning_rate": 1.9510582010582013e-05,
+      "loss": 0.3052,
+      "step": 546
+    },
+    {
+      "epoch": 1.9483065953654188,
+      "grad_norm": 0.19434036724674142,
+      "learning_rate": 1.9444444444444445e-05,
+      "loss": 0.3003,
+      "step": 547
+    },
+    {
+      "epoch": 1.9518716577540107,
+      "grad_norm": 0.20255010966835305,
+      "learning_rate": 1.9378306878306878e-05,
+      "loss": 0.3056,
+      "step": 548
+    },
+    {
+      "epoch": 1.9554367201426026,
+      "grad_norm": 0.2259647793425741,
+      "learning_rate": 1.9312169312169313e-05,
+      "loss": 0.3182,
+      "step": 549
+    },
+    {
+      "epoch": 1.9590017825311943,
+      "grad_norm": 0.19904883886246458,
+      "learning_rate": 1.924603174603175e-05,
+      "loss": 0.312,
+      "step": 550
+    },
+    {
+      "epoch": 1.962566844919786,
+      "grad_norm": 0.21291173684606893,
+      "learning_rate": 1.917989417989418e-05,
+      "loss": 0.2919,
+      "step": 551
+    },
+    {
+      "epoch": 1.966131907308378,
+      "grad_norm": 0.2178214941672347,
+      "learning_rate": 1.9113756613756614e-05,
+      "loss": 0.3048,
+      "step": 552
+    },
+    {
+      "epoch": 1.9696969696969697,
+      "grad_norm": 0.21717333547050918,
+      "learning_rate": 1.9047619047619046e-05,
+      "loss": 0.3128,
+      "step": 553
+    },
+    {
+      "epoch": 1.9732620320855614,
+      "grad_norm": 0.19133325868008694,
+      "learning_rate": 1.8981481481481482e-05,
+      "loss": 0.3069,
+      "step": 554
+    },
+    {
+      "epoch": 1.9768270944741533,
+      "grad_norm": 0.20812501603574565,
+      "learning_rate": 1.8915343915343918e-05,
+      "loss": 0.3051,
+      "step": 555
+    },
+    {
+      "epoch": 1.9803921568627452,
+      "grad_norm": 0.1918324052335252,
+      "learning_rate": 1.884920634920635e-05,
+      "loss": 0.3008,
+      "step": 556
+    },
+    {
+      "epoch": 1.9839572192513368,
+      "grad_norm": 0.1983761238735324,
+      "learning_rate": 1.8783068783068782e-05,
+      "loss": 0.309,
+      "step": 557
+    },
+    {
+      "epoch": 1.9875222816399287,
+      "grad_norm": 0.19805315575741023,
+      "learning_rate": 1.8716931216931218e-05,
+      "loss": 0.2997,
+      "step": 558
+    },
+    {
+      "epoch": 1.9910873440285206,
+      "grad_norm": 0.2017946049744733,
+      "learning_rate": 1.8650793650793654e-05,
+      "loss": 0.3133,
+      "step": 559
+    },
+    {
+      "epoch": 1.9946524064171123,
+      "grad_norm": 0.19089206072469186,
+      "learning_rate": 1.8584656084656086e-05,
+      "loss": 0.3249,
+      "step": 560
+    },
+    {
+      "epoch": 1.998217468805704,
+      "grad_norm": 0.18820984197910715,
+      "learning_rate": 1.8518518518518518e-05,
+      "loss": 0.3095,
+      "step": 561
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.27500032006878283,
+      "learning_rate": 1.8452380952380954e-05,
+      "loss": 0.2648,
+      "step": 562
+    },
+    {
+      "epoch": 2.0035650623885917,
+      "grad_norm": 0.28379119971961936,
+      "learning_rate": 1.8386243386243386e-05,
+      "loss": 0.2342,
+      "step": 563
+    },
+    {
+      "epoch": 2.0071301247771838,
+      "grad_norm": 0.22234230839094554,
+      "learning_rate": 1.8320105820105822e-05,
+      "loss": 0.2489,
+      "step": 564
+    },
+    {
+      "epoch": 2.0106951871657754,
+      "grad_norm": 0.25559758497646856,
+      "learning_rate": 1.8253968253968254e-05,
+      "loss": 0.2288,
+      "step": 565
+    },
+    {
+      "epoch": 2.014260249554367,
+      "grad_norm": 0.3241997018103919,
+      "learning_rate": 1.818783068783069e-05,
+      "loss": 0.2423,
+      "step": 566
+    },
+    {
+      "epoch": 2.017825311942959,
+      "grad_norm": 0.2307267277196982,
+      "learning_rate": 1.8121693121693122e-05,
+      "loss": 0.2367,
+      "step": 567
+    },
+    {
+      "epoch": 2.021390374331551,
+      "grad_norm": 0.22582235211301108,
+      "learning_rate": 1.8055555555555555e-05,
+      "loss": 0.2276,
+      "step": 568
+    },
+    {
+      "epoch": 2.0249554367201426,
+      "grad_norm": 0.26665536394958717,
+      "learning_rate": 1.798941798941799e-05,
+      "loss": 0.2346,
+      "step": 569
+    },
+    {
+      "epoch": 2.028520499108734,
+      "grad_norm": 0.2462916195881756,
+      "learning_rate": 1.7923280423280426e-05,
+      "loss": 0.2346,
+      "step": 570
+    },
+    {
+      "epoch": 2.0320855614973263,
+      "grad_norm": 0.23387671219442333,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 0.2408,
+      "step": 571
+    },
+    {
+      "epoch": 2.035650623885918,
+      "grad_norm": 0.23347623800125808,
+      "learning_rate": 1.779100529100529e-05,
+      "loss": 0.2159,
+      "step": 572
+    },
+    {
+      "epoch": 2.0392156862745097,
+      "grad_norm": 0.23465410615171595,
+      "learning_rate": 1.7724867724867723e-05,
+      "loss": 0.2306,
+      "step": 573
+    },
+    {
+      "epoch": 2.0427807486631018,
+      "grad_norm": 0.2140246754992095,
+      "learning_rate": 1.7658730158730162e-05,
+      "loss": 0.237,
+      "step": 574
+    },
+    {
+      "epoch": 2.0463458110516934,
+      "grad_norm": 0.2001059867340575,
+      "learning_rate": 1.7592592592592595e-05,
+      "loss": 0.2255,
+      "step": 575
+    },
+    {
+      "epoch": 2.049910873440285,
+      "grad_norm": 0.2248151672531242,
+      "learning_rate": 1.7526455026455027e-05,
+      "loss": 0.2195,
+      "step": 576
+    },
+    {
+      "epoch": 2.053475935828877,
+      "grad_norm": 0.21162463059429254,
+      "learning_rate": 1.746031746031746e-05,
+      "loss": 0.2551,
+      "step": 577
+    },
+    {
+      "epoch": 2.057040998217469,
+      "grad_norm": 0.19715146941413733,
+      "learning_rate": 1.7394179894179895e-05,
+      "loss": 0.2317,
+      "step": 578
+    },
+    {
+      "epoch": 2.0606060606060606,
+      "grad_norm": 0.19825975704264387,
+      "learning_rate": 1.732804232804233e-05,
+      "loss": 0.2234,
+      "step": 579
+    },
+    {
+      "epoch": 2.064171122994652,
+      "grad_norm": 0.20193181402375224,
+      "learning_rate": 1.7261904761904763e-05,
+      "loss": 0.2295,
+      "step": 580
+    },
+    {
+      "epoch": 2.0677361853832443,
+      "grad_norm": 0.2101761210696386,
+      "learning_rate": 1.7195767195767195e-05,
+      "loss": 0.2256,
+      "step": 581
+    },
+    {
+      "epoch": 2.071301247771836,
+      "grad_norm": 0.2268071999388428,
+      "learning_rate": 1.712962962962963e-05,
+      "loss": 0.2224,
+      "step": 582
+    },
+    {
+      "epoch": 2.0748663101604277,
+      "grad_norm": 0.20334057533400052,
+      "learning_rate": 1.7063492063492063e-05,
+      "loss": 0.2324,
+      "step": 583
+    },
+    {
+      "epoch": 2.0784313725490198,
+      "grad_norm": 0.18606476793764393,
+      "learning_rate": 1.69973544973545e-05,
+      "loss": 0.2344,
+      "step": 584
+    },
+    {
+      "epoch": 2.0819964349376114,
+      "grad_norm": 0.19512192318794336,
+      "learning_rate": 1.693121693121693e-05,
+      "loss": 0.2128,
+      "step": 585
+    },
+    {
+      "epoch": 2.085561497326203,
+      "grad_norm": 0.2056322646847476,
+      "learning_rate": 1.6865079365079367e-05,
+      "loss": 0.2438,
+      "step": 586
+    },
+    {
+      "epoch": 2.089126559714795,
+      "grad_norm": 0.18958077307675372,
+      "learning_rate": 1.67989417989418e-05,
+      "loss": 0.2347,
+      "step": 587
+    },
+    {
+      "epoch": 2.092691622103387,
+      "grad_norm": 0.18216691759050196,
+      "learning_rate": 1.6732804232804232e-05,
+      "loss": 0.2295,
+      "step": 588
+    },
+    {
+      "epoch": 2.0962566844919786,
+      "grad_norm": 0.197243755936361,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.2242,
+      "step": 589
+    },
+    {
+      "epoch": 2.09982174688057,
+      "grad_norm": 0.1940728155593685,
+      "learning_rate": 1.6600529100529103e-05,
+      "loss": 0.2401,
+      "step": 590
+    },
+    {
+      "epoch": 2.1033868092691623,
+      "grad_norm": 0.18989238084760826,
+      "learning_rate": 1.6534391534391536e-05,
+      "loss": 0.2348,
+      "step": 591
+    },
+    {
+      "epoch": 2.106951871657754,
+      "grad_norm": 0.17288669957590547,
+      "learning_rate": 1.6468253968253968e-05,
+      "loss": 0.2144,
+      "step": 592
+    },
+    {
+      "epoch": 2.1105169340463457,
+      "grad_norm": 0.20050854786832728,
+      "learning_rate": 1.6402116402116404e-05,
+      "loss": 0.2342,
+      "step": 593
+    },
+    {
+      "epoch": 2.1140819964349378,
+      "grad_norm": 0.18413076268074716,
+      "learning_rate": 1.633597883597884e-05,
+      "loss": 0.2281,
+      "step": 594
+    },
+    {
+      "epoch": 2.1176470588235294,
+      "grad_norm": 0.180802862803312,
+      "learning_rate": 1.626984126984127e-05,
+      "loss": 0.2326,
+      "step": 595
+    },
+    {
+      "epoch": 2.121212121212121,
+      "grad_norm": 0.18101245554730538,
+      "learning_rate": 1.6203703703703704e-05,
+      "loss": 0.2172,
+      "step": 596
+    },
+    {
+      "epoch": 2.124777183600713,
+      "grad_norm": 0.17814013076760596,
+      "learning_rate": 1.6137566137566136e-05,
+      "loss": 0.235,
+      "step": 597
+    },
+    {
+      "epoch": 2.128342245989305,
+      "grad_norm": 0.1872307915319883,
+      "learning_rate": 1.6071428571428572e-05,
+      "loss": 0.2278,
+      "step": 598
+    },
+    {
+      "epoch": 2.1319073083778965,
+      "grad_norm": 0.1748629321758084,
+      "learning_rate": 1.6005291005291008e-05,
+      "loss": 0.2215,
+      "step": 599
+    },
+    {
+      "epoch": 2.135472370766488,
+      "grad_norm": 0.21611689518109178,
+      "learning_rate": 1.593915343915344e-05,
+      "loss": 0.2496,
+      "step": 600
+    },
+    {
+      "epoch": 2.1390374331550803,
+      "grad_norm": 0.2015860874321857,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 0.2411,
+      "step": 601
+    },
+    {
+      "epoch": 2.142602495543672,
+      "grad_norm": 0.18080973539415182,
+      "learning_rate": 1.5806878306878308e-05,
+      "loss": 0.2289,
+      "step": 602
+    },
+    {
+      "epoch": 2.1461675579322637,
+      "grad_norm": 0.1807817635387203,
+      "learning_rate": 1.574074074074074e-05,
+      "loss": 0.2391,
+      "step": 603
+    },
+    {
+      "epoch": 2.1497326203208558,
+      "grad_norm": 0.1860516054659944,
+      "learning_rate": 1.5674603174603176e-05,
+      "loss": 0.2383,
+      "step": 604
+    },
+    {
+      "epoch": 2.1532976827094474,
+      "grad_norm": 0.18914854384049476,
+      "learning_rate": 1.560846560846561e-05,
+      "loss": 0.2303,
+      "step": 605
+    },
+    {
+      "epoch": 2.156862745098039,
+      "grad_norm": 0.19383086056215917,
+      "learning_rate": 1.5542328042328044e-05,
+      "loss": 0.2375,
+      "step": 606
+    },
+    {
+      "epoch": 2.160427807486631,
+      "grad_norm": 0.19877678945645497,
+      "learning_rate": 1.5476190476190476e-05,
+      "loss": 0.2279,
+      "step": 607
+    },
+    {
+      "epoch": 2.163992869875223,
+      "grad_norm": 0.19683898126517263,
+      "learning_rate": 1.5410052910052912e-05,
+      "loss": 0.2281,
+      "step": 608
+    },
+    {
+      "epoch": 2.1675579322638145,
+      "grad_norm": 0.1935193505959923,
+      "learning_rate": 1.5343915343915344e-05,
+      "loss": 0.2335,
+      "step": 609
+    },
+    {
+      "epoch": 2.171122994652406,
+      "grad_norm": 0.1941104079122024,
+      "learning_rate": 1.527777777777778e-05,
+      "loss": 0.2369,
+      "step": 610
+    },
+    {
+      "epoch": 2.1746880570409983,
+      "grad_norm": 0.1994900091106579,
+      "learning_rate": 1.5211640211640213e-05,
+      "loss": 0.2315,
+      "step": 611
+    },
+    {
+      "epoch": 2.17825311942959,
+      "grad_norm": 0.18942474892391648,
+      "learning_rate": 1.5145502645502647e-05,
+      "loss": 0.2322,
+      "step": 612
+    },
+    {
+      "epoch": 2.1818181818181817,
+      "grad_norm": 0.20990059711006312,
+      "learning_rate": 1.5079365079365079e-05,
+      "loss": 0.2361,
+      "step": 613
+    },
+    {
+      "epoch": 2.1853832442067738,
+      "grad_norm": 0.19177453740511974,
+      "learning_rate": 1.5013227513227515e-05,
+      "loss": 0.2268,
+      "step": 614
+    },
+    {
+      "epoch": 2.1889483065953654,
+      "grad_norm": 0.19741258199289288,
+      "learning_rate": 1.4947089947089949e-05,
+      "loss": 0.2379,
+      "step": 615
+    },
+    {
+      "epoch": 2.192513368983957,
+      "grad_norm": 0.1971100624793935,
+      "learning_rate": 1.4880952380952381e-05,
+      "loss": 0.2261,
+      "step": 616
+    },
+    {
+      "epoch": 2.196078431372549,
+      "grad_norm": 0.19296227818553785,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 0.2369,
+      "step": 617
+    },
+    {
+      "epoch": 2.199643493761141,
+      "grad_norm": 0.20466864855607125,
+      "learning_rate": 1.474867724867725e-05,
+      "loss": 0.2291,
+      "step": 618
+    },
+    {
+      "epoch": 2.2032085561497325,
+      "grad_norm": 0.19770566274237758,
+      "learning_rate": 1.4682539682539683e-05,
+      "loss": 0.2335,
+      "step": 619
+    },
+    {
+      "epoch": 2.2067736185383247,
+      "grad_norm": 0.17832271484577494,
+      "learning_rate": 1.4616402116402117e-05,
+      "loss": 0.2187,
+      "step": 620
+    },
+    {
+      "epoch": 2.2103386809269163,
+      "grad_norm": 0.18519127772966246,
+      "learning_rate": 1.455026455026455e-05,
+      "loss": 0.2284,
+      "step": 621
+    },
+    {
+      "epoch": 2.213903743315508,
+      "grad_norm": 0.19481366379015339,
+      "learning_rate": 1.4484126984126987e-05,
+      "loss": 0.2372,
+      "step": 622
+    },
+    {
+      "epoch": 2.2174688057040997,
+      "grad_norm": 0.18999585728264068,
+      "learning_rate": 1.4417989417989419e-05,
+      "loss": 0.2337,
+      "step": 623
+    },
+    {
+      "epoch": 2.2210338680926918,
+      "grad_norm": 0.1806066577870848,
+      "learning_rate": 1.4351851851851853e-05,
+      "loss": 0.2316,
+      "step": 624
+    },
+    {
+      "epoch": 2.2245989304812834,
+      "grad_norm": 0.18216417210880095,
+      "learning_rate": 1.4285714285714285e-05,
+      "loss": 0.2201,
+      "step": 625
+    },
+    {
+      "epoch": 2.228163992869875,
+      "grad_norm": 0.18514557397244763,
+      "learning_rate": 1.4219576719576721e-05,
+      "loss": 0.2278,
+      "step": 626
+    },
+    {
+      "epoch": 2.231729055258467,
+      "grad_norm": 0.19198609211235954,
+      "learning_rate": 1.4153439153439155e-05,
+      "loss": 0.2434,
+      "step": 627
+    },
+    {
+      "epoch": 2.235294117647059,
+      "grad_norm": 0.1924100926118934,
+      "learning_rate": 1.4087301587301587e-05,
+      "loss": 0.2433,
+      "step": 628
+    },
+    {
+      "epoch": 2.2388591800356505,
+      "grad_norm": 0.20382617966737793,
+      "learning_rate": 1.4021164021164022e-05,
+      "loss": 0.2374,
+      "step": 629
+    },
+    {
+      "epoch": 2.242424242424242,
+      "grad_norm": 0.18386300942877712,
+      "learning_rate": 1.3955026455026457e-05,
+      "loss": 0.2325,
+      "step": 630
+    },
+    {
+      "epoch": 2.2459893048128343,
+      "grad_norm": 0.19259119093828594,
+      "learning_rate": 1.388888888888889e-05,
+      "loss": 0.2362,
+      "step": 631
+    },
+    {
+      "epoch": 2.249554367201426,
+      "grad_norm": 0.18247922659221158,
+      "learning_rate": 1.3822751322751324e-05,
+      "loss": 0.2282,
+      "step": 632
+    },
+    {
+      "epoch": 2.2531194295900177,
+      "grad_norm": 0.20248769627119753,
+      "learning_rate": 1.3756613756613756e-05,
+      "loss": 0.2523,
+      "step": 633
+    },
+    {
+      "epoch": 2.2566844919786098,
+      "grad_norm": 0.19445298221569365,
+      "learning_rate": 1.3690476190476192e-05,
+      "loss": 0.2173,
+      "step": 634
+    },
+    {
+      "epoch": 2.2602495543672014,
+      "grad_norm": 0.18954404113743362,
+      "learning_rate": 1.3624338624338626e-05,
+      "loss": 0.2321,
+      "step": 635
+    },
+    {
+      "epoch": 2.263814616755793,
+      "grad_norm": 0.20547864680481315,
+      "learning_rate": 1.3558201058201058e-05,
+      "loss": 0.2309,
+      "step": 636
+    },
+    {
+      "epoch": 2.267379679144385,
+      "grad_norm": 0.192064923010938,
+      "learning_rate": 1.3492063492063492e-05,
+      "loss": 0.2269,
+      "step": 637
+    },
+    {
+      "epoch": 2.270944741532977,
+      "grad_norm": 0.1930203979106369,
+      "learning_rate": 1.3425925925925928e-05,
+      "loss": 0.237,
+      "step": 638
+    },
+    {
+      "epoch": 2.2745098039215685,
+      "grad_norm": 0.1894686830595722,
+      "learning_rate": 1.3359788359788362e-05,
+      "loss": 0.2309,
+      "step": 639
+    },
+    {
+      "epoch": 2.2780748663101607,
+      "grad_norm": 0.1821009316469466,
+      "learning_rate": 1.3293650793650794e-05,
+      "loss": 0.229,
+      "step": 640
+    },
+    {
+      "epoch": 2.2816399286987523,
+      "grad_norm": 0.19423021723183728,
+      "learning_rate": 1.3227513227513228e-05,
+      "loss": 0.2287,
+      "step": 641
+    },
+    {
+      "epoch": 2.285204991087344,
+      "grad_norm": 0.1785078372370465,
+      "learning_rate": 1.3161375661375664e-05,
+      "loss": 0.2222,
+      "step": 642
+    },
+    {
+      "epoch": 2.2887700534759357,
+      "grad_norm": 0.17397995006953645,
+      "learning_rate": 1.3095238095238096e-05,
+      "loss": 0.2165,
+      "step": 643
+    },
+    {
+      "epoch": 2.2923351158645278,
+      "grad_norm": 0.18005696399999194,
+      "learning_rate": 1.302910052910053e-05,
+      "loss": 0.2284,
+      "step": 644
+    },
+    {
+      "epoch": 2.2959001782531194,
+      "grad_norm": 0.19853073168135904,
+      "learning_rate": 1.2962962962962962e-05,
+      "loss": 0.2321,
+      "step": 645
+    },
+    {
+      "epoch": 2.299465240641711,
+      "grad_norm": 0.18851558124740694,
+      "learning_rate": 1.2896825396825398e-05,
+      "loss": 0.2413,
+      "step": 646
+    },
+    {
+      "epoch": 2.303030303030303,
+      "grad_norm": 0.18597406182344162,
+      "learning_rate": 1.2830687830687832e-05,
+      "loss": 0.238,
+      "step": 647
+    },
+    {
+      "epoch": 2.306595365418895,
+      "grad_norm": 0.2457727085124803,
+      "learning_rate": 1.2764550264550265e-05,
+      "loss": 0.2199,
+      "step": 648
+    },
+    {
+      "epoch": 2.3101604278074865,
+      "grad_norm": 0.1899712262722189,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 0.2309,
+      "step": 649
+    },
+    {
+      "epoch": 2.313725490196078,
+      "grad_norm": 0.1866545396554979,
+      "learning_rate": 1.2632275132275134e-05,
+      "loss": 0.2381,
+      "step": 650
+    },
+    {
+      "epoch": 2.3172905525846703,
+      "grad_norm": 0.18350208668293394,
+      "learning_rate": 1.2566137566137568e-05,
+      "loss": 0.216,
+      "step": 651
+    },
+    {
+      "epoch": 2.320855614973262,
+      "grad_norm": 0.1929901716297844,
+      "learning_rate": 1.25e-05,
+      "loss": 0.2369,
+      "step": 652
+    },
+    {
+      "epoch": 2.3244206773618536,
+      "grad_norm": 0.19946599616279062,
+      "learning_rate": 1.2433862433862433e-05,
+      "loss": 0.222,
+      "step": 653
+    },
+    {
+      "epoch": 2.3279857397504458,
+      "grad_norm": 0.18164823410250522,
+      "learning_rate": 1.2367724867724869e-05,
+      "loss": 0.2389,
+      "step": 654
+    },
+    {
+      "epoch": 2.3315508021390374,
+      "grad_norm": 0.18649337080693953,
+      "learning_rate": 1.2301587301587301e-05,
+      "loss": 0.218,
+      "step": 655
+    },
+    {
+      "epoch": 2.335115864527629,
+      "grad_norm": 0.18955525609147536,
+      "learning_rate": 1.2235449735449737e-05,
+      "loss": 0.2288,
+      "step": 656
+    },
+    {
+      "epoch": 2.338680926916221,
+      "grad_norm": 0.17951272974302915,
+      "learning_rate": 1.2169312169312169e-05,
+      "loss": 0.2234,
+      "step": 657
+    },
+    {
+      "epoch": 2.342245989304813,
+      "grad_norm": 0.20214921531949875,
+      "learning_rate": 1.2103174603174603e-05,
+      "loss": 0.2466,
+      "step": 658
+    },
+    {
+      "epoch": 2.3458110516934045,
+      "grad_norm": 0.19427765538373828,
+      "learning_rate": 1.2037037037037037e-05,
+      "loss": 0.2237,
+      "step": 659
+    },
+    {
+      "epoch": 2.3493761140819966,
+      "grad_norm": 0.20221747320603128,
+      "learning_rate": 1.1970899470899471e-05,
+      "loss": 0.2268,
+      "step": 660
+    },
+    {
+      "epoch": 2.3529411764705883,
+      "grad_norm": 0.18693268622521497,
+      "learning_rate": 1.1904761904761905e-05,
+      "loss": 0.2335,
+      "step": 661
+    },
+    {
+      "epoch": 2.35650623885918,
+      "grad_norm": 0.18231702039609785,
+      "learning_rate": 1.1838624338624339e-05,
+      "loss": 0.23,
+      "step": 662
+    },
+    {
+      "epoch": 2.3600713012477716,
+      "grad_norm": 0.19398201699865386,
+      "learning_rate": 1.1772486772486773e-05,
+      "loss": 0.2291,
+      "step": 663
+    },
+    {
+      "epoch": 2.3636363636363638,
+      "grad_norm": 0.20413236489622605,
+      "learning_rate": 1.1706349206349207e-05,
+      "loss": 0.2266,
+      "step": 664
+    },
+    {
+      "epoch": 2.3672014260249554,
+      "grad_norm": 0.17733320869365748,
+      "learning_rate": 1.164021164021164e-05,
+      "loss": 0.2227,
+      "step": 665
+    },
+    {
+      "epoch": 2.370766488413547,
+      "grad_norm": 0.19780332527386077,
+      "learning_rate": 1.1574074074074075e-05,
+      "loss": 0.2391,
+      "step": 666
+    },
+    {
+      "epoch": 2.374331550802139,
+      "grad_norm": 0.2279342537254599,
+      "learning_rate": 1.1507936507936508e-05,
+      "loss": 0.2323,
+      "step": 667
+    },
+    {
+      "epoch": 2.377896613190731,
+      "grad_norm": 0.1904659111734331,
+      "learning_rate": 1.1441798941798943e-05,
+      "loss": 0.2184,
+      "step": 668
+    },
+    {
+      "epoch": 2.3814616755793225,
+      "grad_norm": 0.19712647015899476,
+      "learning_rate": 1.1375661375661376e-05,
+      "loss": 0.2407,
+      "step": 669
+    },
+    {
+      "epoch": 2.385026737967914,
+      "grad_norm": 0.1869566048963991,
+      "learning_rate": 1.130952380952381e-05,
+      "loss": 0.2274,
+      "step": 670
+    },
+    {
+      "epoch": 2.3885918003565063,
+      "grad_norm": 0.20468476024619,
+      "learning_rate": 1.1243386243386244e-05,
+      "loss": 0.2348,
+      "step": 671
+    },
+    {
+      "epoch": 2.392156862745098,
+      "grad_norm": 0.19307424272825005,
+      "learning_rate": 1.1177248677248678e-05,
+      "loss": 0.2189,
+      "step": 672
+    },
+    {
+      "epoch": 2.3957219251336896,
+      "grad_norm": 0.18073077136063678,
+      "learning_rate": 1.1111111111111112e-05,
+      "loss": 0.235,
+      "step": 673
+    },
+    {
+      "epoch": 2.3992869875222818,
+      "grad_norm": 0.1828713899435911,
+      "learning_rate": 1.1044973544973546e-05,
+      "loss": 0.2225,
+      "step": 674
+    },
+    {
+      "epoch": 2.4028520499108734,
+      "grad_norm": 0.18598213054968707,
+      "learning_rate": 1.0978835978835978e-05,
+      "loss": 0.2403,
+      "step": 675
+    },
+    {
+      "epoch": 2.406417112299465,
+      "grad_norm": 0.19662326611067624,
+      "learning_rate": 1.0912698412698414e-05,
+      "loss": 0.2344,
+      "step": 676
+    },
+    {
+      "epoch": 2.409982174688057,
+      "grad_norm": 0.187200607930686,
+      "learning_rate": 1.0846560846560846e-05,
+      "loss": 0.225,
+      "step": 677
+    },
+    {
+      "epoch": 2.413547237076649,
+      "grad_norm": 0.1910800131613057,
+      "learning_rate": 1.0780423280423282e-05,
+      "loss": 0.2265,
+      "step": 678
+    },
+    {
+      "epoch": 2.4171122994652405,
+      "grad_norm": 0.17157478358791453,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 0.2149,
+      "step": 679
+    },
+    {
+      "epoch": 2.4206773618538326,
+      "grad_norm": 0.188282022639635,
+      "learning_rate": 1.0648148148148148e-05,
+      "loss": 0.2246,
+      "step": 680
+    },
+    {
+      "epoch": 2.4242424242424243,
+      "grad_norm": 0.20798389109565102,
+      "learning_rate": 1.0582010582010582e-05,
+      "loss": 0.2214,
+      "step": 681
+    },
+    {
+      "epoch": 2.427807486631016,
+      "grad_norm": 0.1904296557971635,
+      "learning_rate": 1.0515873015873016e-05,
+      "loss": 0.2407,
+      "step": 682
+    },
+    {
+      "epoch": 2.431372549019608,
+      "grad_norm": 0.19378904601861913,
+      "learning_rate": 1.044973544973545e-05,
+      "loss": 0.2337,
+      "step": 683
+    },
+    {
+      "epoch": 2.4349376114081998,
+      "grad_norm": 0.18423918729494884,
+      "learning_rate": 1.0383597883597884e-05,
+      "loss": 0.226,
+      "step": 684
+    },
+    {
+      "epoch": 2.4385026737967914,
+      "grad_norm": 0.19267681605314207,
+      "learning_rate": 1.0317460317460318e-05,
+      "loss": 0.223,
+      "step": 685
+    },
+    {
+      "epoch": 2.442067736185383,
+      "grad_norm": 0.18559552068387486,
+      "learning_rate": 1.0251322751322752e-05,
+      "loss": 0.2347,
+      "step": 686
+    },
+    {
+      "epoch": 2.445632798573975,
+      "grad_norm": 0.17806027872902971,
+      "learning_rate": 1.0185185185185185e-05,
+      "loss": 0.2222,
+      "step": 687
+    },
+    {
+      "epoch": 2.449197860962567,
+      "grad_norm": 0.17485815448826222,
+      "learning_rate": 1.011904761904762e-05,
+      "loss": 0.2252,
+      "step": 688
+    },
+    {
+      "epoch": 2.4527629233511585,
+      "grad_norm": 0.18540229933329072,
+      "learning_rate": 1.0052910052910053e-05,
+      "loss": 0.2392,
+      "step": 689
+    },
+    {
+      "epoch": 2.45632798573975,
+      "grad_norm": 0.1767807593133908,
+      "learning_rate": 9.986772486772487e-06,
+      "loss": 0.2389,
+      "step": 690
+    },
+    {
+      "epoch": 2.4598930481283423,
+      "grad_norm": 0.18075588403612386,
+      "learning_rate": 9.92063492063492e-06,
+      "loss": 0.224,
+      "step": 691
+    },
+    {
+      "epoch": 2.463458110516934,
+      "grad_norm": 0.17959964243921345,
+      "learning_rate": 9.854497354497355e-06,
+      "loss": 0.2317,
+      "step": 692
+    },
+    {
+      "epoch": 2.4670231729055256,
+      "grad_norm": 0.20065335900106432,
+      "learning_rate": 9.788359788359789e-06,
+      "loss": 0.2399,
+      "step": 693
+    },
+    {
+      "epoch": 2.4705882352941178,
+      "grad_norm": 0.1786894123766744,
+      "learning_rate": 9.722222222222223e-06,
+      "loss": 0.2246,
+      "step": 694
+    },
+    {
+      "epoch": 2.4741532976827094,
+      "grad_norm": 0.17638835052251736,
+      "learning_rate": 9.656084656084657e-06,
+      "loss": 0.2153,
+      "step": 695
+    },
+    {
+      "epoch": 2.477718360071301,
+      "grad_norm": 0.1868836117931924,
+      "learning_rate": 9.58994708994709e-06,
+      "loss": 0.2299,
+      "step": 696
+    },
+    {
+      "epoch": 2.481283422459893,
+      "grad_norm": 0.18982425969111952,
+      "learning_rate": 9.523809523809523e-06,
+      "loss": 0.2374,
+      "step": 697
+    },
+    {
+      "epoch": 2.484848484848485,
+      "grad_norm": 0.17609185628718083,
+      "learning_rate": 9.457671957671959e-06,
+      "loss": 0.2243,
+      "step": 698
+    },
+    {
+      "epoch": 2.4884135472370765,
+      "grad_norm": 0.17499571402448216,
+      "learning_rate": 9.391534391534391e-06,
+      "loss": 0.2382,
+      "step": 699
+    },
+    {
+      "epoch": 2.4919786096256686,
+      "grad_norm": 0.18360505899213056,
+      "learning_rate": 9.325396825396827e-06,
+      "loss": 0.2384,
+      "step": 700
+    },
+    {
+      "epoch": 2.4955436720142603,
+      "grad_norm": 0.18007821788634207,
+      "learning_rate": 9.259259259259259e-06,
+      "loss": 0.2309,
+      "step": 701
+    },
+    {
+      "epoch": 2.499108734402852,
+      "grad_norm": 0.18051789194339127,
+      "learning_rate": 9.193121693121693e-06,
+      "loss": 0.2395,
+      "step": 702
+    },
+    {
+      "epoch": 2.502673796791444,
+      "grad_norm": 0.18082385991467867,
+      "learning_rate": 9.126984126984127e-06,
+      "loss": 0.2274,
+      "step": 703
+    },
+    {
+      "epoch": 2.5062388591800357,
+      "grad_norm": 0.1798947221376965,
+      "learning_rate": 9.060846560846561e-06,
+      "loss": 0.2356,
+      "step": 704
+    },
+    {
+      "epoch": 2.5098039215686274,
+      "grad_norm": 0.1816759855580067,
+      "learning_rate": 8.994708994708995e-06,
+      "loss": 0.2314,
+      "step": 705
+    },
+    {
+      "epoch": 2.5133689839572195,
+      "grad_norm": 0.18998422988985153,
+      "learning_rate": 8.92857142857143e-06,
+      "loss": 0.225,
+      "step": 706
+    },
+    {
+      "epoch": 2.516934046345811,
+      "grad_norm": 0.18922430429233833,
+      "learning_rate": 8.862433862433862e-06,
+      "loss": 0.2228,
+      "step": 707
+    },
+    {
+      "epoch": 2.520499108734403,
+      "grad_norm": 0.17677242790457534,
+      "learning_rate": 8.796296296296297e-06,
+      "loss": 0.2262,
+      "step": 708
+    },
+    {
+      "epoch": 2.5240641711229945,
+      "grad_norm": 0.18634776553043544,
+      "learning_rate": 8.73015873015873e-06,
+      "loss": 0.249,
+      "step": 709
+    },
+    {
+      "epoch": 2.527629233511586,
+      "grad_norm": 0.18215688566739707,
+      "learning_rate": 8.664021164021165e-06,
+      "loss": 0.2294,
+      "step": 710
+    },
+    {
+      "epoch": 2.5311942959001783,
+      "grad_norm": 0.1782300566233256,
+      "learning_rate": 8.597883597883598e-06,
+      "loss": 0.2245,
+      "step": 711
+    },
+    {
+      "epoch": 2.53475935828877,
+      "grad_norm": 0.18404520117414028,
+      "learning_rate": 8.531746031746032e-06,
+      "loss": 0.21,
+      "step": 712
+    },
+    {
+      "epoch": 2.5383244206773616,
+      "grad_norm": 0.1826690498337214,
+      "learning_rate": 8.465608465608466e-06,
+      "loss": 0.2236,
+      "step": 713
+    },
+    {
+      "epoch": 2.5418894830659537,
+      "grad_norm": 0.18651903771948233,
+      "learning_rate": 8.3994708994709e-06,
+      "loss": 0.2245,
+      "step": 714
+    },
+    {
+      "epoch": 2.5454545454545454,
+      "grad_norm": 0.17678957016621794,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 0.2301,
+      "step": 715
+    },
+    {
+      "epoch": 2.549019607843137,
+      "grad_norm": 0.1742638923785468,
+      "learning_rate": 8.267195767195768e-06,
+      "loss": 0.2235,
+      "step": 716
+    },
+    {
+      "epoch": 2.552584670231729,
+      "grad_norm": 0.17641249713440432,
+      "learning_rate": 8.201058201058202e-06,
+      "loss": 0.2236,
+      "step": 717
+    },
+    {
+      "epoch": 2.556149732620321,
+      "grad_norm": 0.1824073213389634,
+      "learning_rate": 8.134920634920636e-06,
+      "loss": 0.2291,
+      "step": 718
+    },
+    {
+      "epoch": 2.5597147950089125,
+      "grad_norm": 0.199949423716211,
+      "learning_rate": 8.068783068783068e-06,
+      "loss": 0.2461,
+      "step": 719
+    },
+    {
+      "epoch": 2.5632798573975046,
+      "grad_norm": 0.2100007813304115,
+      "learning_rate": 8.002645502645504e-06,
+      "loss": 0.2374,
+      "step": 720
+    },
+    {
+      "epoch": 2.5668449197860963,
+      "grad_norm": 0.1835626211674465,
+      "learning_rate": 7.936507936507936e-06,
+      "loss": 0.2292,
+      "step": 721
+    },
+    {
+      "epoch": 2.570409982174688,
+      "grad_norm": 0.1938682465133642,
+      "learning_rate": 7.87037037037037e-06,
+      "loss": 0.2243,
+      "step": 722
+    },
+    {
+      "epoch": 2.57397504456328,
+      "grad_norm": 0.17734835088949738,
+      "learning_rate": 7.804232804232804e-06,
+      "loss": 0.2167,
+      "step": 723
+    },
+    {
+      "epoch": 2.5775401069518717,
+      "grad_norm": 0.17686788914695675,
+      "learning_rate": 7.738095238095238e-06,
+      "loss": 0.2318,
+      "step": 724
+    },
+    {
+      "epoch": 2.5811051693404634,
+      "grad_norm": 0.1851354749806612,
+      "learning_rate": 7.671957671957672e-06,
+      "loss": 0.2321,
+      "step": 725
+    },
+    {
+      "epoch": 2.5846702317290555,
+      "grad_norm": 0.17820285318111276,
+      "learning_rate": 7.605820105820106e-06,
+      "loss": 0.2179,
+      "step": 726
+    },
+    {
+      "epoch": 2.588235294117647,
+      "grad_norm": 0.18201985881703542,
+      "learning_rate": 7.5396825396825394e-06,
+      "loss": 0.2274,
+      "step": 727
+    },
+    {
+      "epoch": 2.591800356506239,
+      "grad_norm": 0.18531974862744746,
+      "learning_rate": 7.473544973544974e-06,
+      "loss": 0.2322,
+      "step": 728
+    },
+    {
+      "epoch": 2.5953654188948305,
+      "grad_norm": 0.19136900804472998,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 0.235,
+      "step": 729
+    },
+    {
+      "epoch": 2.598930481283422,
+      "grad_norm": 0.17410574828341133,
+      "learning_rate": 7.3412698412698415e-06,
+      "loss": 0.2356,
+      "step": 730
+    },
+    {
+      "epoch": 2.6024955436720143,
+      "grad_norm": 0.18741154352061531,
+      "learning_rate": 7.275132275132275e-06,
+      "loss": 0.248,
+      "step": 731
+    },
+    {
+      "epoch": 2.606060606060606,
+      "grad_norm": 0.1830694230341296,
+      "learning_rate": 7.2089947089947095e-06,
+      "loss": 0.2349,
+      "step": 732
+    },
+    {
+      "epoch": 2.6096256684491976,
+      "grad_norm": 0.17591609983544185,
+      "learning_rate": 7.142857142857143e-06,
+      "loss": 0.2353,
+      "step": 733
+    },
+    {
+      "epoch": 2.6131907308377897,
+      "grad_norm": 0.17637526754601035,
+      "learning_rate": 7.076719576719578e-06,
+      "loss": 0.2379,
+      "step": 734
+    },
+    {
+      "epoch": 2.6167557932263814,
+      "grad_norm": 0.1736861744114466,
+      "learning_rate": 7.010582010582011e-06,
+      "loss": 0.2266,
+      "step": 735
+    },
+    {
+      "epoch": 2.620320855614973,
+      "grad_norm": 0.1730002572927092,
+      "learning_rate": 6.944444444444445e-06,
+      "loss": 0.2364,
+      "step": 736
+    },
+    {
+      "epoch": 2.623885918003565,
+      "grad_norm": 0.18970493987654385,
+      "learning_rate": 6.878306878306878e-06,
+      "loss": 0.2348,
+      "step": 737
+    },
+    {
+      "epoch": 2.627450980392157,
+      "grad_norm": 0.17489351524441868,
+      "learning_rate": 6.812169312169313e-06,
+      "loss": 0.2384,
+      "step": 738
+    },
+    {
+      "epoch": 2.6310160427807485,
+      "grad_norm": 0.17785149205174128,
+      "learning_rate": 6.746031746031746e-06,
+      "loss": 0.2365,
+      "step": 739
+    },
+    {
+      "epoch": 2.6345811051693406,
+      "grad_norm": 0.1756268888277164,
+      "learning_rate": 6.679894179894181e-06,
+      "loss": 0.2212,
+      "step": 740
+    },
+    {
+      "epoch": 2.6381461675579323,
+      "grad_norm": 0.1744771538591329,
+      "learning_rate": 6.613756613756614e-06,
+      "loss": 0.2299,
+      "step": 741
+    },
+    {
+      "epoch": 2.641711229946524,
+      "grad_norm": 0.17321339635582686,
+      "learning_rate": 6.547619047619048e-06,
+      "loss": 0.2355,
+      "step": 742
+    },
+    {
+      "epoch": 2.645276292335116,
+      "grad_norm": 0.18115043073224243,
+      "learning_rate": 6.481481481481481e-06,
+      "loss": 0.2465,
+      "step": 743
+    },
+    {
+      "epoch": 2.6488413547237077,
+      "grad_norm": 0.17181367719425075,
+      "learning_rate": 6.415343915343916e-06,
+      "loss": 0.2257,
+      "step": 744
+    },
+    {
+      "epoch": 2.6524064171122994,
+      "grad_norm": 0.16664562195105662,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 0.2255,
+      "step": 745
+    },
+    {
+      "epoch": 2.6559714795008915,
+      "grad_norm": 0.17495096235725907,
+      "learning_rate": 6.283068783068784e-06,
+      "loss": 0.2224,
+      "step": 746
+    },
+    {
+      "epoch": 2.659536541889483,
+      "grad_norm": 0.1694911456148541,
+      "learning_rate": 6.2169312169312165e-06,
+      "loss": 0.2139,
+      "step": 747
+    },
+    {
+      "epoch": 2.663101604278075,
+      "grad_norm": 0.17164539816907423,
+      "learning_rate": 6.1507936507936505e-06,
+      "loss": 0.2339,
+      "step": 748
+    },
+    {
+      "epoch": 2.6666666666666665,
+      "grad_norm": 0.17118965281305873,
+      "learning_rate": 6.0846560846560845e-06,
+      "loss": 0.2287,
+      "step": 749
+    },
+    {
+      "epoch": 2.6702317290552586,
+      "grad_norm": 0.16911804540349315,
+      "learning_rate": 6.0185185185185185e-06,
+      "loss": 0.2248,
+      "step": 750
+    },
+    {
+      "epoch": 2.6737967914438503,
+      "grad_norm": 0.17417098754204538,
+      "learning_rate": 5.9523809523809525e-06,
+      "loss": 0.2446,
+      "step": 751
+    },
+    {
+      "epoch": 2.677361853832442,
+      "grad_norm": 0.17866864349344216,
+      "learning_rate": 5.8862433862433866e-06,
+      "loss": 0.2262,
+      "step": 752
+    },
+    {
+      "epoch": 2.6809269162210336,
+      "grad_norm": 0.17050072413278544,
+      "learning_rate": 5.82010582010582e-06,
+      "loss": 0.2288,
+      "step": 753
+    },
+    {
+      "epoch": 2.6844919786096257,
+      "grad_norm": 0.17851130240663474,
+      "learning_rate": 5.753968253968254e-06,
+      "loss": 0.2329,
+      "step": 754
+    },
+    {
+      "epoch": 2.6880570409982174,
+      "grad_norm": 0.1638316108509075,
+      "learning_rate": 5.687830687830688e-06,
+      "loss": 0.2212,
+      "step": 755
+    },
+    {
+      "epoch": 2.691622103386809,
+      "grad_norm": 0.17779212830289176,
+      "learning_rate": 5.621693121693122e-06,
+      "loss": 0.2425,
+      "step": 756
+    },
+    {
+      "epoch": 2.695187165775401,
+      "grad_norm": 0.17185330301412938,
+      "learning_rate": 5.555555555555556e-06,
+      "loss": 0.2285,
+      "step": 757
+    },
+    {
+      "epoch": 2.698752228163993,
+      "grad_norm": 0.17868847113604616,
+      "learning_rate": 5.489417989417989e-06,
+      "loss": 0.2434,
+      "step": 758
+    },
+    {
+      "epoch": 2.7023172905525845,
+      "grad_norm": 0.18296634350156232,
+      "learning_rate": 5.423280423280423e-06,
+      "loss": 0.2499,
+      "step": 759
+    },
+    {
+      "epoch": 2.7058823529411766,
+      "grad_norm": 0.17210163057924824,
+      "learning_rate": 5.357142857142857e-06,
+      "loss": 0.2286,
+      "step": 760
+    },
+    {
+      "epoch": 2.7094474153297683,
+      "grad_norm": 0.18760403668715556,
+      "learning_rate": 5.291005291005291e-06,
+      "loss": 0.2349,
+      "step": 761
+    },
+    {
+      "epoch": 2.71301247771836,
+      "grad_norm": 0.18012917731310804,
+      "learning_rate": 5.224867724867725e-06,
+      "loss": 0.2387,
+      "step": 762
+    },
+    {
+      "epoch": 2.716577540106952,
+      "grad_norm": 0.17564038855983413,
+      "learning_rate": 5.158730158730159e-06,
+      "loss": 0.2246,
+      "step": 763
+    },
+    {
+      "epoch": 2.7201426024955437,
+      "grad_norm": 0.17515833992080637,
+      "learning_rate": 5.092592592592592e-06,
+      "loss": 0.226,
+      "step": 764
+    },
+    {
+      "epoch": 2.7237076648841354,
+      "grad_norm": 0.18476191389936114,
+      "learning_rate": 5.026455026455026e-06,
+      "loss": 0.2315,
+      "step": 765
+    },
+    {
+      "epoch": 2.7272727272727275,
+      "grad_norm": 0.1846230143988715,
+      "learning_rate": 4.96031746031746e-06,
+      "loss": 0.2262,
+      "step": 766
+    },
+    {
+      "epoch": 2.730837789661319,
+      "grad_norm": 0.18269261643153004,
+      "learning_rate": 4.894179894179894e-06,
+      "loss": 0.2337,
+      "step": 767
+    },
+    {
+      "epoch": 2.734402852049911,
+      "grad_norm": 0.1749260045035671,
+      "learning_rate": 4.828042328042328e-06,
+      "loss": 0.2197,
+      "step": 768
+    },
+    {
+      "epoch": 2.7379679144385025,
+      "grad_norm": 0.1731387779253303,
+      "learning_rate": 4.7619047619047615e-06,
+      "loss": 0.2179,
+      "step": 769
+    },
+    {
+      "epoch": 2.7415329768270946,
+      "grad_norm": 0.17405193437146413,
+      "learning_rate": 4.6957671957671955e-06,
+      "loss": 0.2245,
+      "step": 770
+    },
+    {
+      "epoch": 2.7450980392156863,
+      "grad_norm": 0.17864180230187682,
+      "learning_rate": 4.6296296296296296e-06,
+      "loss": 0.2373,
+      "step": 771
+    },
+    {
+      "epoch": 2.748663101604278,
+      "grad_norm": 0.1664556908055509,
+      "learning_rate": 4.563492063492064e-06,
+      "loss": 0.2342,
+      "step": 772
+    },
+    {
+      "epoch": 2.7522281639928696,
+      "grad_norm": 0.17130186895062516,
+      "learning_rate": 4.497354497354498e-06,
+      "loss": 0.232,
+      "step": 773
+    },
+    {
+      "epoch": 2.7557932263814617,
+      "grad_norm": 0.17352300983572716,
+      "learning_rate": 4.431216931216931e-06,
+      "loss": 0.2275,
+      "step": 774
+    },
+    {
+      "epoch": 2.7593582887700534,
+      "grad_norm": 0.17811070951303326,
+      "learning_rate": 4.365079365079365e-06,
+      "loss": 0.2371,
+      "step": 775
+    },
+    {
+      "epoch": 2.762923351158645,
+      "grad_norm": 0.18908712670769268,
+      "learning_rate": 4.298941798941799e-06,
+      "loss": 0.2373,
+      "step": 776
+    },
+    {
+      "epoch": 2.766488413547237,
+      "grad_norm": 0.1859153491610666,
+      "learning_rate": 4.232804232804233e-06,
+      "loss": 0.2453,
+      "step": 777
+    },
+    {
+      "epoch": 2.770053475935829,
+      "grad_norm": 0.17342317924360504,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 0.2205,
+      "step": 778
+    },
+    {
+      "epoch": 2.7736185383244205,
+      "grad_norm": 0.16898772011686414,
+      "learning_rate": 4.100529100529101e-06,
+      "loss": 0.2291,
+      "step": 779
+    },
+    {
+      "epoch": 2.7771836007130126,
+      "grad_norm": 0.1683077731924244,
+      "learning_rate": 4.034391534391534e-06,
+      "loss": 0.2224,
+      "step": 780
+    },
+    {
+      "epoch": 2.7807486631016043,
+      "grad_norm": 0.17012960640827113,
+      "learning_rate": 3.968253968253968e-06,
+      "loss": 0.235,
+      "step": 781
+    },
+    {
+      "epoch": 2.784313725490196,
+      "grad_norm": 0.17480045677993955,
+      "learning_rate": 3.902116402116402e-06,
+      "loss": 0.2279,
+      "step": 782
+    },
+    {
+      "epoch": 2.787878787878788,
+      "grad_norm": 0.19387061916627255,
+      "learning_rate": 3.835978835978836e-06,
+      "loss": 0.2304,
+      "step": 783
+    },
+    {
+      "epoch": 2.7914438502673797,
+      "grad_norm": 0.17313512293393837,
+      "learning_rate": 3.7698412698412697e-06,
+      "loss": 0.2244,
+      "step": 784
+    },
+    {
+      "epoch": 2.7950089126559714,
+      "grad_norm": 0.16557070449949537,
+      "learning_rate": 3.7037037037037037e-06,
+      "loss": 0.2331,
+      "step": 785
+    },
+    {
+      "epoch": 2.7985739750445635,
+      "grad_norm": 0.17182792776199704,
+      "learning_rate": 3.6375661375661373e-06,
+      "loss": 0.2188,
+      "step": 786
+    },
+    {
+      "epoch": 2.802139037433155,
+      "grad_norm": 0.1719065624378341,
+      "learning_rate": 3.5714285714285714e-06,
+      "loss": 0.2226,
+      "step": 787
+    },
+    {
+      "epoch": 2.805704099821747,
+      "grad_norm": 0.17553321488296028,
+      "learning_rate": 3.5052910052910054e-06,
+      "loss": 0.2404,
+      "step": 788
+    },
+    {
+      "epoch": 2.809269162210339,
+      "grad_norm": 0.17572800129876848,
+      "learning_rate": 3.439153439153439e-06,
+      "loss": 0.2405,
+      "step": 789
+    },
+    {
+      "epoch": 2.8128342245989306,
+      "grad_norm": 0.1674610477724161,
+      "learning_rate": 3.373015873015873e-06,
+      "loss": 0.2261,
+      "step": 790
+    },
+    {
+      "epoch": 2.8163992869875223,
+      "grad_norm": 0.1674498324318124,
+      "learning_rate": 3.306878306878307e-06,
+      "loss": 0.2168,
+      "step": 791
+    },
+    {
+      "epoch": 2.819964349376114,
+      "grad_norm": 0.17425340451012036,
+      "learning_rate": 3.2407407407407406e-06,
+      "loss": 0.2284,
+      "step": 792
+    },
+    {
+      "epoch": 2.8235294117647056,
+      "grad_norm": 0.1757959553891028,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 0.2248,
+      "step": 793
+    },
+    {
+      "epoch": 2.8270944741532977,
+      "grad_norm": 0.16728578659694682,
+      "learning_rate": 3.1084656084656082e-06,
+      "loss": 0.2208,
+      "step": 794
+    },
+    {
+      "epoch": 2.8306595365418894,
+      "grad_norm": 0.16795858484116724,
+      "learning_rate": 3.0423280423280423e-06,
+      "loss": 0.2305,
+      "step": 795
+    },
+    {
+      "epoch": 2.834224598930481,
+      "grad_norm": 0.16686880371583485,
+      "learning_rate": 2.9761904761904763e-06,
+      "loss": 0.2248,
+      "step": 796
+    },
+    {
+      "epoch": 2.837789661319073,
+      "grad_norm": 0.17289022440316487,
+      "learning_rate": 2.91005291005291e-06,
+      "loss": 0.2299,
+      "step": 797
+    },
+    {
+      "epoch": 2.841354723707665,
+      "grad_norm": 0.16684396843458116,
+      "learning_rate": 2.843915343915344e-06,
+      "loss": 0.2192,
+      "step": 798
+    },
+    {
+      "epoch": 2.8449197860962565,
+      "grad_norm": 0.17506808756570338,
+      "learning_rate": 2.777777777777778e-06,
+      "loss": 0.2254,
+      "step": 799
+    },
+    {
+      "epoch": 2.8484848484848486,
+      "grad_norm": 0.18586133600559546,
+      "learning_rate": 2.7116402116402115e-06,
+      "loss": 0.2418,
+      "step": 800
+    },
+    {
+      "epoch": 2.8520499108734403,
+      "grad_norm": 0.16925432146263436,
+      "learning_rate": 2.6455026455026455e-06,
+      "loss": 0.2186,
+      "step": 801
+    },
+    {
+      "epoch": 2.855614973262032,
+      "grad_norm": 0.19224549166844843,
+      "learning_rate": 2.5793650793650795e-06,
+      "loss": 0.244,
+      "step": 802
+    },
+    {
+      "epoch": 2.859180035650624,
+      "grad_norm": 0.16721671419133316,
+      "learning_rate": 2.513227513227513e-06,
+      "loss": 0.2099,
+      "step": 803
+    },
+    {
+      "epoch": 2.8627450980392157,
+      "grad_norm": 0.17509323914185104,
+      "learning_rate": 2.447089947089947e-06,
+      "loss": 0.2245,
+      "step": 804
+    },
+    {
+      "epoch": 2.8663101604278074,
+      "grad_norm": 0.17230555871180658,
+      "learning_rate": 2.3809523809523808e-06,
+      "loss": 0.2357,
+      "step": 805
+    },
+    {
+      "epoch": 2.8698752228163995,
+      "grad_norm": 0.16330934816319997,
+      "learning_rate": 2.3148148148148148e-06,
+      "loss": 0.2188,
+      "step": 806
+    },
+    {
+      "epoch": 2.873440285204991,
+      "grad_norm": 0.17212573773879028,
+      "learning_rate": 2.248677248677249e-06,
+      "loss": 0.2328,
+      "step": 807
+    },
+    {
+      "epoch": 2.877005347593583,
+      "grad_norm": 0.17021151891234074,
+      "learning_rate": 2.1825396825396824e-06,
+      "loss": 0.2203,
+      "step": 808
+    },
+    {
+      "epoch": 2.880570409982175,
+      "grad_norm": 0.16871079862480406,
+      "learning_rate": 2.1164021164021164e-06,
+      "loss": 0.2265,
+      "step": 809
+    },
+    {
+      "epoch": 2.8841354723707666,
+      "grad_norm": 0.17771570836178172,
+      "learning_rate": 2.0502645502645504e-06,
+      "loss": 0.2351,
+      "step": 810
+    },
+    {
+      "epoch": 2.8877005347593583,
+      "grad_norm": 0.1630776789487284,
+      "learning_rate": 1.984126984126984e-06,
+      "loss": 0.2297,
+      "step": 811
+    },
+    {
+      "epoch": 2.89126559714795,
+      "grad_norm": 0.17756193806200654,
+      "learning_rate": 1.917989417989418e-06,
+      "loss": 0.2308,
+      "step": 812
+    },
+    {
+      "epoch": 2.8948306595365416,
+      "grad_norm": 0.17692638236838543,
+      "learning_rate": 1.8518518518518519e-06,
+      "loss": 0.223,
+      "step": 813
+    },
+    {
+      "epoch": 2.8983957219251337,
+      "grad_norm": 0.16846355175079644,
+      "learning_rate": 1.7857142857142857e-06,
+      "loss": 0.2345,
+      "step": 814
+    },
+    {
+      "epoch": 2.9019607843137254,
+      "grad_norm": 0.19061321798917505,
+      "learning_rate": 1.7195767195767195e-06,
+      "loss": 0.2265,
+      "step": 815
+    },
+    {
+      "epoch": 2.905525846702317,
+      "grad_norm": 0.16728359890776126,
+      "learning_rate": 1.6534391534391535e-06,
+      "loss": 0.2274,
+      "step": 816
+    },
+    {
+      "epoch": 2.909090909090909,
+      "grad_norm": 0.16497122067042994,
+      "learning_rate": 1.5873015873015873e-06,
+      "loss": 0.2269,
+      "step": 817
+    },
+    {
+      "epoch": 2.912655971479501,
+      "grad_norm": 0.19286574194808898,
+      "learning_rate": 1.5211640211640211e-06,
+      "loss": 0.2277,
+      "step": 818
+    },
+    {
+      "epoch": 2.9162210338680925,
+      "grad_norm": 0.16134761192893696,
+      "learning_rate": 1.455026455026455e-06,
+      "loss": 0.2374,
+      "step": 819
+    },
+    {
+      "epoch": 2.9197860962566846,
+      "grad_norm": 0.16418743529691407,
+      "learning_rate": 1.388888888888889e-06,
+      "loss": 0.226,
+      "step": 820
+    },
+    {
+      "epoch": 2.9233511586452763,
+      "grad_norm": 0.16485290239210607,
+      "learning_rate": 1.3227513227513228e-06,
+      "loss": 0.2267,
+      "step": 821
+    },
+    {
+      "epoch": 2.926916221033868,
+      "grad_norm": 0.21014287441242274,
+      "learning_rate": 1.2566137566137566e-06,
+      "loss": 0.2328,
+      "step": 822
+    },
+    {
+      "epoch": 2.93048128342246,
+      "grad_norm": 0.1640957230151931,
+      "learning_rate": 1.1904761904761904e-06,
+      "loss": 0.2284,
+      "step": 823
+    },
+    {
+      "epoch": 2.9340463458110517,
+      "grad_norm": 0.16223078054846984,
+      "learning_rate": 1.1243386243386244e-06,
+      "loss": 0.2126,
+      "step": 824
+    },
+    {
+      "epoch": 2.9376114081996434,
+      "grad_norm": 0.16755868866918178,
+      "learning_rate": 1.0582010582010582e-06,
+      "loss": 0.2319,
+      "step": 825
+    },
+    {
+      "epoch": 2.9411764705882355,
+      "grad_norm": 0.1749102833736808,
+      "learning_rate": 9.92063492063492e-07,
+      "loss": 0.2411,
+      "step": 826
+    },
+    {
+      "epoch": 2.944741532976827,
+      "grad_norm": 0.16421403391532102,
+      "learning_rate": 9.259259259259259e-07,
+      "loss": 0.2312,
+      "step": 827
+    },
+    {
+      "epoch": 2.948306595365419,
+      "grad_norm": 0.16489607284429073,
+      "learning_rate": 8.597883597883597e-07,
+      "loss": 0.2205,
+      "step": 828
+    },
+    {
+      "epoch": 2.951871657754011,
+      "grad_norm": 0.16508324263880042,
+      "learning_rate": 7.936507936507937e-07,
+      "loss": 0.2172,
+      "step": 829
+    },
+    {
+      "epoch": 2.9554367201426026,
+      "grad_norm": 0.17154348766055802,
+      "learning_rate": 7.275132275132275e-07,
+      "loss": 0.2196,
+      "step": 830
+    },
+    {
+      "epoch": 2.9590017825311943,
+      "grad_norm": 0.1664099808800128,
+      "learning_rate": 6.613756613756614e-07,
+      "loss": 0.2338,
+      "step": 831
+    },
+    {
+      "epoch": 2.962566844919786,
+      "grad_norm": 0.16743598641371263,
+      "learning_rate": 5.952380952380952e-07,
+      "loss": 0.2226,
+      "step": 832
+    },
+    {
+      "epoch": 2.966131907308378,
+      "grad_norm": 0.1775200490143428,
+      "learning_rate": 5.291005291005291e-07,
+      "loss": 0.2315,
+      "step": 833
+    },
+    {
+      "epoch": 2.9696969696969697,
+      "grad_norm": 0.16295772883951282,
+      "learning_rate": 4.6296296296296297e-07,
+      "loss": 0.2376,
+      "step": 834
+    },
+    {
+      "epoch": 2.9732620320855614,
+      "grad_norm": 0.1632816048950053,
+      "learning_rate": 3.9682539682539683e-07,
+      "loss": 0.2169,
+      "step": 835
+    },
+    {
+      "epoch": 2.976827094474153,
+      "grad_norm": 0.16525719276886336,
+      "learning_rate": 3.306878306878307e-07,
+      "loss": 0.2257,
+      "step": 836
+    },
+    {
+      "epoch": 2.980392156862745,
+      "grad_norm": 0.1651323463005636,
+      "learning_rate": 2.6455026455026455e-07,
+      "loss": 0.2292,
+      "step": 837
+    },
+    {
+      "epoch": 2.983957219251337,
+      "grad_norm": 0.16872336908304486,
+      "learning_rate": 1.9841269841269841e-07,
+      "loss": 0.2312,
+      "step": 838
+    },
+    {
+      "epoch": 2.9875222816399285,
+      "grad_norm": 0.1721854658602615,
+      "learning_rate": 1.3227513227513228e-07,
+      "loss": 0.2255,
+      "step": 839
+    },
+    {
+      "epoch": 2.9910873440285206,
+      "grad_norm": 0.16062477924559854,
+      "learning_rate": 6.613756613756614e-08,
+      "loss": 0.2333,
+      "step": 840
+    },
+    {
+      "epoch": 2.9910873440285206,
+      "step": 840,
+      "total_flos": 9.33106100089756e+18,
+      "train_loss": 0.43586345953600747,
+      "train_runtime": 28322.0165,
+      "train_samples_per_second": 0.475,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 840,
+  "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": 9.33106100089756e+18,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}