{
  "best_global_step": 4320,
  "best_metric": 2.488868474960327,
  "best_model_checkpoint": "./output_dir/Video_Games_full_reject_sampled/checkpoint-4320",
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 4320,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00046307350568350374,
      "grad_norm": 19.0,
      "learning_rate": 0.0,
      "loss": 1.5503,
      "step": 1
    },
    {
      "epoch": 0.0009261470113670075,
      "grad_norm": 14.1875,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 1.2954,
      "step": 2
    },
    {
      "epoch": 0.0013892205170505112,
      "grad_norm": 12.375,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 1.3655,
      "step": 3
    },
    {
      "epoch": 0.001852294022734015,
      "grad_norm": 5.90625,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 1.2177,
      "step": 4
    },
    {
      "epoch": 0.0023153675284175185,
      "grad_norm": 2.96875,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.9762,
      "step": 5
    },
    {
      "epoch": 0.0027784410341010223,
      "grad_norm": 2.71875,
      "learning_rate": 7.5e-05,
      "loss": 0.9528,
      "step": 6
    },
    {
      "epoch": 0.003241514539784526,
      "grad_norm": 2.390625,
      "learning_rate": 8.999999999999999e-05,
      "loss": 1.0856,
      "step": 7
    },
    {
      "epoch": 0.00370458804546803,
      "grad_norm": 1.515625,
      "learning_rate": 0.00010499999999999999,
      "loss": 0.9993,
      "step": 8
    },
    {
      "epoch": 0.004167661551151534,
      "grad_norm": 1.7109375,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.957,
      "step": 9
    },
    {
      "epoch": 0.004630735056835037,
      "grad_norm": 1.5390625,
      "learning_rate": 0.000135,
      "loss": 0.9725,
      "step": 10
    },
    {
      "epoch": 0.005093808562518541,
      "grad_norm": 1.59375,
      "learning_rate": 0.00015,
      "loss": 0.9298,
      "step": 11
    },
    {
      "epoch": 0.005556882068202045,
      "grad_norm": 1.4765625,
      "learning_rate": 0.000165,
      "loss": 0.9297,
      "step": 12
    },
    {
      "epoch": 0.006019955573885549,
      "grad_norm": 1.375,
      "learning_rate": 0.00017999999999999998,
      "loss": 1.0551,
      "step": 13
    },
    {
      "epoch": 0.006483029079569052,
      "grad_norm": 1.4375,
      "learning_rate": 0.000195,
      "loss": 1.029,
      "step": 14
    },
    {
      "epoch": 0.006946102585252556,
      "grad_norm": 1.40625,
      "learning_rate": 0.00020999999999999998,
      "loss": 0.9825,
      "step": 15
    },
    {
      "epoch": 0.00740917609093606,
      "grad_norm": 1.4375,
      "learning_rate": 0.000225,
      "loss": 0.9383,
      "step": 16
    },
    {
      "epoch": 0.007872249596619564,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.9269,
      "step": 17
    },
    {
      "epoch": 0.008335323102303067,
      "grad_norm": 1.40625,
      "learning_rate": 0.00025499999999999996,
      "loss": 0.9304,
      "step": 18
    },
    {
      "epoch": 0.00879839660798657,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00027,
      "loss": 0.9309,
      "step": 19
    },
    {
      "epoch": 0.009261470113670074,
      "grad_norm": 1.4375,
      "learning_rate": 0.000285,
      "loss": 0.992,
      "step": 20
    },
    {
      "epoch": 0.009724543619353578,
      "grad_norm": 1.4140625,
      "learning_rate": 0.0003,
      "loss": 0.9705,
      "step": 21
    },
    {
      "epoch": 0.010187617125037083,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00029997217068645636,
      "loss": 0.9102,
      "step": 22
    },
    {
      "epoch": 0.010650690630720586,
      "grad_norm": 1.265625,
      "learning_rate": 0.00029994434137291275,
      "loss": 1.124,
      "step": 23
    },
    {
      "epoch": 0.01111376413640409,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00029991651205936914,
      "loss": 0.9482,
      "step": 24
    },
    {
      "epoch": 0.011576837642087593,
      "grad_norm": 1.21875,
      "learning_rate": 0.0002998886827458256,
      "loss": 0.9206,
      "step": 25
    },
    {
      "epoch": 0.012039911147771098,
      "grad_norm": 1.3359375,
      "learning_rate": 0.000299860853432282,
      "loss": 1.0762,
      "step": 26
    },
    {
      "epoch": 0.012502984653454601,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00029983302411873837,
      "loss": 1.0507,
      "step": 27
    },
    {
      "epoch": 0.012966058159138105,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029980519480519476,
      "loss": 0.869,
      "step": 28
    },
    {
      "epoch": 0.013429131664821608,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002997773654916512,
      "loss": 0.8575,
      "step": 29
    },
    {
      "epoch": 0.013892205170505111,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002997495361781076,
      "loss": 0.9878,
      "step": 30
    },
    {
      "epoch": 0.014355278676188616,
      "grad_norm": 1.171875,
      "learning_rate": 0.000299721706864564,
      "loss": 1.1299,
      "step": 31
    },
    {
      "epoch": 0.01481835218187212,
      "grad_norm": 1.0625,
      "learning_rate": 0.00029969387755102037,
      "loss": 0.9897,
      "step": 32
    },
    {
      "epoch": 0.015281425687555623,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002996660482374768,
      "loss": 1.0583,
      "step": 33
    },
    {
      "epoch": 0.015744499193239128,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0002996382189239332,
      "loss": 1.0911,
      "step": 34
    },
    {
      "epoch": 0.01620757269892263,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002996103896103896,
      "loss": 0.9597,
      "step": 35
    },
    {
      "epoch": 0.016670646204606135,
      "grad_norm": 1.546875,
      "learning_rate": 0.000299582560296846,
      "loss": 1.0756,
      "step": 36
    },
    {
      "epoch": 0.017133719710289638,
      "grad_norm": 1.484375,
      "learning_rate": 0.00029955473098330237,
      "loss": 1.1642,
      "step": 37
    },
    {
      "epoch": 0.01759679321597314,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0002995269016697588,
      "loss": 0.9623,
      "step": 38
    },
    {
      "epoch": 0.018059866721656645,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00029949907235621515,
      "loss": 1.0269,
      "step": 39
    },
    {
      "epoch": 0.01852294022734015,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002994712430426716,
      "loss": 1.1617,
      "step": 40
    },
    {
      "epoch": 0.01898601373302365,
      "grad_norm": 1.328125,
      "learning_rate": 0.000299443413729128,
      "loss": 1.1297,
      "step": 41
    },
    {
      "epoch": 0.019449087238707155,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00029941558441558437,
      "loss": 1.1346,
      "step": 42
    },
    {
      "epoch": 0.019912160744390662,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00029938775510204076,
      "loss": 1.0287,
      "step": 43
    },
    {
      "epoch": 0.020375234250074165,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002993599257884972,
      "loss": 0.9627,
      "step": 44
    },
    {
      "epoch": 0.02083830775575767,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002993320964749536,
      "loss": 1.0273,
      "step": 45
    },
    {
      "epoch": 0.021301381261441172,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00029930426716141,
      "loss": 1.0832,
      "step": 46
    },
    {
      "epoch": 0.021764454767124675,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002992764378478664,
      "loss": 0.978,
      "step": 47
    },
    {
      "epoch": 0.02222752827280818,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002992486085343228,
      "loss": 1.0208,
      "step": 48
    },
    {
      "epoch": 0.022690601778491682,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002992207792207792,
      "loss": 1.0075,
      "step": 49
    },
    {
      "epoch": 0.023153675284175185,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002991929499072356,
      "loss": 0.9251,
      "step": 50
    },
    {
      "epoch": 0.02361674878985869,
      "grad_norm": 1.171875,
      "learning_rate": 0.000299165120593692,
      "loss": 1.0456,
      "step": 51
    },
    {
      "epoch": 0.024079822295542196,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002991372912801484,
      "loss": 1.0414,
      "step": 52
    },
    {
      "epoch": 0.0245428958012257,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002991094619666048,
      "loss": 0.9766,
      "step": 53
    },
    {
      "epoch": 0.025005969306909202,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002990816326530612,
      "loss": 1.1055,
      "step": 54
    },
    {
      "epoch": 0.025469042812592706,
      "grad_norm": 1.046875,
      "learning_rate": 0.0002990538033395176,
      "loss": 1.1284,
      "step": 55
    },
    {
      "epoch": 0.02593211631827621,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000299025974025974,
      "loss": 1.0519,
      "step": 56
    },
    {
      "epoch": 0.026395189823959712,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00029899814471243043,
      "loss": 1.0494,
      "step": 57
    },
    {
      "epoch": 0.026858263329643216,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00029897031539888677,
      "loss": 0.9979,
      "step": 58
    },
    {
      "epoch": 0.02732133683532672,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0002989424860853432,
      "loss": 0.9821,
      "step": 59
    },
    {
      "epoch": 0.027784410341010223,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0002989146567717996,
      "loss": 1.0424,
      "step": 60
    },
    {
      "epoch": 0.028247483846693726,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000298886827458256,
      "loss": 1.1106,
      "step": 61
    },
    {
      "epoch": 0.028710557352377233,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002988589981447124,
      "loss": 1.0478,
      "step": 62
    },
    {
      "epoch": 0.029173630858060736,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002988311688311688,
      "loss": 1.1169,
      "step": 63
    },
    {
      "epoch": 0.02963670436374424,
      "grad_norm": 1.03125,
      "learning_rate": 0.0002988033395176252,
      "loss": 1.1234,
      "step": 64
    },
    {
      "epoch": 0.030099777869427743,
      "grad_norm": 1.125,
      "learning_rate": 0.0002987755102040816,
      "loss": 1.0798,
      "step": 65
    },
    {
      "epoch": 0.030562851375111246,
      "grad_norm": 1.109375,
      "learning_rate": 0.000298747680890538,
      "loss": 1.1014,
      "step": 66
    },
    {
      "epoch": 0.03102592488079475,
      "grad_norm": 1.0625,
      "learning_rate": 0.0002987198515769944,
      "loss": 1.1015,
      "step": 67
    },
    {
      "epoch": 0.031488998386478256,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00029869202226345083,
      "loss": 0.9282,
      "step": 68
    },
    {
      "epoch": 0.03195207189216176,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002986641929499072,
      "loss": 1.0149,
      "step": 69
    },
    {
      "epoch": 0.03241514539784526,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002986363636363636,
      "loss": 0.9598,
      "step": 70
    },
    {
      "epoch": 0.032878218903528766,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00029860853432282,
      "loss": 0.957,
      "step": 71
    },
    {
      "epoch": 0.03334129240921227,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00029858070500927644,
      "loss": 0.9885,
      "step": 72
    },
    {
      "epoch": 0.03380436591489577,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00029855287569573283,
      "loss": 1.1079,
      "step": 73
    },
    {
      "epoch": 0.034267439420579276,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002985250463821892,
      "loss": 0.9396,
      "step": 74
    },
    {
      "epoch": 0.03473051292626278,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002984972170686456,
      "loss": 1.0169,
      "step": 75
    },
    {
      "epoch": 0.03519358643194628,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00029846938775510205,
      "loss": 1.0432,
      "step": 76
    },
    {
      "epoch": 0.03565665993762979,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0002984415584415584,
      "loss": 1.1186,
      "step": 77
    },
    {
      "epoch": 0.03611973344331329,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002984137291280148,
      "loss": 1.1016,
      "step": 78
    },
    {
      "epoch": 0.03658280694899679,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0002983858998144712,
      "loss": 1.0713,
      "step": 79
    },
    {
      "epoch": 0.0370458804546803,
      "grad_norm": 0.875,
      "learning_rate": 0.0002983580705009276,
      "loss": 1.0159,
      "step": 80
    },
    {
      "epoch": 0.0375089539603638,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000298330241187384,
      "loss": 1.0061,
      "step": 81
    },
    {
      "epoch": 0.0379720274660473,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002983024118738404,
      "loss": 0.9938,
      "step": 82
    },
    {
      "epoch": 0.03843510097173081,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00029827458256029683,
      "loss": 1.2419,
      "step": 83
    },
    {
      "epoch": 0.03889817447741431,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002982467532467532,
      "loss": 1.0958,
      "step": 84
    },
    {
      "epoch": 0.03936124798309782,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002982189239332096,
      "loss": 1.1221,
      "step": 85
    },
    {
      "epoch": 0.039824321488781324,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000298191094619666,
      "loss": 1.1391,
      "step": 86
    },
    {
      "epoch": 0.04028739499446483,
      "grad_norm": 0.8125,
      "learning_rate": 0.00029816326530612245,
      "loss": 1.08,
      "step": 87
    },
    {
      "epoch": 0.04075046850014833,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00029813543599257884,
      "loss": 0.9507,
      "step": 88
    },
    {
      "epoch": 0.041213542005831834,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002981076066790352,
      "loss": 1.1041,
      "step": 89
    },
    {
      "epoch": 0.04167661551151534,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002980797773654916,
      "loss": 0.9386,
      "step": 90
    },
    {
      "epoch": 0.04213968901719884,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00029805194805194806,
      "loss": 1.1573,
      "step": 91
    },
    {
      "epoch": 0.042602762522882344,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00029802411873840445,
      "loss": 1.0445,
      "step": 92
    },
    {
      "epoch": 0.04306583602856585,
      "grad_norm": 0.875,
      "learning_rate": 0.00029799628942486084,
      "loss": 1.1435,
      "step": 93
    },
    {
      "epoch": 0.04352890953424935,
      "grad_norm": 0.921875,
      "learning_rate": 0.00029796846011131723,
      "loss": 0.9361,
      "step": 94
    },
    {
      "epoch": 0.043991983039932854,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002979406307977736,
      "loss": 0.915,
      "step": 95
    },
    {
      "epoch": 0.04445505654561636,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00029791280148423006,
      "loss": 1.0198,
      "step": 96
    },
    {
      "epoch": 0.04491813005129986,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002978849721706864,
      "loss": 1.0305,
      "step": 97
    },
    {
      "epoch": 0.045381203556983364,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00029785714285714284,
      "loss": 1.1784,
      "step": 98
    },
    {
      "epoch": 0.04584427706266687,
      "grad_norm": 0.75,
      "learning_rate": 0.00029782931354359923,
      "loss": 0.9419,
      "step": 99
    },
    {
      "epoch": 0.04630735056835037,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002978014842300556,
      "loss": 1.0949,
      "step": 100
    },
    {
      "epoch": 0.046770424074033874,
      "grad_norm": 0.765625,
      "learning_rate": 0.000297773654916512,
      "loss": 0.9589,
      "step": 101
    },
    {
      "epoch": 0.04723349757971738,
      "grad_norm": 0.84375,
      "learning_rate": 0.00029774582560296845,
      "loss": 1.132,
      "step": 102
    },
    {
      "epoch": 0.04769657108540088,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00029771799628942484,
      "loss": 1.081,
      "step": 103
    },
    {
      "epoch": 0.04815964459108439,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00029769016697588123,
      "loss": 1.0645,
      "step": 104
    },
    {
      "epoch": 0.048622718096767895,
      "grad_norm": 0.90625,
      "learning_rate": 0.0002976623376623376,
      "loss": 0.9226,
      "step": 105
    },
    {
      "epoch": 0.0490857916024514,
      "grad_norm": 0.828125,
      "learning_rate": 0.000297634508348794,
      "loss": 1.0072,
      "step": 106
    },
    {
      "epoch": 0.0495488651081349,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00029760667903525046,
      "loss": 0.9508,
      "step": 107
    },
    {
      "epoch": 0.050011938613818405,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00029757884972170685,
      "loss": 1.0899,
      "step": 108
    },
    {
      "epoch": 0.05047501211950191,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00029755102040816323,
      "loss": 1.0214,
      "step": 109
    },
    {
      "epoch": 0.05093808562518541,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002975231910946196,
      "loss": 1.0518,
      "step": 110
    },
    {
      "epoch": 0.051401159130868915,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00029749536178107607,
      "loss": 1.0143,
      "step": 111
    },
    {
      "epoch": 0.05186423263655242,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00029746753246753246,
      "loss": 1.0133,
      "step": 112
    },
    {
      "epoch": 0.05232730614223592,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00029743970315398885,
      "loss": 1.0376,
      "step": 113
    },
    {
      "epoch": 0.052790379647919425,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00029741187384044524,
      "loss": 1.0518,
      "step": 114
    },
    {
      "epoch": 0.05325345315360293,
      "grad_norm": 0.9375,
      "learning_rate": 0.0002973840445269017,
      "loss": 1.0614,
      "step": 115
    },
    {
      "epoch": 0.05371652665928643,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000297356215213358,
      "loss": 1.0615,
      "step": 116
    },
    {
      "epoch": 0.054179600164969935,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00029732838589981446,
      "loss": 0.9695,
      "step": 117
    },
    {
      "epoch": 0.05464267367065344,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00029730055658627085,
      "loss": 1.1936,
      "step": 118
    },
    {
      "epoch": 0.05510574717633694,
      "grad_norm": 0.8125,
      "learning_rate": 0.00029727272727272724,
      "loss": 0.9113,
      "step": 119
    },
    {
      "epoch": 0.055568820682020445,
      "grad_norm": 0.921875,
      "learning_rate": 0.00029724489795918363,
      "loss": 1.115,
      "step": 120
    },
    {
      "epoch": 0.05603189418770395,
      "grad_norm": 1.046875,
      "learning_rate": 0.00029721706864564,
      "loss": 0.9966,
      "step": 121
    },
    {
      "epoch": 0.05649496769338745,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00029718923933209646,
      "loss": 1.0117,
      "step": 122
    },
    {
      "epoch": 0.05695804119907096,
      "grad_norm": 0.78125,
      "learning_rate": 0.00029716141001855285,
      "loss": 0.9598,
      "step": 123
    },
    {
      "epoch": 0.057421114704754465,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00029713358070500924,
      "loss": 0.973,
      "step": 124
    },
    {
      "epoch": 0.05788418821043797,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00029710575139146563,
      "loss": 1.0158,
      "step": 125
    },
    {
      "epoch": 0.05834726171612147,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002970779220779221,
      "loss": 1.0822,
      "step": 126
    },
    {
      "epoch": 0.058810335221804975,
      "grad_norm": 0.953125,
      "learning_rate": 0.00029705009276437846,
      "loss": 1.0725,
      "step": 127
    },
    {
      "epoch": 0.05927340872748848,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00029702226345083485,
      "loss": 0.9666,
      "step": 128
    },
    {
      "epoch": 0.05973648223317198,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00029699443413729124,
      "loss": 1.0873,
      "step": 129
    },
    {
      "epoch": 0.060199555738855486,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002969666048237477,
      "loss": 0.917,
      "step": 130
    },
    {
      "epoch": 0.06066262924453899,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0002969387755102041,
      "loss": 1.0861,
      "step": 131
    },
    {
      "epoch": 0.06112570275022249,
      "grad_norm": 0.875,
      "learning_rate": 0.00029691094619666047,
      "loss": 0.895,
      "step": 132
    },
    {
      "epoch": 0.061588776255905996,
      "grad_norm": 1.015625,
      "learning_rate": 0.00029688311688311686,
      "loss": 1.0848,
      "step": 133
    },
    {
      "epoch": 0.0620518497615895,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00029685528756957325,
      "loss": 1.0113,
      "step": 134
    },
    {
      "epoch": 0.06251492326727301,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00029682745825602964,
      "loss": 1.0166,
      "step": 135
    },
    {
      "epoch": 0.06297799677295651,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000296799628942486,
      "loss": 0.8825,
      "step": 136
    },
    {
      "epoch": 0.06344107027864002,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00029677179962894247,
      "loss": 0.999,
      "step": 137
    },
    {
      "epoch": 0.06390414378432352,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00029674397031539886,
      "loss": 0.9163,
      "step": 138
    },
    {
      "epoch": 0.06436721729000702,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00029671614100185525,
      "loss": 1.0529,
      "step": 139
    },
    {
      "epoch": 0.06483029079569053,
      "grad_norm": 0.875,
      "learning_rate": 0.00029668831168831164,
      "loss": 0.9335,
      "step": 140
    },
    {
      "epoch": 0.06529336430137403,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002966604823747681,
      "loss": 0.989,
      "step": 141
    },
    {
      "epoch": 0.06575643780705753,
      "grad_norm": 0.75,
      "learning_rate": 0.00029663265306122447,
      "loss": 1.0173,
      "step": 142
    },
    {
      "epoch": 0.06621951131274104,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00029660482374768086,
      "loss": 1.0119,
      "step": 143
    },
    {
      "epoch": 0.06668258481842454,
      "grad_norm": 0.78125,
      "learning_rate": 0.00029657699443413725,
      "loss": 0.9784,
      "step": 144
    },
    {
      "epoch": 0.06714565832410804,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002965491651205937,
      "loss": 1.0188,
      "step": 145
    },
    {
      "epoch": 0.06760873182979155,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002965213358070501,
      "loss": 1.0843,
      "step": 146
    },
    {
      "epoch": 0.06807180533547505,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002964935064935065,
      "loss": 1.198,
      "step": 147
    },
    {
      "epoch": 0.06853487884115855,
      "grad_norm": 1.46875,
      "learning_rate": 0.00029646567717996286,
      "loss": 0.9267,
      "step": 148
    },
    {
      "epoch": 0.06899795234684206,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00029643784786641925,
      "loss": 1.0089,
      "step": 149
    },
    {
      "epoch": 0.06946102585252556,
      "grad_norm": 7.625,
      "learning_rate": 0.0002964100185528757,
      "loss": 1.1468,
      "step": 150
    },
    {
      "epoch": 0.06992409935820906,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00029638218923933203,
      "loss": 1.0433,
      "step": 151
    },
    {
      "epoch": 0.07038717286389257,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002963543599257885,
      "loss": 0.9452,
      "step": 152
    },
    {
      "epoch": 0.07085024636957607,
      "grad_norm": 0.75,
      "learning_rate": 0.00029632653061224487,
      "loss": 0.9864,
      "step": 153
    },
    {
      "epoch": 0.07131331987525957,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00029629870129870125,
      "loss": 1.1263,
      "step": 154
    },
    {
      "epoch": 0.07177639338094308,
      "grad_norm": 0.765625,
      "learning_rate": 0.00029627087198515764,
      "loss": 1.053,
      "step": 155
    },
    {
      "epoch": 0.07223946688662658,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0002962430426716141,
      "loss": 0.999,
      "step": 156
    },
    {
      "epoch": 0.07270254039231008,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002962152133580705,
      "loss": 0.9808,
      "step": 157
    },
    {
      "epoch": 0.07316561389799359,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00029618738404452687,
      "loss": 1.1833,
      "step": 158
    },
    {
      "epoch": 0.07362868740367709,
      "grad_norm": 0.765625,
      "learning_rate": 0.00029615955473098326,
      "loss": 0.9561,
      "step": 159
    },
    {
      "epoch": 0.0740917609093606,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002961317254174397,
      "loss": 1.0138,
      "step": 160
    },
    {
      "epoch": 0.0745548344150441,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002961038961038961,
      "loss": 0.9379,
      "step": 161
    },
    {
      "epoch": 0.0750179079207276,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002960760667903525,
      "loss": 0.9597,
      "step": 162
    },
    {
      "epoch": 0.0754809814264111,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00029604823747680887,
      "loss": 1.042,
      "step": 163
    },
    {
      "epoch": 0.0759440549320946,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00029602040816326526,
      "loss": 0.944,
      "step": 164
    },
    {
      "epoch": 0.07640712843777811,
      "grad_norm": 0.875,
      "learning_rate": 0.0002959925788497217,
      "loss": 0.8604,
      "step": 165
    },
    {
      "epoch": 0.07687020194346161,
      "grad_norm": 1.375,
      "learning_rate": 0.0002959647495361781,
      "loss": 0.9329,
      "step": 166
    },
    {
      "epoch": 0.07733327544914512,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002959369202226345,
      "loss": 0.8131,
      "step": 167
    },
    {
      "epoch": 0.07779634895482862,
      "grad_norm": 0.953125,
      "learning_rate": 0.00029590909090909087,
      "loss": 1.1072,
      "step": 168
    },
    {
      "epoch": 0.07825942246051212,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002958812615955473,
      "loss": 0.9636,
      "step": 169
    },
    {
      "epoch": 0.07872249596619564,
      "grad_norm": 0.859375,
      "learning_rate": 0.00029585343228200365,
      "loss": 1.0443,
      "step": 170
    },
    {
      "epoch": 0.07918556947187914,
      "grad_norm": 0.84375,
      "learning_rate": 0.0002958256029684601,
      "loss": 1.1129,
      "step": 171
    },
    {
      "epoch": 0.07964864297756265,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002957977736549165,
      "loss": 0.9087,
      "step": 172
    },
    {
      "epoch": 0.08011171648324615,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0002957699443413729,
      "loss": 1.0317,
      "step": 173
    },
    {
      "epoch": 0.08057478998892965,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00029574211502782926,
      "loss": 1.1567,
      "step": 174
    },
    {
      "epoch": 0.08103786349461316,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002957142857142857,
      "loss": 0.9944,
      "step": 175
    },
    {
      "epoch": 0.08150093700029666,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002956864564007421,
      "loss": 0.9193,
      "step": 176
    },
    {
      "epoch": 0.08196401050598016,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002956586270871985,
      "loss": 0.9851,
      "step": 177
    },
    {
      "epoch": 0.08242708401166367,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002956307977736549,
      "loss": 0.9443,
      "step": 178
    },
    {
      "epoch": 0.08289015751734717,
      "grad_norm": 0.9375,
      "learning_rate": 0.00029560296846011127,
      "loss": 0.987,
      "step": 179
    },
    {
      "epoch": 0.08335323102303067,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0002955751391465677,
      "loss": 1.1265,
      "step": 180
    },
    {
      "epoch": 0.08381630452871418,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0002955473098330241,
      "loss": 0.9103,
      "step": 181
    },
    {
      "epoch": 0.08427937803439768,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002955194805194805,
      "loss": 1.0339,
      "step": 182
    },
    {
      "epoch": 0.08474245154008118,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002954916512059369,
      "loss": 0.99,
      "step": 183
    },
    {
      "epoch": 0.08520552504576469,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002954638218923933,
      "loss": 0.9242,
      "step": 184
    },
    {
      "epoch": 0.08566859855144819,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002954359925788497,
      "loss": 0.9522,
      "step": 185
    },
    {
      "epoch": 0.0861316720571317,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002954081632653061,
      "loss": 1.0146,
      "step": 186
    },
    {
      "epoch": 0.0865947455628152,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002953803339517625,
      "loss": 0.975,
      "step": 187
    },
    {
      "epoch": 0.0870578190684987,
      "grad_norm": 0.765625,
      "learning_rate": 0.00029535250463821894,
      "loss": 0.9676,
      "step": 188
    },
    {
      "epoch": 0.0875208925741822,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00029532467532467527,
      "loss": 0.9017,
      "step": 189
    },
    {
      "epoch": 0.08798396607986571,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002952968460111317,
      "loss": 1.1146,
      "step": 190
    },
    {
      "epoch": 0.08844703958554921,
      "grad_norm": 1.015625,
      "learning_rate": 0.0002952690166975881,
      "loss": 1.0532,
      "step": 191
    },
    {
      "epoch": 0.08891011309123271,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002952411873840445,
      "loss": 0.9422,
      "step": 192
    },
    {
      "epoch": 0.08937318659691622,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002952133580705009,
      "loss": 0.9428,
      "step": 193
    },
    {
      "epoch": 0.08983626010259972,
      "grad_norm": 0.734375,
      "learning_rate": 0.00029518552875695727,
      "loss": 0.9768,
      "step": 194
    },
    {
      "epoch": 0.09029933360828322,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002951576994434137,
      "loss": 0.9984,
      "step": 195
    },
    {
      "epoch": 0.09076240711396673,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002951298701298701,
      "loss": 1.0075,
      "step": 196
    },
    {
      "epoch": 0.09122548061965023,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002951020408163265,
      "loss": 0.9195,
      "step": 197
    },
    {
      "epoch": 0.09168855412533373,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002950742115027829,
      "loss": 1.0536,
      "step": 198
    },
    {
      "epoch": 0.09215162763101724,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00029504638218923933,
      "loss": 1.0147,
      "step": 199
    },
    {
      "epoch": 0.09261470113670074,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002950185528756957,
      "loss": 0.8806,
      "step": 200
    },
    {
      "epoch": 0.09307777464238425,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002949907235621521,
      "loss": 0.9911,
      "step": 201
    },
    {
      "epoch": 0.09354084814806775,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002949628942486085,
      "loss": 0.9658,
      "step": 202
    },
    {
      "epoch": 0.09400392165375125,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00029493506493506494,
      "loss": 0.8779,
      "step": 203
    },
    {
      "epoch": 0.09446699515943476,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00029490723562152133,
      "loss": 1.0559,
      "step": 204
    },
    {
      "epoch": 0.09493006866511826,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002948794063079777,
      "loss": 1.0523,
      "step": 205
    },
    {
      "epoch": 0.09539314217080176,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0002948515769944341,
      "loss": 1.1094,
      "step": 206
    },
    {
      "epoch": 0.09585621567648527,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002948237476808905,
      "loss": 0.9641,
      "step": 207
    },
    {
      "epoch": 0.09631928918216878,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00029479591836734694,
      "loss": 1.0567,
      "step": 208
    },
    {
      "epoch": 0.09678236268785229,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002947680890538033,
      "loss": 1.1497,
      "step": 209
    },
    {
      "epoch": 0.09724543619353579,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002947402597402597,
      "loss": 1.0827,
      "step": 210
    },
    {
      "epoch": 0.09770850969921929,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002947124304267161,
      "loss": 0.7755,
      "step": 211
    },
    {
      "epoch": 0.0981715832049028,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002946846011131725,
      "loss": 1.1384,
      "step": 212
    },
    {
      "epoch": 0.0986346567105863,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002946567717996289,
      "loss": 0.9872,
      "step": 213
    },
    {
      "epoch": 0.0990977302162698,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00029462894248608534,
      "loss": 1.0381,
      "step": 214
    },
    {
      "epoch": 0.0995608037219533,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002946011131725417,
      "loss": 0.9868,
      "step": 215
    },
    {
      "epoch": 0.10002387722763681,
      "grad_norm": 0.921875,
      "learning_rate": 0.0002945732838589981,
      "loss": 1.0917,
      "step": 216
    },
    {
      "epoch": 0.10048695073332031,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002945454545454545,
      "loss": 0.9566,
      "step": 217
    },
    {
      "epoch": 0.10095002423900382,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00029451762523191095,
      "loss": 0.9335,
      "step": 218
    },
    {
      "epoch": 0.10141309774468732,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00029448979591836734,
      "loss": 0.9509,
      "step": 219
    },
    {
      "epoch": 0.10187617125037082,
      "grad_norm": 0.796875,
      "learning_rate": 0.00029446196660482373,
      "loss": 0.9465,
      "step": 220
    },
    {
      "epoch": 0.10233924475605433,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002944341372912801,
      "loss": 1.0751,
      "step": 221
    },
    {
      "epoch": 0.10280231826173783,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002944063079777365,
      "loss": 0.7926,
      "step": 222
    },
    {
      "epoch": 0.10326539176742133,
      "grad_norm": 0.703125,
      "learning_rate": 0.00029437847866419295,
      "loss": 0.9567,
      "step": 223
    },
    {
      "epoch": 0.10372846527310484,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00029435064935064934,
      "loss": 1.0495,
      "step": 224
    },
    {
      "epoch": 0.10419153877878834,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00029432282003710573,
      "loss": 1.072,
      "step": 225
    },
    {
      "epoch": 0.10465461228447184,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002942949907235621,
      "loss": 0.9545,
      "step": 226
    },
    {
      "epoch": 0.10511768579015535,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00029426716141001856,
      "loss": 0.9621,
      "step": 227
    },
    {
      "epoch": 0.10558075929583885,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002942393320964749,
      "loss": 1.0985,
      "step": 228
    },
    {
      "epoch": 0.10604383280152235,
      "grad_norm": 0.734375,
      "learning_rate": 0.00029421150278293134,
      "loss": 0.9616,
      "step": 229
    },
    {
      "epoch": 0.10650690630720586,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00029418367346938773,
      "loss": 0.9752,
      "step": 230
    },
    {
      "epoch": 0.10696997981288936,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002941558441558441,
      "loss": 0.9322,
      "step": 231
    },
    {
      "epoch": 0.10743305331857286,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002941280148423005,
      "loss": 0.9553,
      "step": 232
    },
    {
      "epoch": 0.10789612682425637,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002941001855287569,
      "loss": 1.1748,
      "step": 233
    },
    {
      "epoch": 0.10835920032993987,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00029407235621521335,
      "loss": 0.9486,
      "step": 234
    },
    {
      "epoch": 0.10882227383562337,
      "grad_norm": 0.84375,
      "learning_rate": 0.00029404452690166973,
      "loss": 0.9827,
      "step": 235
    },
    {
      "epoch": 0.10928534734130688,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002940166975881261,
      "loss": 0.8266,
      "step": 236
    },
    {
      "epoch": 0.10974842084699038,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002939888682745825,
      "loss": 1.0792,
      "step": 237
    },
    {
      "epoch": 0.11021149435267388,
      "grad_norm": 0.890625,
      "learning_rate": 0.00029396103896103896,
      "loss": 0.9189,
      "step": 238
    },
    {
      "epoch": 0.11067456785835739,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00029393320964749535,
      "loss": 0.9003,
      "step": 239
    },
    {
      "epoch": 0.11113764136404089,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00029390538033395174,
      "loss": 0.9644,
      "step": 240
    },
    {
      "epoch": 0.1116007148697244,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002938775510204081,
      "loss": 0.9852,
      "step": 241
    },
    {
      "epoch": 0.1120637883754079,
      "grad_norm": 0.859375,
      "learning_rate": 0.00029384972170686457,
      "loss": 1.0832,
      "step": 242
    },
    {
      "epoch": 0.1125268618810914,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00029382189239332096,
      "loss": 0.9616,
      "step": 243
    },
    {
      "epoch": 0.1129899353867749,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00029379406307977735,
      "loss": 1.0252,
      "step": 244
    },
    {
      "epoch": 0.11345300889245842,
      "grad_norm": 0.8125,
      "learning_rate": 0.00029376623376623374,
      "loss": 0.8966,
      "step": 245
    },
    {
      "epoch": 0.11391608239814192,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002937384044526902,
      "loss": 0.8485,
      "step": 246
    },
    {
      "epoch": 0.11437915590382543,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002937105751391465,
      "loss": 1.0559,
      "step": 247
    },
    {
      "epoch": 0.11484222940950893,
      "grad_norm": 0.890625,
      "learning_rate": 0.0002936827458256029,
      "loss": 1.1229,
      "step": 248
    },
    {
      "epoch": 0.11530530291519243,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00029365491651205935,
      "loss": 0.9838,
      "step": 249
    },
    {
      "epoch": 0.11576837642087594,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00029362708719851574,
      "loss": 1.0229,
      "step": 250
    },
    {
      "epoch": 0.11623144992655944,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00029359925788497213,
      "loss": 1.0496,
      "step": 251
    },
    {
      "epoch": 0.11669452343224294,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002935714285714285,
      "loss": 0.995,
      "step": 252
    },
    {
      "epoch": 0.11715759693792645,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00029354359925788496,
      "loss": 1.0274,
      "step": 253
    },
    {
      "epoch": 0.11762067044360995,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00029351576994434135,
      "loss": 0.9493,
      "step": 254
    },
    {
      "epoch": 0.11808374394929345,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00029348794063079774,
      "loss": 1.0747,
      "step": 255
    },
    {
      "epoch": 0.11854681745497696,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00029346011131725413,
      "loss": 0.9409,
      "step": 256
    },
    {
      "epoch": 0.11900989096066046,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002934322820037106,
      "loss": 0.8793,
      "step": 257
    },
    {
      "epoch": 0.11947296446634396,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00029340445269016697,
      "loss": 0.9942,
      "step": 258
    },
    {
      "epoch": 0.11993603797202747,
      "grad_norm": 0.640625,
      "learning_rate": 0.00029337662337662336,
      "loss": 0.9653,
      "step": 259
    },
    {
      "epoch": 0.12039911147771097,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00029334879406307975,
      "loss": 0.9621,
      "step": 260
    },
    {
      "epoch": 0.12086218498339447,
      "grad_norm": 0.6875,
      "learning_rate": 0.00029332096474953614,
      "loss": 0.9008,
      "step": 261
    },
    {
      "epoch": 0.12132525848907798,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002932931354359926,
      "loss": 1.0006,
      "step": 262
    },
    {
      "epoch": 0.12178833199476148,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002932653061224489,
      "loss": 0.9583,
      "step": 263
    },
    {
      "epoch": 0.12225140550044498,
      "grad_norm": 0.875,
      "learning_rate": 0.00029323747680890536,
      "loss": 0.9004,
      "step": 264
    },
    {
      "epoch": 0.12271447900612849,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00029320964749536175,
      "loss": 0.8936,
      "step": 265
    },
    {
      "epoch": 0.12317755251181199,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00029318181818181814,
      "loss": 0.9398,
      "step": 266
    },
    {
      "epoch": 0.1236406260174955,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00029315398886827453,
      "loss": 0.8424,
      "step": 267
    },
    {
      "epoch": 0.124103699523179,
      "grad_norm": 1.171875,
      "learning_rate": 0.00029312615955473097,
      "loss": 1.2015,
      "step": 268
    },
    {
      "epoch": 0.1245667730288625,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00029309833024118736,
      "loss": 0.8727,
      "step": 269
    },
    {
      "epoch": 0.12502984653454602,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00029307050092764375,
      "loss": 0.8303,
      "step": 270
    },
    {
      "epoch": 0.12549292004022952,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00029304267161410014,
      "loss": 0.9809,
      "step": 271
    },
    {
      "epoch": 0.12595599354591303,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002930148423005566,
      "loss": 0.846,
      "step": 272
    },
    {
      "epoch": 0.12641906705159653,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000292987012987013,
      "loss": 1.046,
      "step": 273
    },
    {
      "epoch": 0.12688214055728003,
      "grad_norm": 0.65625,
      "learning_rate": 0.00029295918367346936,
      "loss": 0.9213,
      "step": 274
    },
    {
      "epoch": 0.12734521406296354,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00029293135435992575,
      "loss": 0.9889,
      "step": 275
    },
    {
      "epoch": 0.12780828756864704,
      "grad_norm": 0.71875,
      "learning_rate": 0.00029290352504638214,
      "loss": 0.9528,
      "step": 276
    },
    {
      "epoch": 0.12827136107433054,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002928756957328386,
      "loss": 0.9436,
      "step": 277
    },
    {
      "epoch": 0.12873443458001405,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000292847866419295,
      "loss": 0.9699,
      "step": 278
    },
    {
      "epoch": 0.12919750808569755,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00029282003710575137,
      "loss": 0.9604,
      "step": 279
    },
    {
      "epoch": 0.12966058159138105,
      "grad_norm": 0.6875,
      "learning_rate": 0.00029279220779220775,
      "loss": 0.9543,
      "step": 280
    },
    {
      "epoch": 0.13012365509706456,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002927643784786642,
      "loss": 0.9336,
      "step": 281
    },
    {
      "epoch": 0.13058672860274806,
      "grad_norm": 0.65625,
      "learning_rate": 0.00029273654916512053,
      "loss": 0.8744,
      "step": 282
    },
    {
      "epoch": 0.13104980210843156,
      "grad_norm": 0.859375,
      "learning_rate": 0.000292708719851577,
      "loss": 0.9497,
      "step": 283
    },
    {
      "epoch": 0.13151287561411507,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00029268089053803337,
      "loss": 1.0254,
      "step": 284
    },
    {
      "epoch": 0.13197594911979857,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00029265306122448976,
      "loss": 1.1114,
      "step": 285
    },
    {
      "epoch": 0.13243902262548207,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00029262523191094615,
      "loss": 0.9706,
      "step": 286
    },
    {
      "epoch": 0.13290209613116558,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002925974025974026,
      "loss": 0.9602,
      "step": 287
    },
    {
      "epoch": 0.13336516963684908,
      "grad_norm": 0.6640625,
      "learning_rate": 0.000292569573283859,
      "loss": 0.8833,
      "step": 288
    },
    {
      "epoch": 0.13382824314253258,
      "grad_norm": 0.75,
      "learning_rate": 0.00029254174397031537,
      "loss": 0.9812,
      "step": 289
    },
    {
      "epoch": 0.13429131664821609,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00029251391465677176,
      "loss": 1.0522,
      "step": 290
    },
    {
      "epoch": 0.1347543901538996,
      "grad_norm": 0.703125,
      "learning_rate": 0.00029248608534322815,
      "loss": 0.8918,
      "step": 291
    },
    {
      "epoch": 0.1352174636595831,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002924582560296846,
      "loss": 1.0237,
      "step": 292
    },
    {
      "epoch": 0.1356805371652666,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000292430426716141,
      "loss": 0.8514,
      "step": 293
    },
    {
      "epoch": 0.1361436106709501,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00029240259740259737,
      "loss": 1.1273,
      "step": 294
    },
    {
      "epoch": 0.1366066841766336,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00029237476808905376,
      "loss": 0.9821,
      "step": 295
    },
    {
      "epoch": 0.1370697576823171,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0002923469387755102,
      "loss": 1.0868,
      "step": 296
    },
    {
      "epoch": 0.1375328311880006,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002923191094619666,
      "loss": 0.983,
      "step": 297
    },
    {
      "epoch": 0.1379959046936841,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000292291280148423,
      "loss": 1.1093,
      "step": 298
    },
    {
      "epoch": 0.13845897819936762,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002922634508348794,
      "loss": 0.9067,
      "step": 299
    },
    {
      "epoch": 0.13892205170505112,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002922356215213358,
      "loss": 0.9312,
      "step": 300
    },
    {
      "epoch": 0.13938512521073462,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00029220779220779215,
      "loss": 0.9148,
      "step": 301
    },
    {
      "epoch": 0.13984819871641813,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002921799628942486,
      "loss": 0.8734,
      "step": 302
    },
    {
      "epoch": 0.14031127222210163,
      "grad_norm": 0.734375,
      "learning_rate": 0.000292152133580705,
      "loss": 1.0042,
      "step": 303
    },
    {
      "epoch": 0.14077434572778513,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002921243042671614,
      "loss": 0.9851,
      "step": 304
    },
    {
      "epoch": 0.14123741923346864,
      "grad_norm": 0.8125,
      "learning_rate": 0.00029209647495361777,
      "loss": 1.1021,
      "step": 305
    },
    {
      "epoch": 0.14170049273915214,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00029206864564007416,
      "loss": 0.918,
      "step": 306
    },
    {
      "epoch": 0.14216356624483564,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002920408163265306,
      "loss": 0.8568,
      "step": 307
    },
    {
      "epoch": 0.14262663975051915,
      "grad_norm": 0.65625,
      "learning_rate": 0.000292012987012987,
      "loss": 0.8974,
      "step": 308
    },
    {
      "epoch": 0.14308971325620265,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002919851576994434,
      "loss": 0.9691,
      "step": 309
    },
    {
      "epoch": 0.14355278676188615,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00029195732838589977,
      "loss": 1.1756,
      "step": 310
    },
    {
      "epoch": 0.14401586026756966,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002919294990723562,
      "loss": 0.8741,
      "step": 311
    },
    {
      "epoch": 0.14447893377325316,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002919016697588126,
      "loss": 0.9727,
      "step": 312
    },
    {
      "epoch": 0.14494200727893666,
      "grad_norm": 0.6796875,
      "learning_rate": 0.000291873840445269,
      "loss": 0.949,
      "step": 313
    },
    {
      "epoch": 0.14540508078462017,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002918460111317254,
      "loss": 0.8951,
      "step": 314
    },
    {
      "epoch": 0.14586815429030367,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002918181818181818,
      "loss": 1.0549,
      "step": 315
    },
    {
      "epoch": 0.14633122779598717,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002917903525046382,
      "loss": 1.0321,
      "step": 316
    },
    {
      "epoch": 0.14679430130167068,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002917625231910946,
      "loss": 0.8376,
      "step": 317
    },
    {
      "epoch": 0.14725737480735418,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000291734693877551,
      "loss": 0.8745,
      "step": 318
    },
    {
      "epoch": 0.14772044831303768,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002917068645640074,
      "loss": 1.0206,
      "step": 319
    },
    {
      "epoch": 0.1481835218187212,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00029167903525046377,
      "loss": 1.1584,
      "step": 320
    },
    {
      "epoch": 0.1486465953244047,
      "grad_norm": 0.671875,
      "learning_rate": 0.00029165120593692016,
      "loss": 0.8989,
      "step": 321
    },
    {
      "epoch": 0.1491096688300882,
      "grad_norm": 1.734375,
      "learning_rate": 0.0002916233766233766,
      "loss": 0.879,
      "step": 322
    },
    {
      "epoch": 0.1495727423357717,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000291595547309833,
      "loss": 0.8278,
      "step": 323
    },
    {
      "epoch": 0.1500358158414552,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002915677179962894,
      "loss": 0.876,
      "step": 324
    },
    {
      "epoch": 0.1504988893471387,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0002915398886827458,
      "loss": 1.1477,
      "step": 325
    },
    {
      "epoch": 0.1509619628528222,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002915120593692022,
      "loss": 0.9986,
      "step": 326
    },
    {
      "epoch": 0.1514250363585057,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002914842300556586,
      "loss": 1.0274,
      "step": 327
    },
    {
      "epoch": 0.1518881098641892,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000291456400742115,
      "loss": 0.9265,
      "step": 328
    },
    {
      "epoch": 0.15235118336987272,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002914285714285714,
      "loss": 0.938,
      "step": 329
    },
    {
      "epoch": 0.15281425687555622,
      "grad_norm": 0.75,
      "learning_rate": 0.00029140074211502783,
      "loss": 0.9381,
      "step": 330
    },
    {
      "epoch": 0.15327733038123972,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002913729128014842,
      "loss": 0.9139,
      "step": 331
    },
    {
      "epoch": 0.15374040388692323,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002913450834879406,
      "loss": 0.9669,
      "step": 332
    },
    {
      "epoch": 0.15420347739260673,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000291317254174397,
      "loss": 0.8535,
      "step": 333
    },
    {
      "epoch": 0.15466655089829023,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002912894248608534,
      "loss": 0.97,
      "step": 334
    },
    {
      "epoch": 0.15512962440397374,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00029126159554730983,
      "loss": 0.9617,
      "step": 335
    },
    {
      "epoch": 0.15559269790965724,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002912337662337662,
      "loss": 1.0698,
      "step": 336
    },
    {
      "epoch": 0.15605577141534074,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002912059369202226,
      "loss": 1.0265,
      "step": 337
    },
    {
      "epoch": 0.15651884492102425,
      "grad_norm": 4.46875,
      "learning_rate": 0.000291178107606679,
      "loss": 1.1788,
      "step": 338
    },
    {
      "epoch": 0.15698191842670778,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00029115027829313545,
      "loss": 0.9352,
      "step": 339
    },
    {
      "epoch": 0.15744499193239128,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002911224489795918,
      "loss": 0.9715,
      "step": 340
    },
    {
      "epoch": 0.15790806543807479,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002910946196660482,
      "loss": 0.9566,
      "step": 341
    },
    {
      "epoch": 0.1583711389437583,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002910667903525046,
      "loss": 1.0227,
      "step": 342
    },
    {
      "epoch": 0.1588342124494418,
      "grad_norm": 0.78125,
      "learning_rate": 0.000291038961038961,
      "loss": 0.9852,
      "step": 343
    },
    {
      "epoch": 0.1592972859551253,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002910111317254174,
      "loss": 1.0074,
      "step": 344
    },
    {
      "epoch": 0.1597603594608088,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00029098330241187384,
      "loss": 0.9776,
      "step": 345
    },
    {
      "epoch": 0.1602234329664923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00029095547309833023,
      "loss": 0.9672,
      "step": 346
    },
    {
      "epoch": 0.1606865064721758,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002909276437847866,
      "loss": 0.9041,
      "step": 347
    },
    {
      "epoch": 0.1611495799778593,
      "grad_norm": 0.765625,
      "learning_rate": 0.000290899814471243,
      "loss": 0.9295,
      "step": 348
    },
    {
      "epoch": 0.1616126534835428,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002908719851576994,
      "loss": 0.929,
      "step": 349
    },
    {
      "epoch": 0.16207572698922632,
      "grad_norm": 0.859375,
      "learning_rate": 0.00029084415584415584,
      "loss": 1.0589,
      "step": 350
    },
    {
      "epoch": 0.16253880049490982,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00029081632653061223,
      "loss": 0.8622,
      "step": 351
    },
    {
      "epoch": 0.16300187400059332,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002907884972170686,
      "loss": 0.9786,
      "step": 352
    },
    {
      "epoch": 0.16346494750627683,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000290760667903525,
      "loss": 0.9872,
      "step": 353
    },
    {
      "epoch": 0.16392802101196033,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00029073283858998145,
      "loss": 0.8679,
      "step": 354
    },
    {
      "epoch": 0.16439109451764383,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00029070500927643784,
      "loss": 0.982,
      "step": 355
    },
    {
      "epoch": 0.16485416802332734,
      "grad_norm": 0.796875,
      "learning_rate": 0.00029067717996289423,
      "loss": 1.0276,
      "step": 356
    },
    {
      "epoch": 0.16531724152901084,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002906493506493506,
      "loss": 0.8621,
      "step": 357
    },
    {
      "epoch": 0.16578031503469434,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00029062152133580707,
      "loss": 1.0653,
      "step": 358
    },
    {
      "epoch": 0.16624338854037785,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002905936920222634,
      "loss": 0.9457,
      "step": 359
    },
    {
      "epoch": 0.16670646204606135,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002905658627087198,
      "loss": 1.0127,
      "step": 360
    },
    {
      "epoch": 0.16716953555174485,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00029053803339517623,
      "loss": 0.8846,
      "step": 361
    },
    {
      "epoch": 0.16763260905742836,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002905102040816326,
      "loss": 0.9339,
      "step": 362
    },
    {
      "epoch": 0.16809568256311186,
      "grad_norm": 0.828125,
      "learning_rate": 0.000290482374768089,
      "loss": 1.0881,
      "step": 363
    },
    {
      "epoch": 0.16855875606879536,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002904545454545454,
      "loss": 1.0204,
      "step": 364
    },
    {
      "epoch": 0.16902182957447887,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00029042671614100185,
      "loss": 1.0808,
      "step": 365
    },
    {
      "epoch": 0.16948490308016237,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00029039888682745824,
      "loss": 0.9777,
      "step": 366
    },
    {
      "epoch": 0.16994797658584587,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002903710575139146,
      "loss": 0.8934,
      "step": 367
    },
    {
      "epoch": 0.17041105009152938,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000290343228200371,
      "loss": 0.9424,
      "step": 368
    },
    {
      "epoch": 0.17087412359721288,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00029031539888682746,
      "loss": 0.9804,
      "step": 369
    },
    {
      "epoch": 0.17133719710289638,
      "grad_norm": 0.875,
      "learning_rate": 0.00029028756957328385,
      "loss": 1.0616,
      "step": 370
    },
    {
      "epoch": 0.17180027060857989,
      "grad_norm": 0.671875,
      "learning_rate": 0.00029025974025974024,
      "loss": 0.9568,
      "step": 371
    },
    {
      "epoch": 0.1722633441142634,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00029023191094619663,
      "loss": 0.8847,
      "step": 372
    },
    {
      "epoch": 0.1727264176199469,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00029020408163265307,
      "loss": 0.9427,
      "step": 373
    },
    {
      "epoch": 0.1731894911256304,
      "grad_norm": 0.671875,
      "learning_rate": 0.00029017625231910946,
      "loss": 0.8173,
      "step": 374
    },
    {
      "epoch": 0.1736525646313139,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002901484230055658,
      "loss": 0.8983,
      "step": 375
    },
    {
      "epoch": 0.1741156381369974,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00029012059369202224,
      "loss": 0.97,
      "step": 376
    },
    {
      "epoch": 0.1745787116426809,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00029009276437847863,
      "loss": 1.019,
      "step": 377
    },
    {
      "epoch": 0.1750417851483644,
      "grad_norm": 0.703125,
      "learning_rate": 0.000290064935064935,
      "loss": 1.054,
      "step": 378
    },
    {
      "epoch": 0.1755048586540479,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002900371057513914,
      "loss": 0.9867,
      "step": 379
    },
    {
      "epoch": 0.17596793215973142,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00029000927643784785,
      "loss": 0.9887,
      "step": 380
    },
    {
      "epoch": 0.17643100566541492,
      "grad_norm": 0.859375,
      "learning_rate": 0.00028998144712430424,
      "loss": 0.9736,
      "step": 381
    },
    {
      "epoch": 0.17689407917109842,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00028995361781076063,
      "loss": 0.9269,
      "step": 382
    },
    {
      "epoch": 0.17735715267678193,
      "grad_norm": 0.76953125,
      "learning_rate": 0.000289925788497217,
      "loss": 0.9603,
      "step": 383
    },
    {
      "epoch": 0.17782022618246543,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00028989795918367347,
      "loss": 1.0806,
      "step": 384
    },
    {
      "epoch": 0.17828329968814893,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00028987012987012986,
      "loss": 0.9076,
      "step": 385
    },
    {
      "epoch": 0.17874637319383244,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00028984230055658625,
      "loss": 0.8321,
      "step": 386
    },
    {
      "epoch": 0.17920944669951594,
      "grad_norm": 0.734375,
      "learning_rate": 0.00028981447124304264,
      "loss": 1.0516,
      "step": 387
    },
    {
      "epoch": 0.17967252020519944,
      "grad_norm": 0.8046875,
      "learning_rate": 0.000289786641929499,
      "loss": 0.9506,
      "step": 388
    },
    {
      "epoch": 0.18013559371088295,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028975881261595547,
      "loss": 0.9266,
      "step": 389
    },
    {
      "epoch": 0.18059866721656645,
      "grad_norm": 0.59375,
      "learning_rate": 0.00028973098330241186,
      "loss": 0.8879,
      "step": 390
    },
    {
      "epoch": 0.18106174072224995,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00028970315398886825,
      "loss": 0.9858,
      "step": 391
    },
    {
      "epoch": 0.18152481422793346,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00028967532467532464,
      "loss": 0.972,
      "step": 392
    },
    {
      "epoch": 0.18198788773361696,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002896474953617811,
      "loss": 1.1511,
      "step": 393
    },
    {
      "epoch": 0.18245096123930046,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002896196660482374,
      "loss": 1.003,
      "step": 394
    },
    {
      "epoch": 0.18291403474498397,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00028959183673469386,
      "loss": 0.951,
      "step": 395
    },
    {
      "epoch": 0.18337710825066747,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00028956400742115025,
      "loss": 1.0814,
      "step": 396
    },
    {
      "epoch": 0.18384018175635097,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00028953617810760664,
      "loss": 1.1192,
      "step": 397
    },
    {
      "epoch": 0.18430325526203448,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00028950834879406303,
      "loss": 1.1619,
      "step": 398
    },
    {
      "epoch": 0.18476632876771798,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002894805194805195,
      "loss": 1.0492,
      "step": 399
    },
    {
      "epoch": 0.18522940227340148,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00028945269016697586,
      "loss": 1.0257,
      "step": 400
    },
    {
      "epoch": 0.185692475779085,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00028942486085343225,
      "loss": 1.0163,
      "step": 401
    },
    {
      "epoch": 0.1861555492847685,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00028939703153988864,
      "loss": 0.9235,
      "step": 402
    },
    {
      "epoch": 0.186618622790452,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00028936920222634503,
      "loss": 0.9506,
      "step": 403
    },
    {
      "epoch": 0.1870816962961355,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002893413729128015,
      "loss": 0.9282,
      "step": 404
    },
    {
      "epoch": 0.187544769801819,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00028931354359925786,
      "loss": 0.9394,
      "step": 405
    },
    {
      "epoch": 0.1880078433075025,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00028928571428571425,
      "loss": 0.8235,
      "step": 406
    },
    {
      "epoch": 0.188470916813186,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00028925788497217064,
      "loss": 0.9942,
      "step": 407
    },
    {
      "epoch": 0.1889339903188695,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002892300556586271,
      "loss": 0.8738,
      "step": 408
    },
    {
      "epoch": 0.189397063824553,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002892022263450835,
      "loss": 0.8662,
      "step": 409
    },
    {
      "epoch": 0.18986013733023652,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00028917439703153987,
      "loss": 1.0247,
      "step": 410
    },
    {
      "epoch": 0.19032321083592002,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00028914656771799626,
      "loss": 0.9651,
      "step": 411
    },
    {
      "epoch": 0.19078628434160352,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0002891187384044527,
      "loss": 1.009,
      "step": 412
    },
    {
      "epoch": 0.19124935784728703,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028909090909090904,
      "loss": 0.8342,
      "step": 413
    },
    {
      "epoch": 0.19171243135297053,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002890630797773655,
      "loss": 0.9347,
      "step": 414
    },
    {
      "epoch": 0.19217550485865406,
      "grad_norm": 0.78125,
      "learning_rate": 0.00028903525046382187,
      "loss": 0.8669,
      "step": 415
    },
    {
      "epoch": 0.19263857836433756,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00028900742115027826,
      "loss": 0.8532,
      "step": 416
    },
    {
      "epoch": 0.19310165187002107,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028897959183673465,
      "loss": 0.8464,
      "step": 417
    },
    {
      "epoch": 0.19356472537570457,
      "grad_norm": 0.78125,
      "learning_rate": 0.00028895176252319104,
      "loss": 1.0372,
      "step": 418
    },
    {
      "epoch": 0.19402779888138807,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002889239332096475,
      "loss": 1.0232,
      "step": 419
    },
    {
      "epoch": 0.19449087238707158,
      "grad_norm": 0.65625,
      "learning_rate": 0.00028889610389610387,
      "loss": 1.0054,
      "step": 420
    },
    {
      "epoch": 0.19495394589275508,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00028886827458256026,
      "loss": 0.9524,
      "step": 421
    },
    {
      "epoch": 0.19541701939843859,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00028884044526901665,
      "loss": 0.81,
      "step": 422
    },
    {
      "epoch": 0.1958800929041221,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002888126159554731,
      "loss": 0.8813,
      "step": 423
    },
    {
      "epoch": 0.1963431664098056,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002887847866419295,
      "loss": 0.9159,
      "step": 424
    },
    {
      "epoch": 0.1968062399154891,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002887569573283859,
      "loss": 1.0256,
      "step": 425
    },
    {
      "epoch": 0.1972693134211726,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028872912801484226,
      "loss": 0.9513,
      "step": 426
    },
    {
      "epoch": 0.1977323869268561,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0002887012987012987,
      "loss": 1.0065,
      "step": 427
    },
    {
      "epoch": 0.1981954604325396,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002886734693877551,
      "loss": 0.8704,
      "step": 428
    },
    {
      "epoch": 0.1986585339382231,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002886456400742115,
      "loss": 0.9391,
      "step": 429
    },
    {
      "epoch": 0.1991216074439066,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0002886178107606679,
      "loss": 0.9157,
      "step": 430
    },
    {
      "epoch": 0.19958468094959012,
      "grad_norm": 0.640625,
      "learning_rate": 0.00028858998144712427,
      "loss": 0.9773,
      "step": 431
    },
    {
      "epoch": 0.20004775445527362,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028856215213358066,
      "loss": 1.0171,
      "step": 432
    },
    {
      "epoch": 0.20051082796095712,
      "grad_norm": 5.90625,
      "learning_rate": 0.00028853432282003704,
      "loss": 1.1327,
      "step": 433
    },
    {
      "epoch": 0.20097390146664063,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002885064935064935,
      "loss": 0.9582,
      "step": 434
    },
    {
      "epoch": 0.20143697497232413,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0002884786641929499,
      "loss": 1.036,
      "step": 435
    },
    {
      "epoch": 0.20190004847800763,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00028845083487940627,
      "loss": 0.9,
      "step": 436
    },
    {
      "epoch": 0.20236312198369114,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00028842300556586266,
      "loss": 0.8288,
      "step": 437
    },
    {
      "epoch": 0.20282619548937464,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002883951762523191,
      "loss": 1.1334,
      "step": 438
    },
    {
      "epoch": 0.20328926899505814,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002883673469387755,
      "loss": 1.0604,
      "step": 439
    },
    {
      "epoch": 0.20375234250074165,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002883395176252319,
      "loss": 0.9234,
      "step": 440
    },
    {
      "epoch": 0.20421541600642515,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00028831168831168827,
      "loss": 0.7726,
      "step": 441
    },
    {
      "epoch": 0.20467848951210865,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002882838589981447,
      "loss": 0.9824,
      "step": 442
    },
    {
      "epoch": 0.20514156301779216,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002882560296846011,
      "loss": 1.0832,
      "step": 443
    },
    {
      "epoch": 0.20560463652347566,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002882282003710575,
      "loss": 0.9744,
      "step": 444
    },
    {
      "epoch": 0.20606771002915916,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002882003710575139,
      "loss": 0.9841,
      "step": 445
    },
    {
      "epoch": 0.20653078353484267,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00028817254174397027,
      "loss": 0.9728,
      "step": 446
    },
    {
      "epoch": 0.20699385704052617,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002881447124304267,
      "loss": 1.0376,
      "step": 447
    },
    {
      "epoch": 0.20745693054620967,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002881168831168831,
      "loss": 0.8902,
      "step": 448
    },
    {
      "epoch": 0.20792000405189318,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002880890538033395,
      "loss": 1.0244,
      "step": 449
    },
    {
      "epoch": 0.20838307755757668,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002880612244897959,
      "loss": 0.9316,
      "step": 450
    },
    {
      "epoch": 0.20884615106326018,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002880333951762523,
      "loss": 1.0023,
      "step": 451
    },
    {
      "epoch": 0.20930922456894369,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028800556586270866,
      "loss": 0.9696,
      "step": 452
    },
    {
      "epoch": 0.2097722980746272,
      "grad_norm": 0.8125,
      "learning_rate": 0.0002879777365491651,
      "loss": 1.0059,
      "step": 453
    },
    {
      "epoch": 0.2102353715803107,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002879499072356215,
      "loss": 0.9941,
      "step": 454
    },
    {
      "epoch": 0.2106984450859942,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002879220779220779,
      "loss": 1.0402,
      "step": 455
    },
    {
      "epoch": 0.2111615185916777,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002878942486085343,
      "loss": 1.0471,
      "step": 456
    },
    {
      "epoch": 0.2116245920973612,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002878664192949907,
      "loss": 0.9333,
      "step": 457
    },
    {
      "epoch": 0.2120876656030447,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002878385899814471,
      "loss": 1.0349,
      "step": 458
    },
    {
      "epoch": 0.2125507391087282,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002878107606679035,
      "loss": 1.0012,
      "step": 459
    },
    {
      "epoch": 0.2130138126144117,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002877829313543599,
      "loss": 0.9442,
      "step": 460
    },
    {
      "epoch": 0.21347688612009522,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002877551020408163,
      "loss": 1.0865,
      "step": 461
    },
    {
      "epoch": 0.21393995962577872,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002877272727272727,
      "loss": 1.0017,
      "step": 462
    },
    {
      "epoch": 0.21440303313146222,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002876994434137291,
      "loss": 0.9094,
      "step": 463
    },
    {
      "epoch": 0.21486610663714573,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002876716141001855,
      "loss": 0.9956,
      "step": 464
    },
    {
      "epoch": 0.21532918014282923,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002876437847866419,
      "loss": 1.0081,
      "step": 465
    },
    {
      "epoch": 0.21579225364851273,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00028761595547309834,
      "loss": 0.9665,
      "step": 466
    },
    {
      "epoch": 0.21625532715419624,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002875881261595547,
      "loss": 0.872,
      "step": 467
    },
    {
      "epoch": 0.21671840065987974,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002875602968460111,
      "loss": 0.9229,
      "step": 468
    },
    {
      "epoch": 0.21718147416556324,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0002875324675324675,
      "loss": 0.9466,
      "step": 469
    },
    {
      "epoch": 0.21764454767124675,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00028750463821892395,
      "loss": 1.0683,
      "step": 470
    },
    {
      "epoch": 0.21810762117693025,
      "grad_norm": 0.75,
      "learning_rate": 0.0002874768089053803,
      "loss": 0.8664,
      "step": 471
    },
    {
      "epoch": 0.21857069468261375,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00028744897959183673,
      "loss": 1.0192,
      "step": 472
    },
    {
      "epoch": 0.21903376818829726,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002874211502782931,
      "loss": 0.9251,
      "step": 473
    },
    {
      "epoch": 0.21949684169398076,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002873933209647495,
      "loss": 0.9725,
      "step": 474
    },
    {
      "epoch": 0.21995991519966426,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0002873654916512059,
      "loss": 1.1552,
      "step": 475
    },
    {
      "epoch": 0.22042298870534777,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002873376623376623,
      "loss": 0.8509,
      "step": 476
    },
    {
      "epoch": 0.22088606221103127,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00028730983302411873,
      "loss": 0.9902,
      "step": 477
    },
    {
      "epoch": 0.22134913571671477,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002872820037105751,
      "loss": 1.0346,
      "step": 478
    },
    {
      "epoch": 0.22181220922239828,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002872541743970315,
      "loss": 0.8699,
      "step": 479
    },
    {
      "epoch": 0.22227528272808178,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002872263450834879,
      "loss": 0.9127,
      "step": 480
    },
    {
      "epoch": 0.22273835623376528,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00028719851576994434,
      "loss": 0.8371,
      "step": 481
    },
    {
      "epoch": 0.2232014297394488,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00028717068645640073,
      "loss": 0.9005,
      "step": 482
    },
    {
      "epoch": 0.2236645032451323,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002871428571428571,
      "loss": 0.8602,
      "step": 483
    },
    {
      "epoch": 0.2241275767508158,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002871150278293135,
      "loss": 1.0567,
      "step": 484
    },
    {
      "epoch": 0.2245906502564993,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00028708719851576995,
      "loss": 1.1935,
      "step": 485
    },
    {
      "epoch": 0.2250537237621828,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00028705936920222634,
      "loss": 0.8947,
      "step": 486
    },
    {
      "epoch": 0.2255167972678663,
      "grad_norm": 0.859375,
      "learning_rate": 0.0002870315398886827,
      "loss": 0.9319,
      "step": 487
    },
    {
      "epoch": 0.2259798707735498,
      "grad_norm": 1.328125,
      "learning_rate": 0.0002870037105751391,
      "loss": 1.0117,
      "step": 488
    },
    {
      "epoch": 0.2264429442792333,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002869758812615955,
      "loss": 0.9146,
      "step": 489
    },
    {
      "epoch": 0.22690601778491684,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002869480519480519,
      "loss": 0.9468,
      "step": 490
    },
    {
      "epoch": 0.22736909129060034,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002869202226345083,
      "loss": 1.024,
      "step": 491
    },
    {
      "epoch": 0.22783216479628385,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00028689239332096474,
      "loss": 0.8968,
      "step": 492
    },
    {
      "epoch": 0.22829523830196735,
      "grad_norm": 1.171875,
      "learning_rate": 0.0002868645640074211,
      "loss": 0.8531,
      "step": 493
    },
    {
      "epoch": 0.22875831180765085,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002868367346938775,
      "loss": 0.8228,
      "step": 494
    },
    {
      "epoch": 0.22922138531333436,
      "grad_norm": 1.6328125,
      "learning_rate": 0.0002868089053803339,
      "loss": 0.9772,
      "step": 495
    },
    {
      "epoch": 0.22968445881901786,
      "grad_norm": 0.828125,
      "learning_rate": 0.00028678107606679035,
      "loss": 0.9903,
      "step": 496
    },
    {
      "epoch": 0.23014753232470136,
      "grad_norm": 0.75,
      "learning_rate": 0.00028675324675324674,
      "loss": 0.9282,
      "step": 497
    },
    {
      "epoch": 0.23061060583038487,
      "grad_norm": 2.6875,
      "learning_rate": 0.00028672541743970313,
      "loss": 1.1644,
      "step": 498
    },
    {
      "epoch": 0.23107367933606837,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002866975881261595,
      "loss": 0.9317,
      "step": 499
    },
    {
      "epoch": 0.23153675284175187,
      "grad_norm": 0.65625,
      "learning_rate": 0.00028666975881261596,
      "loss": 0.897,
      "step": 500
    },
    {
      "epoch": 0.23199982634743538,
      "grad_norm": 1.171875,
      "learning_rate": 0.00028664192949907235,
      "loss": 1.0969,
      "step": 501
    },
    {
      "epoch": 0.23246289985311888,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00028661410018552874,
      "loss": 1.0516,
      "step": 502
    },
    {
      "epoch": 0.23292597335880239,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00028658627087198513,
      "loss": 1.0235,
      "step": 503
    },
    {
      "epoch": 0.2333890468644859,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002865584415584415,
      "loss": 0.9336,
      "step": 504
    },
    {
      "epoch": 0.2338521203701694,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00028653061224489796,
      "loss": 0.9861,
      "step": 505
    },
    {
      "epoch": 0.2343151938758529,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002865027829313543,
      "loss": 1.0147,
      "step": 506
    },
    {
      "epoch": 0.2347782673815364,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00028647495361781074,
      "loss": 0.9472,
      "step": 507
    },
    {
      "epoch": 0.2352413408872199,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00028644712430426713,
      "loss": 0.896,
      "step": 508
    },
    {
      "epoch": 0.2357044143929034,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002864192949907235,
      "loss": 0.9465,
      "step": 509
    },
    {
      "epoch": 0.2361674878985869,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002863914656771799,
      "loss": 0.8535,
      "step": 510
    },
    {
      "epoch": 0.2366305614042704,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00028636363636363636,
      "loss": 0.8625,
      "step": 511
    },
    {
      "epoch": 0.23709363490995392,
      "grad_norm": 0.65625,
      "learning_rate": 0.00028633580705009275,
      "loss": 0.8945,
      "step": 512
    },
    {
      "epoch": 0.23755670841563742,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00028630797773654914,
      "loss": 0.9874,
      "step": 513
    },
    {
      "epoch": 0.23801978192132092,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002862801484230055,
      "loss": 0.963,
      "step": 514
    },
    {
      "epoch": 0.23848285542700443,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002862523191094619,
      "loss": 0.8447,
      "step": 515
    },
    {
      "epoch": 0.23894592893268793,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00028622448979591836,
      "loss": 0.9617,
      "step": 516
    },
    {
      "epoch": 0.23940900243837143,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00028619666048237475,
      "loss": 0.8961,
      "step": 517
    },
    {
      "epoch": 0.23987207594405494,
      "grad_norm": 0.71875,
      "learning_rate": 0.00028616883116883114,
      "loss": 0.9745,
      "step": 518
    },
    {
      "epoch": 0.24033514944973844,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002861410018552875,
      "loss": 1.11,
      "step": 519
    },
    {
      "epoch": 0.24079822295542194,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00028611317254174397,
      "loss": 0.8799,
      "step": 520
    },
    {
      "epoch": 0.24126129646110545,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00028608534322820036,
      "loss": 0.9389,
      "step": 521
    },
    {
      "epoch": 0.24172436996678895,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00028605751391465675,
      "loss": 0.8806,
      "step": 522
    },
    {
      "epoch": 0.24218744347247245,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00028602968460111314,
      "loss": 0.9548,
      "step": 523
    },
    {
      "epoch": 0.24265051697815596,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002860018552875696,
      "loss": 1.0325,
      "step": 524
    },
    {
      "epoch": 0.24311359048383946,
      "grad_norm": 0.625,
      "learning_rate": 0.0002859740259740259,
      "loss": 0.9073,
      "step": 525
    },
    {
      "epoch": 0.24357666398952296,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00028594619666048236,
      "loss": 0.8572,
      "step": 526
    },
    {
      "epoch": 0.24403973749520647,
      "grad_norm": 0.625,
      "learning_rate": 0.00028591836734693875,
      "loss": 0.7897,
      "step": 527
    },
    {
      "epoch": 0.24450281100088997,
      "grad_norm": 0.71875,
      "learning_rate": 0.00028589053803339514,
      "loss": 0.959,
      "step": 528
    },
    {
      "epoch": 0.24496588450657347,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00028586270871985153,
      "loss": 0.8472,
      "step": 529
    },
    {
      "epoch": 0.24542895801225698,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002858348794063079,
      "loss": 0.9203,
      "step": 530
    },
    {
      "epoch": 0.24589203151794048,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028580705009276436,
      "loss": 0.9282,
      "step": 531
    },
    {
      "epoch": 0.24635510502362398,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028577922077922075,
      "loss": 0.9359,
      "step": 532
    },
    {
      "epoch": 0.24681817852930749,
      "grad_norm": 0.59375,
      "learning_rate": 0.00028575139146567714,
      "loss": 0.941,
      "step": 533
    },
    {
      "epoch": 0.247281252034991,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028572356215213353,
      "loss": 0.9962,
      "step": 534
    },
    {
      "epoch": 0.2477443255406745,
      "grad_norm": 0.765625,
      "learning_rate": 0.00028569573283859,
      "loss": 0.9388,
      "step": 535
    },
    {
      "epoch": 0.248207399046358,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00028566790352504637,
      "loss": 0.9559,
      "step": 536
    },
    {
      "epoch": 0.2486704725520415,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00028564007421150276,
      "loss": 0.9493,
      "step": 537
    },
    {
      "epoch": 0.249133546057725,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00028561224489795915,
      "loss": 1.1613,
      "step": 538
    },
    {
      "epoch": 0.2495966195634085,
      "grad_norm": 1.6796875,
      "learning_rate": 0.0002855844155844156,
      "loss": 1.0152,
      "step": 539
    },
    {
      "epoch": 0.25005969306909204,
      "grad_norm": 0.6640625,
      "learning_rate": 0.000285556586270872,
      "loss": 1.0636,
      "step": 540
    },
    {
      "epoch": 0.2505227665747755,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00028552875695732837,
      "loss": 0.8507,
      "step": 541
    },
    {
      "epoch": 0.25098584008045904,
      "grad_norm": 0.703125,
      "learning_rate": 0.00028550092764378476,
      "loss": 1.0276,
      "step": 542
    },
    {
      "epoch": 0.2514489135861425,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0002854730983302412,
      "loss": 0.9843,
      "step": 543
    },
    {
      "epoch": 0.25191198709182605,
      "grad_norm": 0.65625,
      "learning_rate": 0.00028544526901669754,
      "loss": 0.9163,
      "step": 544
    },
    {
      "epoch": 0.2523750605975095,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00028541743970315393,
      "loss": 0.7777,
      "step": 545
    },
    {
      "epoch": 0.25283813410319306,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00028538961038961037,
      "loss": 1.0524,
      "step": 546
    },
    {
      "epoch": 0.25330120760887653,
      "grad_norm": 0.78125,
      "learning_rate": 0.00028536178107606676,
      "loss": 1.0272,
      "step": 547
    },
    {
      "epoch": 0.25376428111456006,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00028533395176252315,
      "loss": 0.957,
      "step": 548
    },
    {
      "epoch": 0.25422735462024354,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00028530612244897954,
      "loss": 1.1143,
      "step": 549
    },
    {
      "epoch": 0.25469042812592707,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000285278293135436,
      "loss": 1.0471,
      "step": 550
    },
    {
      "epoch": 0.25515350163161055,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0002852504638218924,
      "loss": 1.0493,
      "step": 551
    },
    {
      "epoch": 0.2556165751372941,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00028522263450834876,
      "loss": 0.9347,
      "step": 552
    },
    {
      "epoch": 0.25607964864297755,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028519480519480515,
      "loss": 1.0352,
      "step": 553
    },
    {
      "epoch": 0.2565427221486611,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002851669758812616,
      "loss": 0.9731,
      "step": 554
    },
    {
      "epoch": 0.25700579565434456,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000285139146567718,
      "loss": 1.2117,
      "step": 555
    },
    {
      "epoch": 0.2574688691600281,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002851113172541744,
      "loss": 0.9915,
      "step": 556
    },
    {
      "epoch": 0.25793194266571157,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00028508348794063077,
      "loss": 1.0359,
      "step": 557
    },
    {
      "epoch": 0.2583950161713951,
      "grad_norm": 0.65625,
      "learning_rate": 0.00028505565862708716,
      "loss": 0.8684,
      "step": 558
    },
    {
      "epoch": 0.2588580896770786,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002850278293135436,
      "loss": 0.9146,
      "step": 559
    },
    {
      "epoch": 0.2593211631827621,
      "grad_norm": 0.6875,
      "learning_rate": 0.000285,
      "loss": 0.932,
      "step": 560
    },
    {
      "epoch": 0.2597842366884456,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002849721706864564,
      "loss": 0.9302,
      "step": 561
    },
    {
      "epoch": 0.2602473101941291,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00028494434137291277,
      "loss": 0.9296,
      "step": 562
    },
    {
      "epoch": 0.2607103836998126,
      "grad_norm": 0.609375,
      "learning_rate": 0.00028491651205936916,
      "loss": 0.8862,
      "step": 563
    },
    {
      "epoch": 0.2611734572054961,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00028488868274582555,
      "loss": 0.9298,
      "step": 564
    },
    {
      "epoch": 0.2616365307111796,
      "grad_norm": 2.546875,
      "learning_rate": 0.000284860853432282,
      "loss": 1.1566,
      "step": 565
    },
    {
      "epoch": 0.2620996042168631,
      "grad_norm": 1.2890625,
      "learning_rate": 0.0002848330241187384,
      "loss": 0.969,
      "step": 566
    },
    {
      "epoch": 0.2625626777225466,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028480519480519477,
      "loss": 0.8677,
      "step": 567
    },
    {
      "epoch": 0.26302575122823013,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00028477736549165116,
      "loss": 0.974,
      "step": 568
    },
    {
      "epoch": 0.2634888247339136,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002847495361781076,
      "loss": 0.9683,
      "step": 569
    },
    {
      "epoch": 0.26395189823959714,
      "grad_norm": 1.03125,
      "learning_rate": 0.000284721706864564,
      "loss": 1.0504,
      "step": 570
    },
    {
      "epoch": 0.2644149717452806,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002846938775510204,
      "loss": 0.9953,
      "step": 571
    },
    {
      "epoch": 0.26487804525096414,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00028466604823747677,
      "loss": 1.0463,
      "step": 572
    },
    {
      "epoch": 0.2653411187566476,
      "grad_norm": 1.28125,
      "learning_rate": 0.00028463821892393316,
      "loss": 1.0158,
      "step": 573
    },
    {
      "epoch": 0.26580419226233115,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002846103896103896,
      "loss": 1.0881,
      "step": 574
    },
    {
      "epoch": 0.2662672657680146,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000284582560296846,
      "loss": 1.0603,
      "step": 575
    },
    {
      "epoch": 0.26673033927369816,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002845547309833024,
      "loss": 0.9277,
      "step": 576
    },
    {
      "epoch": 0.26719341277938163,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002845269016697588,
      "loss": 0.8429,
      "step": 577
    },
    {
      "epoch": 0.26765648628506516,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002844990723562152,
      "loss": 0.9648,
      "step": 578
    },
    {
      "epoch": 0.26811955979074864,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002844712430426716,
      "loss": 1.0107,
      "step": 579
    },
    {
      "epoch": 0.26858263329643217,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000284443413729128,
      "loss": 0.9312,
      "step": 580
    },
    {
      "epoch": 0.26904570680211565,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002844155844155844,
      "loss": 0.9659,
      "step": 581
    },
    {
      "epoch": 0.2695087803077992,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00028438775510204083,
      "loss": 0.9961,
      "step": 582
    },
    {
      "epoch": 0.26997185381348265,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00028435992578849717,
      "loss": 0.9759,
      "step": 583
    },
    {
      "epoch": 0.2704349273191662,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002843320964749536,
      "loss": 0.9053,
      "step": 584
    },
    {
      "epoch": 0.27089800082484966,
      "grad_norm": 0.984375,
      "learning_rate": 0.00028430426716141,
      "loss": 0.9645,
      "step": 585
    },
    {
      "epoch": 0.2713610743305332,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002842764378478664,
      "loss": 0.9478,
      "step": 586
    },
    {
      "epoch": 0.27182414783621667,
      "grad_norm": 7.90625,
      "learning_rate": 0.0002842486085343228,
      "loss": 1.0992,
      "step": 587
    },
    {
      "epoch": 0.2722872213419002,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00028422077922077917,
      "loss": 0.9428,
      "step": 588
    },
    {
      "epoch": 0.2727502948475837,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0002841929499072356,
      "loss": 1.1206,
      "step": 589
    },
    {
      "epoch": 0.2732133683532672,
      "grad_norm": 3.0,
      "learning_rate": 0.000284165120593692,
      "loss": 1.0375,
      "step": 590
    },
    {
      "epoch": 0.2736764418589507,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002841372912801484,
      "loss": 0.9377,
      "step": 591
    },
    {
      "epoch": 0.2741395153646342,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002841094619666048,
      "loss": 1.0541,
      "step": 592
    },
    {
      "epoch": 0.2746025888703177,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002840816326530612,
      "loss": 0.9943,
      "step": 593
    },
    {
      "epoch": 0.2750656623760012,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002840538033395176,
      "loss": 1.0059,
      "step": 594
    },
    {
      "epoch": 0.2755287358816847,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000284025974025974,
      "loss": 1.0259,
      "step": 595
    },
    {
      "epoch": 0.2759918093873682,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002839981447124304,
      "loss": 0.9542,
      "step": 596
    },
    {
      "epoch": 0.2764548828930517,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00028397031539888684,
      "loss": 1.0579,
      "step": 597
    },
    {
      "epoch": 0.27691795639873523,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00028394248608534323,
      "loss": 1.0894,
      "step": 598
    },
    {
      "epoch": 0.2773810299044187,
      "grad_norm": 0.625,
      "learning_rate": 0.0002839146567717996,
      "loss": 0.9263,
      "step": 599
    },
    {
      "epoch": 0.27784410341010224,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000283886827458256,
      "loss": 0.8987,
      "step": 600
    },
    {
      "epoch": 0.2783071769157857,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002838589981447124,
      "loss": 0.8994,
      "step": 601
    },
    {
      "epoch": 0.27877025042146925,
      "grad_norm": 0.75,
      "learning_rate": 0.0002838311688311688,
      "loss": 0.8451,
      "step": 602
    },
    {
      "epoch": 0.2792333239271528,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002838033395176252,
      "loss": 0.9383,
      "step": 603
    },
    {
      "epoch": 0.27969639743283625,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002837755102040816,
      "loss": 1.009,
      "step": 604
    },
    {
      "epoch": 0.2801594709385198,
      "grad_norm": 0.58203125,
      "learning_rate": 0.000283747680890538,
      "loss": 0.8599,
      "step": 605
    },
    {
      "epoch": 0.28062254444420326,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002837198515769944,
      "loss": 1.0774,
      "step": 606
    },
    {
      "epoch": 0.2810856179498868,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002836920222634508,
      "loss": 0.9154,
      "step": 607
    },
    {
      "epoch": 0.28154869145557027,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00028366419294990723,
      "loss": 0.8714,
      "step": 608
    },
    {
      "epoch": 0.2820117649612538,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002836363636363636,
      "loss": 0.983,
      "step": 609
    },
    {
      "epoch": 0.2824748384669373,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00028360853432282,
      "loss": 0.9283,
      "step": 610
    },
    {
      "epoch": 0.2829379119726208,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002835807050092764,
      "loss": 0.9708,
      "step": 611
    },
    {
      "epoch": 0.2834009854783043,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00028355287569573284,
      "loss": 0.966,
      "step": 612
    },
    {
      "epoch": 0.2838640589839878,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00028352504638218923,
      "loss": 0.8658,
      "step": 613
    },
    {
      "epoch": 0.2843271324896713,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002834972170686456,
      "loss": 0.9103,
      "step": 614
    },
    {
      "epoch": 0.2847902059953548,
      "grad_norm": 0.69140625,
      "learning_rate": 0.000283469387755102,
      "loss": 0.9089,
      "step": 615
    },
    {
      "epoch": 0.2852532795010383,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002834415584415584,
      "loss": 1.0577,
      "step": 616
    },
    {
      "epoch": 0.2857163530067218,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00028341372912801485,
      "loss": 1.0257,
      "step": 617
    },
    {
      "epoch": 0.2861794265124053,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002833858998144712,
      "loss": 0.9932,
      "step": 618
    },
    {
      "epoch": 0.28664250001808883,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002833580705009276,
      "loss": 1.0201,
      "step": 619
    },
    {
      "epoch": 0.2871055735237723,
      "grad_norm": 0.6015625,
      "learning_rate": 0.000283330241187384,
      "loss": 1.008,
      "step": 620
    },
    {
      "epoch": 0.28756864702945584,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002833024118738404,
      "loss": 0.9598,
      "step": 621
    },
    {
      "epoch": 0.2880317205351393,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002832745825602968,
      "loss": 0.972,
      "step": 622
    },
    {
      "epoch": 0.28849479404082284,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028324675324675324,
      "loss": 0.9765,
      "step": 623
    },
    {
      "epoch": 0.2889578675465063,
      "grad_norm": 2.15625,
      "learning_rate": 0.00028321892393320963,
      "loss": 0.8916,
      "step": 624
    },
    {
      "epoch": 0.28942094105218985,
      "grad_norm": 1.515625,
      "learning_rate": 0.000283191094619666,
      "loss": 1.089,
      "step": 625
    },
    {
      "epoch": 0.2898840145578733,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002831632653061224,
      "loss": 0.8564,
      "step": 626
    },
    {
      "epoch": 0.29034708806355686,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00028313543599257885,
      "loss": 0.8657,
      "step": 627
    },
    {
      "epoch": 0.29081016156924033,
      "grad_norm": 0.875,
      "learning_rate": 0.00028310760667903524,
      "loss": 1.206,
      "step": 628
    },
    {
      "epoch": 0.29127323507492386,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00028307977736549163,
      "loss": 0.9507,
      "step": 629
    },
    {
      "epoch": 0.29173630858060734,
      "grad_norm": 0.671875,
      "learning_rate": 0.000283051948051948,
      "loss": 1.0531,
      "step": 630
    },
    {
      "epoch": 0.29219938208629087,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002830241187384044,
      "loss": 0.8986,
      "step": 631
    },
    {
      "epoch": 0.29266245559197435,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00028299628942486085,
      "loss": 0.827,
      "step": 632
    },
    {
      "epoch": 0.2931255290976579,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028296846011131724,
      "loss": 0.903,
      "step": 633
    },
    {
      "epoch": 0.29358860260334135,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00028294063079777363,
      "loss": 0.8523,
      "step": 634
    },
    {
      "epoch": 0.2940516761090249,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00028291280148423,
      "loss": 0.9486,
      "step": 635
    },
    {
      "epoch": 0.29451474961470836,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00028288497217068647,
      "loss": 0.8462,
      "step": 636
    },
    {
      "epoch": 0.2949778231203919,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002828571428571428,
      "loss": 0.8033,
      "step": 637
    },
    {
      "epoch": 0.29544089662607537,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00028282931354359925,
      "loss": 0.9183,
      "step": 638
    },
    {
      "epoch": 0.2959039701317589,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00028280148423005563,
      "loss": 0.9656,
      "step": 639
    },
    {
      "epoch": 0.2963670436374424,
      "grad_norm": 0.59375,
      "learning_rate": 0.000282773654916512,
      "loss": 0.9611,
      "step": 640
    },
    {
      "epoch": 0.2968301171431259,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002827458256029684,
      "loss": 0.9799,
      "step": 641
    },
    {
      "epoch": 0.2972931906488094,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002827179962894248,
      "loss": 1.0703,
      "step": 642
    },
    {
      "epoch": 0.2977562641544929,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00028269016697588125,
      "loss": 1.047,
      "step": 643
    },
    {
      "epoch": 0.2982193376601764,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00028266233766233764,
      "loss": 0.793,
      "step": 644
    },
    {
      "epoch": 0.2986824111658599,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000282634508348794,
      "loss": 0.9426,
      "step": 645
    },
    {
      "epoch": 0.2991454846715434,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002826066790352504,
      "loss": 0.8848,
      "step": 646
    },
    {
      "epoch": 0.2996085581772269,
      "grad_norm": 0.609375,
      "learning_rate": 0.00028257884972170686,
      "loss": 0.9687,
      "step": 647
    },
    {
      "epoch": 0.3000716316829104,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00028255102040816325,
      "loss": 0.9886,
      "step": 648
    },
    {
      "epoch": 0.30053470518859393,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00028252319109461964,
      "loss": 0.9586,
      "step": 649
    },
    {
      "epoch": 0.3009977786942774,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028249536178107603,
      "loss": 0.9886,
      "step": 650
    },
    {
      "epoch": 0.30146085219996094,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00028246753246753247,
      "loss": 0.9024,
      "step": 651
    },
    {
      "epoch": 0.3019239257056444,
      "grad_norm": 0.59375,
      "learning_rate": 0.00028243970315398886,
      "loss": 0.933,
      "step": 652
    },
    {
      "epoch": 0.30238699921132794,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028241187384044525,
      "loss": 1.035,
      "step": 653
    },
    {
      "epoch": 0.3028500727170114,
      "grad_norm": 0.828125,
      "learning_rate": 0.00028238404452690164,
      "loss": 1.0472,
      "step": 654
    },
    {
      "epoch": 0.30331314622269495,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0002823562152133581,
      "loss": 0.8865,
      "step": 655
    },
    {
      "epoch": 0.3037762197283784,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002823283858998144,
      "loss": 0.8834,
      "step": 656
    },
    {
      "epoch": 0.30423929323406196,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002823005565862708,
      "loss": 0.8584,
      "step": 657
    },
    {
      "epoch": 0.30470236673974543,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00028227272727272725,
      "loss": 1.031,
      "step": 658
    },
    {
      "epoch": 0.30516544024542896,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00028224489795918364,
      "loss": 0.8907,
      "step": 659
    },
    {
      "epoch": 0.30562851375111244,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00028221706864564003,
      "loss": 0.8648,
      "step": 660
    },
    {
      "epoch": 0.30609158725679597,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002821892393320964,
      "loss": 0.837,
      "step": 661
    },
    {
      "epoch": 0.30655466076247945,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00028216141001855287,
      "loss": 0.9834,
      "step": 662
    },
    {
      "epoch": 0.307017734268163,
      "grad_norm": 2.09375,
      "learning_rate": 0.00028213358070500926,
      "loss": 1.0886,
      "step": 663
    },
    {
      "epoch": 0.30748080777384645,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00028210575139146565,
      "loss": 0.879,
      "step": 664
    },
    {
      "epoch": 0.30794388127953,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00028207792207792204,
      "loss": 0.9116,
      "step": 665
    },
    {
      "epoch": 0.30840695478521346,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002820500927643785,
      "loss": 0.8674,
      "step": 666
    },
    {
      "epoch": 0.308870028290897,
      "grad_norm": 1.9375,
      "learning_rate": 0.00028202226345083487,
      "loss": 1.0404,
      "step": 667
    },
    {
      "epoch": 0.30933310179658047,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00028199443413729126,
      "loss": 0.9762,
      "step": 668
    },
    {
      "epoch": 0.309796175302264,
      "grad_norm": 0.6875,
      "learning_rate": 0.00028196660482374765,
      "loss": 0.9229,
      "step": 669
    },
    {
      "epoch": 0.3102592488079475,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002819387755102041,
      "loss": 0.803,
      "step": 670
    },
    {
      "epoch": 0.310722322313631,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002819109461966605,
      "loss": 0.9145,
      "step": 671
    },
    {
      "epoch": 0.3111853958193145,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002818831168831168,
      "loss": 0.9041,
      "step": 672
    },
    {
      "epoch": 0.311648469324998,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00028185528756957326,
      "loss": 0.931,
      "step": 673
    },
    {
      "epoch": 0.3121115428306815,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00028182745825602965,
      "loss": 1.0002,
      "step": 674
    },
    {
      "epoch": 0.312574616336365,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00028179962894248604,
      "loss": 0.9156,
      "step": 675
    },
    {
      "epoch": 0.3130376898420485,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00028177179962894243,
      "loss": 0.9035,
      "step": 676
    },
    {
      "epoch": 0.313500763347732,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002817439703153989,
      "loss": 0.9182,
      "step": 677
    },
    {
      "epoch": 0.31396383685341556,
      "grad_norm": 1.046875,
      "learning_rate": 0.00028171614100185526,
      "loss": 1.1461,
      "step": 678
    },
    {
      "epoch": 0.31442691035909903,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00028168831168831165,
      "loss": 0.9524,
      "step": 679
    },
    {
      "epoch": 0.31488998386478256,
      "grad_norm": 0.625,
      "learning_rate": 0.00028166048237476804,
      "loss": 0.9133,
      "step": 680
    },
    {
      "epoch": 0.31535305737046604,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002816326530612245,
      "loss": 0.8558,
      "step": 681
    },
    {
      "epoch": 0.31581613087614957,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002816048237476809,
      "loss": 0.9423,
      "step": 682
    },
    {
      "epoch": 0.31627920438183305,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00028157699443413727,
      "loss": 0.9633,
      "step": 683
    },
    {
      "epoch": 0.3167422778875166,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00028154916512059365,
      "loss": 0.8916,
      "step": 684
    },
    {
      "epoch": 0.31720535139320005,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00028152133580705004,
      "loss": 1.0286,
      "step": 685
    },
    {
      "epoch": 0.3176684248988836,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002814935064935065,
      "loss": 0.8381,
      "step": 686
    },
    {
      "epoch": 0.31813149840456706,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0002814656771799629,
      "loss": 1.0786,
      "step": 687
    },
    {
      "epoch": 0.3185945719102506,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00028143784786641927,
      "loss": 1.0201,
      "step": 688
    },
    {
      "epoch": 0.31905764541593407,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00028141001855287566,
      "loss": 1.0072,
      "step": 689
    },
    {
      "epoch": 0.3195207189216176,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002813821892393321,
      "loss": 1.0832,
      "step": 690
    },
    {
      "epoch": 0.3199837924273011,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002813543599257885,
      "loss": 0.8668,
      "step": 691
    },
    {
      "epoch": 0.3204468659329846,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002813265306122449,
      "loss": 1.0118,
      "step": 692
    },
    {
      "epoch": 0.3209099394386681,
      "grad_norm": 0.671875,
      "learning_rate": 0.00028129870129870127,
      "loss": 0.9449,
      "step": 693
    },
    {
      "epoch": 0.3213730129443516,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00028127087198515766,
      "loss": 1.0513,
      "step": 694
    },
    {
      "epoch": 0.3218360864500351,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00028124304267161405,
      "loss": 0.9007,
      "step": 695
    },
    {
      "epoch": 0.3222991599557186,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002812152133580705,
      "loss": 0.9408,
      "step": 696
    },
    {
      "epoch": 0.3227622334614021,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002811873840445269,
      "loss": 0.7997,
      "step": 697
    },
    {
      "epoch": 0.3232253069670856,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00028115955473098327,
      "loss": 1.0016,
      "step": 698
    },
    {
      "epoch": 0.3236883804727691,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00028113172541743966,
      "loss": 0.7876,
      "step": 699
    },
    {
      "epoch": 0.32415145397845263,
      "grad_norm": 0.953125,
      "learning_rate": 0.00028110389610389605,
      "loss": 0.9216,
      "step": 700
    },
    {
      "epoch": 0.3246145274841361,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002810760667903525,
      "loss": 0.8788,
      "step": 701
    },
    {
      "epoch": 0.32507760098981964,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002810482374768089,
      "loss": 0.9718,
      "step": 702
    },
    {
      "epoch": 0.3255406744955031,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002810204081632653,
      "loss": 0.8864,
      "step": 703
    },
    {
      "epoch": 0.32600374800118664,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00028099257884972166,
      "loss": 0.938,
      "step": 704
    },
    {
      "epoch": 0.3264668215068701,
      "grad_norm": 1.265625,
      "learning_rate": 0.0002809647495361781,
      "loss": 0.9699,
      "step": 705
    },
    {
      "epoch": 0.32692989501255365,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002809369202226345,
      "loss": 0.9571,
      "step": 706
    },
    {
      "epoch": 0.3273929685182371,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002809090909090909,
      "loss": 0.9135,
      "step": 707
    },
    {
      "epoch": 0.32785604202392066,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002808812615955473,
      "loss": 0.833,
      "step": 708
    },
    {
      "epoch": 0.32831911552960413,
      "grad_norm": 0.75,
      "learning_rate": 0.0002808534322820037,
      "loss": 1.0605,
      "step": 709
    },
    {
      "epoch": 0.32878218903528766,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002808256029684601,
      "loss": 1.0488,
      "step": 710
    },
    {
      "epoch": 0.32924526254097114,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002807977736549165,
      "loss": 0.9353,
      "step": 711
    },
    {
      "epoch": 0.32970833604665467,
      "grad_norm": 1.2578125,
      "learning_rate": 0.0002807699443413729,
      "loss": 0.9295,
      "step": 712
    },
    {
      "epoch": 0.33017140955233815,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002807421150278293,
      "loss": 0.9249,
      "step": 713
    },
    {
      "epoch": 0.3306344830580217,
      "grad_norm": 0.84375,
      "learning_rate": 0.00028071428571428567,
      "loss": 1.0222,
      "step": 714
    },
    {
      "epoch": 0.33109755656370515,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00028068645640074206,
      "loss": 0.9142,
      "step": 715
    },
    {
      "epoch": 0.3315606300693887,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002806586270871985,
      "loss": 0.8175,
      "step": 716
    },
    {
      "epoch": 0.33202370357507216,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002806307977736549,
      "loss": 0.9797,
      "step": 717
    },
    {
      "epoch": 0.3324867770807557,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002806029684601113,
      "loss": 1.0774,
      "step": 718
    },
    {
      "epoch": 0.33294985058643917,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00028057513914656767,
      "loss": 0.8587,
      "step": 719
    },
    {
      "epoch": 0.3334129240921227,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002805473098330241,
      "loss": 0.857,
      "step": 720
    },
    {
      "epoch": 0.3338759975978062,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002805194805194805,
      "loss": 0.9124,
      "step": 721
    },
    {
      "epoch": 0.3343390711034897,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002804916512059369,
      "loss": 0.883,
      "step": 722
    },
    {
      "epoch": 0.3348021446091732,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002804638218923933,
      "loss": 0.8805,
      "step": 723
    },
    {
      "epoch": 0.3352652181148567,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002804359925788497,
      "loss": 0.9368,
      "step": 724
    },
    {
      "epoch": 0.3357282916205402,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002804081632653061,
      "loss": 1.0504,
      "step": 725
    },
    {
      "epoch": 0.3361913651262237,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002803803339517625,
      "loss": 0.8525,
      "step": 726
    },
    {
      "epoch": 0.3366544386319072,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002803525046382189,
      "loss": 0.866,
      "step": 727
    },
    {
      "epoch": 0.3371175121375907,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002803246753246753,
      "loss": 1.0144,
      "step": 728
    },
    {
      "epoch": 0.3375805856432742,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00028029684601113173,
      "loss": 1.0586,
      "step": 729
    },
    {
      "epoch": 0.33804365914895773,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00028026901669758806,
      "loss": 1.0301,
      "step": 730
    },
    {
      "epoch": 0.3385067326546412,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002802411873840445,
      "loss": 0.908,
      "step": 731
    },
    {
      "epoch": 0.33896980616032474,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002802133580705009,
      "loss": 0.9352,
      "step": 732
    },
    {
      "epoch": 0.3394328796660082,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002801855287569573,
      "loss": 1.0529,
      "step": 733
    },
    {
      "epoch": 0.33989595317169174,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002801576994434137,
      "loss": 0.9066,
      "step": 734
    },
    {
      "epoch": 0.3403590266773752,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002801298701298701,
      "loss": 1.0312,
      "step": 735
    },
    {
      "epoch": 0.34082210018305875,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002801020408163265,
      "loss": 1.0892,
      "step": 736
    },
    {
      "epoch": 0.3412851736887422,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002800742115027829,
      "loss": 0.9334,
      "step": 737
    },
    {
      "epoch": 0.34174824719442576,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002800463821892393,
      "loss": 0.9612,
      "step": 738
    },
    {
      "epoch": 0.34221132070010923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00028001855287569573,
      "loss": 0.9281,
      "step": 739
    },
    {
      "epoch": 0.34267439420579276,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0002799907235621521,
      "loss": 0.9419,
      "step": 740
    },
    {
      "epoch": 0.34313746771147624,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002799628942486085,
      "loss": 0.9947,
      "step": 741
    },
    {
      "epoch": 0.34360054121715977,
      "grad_norm": 0.625,
      "learning_rate": 0.0002799350649350649,
      "loss": 0.855,
      "step": 742
    },
    {
      "epoch": 0.34406361472284325,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002799072356215213,
      "loss": 0.805,
      "step": 743
    },
    {
      "epoch": 0.3445266882285268,
      "grad_norm": 9.8125,
      "learning_rate": 0.00027987940630797774,
      "loss": 0.8818,
      "step": 744
    },
    {
      "epoch": 0.34498976173421025,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002798515769944341,
      "loss": 0.9928,
      "step": 745
    },
    {
      "epoch": 0.3454528352398938,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002798237476808905,
      "loss": 0.9475,
      "step": 746
    },
    {
      "epoch": 0.34591590874557726,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002797959183673469,
      "loss": 0.8677,
      "step": 747
    },
    {
      "epoch": 0.3463789822512608,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00027976808905380335,
      "loss": 0.9099,
      "step": 748
    },
    {
      "epoch": 0.34684205575694427,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002797402597402597,
      "loss": 0.9001,
      "step": 749
    },
    {
      "epoch": 0.3473051292626278,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00027971243042671613,
      "loss": 0.9098,
      "step": 750
    },
    {
      "epoch": 0.3477682027683113,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002796846011131725,
      "loss": 0.9059,
      "step": 751
    },
    {
      "epoch": 0.3482312762739948,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002796567717996289,
      "loss": 0.8844,
      "step": 752
    },
    {
      "epoch": 0.3486943497796783,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002796289424860853,
      "loss": 0.9077,
      "step": 753
    },
    {
      "epoch": 0.3491574232853618,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00027960111317254174,
      "loss": 1.0122,
      "step": 754
    },
    {
      "epoch": 0.34962049679104534,
      "grad_norm": 0.71875,
      "learning_rate": 0.00027957328385899813,
      "loss": 0.8393,
      "step": 755
    },
    {
      "epoch": 0.3500835702967288,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002795454545454545,
      "loss": 0.9346,
      "step": 756
    },
    {
      "epoch": 0.35054664380241235,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002795176252319109,
      "loss": 0.9163,
      "step": 757
    },
    {
      "epoch": 0.3510097173080958,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002794897959183673,
      "loss": 0.8727,
      "step": 758
    },
    {
      "epoch": 0.35147279081377936,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00027946196660482374,
      "loss": 0.8047,
      "step": 759
    },
    {
      "epoch": 0.35193586431946283,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00027943413729128013,
      "loss": 0.9634,
      "step": 760
    },
    {
      "epoch": 0.35239893782514636,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002794063079777365,
      "loss": 0.9658,
      "step": 761
    },
    {
      "epoch": 0.35286201133082984,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002793784786641929,
      "loss": 0.7866,
      "step": 762
    },
    {
      "epoch": 0.35332508483651337,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027935064935064936,
      "loss": 0.8346,
      "step": 763
    },
    {
      "epoch": 0.35378815834219685,
      "grad_norm": 0.625,
      "learning_rate": 0.00027932282003710575,
      "loss": 0.857,
      "step": 764
    },
    {
      "epoch": 0.3542512318478804,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00027929499072356213,
      "loss": 1.0596,
      "step": 765
    },
    {
      "epoch": 0.35471430535356385,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002792671614100185,
      "loss": 1.0205,
      "step": 766
    },
    {
      "epoch": 0.3551773788592474,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00027923933209647497,
      "loss": 0.9817,
      "step": 767
    },
    {
      "epoch": 0.35564045236493086,
      "grad_norm": 0.625,
      "learning_rate": 0.0002792115027829313,
      "loss": 0.7743,
      "step": 768
    },
    {
      "epoch": 0.3561035258706144,
      "grad_norm": 0.75,
      "learning_rate": 0.0002791836734693877,
      "loss": 0.9937,
      "step": 769
    },
    {
      "epoch": 0.35656659937629787,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00027915584415584414,
      "loss": 1.1301,
      "step": 770
    },
    {
      "epoch": 0.3570296728819814,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002791280148423005,
      "loss": 0.9968,
      "step": 771
    },
    {
      "epoch": 0.3574927463876649,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002791001855287569,
      "loss": 1.0362,
      "step": 772
    },
    {
      "epoch": 0.3579558198933484,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002790723562152133,
      "loss": 0.9185,
      "step": 773
    },
    {
      "epoch": 0.3584188933990319,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00027904452690166975,
      "loss": 0.9679,
      "step": 774
    },
    {
      "epoch": 0.3588819669047154,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027901669758812614,
      "loss": 0.9704,
      "step": 775
    },
    {
      "epoch": 0.3593450404103989,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00027898886827458253,
      "loss": 0.8815,
      "step": 776
    },
    {
      "epoch": 0.3598081139160824,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002789610389610389,
      "loss": 0.904,
      "step": 777
    },
    {
      "epoch": 0.3602711874217659,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00027893320964749536,
      "loss": 1.0133,
      "step": 778
    },
    {
      "epoch": 0.3607342609274494,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00027890538033395175,
      "loss": 0.9082,
      "step": 779
    },
    {
      "epoch": 0.3611973344331329,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00027887755102040814,
      "loss": 0.9191,
      "step": 780
    },
    {
      "epoch": 0.36166040793881643,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027884972170686453,
      "loss": 0.8675,
      "step": 781
    },
    {
      "epoch": 0.3621234814444999,
      "grad_norm": 0.5,
      "learning_rate": 0.000278821892393321,
      "loss": 0.7816,
      "step": 782
    },
    {
      "epoch": 0.36258655495018344,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00027879406307977736,
      "loss": 0.9203,
      "step": 783
    },
    {
      "epoch": 0.3630496284558669,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002787662337662337,
      "loss": 0.9932,
      "step": 784
    },
    {
      "epoch": 0.36351270196155044,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00027873840445269014,
      "loss": 0.9654,
      "step": 785
    },
    {
      "epoch": 0.3639757754672339,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00027871057513914653,
      "loss": 0.9495,
      "step": 786
    },
    {
      "epoch": 0.36443884897291745,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002786827458256029,
      "loss": 1.025,
      "step": 787
    },
    {
      "epoch": 0.3649019224786009,
      "grad_norm": 2.046875,
      "learning_rate": 0.0002786549165120593,
      "loss": 0.9684,
      "step": 788
    },
    {
      "epoch": 0.36536499598428446,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00027862708719851576,
      "loss": 0.9196,
      "step": 789
    },
    {
      "epoch": 0.36582806948996793,
      "grad_norm": 0.71875,
      "learning_rate": 0.00027859925788497215,
      "loss": 0.9163,
      "step": 790
    },
    {
      "epoch": 0.36629114299565146,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00027857142857142854,
      "loss": 0.8266,
      "step": 791
    },
    {
      "epoch": 0.36675421650133494,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002785435992578849,
      "loss": 0.8966,
      "step": 792
    },
    {
      "epoch": 0.36721729000701847,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00027851576994434137,
      "loss": 1.0179,
      "step": 793
    },
    {
      "epoch": 0.36768036351270195,
      "grad_norm": 7.15625,
      "learning_rate": 0.00027848794063079776,
      "loss": 1.6885,
      "step": 794
    },
    {
      "epoch": 0.3681434370183855,
      "grad_norm": 0.71875,
      "learning_rate": 0.00027846011131725415,
      "loss": 0.9134,
      "step": 795
    },
    {
      "epoch": 0.36860651052406895,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00027843228200371054,
      "loss": 1.0335,
      "step": 796
    },
    {
      "epoch": 0.3690695840297525,
      "grad_norm": 0.578125,
      "learning_rate": 0.000278404452690167,
      "loss": 0.9223,
      "step": 797
    },
    {
      "epoch": 0.36953265753543596,
      "grad_norm": 0.609375,
      "learning_rate": 0.00027837662337662337,
      "loss": 0.9364,
      "step": 798
    },
    {
      "epoch": 0.3699957310411195,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00027834879406307976,
      "loss": 0.8816,
      "step": 799
    },
    {
      "epoch": 0.37045880454680297,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00027832096474953615,
      "loss": 0.6585,
      "step": 800
    },
    {
      "epoch": 0.3709218780524865,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00027829313543599254,
      "loss": 0.9149,
      "step": 801
    },
    {
      "epoch": 0.37138495155817,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000278265306122449,
      "loss": 0.8897,
      "step": 802
    },
    {
      "epoch": 0.3718480250638535,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002782374768089054,
      "loss": 0.7975,
      "step": 803
    },
    {
      "epoch": 0.372311098569537,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00027820964749536176,
      "loss": 0.9013,
      "step": 804
    },
    {
      "epoch": 0.3727741720752205,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00027818181818181815,
      "loss": 0.9962,
      "step": 805
    },
    {
      "epoch": 0.373237245580904,
      "grad_norm": 0.609375,
      "learning_rate": 0.00027815398886827454,
      "loss": 0.8854,
      "step": 806
    },
    {
      "epoch": 0.3737003190865875,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00027812615955473093,
      "loss": 0.8581,
      "step": 807
    },
    {
      "epoch": 0.374163392592271,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002780983302411874,
      "loss": 0.9343,
      "step": 808
    },
    {
      "epoch": 0.3746264660979545,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00027807050092764377,
      "loss": 0.8066,
      "step": 809
    },
    {
      "epoch": 0.375089539603638,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00027804267161410015,
      "loss": 0.9431,
      "step": 810
    },
    {
      "epoch": 0.37555261310932153,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00027801484230055654,
      "loss": 0.9581,
      "step": 811
    },
    {
      "epoch": 0.376015686615005,
      "grad_norm": 0.59375,
      "learning_rate": 0.00027798701298701293,
      "loss": 0.925,
      "step": 812
    },
    {
      "epoch": 0.37647876012068854,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002779591836734694,
      "loss": 0.8762,
      "step": 813
    },
    {
      "epoch": 0.376941833626372,
      "grad_norm": 1.28125,
      "learning_rate": 0.00027793135435992577,
      "loss": 0.882,
      "step": 814
    },
    {
      "epoch": 0.37740490713205554,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00027790352504638216,
      "loss": 0.7994,
      "step": 815
    },
    {
      "epoch": 0.377867980637739,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00027787569573283855,
      "loss": 0.9536,
      "step": 816
    },
    {
      "epoch": 0.37833105414342255,
      "grad_norm": 0.65625,
      "learning_rate": 0.000277847866419295,
      "loss": 0.8753,
      "step": 817
    },
    {
      "epoch": 0.378794127649106,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002778200371057514,
      "loss": 1.046,
      "step": 818
    },
    {
      "epoch": 0.37925720115478956,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00027779220779220777,
      "loss": 0.9109,
      "step": 819
    },
    {
      "epoch": 0.37972027466047303,
      "grad_norm": 0.765625,
      "learning_rate": 0.00027776437847866416,
      "loss": 1.0317,
      "step": 820
    },
    {
      "epoch": 0.38018334816615657,
      "grad_norm": 1.4765625,
      "learning_rate": 0.0002777365491651206,
      "loss": 0.8885,
      "step": 821
    },
    {
      "epoch": 0.38064642167184004,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000277708719851577,
      "loss": 0.9691,
      "step": 822
    },
    {
      "epoch": 0.38110949517752357,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002776808905380334,
      "loss": 1.0319,
      "step": 823
    },
    {
      "epoch": 0.38157256868320705,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00027765306122448977,
      "loss": 0.8745,
      "step": 824
    },
    {
      "epoch": 0.3820356421888906,
      "grad_norm": 0.625,
      "learning_rate": 0.0002776252319109462,
      "loss": 0.8946,
      "step": 825
    },
    {
      "epoch": 0.38249871569457405,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00027759740259740255,
      "loss": 0.8988,
      "step": 826
    },
    {
      "epoch": 0.3829617892002576,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00027756957328385894,
      "loss": 0.8293,
      "step": 827
    },
    {
      "epoch": 0.38342486270594106,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002775417439703154,
      "loss": 0.9226,
      "step": 828
    },
    {
      "epoch": 0.3838879362116246,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002775139146567718,
      "loss": 1.06,
      "step": 829
    },
    {
      "epoch": 0.3843510097173081,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00027748608534322816,
      "loss": 0.8573,
      "step": 830
    },
    {
      "epoch": 0.3848140832229916,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00027745825602968455,
      "loss": 0.7888,
      "step": 831
    },
    {
      "epoch": 0.38527715672867513,
      "grad_norm": 0.828125,
      "learning_rate": 0.000277430426716141,
      "loss": 0.8904,
      "step": 832
    },
    {
      "epoch": 0.3857402302343586,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002774025974025974,
      "loss": 0.9322,
      "step": 833
    },
    {
      "epoch": 0.38620330374004214,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002773747680890538,
      "loss": 0.8884,
      "step": 834
    },
    {
      "epoch": 0.3866663772457256,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027734693877551017,
      "loss": 0.8727,
      "step": 835
    },
    {
      "epoch": 0.38712945075140914,
      "grad_norm": 1.5625,
      "learning_rate": 0.0002773191094619666,
      "loss": 0.9404,
      "step": 836
    },
    {
      "epoch": 0.3875925242570926,
      "grad_norm": 0.5703125,
      "learning_rate": 0.000277291280148423,
      "loss": 0.7936,
      "step": 837
    },
    {
      "epoch": 0.38805559776277615,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002772634508348794,
      "loss": 0.9338,
      "step": 838
    },
    {
      "epoch": 0.3885186712684596,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002772356215213358,
      "loss": 0.9495,
      "step": 839
    },
    {
      "epoch": 0.38898174477414316,
      "grad_norm": 0.5625,
      "learning_rate": 0.00027720779220779217,
      "loss": 0.8312,
      "step": 840
    },
    {
      "epoch": 0.38944481827982663,
      "grad_norm": 2.3125,
      "learning_rate": 0.0002771799628942486,
      "loss": 1.0468,
      "step": 841
    },
    {
      "epoch": 0.38990789178551016,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00027715213358070495,
      "loss": 1.0939,
      "step": 842
    },
    {
      "epoch": 0.39037096529119364,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002771243042671614,
      "loss": 0.9895,
      "step": 843
    },
    {
      "epoch": 0.39083403879687717,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0002770964749536178,
      "loss": 0.8596,
      "step": 844
    },
    {
      "epoch": 0.39129711230256065,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00027706864564007417,
      "loss": 0.8481,
      "step": 845
    },
    {
      "epoch": 0.3917601858082442,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027704081632653056,
      "loss": 0.9755,
      "step": 846
    },
    {
      "epoch": 0.39222325931392765,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000277012987012987,
      "loss": 0.9995,
      "step": 847
    },
    {
      "epoch": 0.3926863328196112,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002769851576994434,
      "loss": 0.9025,
      "step": 848
    },
    {
      "epoch": 0.39314940632529466,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002769573283858998,
      "loss": 1.0171,
      "step": 849
    },
    {
      "epoch": 0.3936124798309782,
      "grad_norm": 0.859375,
      "learning_rate": 0.00027692949907235617,
      "loss": 0.749,
      "step": 850
    },
    {
      "epoch": 0.39407555333666167,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0002769016697588126,
      "loss": 0.999,
      "step": 851
    },
    {
      "epoch": 0.3945386268423452,
      "grad_norm": 0.494140625,
      "learning_rate": 0.000276873840445269,
      "loss": 0.8137,
      "step": 852
    },
    {
      "epoch": 0.3950017003480287,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0002768460111317254,
      "loss": 0.9603,
      "step": 853
    },
    {
      "epoch": 0.3954647738537122,
      "grad_norm": 1.2734375,
      "learning_rate": 0.0002768181818181818,
      "loss": 0.9582,
      "step": 854
    },
    {
      "epoch": 0.3959278473593957,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002767903525046382,
      "loss": 0.8601,
      "step": 855
    },
    {
      "epoch": 0.3963909208650792,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002767625231910946,
      "loss": 0.9711,
      "step": 856
    },
    {
      "epoch": 0.3968539943707627,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000276734693877551,
      "loss": 0.9738,
      "step": 857
    },
    {
      "epoch": 0.3973170678764462,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002767068645640074,
      "loss": 0.9109,
      "step": 858
    },
    {
      "epoch": 0.3977801413821297,
      "grad_norm": 2.015625,
      "learning_rate": 0.0002766790352504638,
      "loss": 1.023,
      "step": 859
    },
    {
      "epoch": 0.3982432148878132,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00027665120593692023,
      "loss": 0.8713,
      "step": 860
    },
    {
      "epoch": 0.3987062883934967,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00027662337662337657,
      "loss": 0.8808,
      "step": 861
    },
    {
      "epoch": 0.39916936189918023,
      "grad_norm": 2.6875,
      "learning_rate": 0.000276595547309833,
      "loss": 0.9879,
      "step": 862
    },
    {
      "epoch": 0.3996324354048637,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002765677179962894,
      "loss": 1.0868,
      "step": 863
    },
    {
      "epoch": 0.40009550891054724,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002765398886827458,
      "loss": 0.9246,
      "step": 864
    },
    {
      "epoch": 0.4005585824162307,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0002765120593692022,
      "loss": 0.9316,
      "step": 865
    },
    {
      "epoch": 0.40102165592191424,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002764842300556586,
      "loss": 0.9043,
      "step": 866
    },
    {
      "epoch": 0.4014847294275977,
      "grad_norm": 0.7890625,
      "learning_rate": 0.000276456400742115,
      "loss": 0.8765,
      "step": 867
    },
    {
      "epoch": 0.40194780293328125,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002764285714285714,
      "loss": 0.8499,
      "step": 868
    },
    {
      "epoch": 0.4024108764389647,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002764007421150278,
      "loss": 0.8102,
      "step": 869
    },
    {
      "epoch": 0.40287394994464826,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002763729128014842,
      "loss": 1.0281,
      "step": 870
    },
    {
      "epoch": 0.40333702345033173,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002763450834879406,
      "loss": 0.709,
      "step": 871
    },
    {
      "epoch": 0.40380009695601526,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000276317254174397,
      "loss": 0.91,
      "step": 872
    },
    {
      "epoch": 0.40426317046169874,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002762894248608534,
      "loss": 0.8671,
      "step": 873
    },
    {
      "epoch": 0.40472624396738227,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002762615955473098,
      "loss": 0.8078,
      "step": 874
    },
    {
      "epoch": 0.40518931747306575,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00027623376623376624,
      "loss": 0.7912,
      "step": 875
    },
    {
      "epoch": 0.4056523909787493,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00027620593692022263,
      "loss": 0.816,
      "step": 876
    },
    {
      "epoch": 0.40611546448443275,
      "grad_norm": 0.609375,
      "learning_rate": 0.000276178107606679,
      "loss": 1.0745,
      "step": 877
    },
    {
      "epoch": 0.4065785379901163,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002761502782931354,
      "loss": 0.9417,
      "step": 878
    },
    {
      "epoch": 0.40704161149579976,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00027612244897959185,
      "loss": 0.9288,
      "step": 879
    },
    {
      "epoch": 0.4075046850014833,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002760946196660482,
      "loss": 1.0386,
      "step": 880
    },
    {
      "epoch": 0.40796775850716677,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00027606679035250463,
      "loss": 0.8137,
      "step": 881
    },
    {
      "epoch": 0.4084308320128503,
      "grad_norm": 0.55078125,
      "learning_rate": 0.000276038961038961,
      "loss": 0.9423,
      "step": 882
    },
    {
      "epoch": 0.4088939055185338,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002760111317254174,
      "loss": 1.03,
      "step": 883
    },
    {
      "epoch": 0.4093569790242173,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002759833024118738,
      "loss": 0.9348,
      "step": 884
    },
    {
      "epoch": 0.4098200525299008,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002759554730983302,
      "loss": 0.911,
      "step": 885
    },
    {
      "epoch": 0.4102831260355843,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00027592764378478663,
      "loss": 0.8568,
      "step": 886
    },
    {
      "epoch": 0.4107461995412678,
      "grad_norm": 0.5703125,
      "learning_rate": 0.000275899814471243,
      "loss": 0.8108,
      "step": 887
    },
    {
      "epoch": 0.4112092730469513,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002758719851576994,
      "loss": 0.9997,
      "step": 888
    },
    {
      "epoch": 0.4116723465526348,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002758441558441558,
      "loss": 0.9632,
      "step": 889
    },
    {
      "epoch": 0.4121354200583183,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00027581632653061224,
      "loss": 1.0424,
      "step": 890
    },
    {
      "epoch": 0.4125984935640018,
      "grad_norm": 0.671875,
      "learning_rate": 0.00027578849721706863,
      "loss": 0.9762,
      "step": 891
    },
    {
      "epoch": 0.41306156706968533,
      "grad_norm": 0.63671875,
      "learning_rate": 0.000275760667903525,
      "loss": 0.8851,
      "step": 892
    },
    {
      "epoch": 0.4135246405753688,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002757328385899814,
      "loss": 0.9297,
      "step": 893
    },
    {
      "epoch": 0.41398771408105234,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00027570500927643786,
      "loss": 0.9085,
      "step": 894
    },
    {
      "epoch": 0.4144507875867358,
      "grad_norm": 0.59375,
      "learning_rate": 0.00027567717996289425,
      "loss": 0.9156,
      "step": 895
    },
    {
      "epoch": 0.41491386109241934,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0002756493506493506,
      "loss": 1.0165,
      "step": 896
    },
    {
      "epoch": 0.4153769345981028,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000275621521335807,
      "loss": 1.0134,
      "step": 897
    },
    {
      "epoch": 0.41584000810378635,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002755936920222634,
      "loss": 0.9382,
      "step": 898
    },
    {
      "epoch": 0.4163030816094698,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002755658627087198,
      "loss": 0.8987,
      "step": 899
    },
    {
      "epoch": 0.41676615511515336,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002755380333951762,
      "loss": 0.8654,
      "step": 900
    },
    {
      "epoch": 0.41722922862083683,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027551020408163264,
      "loss": 1.0245,
      "step": 901
    },
    {
      "epoch": 0.41769230212652037,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027548237476808903,
      "loss": 0.9522,
      "step": 902
    },
    {
      "epoch": 0.41815537563220384,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002754545454545454,
      "loss": 1.0534,
      "step": 903
    },
    {
      "epoch": 0.41861844913788737,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002754267161410018,
      "loss": 1.0193,
      "step": 904
    },
    {
      "epoch": 0.4190815226435709,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027539888682745825,
      "loss": 0.8942,
      "step": 905
    },
    {
      "epoch": 0.4195445961492544,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00027537105751391464,
      "loss": 0.7829,
      "step": 906
    },
    {
      "epoch": 0.4200076696549379,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027534322820037103,
      "loss": 0.9081,
      "step": 907
    },
    {
      "epoch": 0.4204707431606214,
      "grad_norm": 1.453125,
      "learning_rate": 0.0002753153988868274,
      "loss": 0.9934,
      "step": 908
    },
    {
      "epoch": 0.4209338166663049,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00027528756957328386,
      "loss": 0.857,
      "step": 909
    },
    {
      "epoch": 0.4213968901719884,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027525974025974025,
      "loss": 0.8913,
      "step": 910
    },
    {
      "epoch": 0.4218599636776719,
      "grad_norm": 0.515625,
      "learning_rate": 0.00027523191094619664,
      "loss": 0.8281,
      "step": 911
    },
    {
      "epoch": 0.4223230371833554,
      "grad_norm": 1.6328125,
      "learning_rate": 0.00027520408163265303,
      "loss": 0.9737,
      "step": 912
    },
    {
      "epoch": 0.42278611068903893,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002751762523191094,
      "loss": 0.8945,
      "step": 913
    },
    {
      "epoch": 0.4232491841947224,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027514842300556587,
      "loss": 0.8518,
      "step": 914
    },
    {
      "epoch": 0.42371225770040594,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002751205936920222,
      "loss": 0.8799,
      "step": 915
    },
    {
      "epoch": 0.4241753312060894,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00027509276437847865,
      "loss": 0.9194,
      "step": 916
    },
    {
      "epoch": 0.42463840471177294,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00027506493506493504,
      "loss": 1.1086,
      "step": 917
    },
    {
      "epoch": 0.4251014782174564,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002750371057513914,
      "loss": 0.8782,
      "step": 918
    },
    {
      "epoch": 0.42556455172313995,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002750092764378478,
      "loss": 0.9893,
      "step": 919
    },
    {
      "epoch": 0.4260276252288234,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027498144712430426,
      "loss": 0.8976,
      "step": 920
    },
    {
      "epoch": 0.42649069873450696,
      "grad_norm": 0.609375,
      "learning_rate": 0.00027495361781076065,
      "loss": 0.8763,
      "step": 921
    },
    {
      "epoch": 0.42695377224019043,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027492578849721704,
      "loss": 0.9036,
      "step": 922
    },
    {
      "epoch": 0.42741684574587396,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002748979591836734,
      "loss": 0.8272,
      "step": 923
    },
    {
      "epoch": 0.42787991925155744,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00027487012987012987,
      "loss": 0.8817,
      "step": 924
    },
    {
      "epoch": 0.42834299275724097,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00027484230055658626,
      "loss": 0.9564,
      "step": 925
    },
    {
      "epoch": 0.42880606626292445,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00027481447124304265,
      "loss": 0.8133,
      "step": 926
    },
    {
      "epoch": 0.429269139768608,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00027478664192949904,
      "loss": 0.8711,
      "step": 927
    },
    {
      "epoch": 0.42973221327429145,
      "grad_norm": 0.671875,
      "learning_rate": 0.00027475881261595543,
      "loss": 0.9706,
      "step": 928
    },
    {
      "epoch": 0.430195286779975,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002747309833024119,
      "loss": 0.8538,
      "step": 929
    },
    {
      "epoch": 0.43065836028565846,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00027470315398886826,
      "loss": 0.8699,
      "step": 930
    },
    {
      "epoch": 0.431121433791342,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027467532467532465,
      "loss": 0.9298,
      "step": 931
    },
    {
      "epoch": 0.43158450729702547,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00027464749536178104,
      "loss": 0.9407,
      "step": 932
    },
    {
      "epoch": 0.432047580802709,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0002746196660482375,
      "loss": 0.798,
      "step": 933
    },
    {
      "epoch": 0.4325106543083925,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002745918367346939,
      "loss": 0.8689,
      "step": 934
    },
    {
      "epoch": 0.432973727814076,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00027456400742115026,
      "loss": 0.7335,
      "step": 935
    },
    {
      "epoch": 0.4334368013197595,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027453617810760665,
      "loss": 0.9024,
      "step": 936
    },
    {
      "epoch": 0.433899874825443,
      "grad_norm": 0.59375,
      "learning_rate": 0.00027450834879406304,
      "loss": 0.8904,
      "step": 937
    },
    {
      "epoch": 0.4343629483311265,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027448051948051943,
      "loss": 0.8556,
      "step": 938
    },
    {
      "epoch": 0.43482602183681,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002744526901669758,
      "loss": 0.9313,
      "step": 939
    },
    {
      "epoch": 0.4352890953424935,
      "grad_norm": 0.625,
      "learning_rate": 0.00027442486085343227,
      "loss": 0.8814,
      "step": 940
    },
    {
      "epoch": 0.435752168848177,
      "grad_norm": 0.59375,
      "learning_rate": 0.00027439703153988866,
      "loss": 0.9109,
      "step": 941
    },
    {
      "epoch": 0.4362152423538605,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00027436920222634505,
      "loss": 0.8876,
      "step": 942
    },
    {
      "epoch": 0.43667831585954403,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00027434137291280144,
      "loss": 0.8555,
      "step": 943
    },
    {
      "epoch": 0.4371413893652275,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002743135435992579,
      "loss": 0.9331,
      "step": 944
    },
    {
      "epoch": 0.43760446287091104,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00027428571428571427,
      "loss": 1.1068,
      "step": 945
    },
    {
      "epoch": 0.4380675363765945,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00027425788497217066,
      "loss": 0.9349,
      "step": 946
    },
    {
      "epoch": 0.43853060988227804,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00027423005565862705,
      "loss": 0.7331,
      "step": 947
    },
    {
      "epoch": 0.4389936833879615,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002742022263450835,
      "loss": 0.9064,
      "step": 948
    },
    {
      "epoch": 0.43945675689364505,
      "grad_norm": 1.0,
      "learning_rate": 0.0002741743970315399,
      "loss": 0.8638,
      "step": 949
    },
    {
      "epoch": 0.4399198303993285,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027414656771799627,
      "loss": 0.9135,
      "step": 950
    },
    {
      "epoch": 0.44038290390501206,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00027411873840445266,
      "loss": 0.8913,
      "step": 951
    },
    {
      "epoch": 0.44084597741069553,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002740909090909091,
      "loss": 0.8354,
      "step": 952
    },
    {
      "epoch": 0.44130905091637906,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002740630797773655,
      "loss": 0.8877,
      "step": 953
    },
    {
      "epoch": 0.44177212442206254,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027403525046382183,
      "loss": 0.9905,
      "step": 954
    },
    {
      "epoch": 0.44223519792774607,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002740074211502783,
      "loss": 0.9288,
      "step": 955
    },
    {
      "epoch": 0.44269827143342955,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00027397959183673466,
      "loss": 1.0038,
      "step": 956
    },
    {
      "epoch": 0.4431613449391131,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027395176252319105,
      "loss": 0.909,
      "step": 957
    },
    {
      "epoch": 0.44362441844479655,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00027392393320964744,
      "loss": 0.9036,
      "step": 958
    },
    {
      "epoch": 0.4440874919504801,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002738961038961039,
      "loss": 0.9963,
      "step": 959
    },
    {
      "epoch": 0.44455056545616356,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002738682745825603,
      "loss": 0.8635,
      "step": 960
    },
    {
      "epoch": 0.4450136389618471,
      "grad_norm": 0.890625,
      "learning_rate": 0.00027384044526901667,
      "loss": 0.9694,
      "step": 961
    },
    {
      "epoch": 0.44547671246753057,
      "grad_norm": 0.640625,
      "learning_rate": 0.00027381261595547306,
      "loss": 0.9335,
      "step": 962
    },
    {
      "epoch": 0.4459397859732141,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002737847866419295,
      "loss": 0.9084,
      "step": 963
    },
    {
      "epoch": 0.4464028594788976,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002737569573283859,
      "loss": 0.8756,
      "step": 964
    },
    {
      "epoch": 0.4468659329845811,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002737291280148423,
      "loss": 0.8986,
      "step": 965
    },
    {
      "epoch": 0.4473290064902646,
      "grad_norm": 1.609375,
      "learning_rate": 0.00027370129870129867,
      "loss": 1.1057,
      "step": 966
    },
    {
      "epoch": 0.4477920799959481,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00027367346938775506,
      "loss": 0.8449,
      "step": 967
    },
    {
      "epoch": 0.4482551535016316,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002736456400742115,
      "loss": 1.1095,
      "step": 968
    },
    {
      "epoch": 0.4487182270073151,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002736178107606679,
      "loss": 0.9099,
      "step": 969
    },
    {
      "epoch": 0.4491813005129986,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002735899814471243,
      "loss": 1.0538,
      "step": 970
    },
    {
      "epoch": 0.4496443740186821,
      "grad_norm": 0.671875,
      "learning_rate": 0.00027356215213358067,
      "loss": 0.9822,
      "step": 971
    },
    {
      "epoch": 0.4501074475243656,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002735343228200371,
      "loss": 1.0391,
      "step": 972
    },
    {
      "epoch": 0.45057052103004913,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00027350649350649345,
      "loss": 0.9036,
      "step": 973
    },
    {
      "epoch": 0.4510335945357326,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002734786641929499,
      "loss": 0.8579,
      "step": 974
    },
    {
      "epoch": 0.45149666804141614,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002734508348794063,
      "loss": 0.965,
      "step": 975
    },
    {
      "epoch": 0.4519597415470996,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027342300556586267,
      "loss": 0.868,
      "step": 976
    },
    {
      "epoch": 0.45242281505278314,
      "grad_norm": 0.71875,
      "learning_rate": 0.00027339517625231906,
      "loss": 1.0139,
      "step": 977
    },
    {
      "epoch": 0.4528858885584666,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002733673469387755,
      "loss": 0.9397,
      "step": 978
    },
    {
      "epoch": 0.45334896206415015,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002733395176252319,
      "loss": 1.0119,
      "step": 979
    },
    {
      "epoch": 0.4538120355698337,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002733116883116883,
      "loss": 0.9415,
      "step": 980
    },
    {
      "epoch": 0.45427510907551716,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002732838589981447,
      "loss": 0.9749,
      "step": 981
    },
    {
      "epoch": 0.4547381825812007,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00027325602968460106,
      "loss": 0.8143,
      "step": 982
    },
    {
      "epoch": 0.45520125608688417,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002732282003710575,
      "loss": 0.7684,
      "step": 983
    },
    {
      "epoch": 0.4556643295925677,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002732003710575139,
      "loss": 0.8674,
      "step": 984
    },
    {
      "epoch": 0.45612740309825117,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002731725417439703,
      "loss": 0.8558,
      "step": 985
    },
    {
      "epoch": 0.4565904766039347,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002731447124304267,
      "loss": 0.9411,
      "step": 986
    },
    {
      "epoch": 0.4570535501096182,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002731168831168831,
      "loss": 0.9786,
      "step": 987
    },
    {
      "epoch": 0.4575166236153017,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002730890538033395,
      "loss": 0.8279,
      "step": 988
    },
    {
      "epoch": 0.4579796971209852,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002730612244897959,
      "loss": 0.9339,
      "step": 989
    },
    {
      "epoch": 0.4584427706266687,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002730333951762523,
      "loss": 0.9254,
      "step": 990
    },
    {
      "epoch": 0.4589058441323522,
      "grad_norm": 0.515625,
      "learning_rate": 0.00027300556586270873,
      "loss": 0.8234,
      "step": 991
    },
    {
      "epoch": 0.4593689176380357,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00027297773654916507,
      "loss": 0.9311,
      "step": 992
    },
    {
      "epoch": 0.4598319911437192,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002729499072356215,
      "loss": 0.9914,
      "step": 993
    },
    {
      "epoch": 0.46029506464940273,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002729220779220779,
      "loss": 0.8922,
      "step": 994
    },
    {
      "epoch": 0.4607581381550862,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002728942486085343,
      "loss": 0.8724,
      "step": 995
    },
    {
      "epoch": 0.46122121166076974,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002728664192949907,
      "loss": 0.8352,
      "step": 996
    },
    {
      "epoch": 0.4616842851664532,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027283858998144707,
      "loss": 1.0511,
      "step": 997
    },
    {
      "epoch": 0.46214735867213674,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002728107606679035,
      "loss": 0.8278,
      "step": 998
    },
    {
      "epoch": 0.4626104321778202,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002727829313543599,
      "loss": 0.8236,
      "step": 999
    },
    {
      "epoch": 0.46307350568350375,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002727551020408163,
      "loss": 0.8631,
      "step": 1000
    },
    {
      "epoch": 0.4635365791891872,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002727272727272727,
      "loss": 0.8189,
      "step": 1001
    },
    {
      "epoch": 0.46399965269487076,
      "grad_norm": 0.671875,
      "learning_rate": 0.00027269944341372913,
      "loss": 0.9788,
      "step": 1002
    },
    {
      "epoch": 0.46446272620055423,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002726716141001855,
      "loss": 1.0456,
      "step": 1003
    },
    {
      "epoch": 0.46492579970623776,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002726437847866419,
      "loss": 0.9953,
      "step": 1004
    },
    {
      "epoch": 0.46538887321192124,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0002726159554730983,
      "loss": 1.0641,
      "step": 1005
    },
    {
      "epoch": 0.46585194671760477,
      "grad_norm": 0.625,
      "learning_rate": 0.00027258812615955474,
      "loss": 0.9403,
      "step": 1006
    },
    {
      "epoch": 0.46631502022328825,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00027256029684601113,
      "loss": 0.9169,
      "step": 1007
    },
    {
      "epoch": 0.4667780937289718,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002725324675324675,
      "loss": 0.9066,
      "step": 1008
    },
    {
      "epoch": 0.46724116723465525,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002725046382189239,
      "loss": 0.82,
      "step": 1009
    },
    {
      "epoch": 0.4677042407403388,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002724768089053803,
      "loss": 1.0239,
      "step": 1010
    },
    {
      "epoch": 0.46816731424602226,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002724489795918367,
      "loss": 0.8715,
      "step": 1011
    },
    {
      "epoch": 0.4686303877517058,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002724211502782931,
      "loss": 0.9009,
      "step": 1012
    },
    {
      "epoch": 0.46909346125738927,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002723933209647495,
      "loss": 0.9311,
      "step": 1013
    },
    {
      "epoch": 0.4695565347630728,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002723654916512059,
      "loss": 0.8026,
      "step": 1014
    },
    {
      "epoch": 0.4700196082687563,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002723376623376623,
      "loss": 0.8671,
      "step": 1015
    },
    {
      "epoch": 0.4704826817744398,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002723098330241187,
      "loss": 0.7974,
      "step": 1016
    },
    {
      "epoch": 0.4709457552801233,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00027228200371057513,
      "loss": 0.7834,
      "step": 1017
    },
    {
      "epoch": 0.4714088287858068,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002722541743970315,
      "loss": 0.9616,
      "step": 1018
    },
    {
      "epoch": 0.4718719022914903,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002722263450834879,
      "loss": 0.9308,
      "step": 1019
    },
    {
      "epoch": 0.4723349757971738,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002721985157699443,
      "loss": 0.8717,
      "step": 1020
    },
    {
      "epoch": 0.4727980493028573,
      "grad_norm": 0.65625,
      "learning_rate": 0.00027217068645640075,
      "loss": 0.994,
      "step": 1021
    },
    {
      "epoch": 0.4732611228085408,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00027214285714285714,
      "loss": 0.9537,
      "step": 1022
    },
    {
      "epoch": 0.4737241963142243,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002721150278293135,
      "loss": 0.9299,
      "step": 1023
    },
    {
      "epoch": 0.47418726981990783,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002720871985157699,
      "loss": 0.8442,
      "step": 1024
    },
    {
      "epoch": 0.4746503433255913,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002720593692022263,
      "loss": 0.9829,
      "step": 1025
    },
    {
      "epoch": 0.47511341683127484,
      "grad_norm": 0.59375,
      "learning_rate": 0.00027203153988868275,
      "loss": 0.9113,
      "step": 1026
    },
    {
      "epoch": 0.4755764903369583,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002720037105751391,
      "loss": 0.8416,
      "step": 1027
    },
    {
      "epoch": 0.47603956384264184,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00027197588126159553,
      "loss": 0.9821,
      "step": 1028
    },
    {
      "epoch": 0.4765026373483253,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002719480519480519,
      "loss": 0.8415,
      "step": 1029
    },
    {
      "epoch": 0.47696571085400885,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002719202226345083,
      "loss": 0.8415,
      "step": 1030
    },
    {
      "epoch": 0.4774287843596923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002718923933209647,
      "loss": 0.9325,
      "step": 1031
    },
    {
      "epoch": 0.47789185786537586,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027186456400742114,
      "loss": 0.8691,
      "step": 1032
    },
    {
      "epoch": 0.47835493137105933,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00027183673469387753,
      "loss": 0.8659,
      "step": 1033
    },
    {
      "epoch": 0.47881800487674286,
      "grad_norm": 0.796875,
      "learning_rate": 0.0002718089053803339,
      "loss": 0.9396,
      "step": 1034
    },
    {
      "epoch": 0.47928107838242634,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0002717810760667903,
      "loss": 0.9861,
      "step": 1035
    },
    {
      "epoch": 0.47974415188810987,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00027175324675324675,
      "loss": 0.9234,
      "step": 1036
    },
    {
      "epoch": 0.48020722539379335,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00027172541743970314,
      "loss": 0.9445,
      "step": 1037
    },
    {
      "epoch": 0.4806702988994769,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027169758812615953,
      "loss": 0.9884,
      "step": 1038
    },
    {
      "epoch": 0.48113337240516035,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002716697588126159,
      "loss": 0.834,
      "step": 1039
    },
    {
      "epoch": 0.4815964459108439,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002716419294990723,
      "loss": 0.9163,
      "step": 1040
    },
    {
      "epoch": 0.48205951941652736,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00027161410018552876,
      "loss": 0.8475,
      "step": 1041
    },
    {
      "epoch": 0.4825225929222109,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00027158627087198515,
      "loss": 0.9022,
      "step": 1042
    },
    {
      "epoch": 0.48298566642789437,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00027155844155844154,
      "loss": 0.8944,
      "step": 1043
    },
    {
      "epoch": 0.4834487399335779,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002715306122448979,
      "loss": 0.8459,
      "step": 1044
    },
    {
      "epoch": 0.4839118134392614,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00027150278293135437,
      "loss": 0.8633,
      "step": 1045
    },
    {
      "epoch": 0.4843748869449449,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00027147495361781076,
      "loss": 0.8403,
      "step": 1046
    },
    {
      "epoch": 0.4848379604506284,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00027144712430426715,
      "loss": 1.0818,
      "step": 1047
    },
    {
      "epoch": 0.4853010339563119,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00027141929499072354,
      "loss": 0.8475,
      "step": 1048
    },
    {
      "epoch": 0.4857641074619954,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002713914656771799,
      "loss": 0.8779,
      "step": 1049
    },
    {
      "epoch": 0.4862271809676789,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002713636363636363,
      "loss": 0.9531,
      "step": 1050
    },
    {
      "epoch": 0.4866902544733624,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027133580705009276,
      "loss": 0.9035,
      "step": 1051
    },
    {
      "epoch": 0.4871533279790459,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027130797773654915,
      "loss": 1.011,
      "step": 1052
    },
    {
      "epoch": 0.4876164014847294,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027128014842300554,
      "loss": 0.9484,
      "step": 1053
    },
    {
      "epoch": 0.48807947499041293,
      "grad_norm": 0.484375,
      "learning_rate": 0.00027125231910946193,
      "loss": 0.7705,
      "step": 1054
    },
    {
      "epoch": 0.48854254849609646,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002712244897959183,
      "loss": 0.9471,
      "step": 1055
    },
    {
      "epoch": 0.48900562200177994,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027119666048237476,
      "loss": 0.9518,
      "step": 1056
    },
    {
      "epoch": 0.48946869550746347,
      "grad_norm": 0.578125,
      "learning_rate": 0.00027116883116883115,
      "loss": 0.8546,
      "step": 1057
    },
    {
      "epoch": 0.48993176901314694,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00027114100185528754,
      "loss": 0.8353,
      "step": 1058
    },
    {
      "epoch": 0.4903948425188305,
      "grad_norm": 0.546875,
      "learning_rate": 0.00027111317254174393,
      "loss": 0.838,
      "step": 1059
    },
    {
      "epoch": 0.49085791602451395,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002710853432282004,
      "loss": 0.8728,
      "step": 1060
    },
    {
      "epoch": 0.4913209895301975,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00027105751391465676,
      "loss": 0.8273,
      "step": 1061
    },
    {
      "epoch": 0.49178406303588096,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00027102968460111315,
      "loss": 0.912,
      "step": 1062
    },
    {
      "epoch": 0.4922471365415645,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00027100185528756954,
      "loss": 0.9424,
      "step": 1063
    },
    {
      "epoch": 0.49271021004724797,
      "grad_norm": 1.375,
      "learning_rate": 0.000270974025974026,
      "loss": 0.8393,
      "step": 1064
    },
    {
      "epoch": 0.4931732835529315,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002709461966604824,
      "loss": 0.8524,
      "step": 1065
    },
    {
      "epoch": 0.49363635705861497,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0002709183673469387,
      "loss": 0.9686,
      "step": 1066
    },
    {
      "epoch": 0.4940994305642985,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00027089053803339516,
      "loss": 0.8258,
      "step": 1067
    },
    {
      "epoch": 0.494562504069982,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00027086270871985155,
      "loss": 1.0619,
      "step": 1068
    },
    {
      "epoch": 0.4950255775756655,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00027083487940630794,
      "loss": 0.7922,
      "step": 1069
    },
    {
      "epoch": 0.495488651081349,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002708070500927643,
      "loss": 0.869,
      "step": 1070
    },
    {
      "epoch": 0.4959517245870325,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00027077922077922077,
      "loss": 0.9393,
      "step": 1071
    },
    {
      "epoch": 0.496414798092716,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00027075139146567716,
      "loss": 0.9047,
      "step": 1072
    },
    {
      "epoch": 0.4968778715983995,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027072356215213355,
      "loss": 0.9657,
      "step": 1073
    },
    {
      "epoch": 0.497340945104083,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027069573283858994,
      "loss": 0.938,
      "step": 1074
    },
    {
      "epoch": 0.49780401860976653,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002706679035250464,
      "loss": 0.8856,
      "step": 1075
    },
    {
      "epoch": 0.49826709211545,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027064007421150277,
      "loss": 0.8482,
      "step": 1076
    },
    {
      "epoch": 0.49873016562113354,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00027061224489795916,
      "loss": 0.6988,
      "step": 1077
    },
    {
      "epoch": 0.499193239126817,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00027058441558441555,
      "loss": 0.7843,
      "step": 1078
    },
    {
      "epoch": 0.49965631263250054,
      "grad_norm": 0.828125,
      "learning_rate": 0.000270556586270872,
      "loss": 0.9708,
      "step": 1079
    },
    {
      "epoch": 0.5001193861381841,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002705287569573284,
      "loss": 0.8174,
      "step": 1080
    },
    {
      "epoch": 0.5005824596438675,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002705009276437848,
      "loss": 0.8981,
      "step": 1081
    },
    {
      "epoch": 0.501045533149551,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00027047309833024116,
      "loss": 0.8978,
      "step": 1082
    },
    {
      "epoch": 0.5015086066552346,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00027044526901669755,
      "loss": 0.8813,
      "step": 1083
    },
    {
      "epoch": 0.5019716801609181,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000270417439703154,
      "loss": 1.017,
      "step": 1084
    },
    {
      "epoch": 0.5024347536666015,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00027038961038961033,
      "loss": 0.9008,
      "step": 1085
    },
    {
      "epoch": 0.502897827172285,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0002703617810760668,
      "loss": 1.0395,
      "step": 1086
    },
    {
      "epoch": 0.5033609006779686,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00027033395176252317,
      "loss": 0.8292,
      "step": 1087
    },
    {
      "epoch": 0.5038239741836521,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00027030612244897956,
      "loss": 0.8458,
      "step": 1088
    },
    {
      "epoch": 0.5042870476893355,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00027027829313543594,
      "loss": 0.7412,
      "step": 1089
    },
    {
      "epoch": 0.504750121195019,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002702504638218924,
      "loss": 1.04,
      "step": 1090
    },
    {
      "epoch": 0.5052131947007026,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002702226345083488,
      "loss": 1.0405,
      "step": 1091
    },
    {
      "epoch": 0.5056762682063861,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00027019480519480517,
      "loss": 0.9558,
      "step": 1092
    },
    {
      "epoch": 0.5061393417120695,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00027016697588126156,
      "loss": 0.9017,
      "step": 1093
    },
    {
      "epoch": 0.5066024152177531,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00027013914656771795,
      "loss": 0.9541,
      "step": 1094
    },
    {
      "epoch": 0.5070654887234366,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002701113172541744,
      "loss": 0.9054,
      "step": 1095
    },
    {
      "epoch": 0.5075285622291201,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002700834879406308,
      "loss": 0.8386,
      "step": 1096
    },
    {
      "epoch": 0.5079916357348035,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00027005565862708717,
      "loss": 0.9676,
      "step": 1097
    },
    {
      "epoch": 0.5084547092404871,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00027002782931354356,
      "loss": 0.8146,
      "step": 1098
    },
    {
      "epoch": 0.5089177827461706,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00027,
      "loss": 0.9791,
      "step": 1099
    },
    {
      "epoch": 0.5093808562518541,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002699721706864564,
      "loss": 0.7496,
      "step": 1100
    },
    {
      "epoch": 0.5098439297575376,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002699443413729128,
      "loss": 0.97,
      "step": 1101
    },
    {
      "epoch": 0.5103070032632211,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026991651205936917,
      "loss": 0.9448,
      "step": 1102
    },
    {
      "epoch": 0.5107700767689046,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002698886827458256,
      "loss": 0.8452,
      "step": 1103
    },
    {
      "epoch": 0.5112331502745882,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00026986085343228195,
      "loss": 0.9063,
      "step": 1104
    },
    {
      "epoch": 0.5116962237802716,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002698330241187384,
      "loss": 0.8754,
      "step": 1105
    },
    {
      "epoch": 0.5121592972859551,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002698051948051948,
      "loss": 0.7791,
      "step": 1106
    },
    {
      "epoch": 0.5126223707916386,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002697773654916512,
      "loss": 0.8042,
      "step": 1107
    },
    {
      "epoch": 0.5130854442973222,
      "grad_norm": 1.03125,
      "learning_rate": 0.00026974953617810756,
      "loss": 1.0401,
      "step": 1108
    },
    {
      "epoch": 0.5135485178030056,
      "grad_norm": 5.1875,
      "learning_rate": 0.00026972170686456395,
      "loss": 0.9968,
      "step": 1109
    },
    {
      "epoch": 0.5140115913086891,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002696938775510204,
      "loss": 0.7618,
      "step": 1110
    },
    {
      "epoch": 0.5144746648143727,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002696660482374768,
      "loss": 0.8352,
      "step": 1111
    },
    {
      "epoch": 0.5149377383200562,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002696382189239332,
      "loss": 0.8848,
      "step": 1112
    },
    {
      "epoch": 0.5154008118257396,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00026961038961038957,
      "loss": 0.8655,
      "step": 1113
    },
    {
      "epoch": 0.5158638853314231,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000269582560296846,
      "loss": 0.955,
      "step": 1114
    },
    {
      "epoch": 0.5163269588371067,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002695547309833024,
      "loss": 0.8693,
      "step": 1115
    },
    {
      "epoch": 0.5167900323427902,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002695269016697588,
      "loss": 0.7835,
      "step": 1116
    },
    {
      "epoch": 0.5172531058484736,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002694990723562152,
      "loss": 1.0212,
      "step": 1117
    },
    {
      "epoch": 0.5177161793541571,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002694712430426716,
      "loss": 0.9802,
      "step": 1118
    },
    {
      "epoch": 0.5181792528598407,
      "grad_norm": 0.640625,
      "learning_rate": 0.000269443413729128,
      "loss": 0.9714,
      "step": 1119
    },
    {
      "epoch": 0.5186423263655242,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002694155844155844,
      "loss": 0.8906,
      "step": 1120
    },
    {
      "epoch": 0.5191053998712076,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002693877551020408,
      "loss": 0.9199,
      "step": 1121
    },
    {
      "epoch": 0.5195684733768912,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002693599257884972,
      "loss": 0.7877,
      "step": 1122
    },
    {
      "epoch": 0.5200315468825747,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00026933209647495357,
      "loss": 1.0793,
      "step": 1123
    },
    {
      "epoch": 0.5204946203882582,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026930426716140996,
      "loss": 0.8332,
      "step": 1124
    },
    {
      "epoch": 0.5209576938939416,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002692764378478664,
      "loss": 0.9282,
      "step": 1125
    },
    {
      "epoch": 0.5214207673996252,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002692486085343228,
      "loss": 0.7855,
      "step": 1126
    },
    {
      "epoch": 0.5218838409053087,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002692207792207792,
      "loss": 0.8301,
      "step": 1127
    },
    {
      "epoch": 0.5223469144109922,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002691929499072356,
      "loss": 0.896,
      "step": 1128
    },
    {
      "epoch": 0.5228099879166757,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000269165120593692,
      "loss": 0.8053,
      "step": 1129
    },
    {
      "epoch": 0.5232730614223592,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002691372912801484,
      "loss": 0.8267,
      "step": 1130
    },
    {
      "epoch": 0.5237361349280427,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002691094619666048,
      "loss": 0.9646,
      "step": 1131
    },
    {
      "epoch": 0.5241992084337262,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002690816326530612,
      "loss": 0.8898,
      "step": 1132
    },
    {
      "epoch": 0.5246622819394098,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00026905380333951763,
      "loss": 0.9585,
      "step": 1133
    },
    {
      "epoch": 0.5251253554450932,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000269025974025974,
      "loss": 0.7948,
      "step": 1134
    },
    {
      "epoch": 0.5255884289507767,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002689981447124304,
      "loss": 0.9105,
      "step": 1135
    },
    {
      "epoch": 0.5260515024564603,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002689703153988868,
      "loss": 0.8968,
      "step": 1136
    },
    {
      "epoch": 0.5265145759621438,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002689424860853432,
      "loss": 0.7854,
      "step": 1137
    },
    {
      "epoch": 0.5269776494678272,
      "grad_norm": 0.5625,
      "learning_rate": 0.00026891465677179963,
      "loss": 0.822,
      "step": 1138
    },
    {
      "epoch": 0.5274407229735107,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00026888682745825597,
      "loss": 0.9382,
      "step": 1139
    },
    {
      "epoch": 0.5279037964791943,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002688589981447124,
      "loss": 0.8639,
      "step": 1140
    },
    {
      "epoch": 0.5283668699848778,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002688311688311688,
      "loss": 0.7947,
      "step": 1141
    },
    {
      "epoch": 0.5288299434905612,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002688033395176252,
      "loss": 1.0982,
      "step": 1142
    },
    {
      "epoch": 0.5292930169962448,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002687755102040816,
      "loss": 0.7896,
      "step": 1143
    },
    {
      "epoch": 0.5297560905019283,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000268747680890538,
      "loss": 0.8804,
      "step": 1144
    },
    {
      "epoch": 0.5302191640076118,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002687198515769944,
      "loss": 0.8142,
      "step": 1145
    },
    {
      "epoch": 0.5306822375132952,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002686920222634508,
      "loss": 0.9316,
      "step": 1146
    },
    {
      "epoch": 0.5311453110189788,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002686641929499072,
      "loss": 0.987,
      "step": 1147
    },
    {
      "epoch": 0.5316083845246623,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00026863636363636364,
      "loss": 0.8099,
      "step": 1148
    },
    {
      "epoch": 0.5320714580303458,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00026860853432282,
      "loss": 0.8934,
      "step": 1149
    },
    {
      "epoch": 0.5325345315360293,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002685807050092764,
      "loss": 0.8356,
      "step": 1150
    },
    {
      "epoch": 0.5329976050417128,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002685528756957328,
      "loss": 0.9791,
      "step": 1151
    },
    {
      "epoch": 0.5334606785473963,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002685250463821892,
      "loss": 0.8509,
      "step": 1152
    },
    {
      "epoch": 0.5339237520530798,
      "grad_norm": 0.625,
      "learning_rate": 0.00026849721706864564,
      "loss": 0.844,
      "step": 1153
    },
    {
      "epoch": 0.5343868255587633,
      "grad_norm": 0.5625,
      "learning_rate": 0.00026846938775510203,
      "loss": 0.9032,
      "step": 1154
    },
    {
      "epoch": 0.5348498990644468,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002684415584415584,
      "loss": 0.9465,
      "step": 1155
    },
    {
      "epoch": 0.5353129725701303,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002684137291280148,
      "loss": 0.8897,
      "step": 1156
    },
    {
      "epoch": 0.5357760460758139,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026838589981447125,
      "loss": 0.8926,
      "step": 1157
    },
    {
      "epoch": 0.5362391195814973,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002683580705009276,
      "loss": 0.9541,
      "step": 1158
    },
    {
      "epoch": 0.5367021930871808,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00026833024118738403,
      "loss": 0.8264,
      "step": 1159
    },
    {
      "epoch": 0.5371652665928643,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002683024118738404,
      "loss": 0.8784,
      "step": 1160
    },
    {
      "epoch": 0.5376283400985479,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002682745825602968,
      "loss": 0.8538,
      "step": 1161
    },
    {
      "epoch": 0.5380914136042313,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002682467532467532,
      "loss": 0.9598,
      "step": 1162
    },
    {
      "epoch": 0.5385544871099148,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00026821892393320964,
      "loss": 0.7706,
      "step": 1163
    },
    {
      "epoch": 0.5390175606155984,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00026819109461966603,
      "loss": 0.8204,
      "step": 1164
    },
    {
      "epoch": 0.5394806341212819,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002681632653061224,
      "loss": 0.7413,
      "step": 1165
    },
    {
      "epoch": 0.5399437076269653,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002681354359925788,
      "loss": 0.8867,
      "step": 1166
    },
    {
      "epoch": 0.5404067811326488,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002681076066790352,
      "loss": 0.9313,
      "step": 1167
    },
    {
      "epoch": 0.5408698546383324,
      "grad_norm": 0.640625,
      "learning_rate": 0.00026807977736549165,
      "loss": 0.8569,
      "step": 1168
    },
    {
      "epoch": 0.5413329281440159,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00026805194805194803,
      "loss": 0.9993,
      "step": 1169
    },
    {
      "epoch": 0.5417960016496993,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002680241187384044,
      "loss": 0.8171,
      "step": 1170
    },
    {
      "epoch": 0.5422590751553829,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002679962894248608,
      "loss": 0.982,
      "step": 1171
    },
    {
      "epoch": 0.5427221486610664,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026796846011131726,
      "loss": 0.9658,
      "step": 1172
    },
    {
      "epoch": 0.5431852221667499,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00026794063079777365,
      "loss": 0.7356,
      "step": 1173
    },
    {
      "epoch": 0.5436482956724333,
      "grad_norm": 0.625,
      "learning_rate": 0.00026791280148423004,
      "loss": 0.9365,
      "step": 1174
    },
    {
      "epoch": 0.5441113691781169,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002678849721706864,
      "loss": 0.8231,
      "step": 1175
    },
    {
      "epoch": 0.5445744426838004,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00026785714285714287,
      "loss": 0.9759,
      "step": 1176
    },
    {
      "epoch": 0.5450375161894839,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00026782931354359926,
      "loss": 1.1614,
      "step": 1177
    },
    {
      "epoch": 0.5455005896951673,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00026780148423005565,
      "loss": 0.8625,
      "step": 1178
    },
    {
      "epoch": 0.5459636632008509,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00026777365491651204,
      "loss": 0.778,
      "step": 1179
    },
    {
      "epoch": 0.5464267367065344,
      "grad_norm": 0.5625,
      "learning_rate": 0.00026774582560296843,
      "loss": 0.822,
      "step": 1180
    },
    {
      "epoch": 0.5468898102122179,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002677179962894248,
      "loss": 0.8626,
      "step": 1181
    },
    {
      "epoch": 0.5473528837179014,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002676901669758812,
      "loss": 0.8432,
      "step": 1182
    },
    {
      "epoch": 0.5478159572235849,
      "grad_norm": 0.53125,
      "learning_rate": 0.00026766233766233765,
      "loss": 0.8798,
      "step": 1183
    },
    {
      "epoch": 0.5482790307292684,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00026763450834879404,
      "loss": 0.888,
      "step": 1184
    },
    {
      "epoch": 0.548742104234952,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00026760667903525043,
      "loss": 0.9464,
      "step": 1185
    },
    {
      "epoch": 0.5492051777406354,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0002675788497217068,
      "loss": 0.985,
      "step": 1186
    },
    {
      "epoch": 0.5496682512463189,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00026755102040816326,
      "loss": 0.9282,
      "step": 1187
    },
    {
      "epoch": 0.5501313247520024,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00026752319109461965,
      "loss": 0.8657,
      "step": 1188
    },
    {
      "epoch": 0.550594398257686,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026749536178107604,
      "loss": 0.9059,
      "step": 1189
    },
    {
      "epoch": 0.5510574717633694,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00026746753246753243,
      "loss": 1.0067,
      "step": 1190
    },
    {
      "epoch": 0.5515205452690529,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002674397031539889,
      "loss": 0.9125,
      "step": 1191
    },
    {
      "epoch": 0.5519836187747365,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026741187384044527,
      "loss": 0.8997,
      "step": 1192
    },
    {
      "epoch": 0.55244669228042,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00026738404452690166,
      "loss": 0.9365,
      "step": 1193
    },
    {
      "epoch": 0.5529097657861034,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026735621521335805,
      "loss": 0.9348,
      "step": 1194
    },
    {
      "epoch": 0.5533728392917869,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00026732838589981444,
      "loss": 0.9599,
      "step": 1195
    },
    {
      "epoch": 0.5538359127974705,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002673005565862709,
      "loss": 0.8858,
      "step": 1196
    },
    {
      "epoch": 0.554298986303154,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002672727272727272,
      "loss": 0.9085,
      "step": 1197
    },
    {
      "epoch": 0.5547620598088374,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00026724489795918366,
      "loss": 0.8787,
      "step": 1198
    },
    {
      "epoch": 0.555225133314521,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026721706864564005,
      "loss": 0.8481,
      "step": 1199
    },
    {
      "epoch": 0.5556882068202045,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00026718923933209644,
      "loss": 0.8799,
      "step": 1200
    },
    {
      "epoch": 0.556151280325888,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00026716141001855283,
      "loss": 0.8713,
      "step": 1201
    },
    {
      "epoch": 0.5566143538315714,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00026713358070500927,
      "loss": 0.8905,
      "step": 1202
    },
    {
      "epoch": 0.557077427337255,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026710575139146566,
      "loss": 0.8594,
      "step": 1203
    },
    {
      "epoch": 0.5575405008429385,
      "grad_norm": 0.46875,
      "learning_rate": 0.00026707792207792205,
      "loss": 0.8045,
      "step": 1204
    },
    {
      "epoch": 0.558003574348622,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00026705009276437844,
      "loss": 0.8789,
      "step": 1205
    },
    {
      "epoch": 0.5584666478543056,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002670222634508349,
      "loss": 0.9747,
      "step": 1206
    },
    {
      "epoch": 0.558929721359989,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0002669944341372913,
      "loss": 0.9798,
      "step": 1207
    },
    {
      "epoch": 0.5593927948656725,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026696660482374766,
      "loss": 0.8012,
      "step": 1208
    },
    {
      "epoch": 0.559855868371356,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026693877551020405,
      "loss": 0.8686,
      "step": 1209
    },
    {
      "epoch": 0.5603189418770396,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026691094619666044,
      "loss": 0.8569,
      "step": 1210
    },
    {
      "epoch": 0.560782015382723,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002668831168831169,
      "loss": 0.9817,
      "step": 1211
    },
    {
      "epoch": 0.5612450888884065,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002668552875695733,
      "loss": 0.8316,
      "step": 1212
    },
    {
      "epoch": 0.56170816239409,
      "grad_norm": 0.640625,
      "learning_rate": 0.00026682745825602967,
      "loss": 0.9221,
      "step": 1213
    },
    {
      "epoch": 0.5621712358997736,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00026679962894248605,
      "loss": 0.8681,
      "step": 1214
    },
    {
      "epoch": 0.562634309405457,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002667717996289425,
      "loss": 0.9259,
      "step": 1215
    },
    {
      "epoch": 0.5630973829111405,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00026674397031539883,
      "loss": 0.8989,
      "step": 1216
    },
    {
      "epoch": 0.5635604564168241,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002667161410018553,
      "loss": 0.8693,
      "step": 1217
    },
    {
      "epoch": 0.5640235299225076,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00026668831168831167,
      "loss": 0.9507,
      "step": 1218
    },
    {
      "epoch": 0.564486603428191,
      "grad_norm": 0.59375,
      "learning_rate": 0.00026666048237476806,
      "loss": 0.9661,
      "step": 1219
    },
    {
      "epoch": 0.5649496769338745,
      "grad_norm": 0.640625,
      "learning_rate": 0.00026663265306122445,
      "loss": 0.8907,
      "step": 1220
    },
    {
      "epoch": 0.5654127504395581,
      "grad_norm": 0.5625,
      "learning_rate": 0.00026660482374768084,
      "loss": 0.898,
      "step": 1221
    },
    {
      "epoch": 0.5658758239452416,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002665769944341373,
      "loss": 0.9651,
      "step": 1222
    },
    {
      "epoch": 0.566338897450925,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026654916512059367,
      "loss": 0.8807,
      "step": 1223
    },
    {
      "epoch": 0.5668019709566086,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00026652133580705006,
      "loss": 0.8309,
      "step": 1224
    },
    {
      "epoch": 0.5672650444622921,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00026649350649350645,
      "loss": 0.8089,
      "step": 1225
    },
    {
      "epoch": 0.5677281179679756,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002664656771799629,
      "loss": 1.0701,
      "step": 1226
    },
    {
      "epoch": 0.568191191473659,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002664378478664193,
      "loss": 0.8425,
      "step": 1227
    },
    {
      "epoch": 0.5686542649793426,
      "grad_norm": 0.625,
      "learning_rate": 0.00026641001855287567,
      "loss": 0.9304,
      "step": 1228
    },
    {
      "epoch": 0.5691173384850261,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00026638218923933206,
      "loss": 0.8905,
      "step": 1229
    },
    {
      "epoch": 0.5695804119907096,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002663543599257885,
      "loss": 0.8555,
      "step": 1230
    },
    {
      "epoch": 0.570043485496393,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002663265306122449,
      "loss": 0.858,
      "step": 1231
    },
    {
      "epoch": 0.5705065590020766,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002662987012987013,
      "loss": 0.9411,
      "step": 1232
    },
    {
      "epoch": 0.5709696325077601,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002662708719851577,
      "loss": 0.8959,
      "step": 1233
    },
    {
      "epoch": 0.5714327060134436,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002662430426716141,
      "loss": 0.8267,
      "step": 1234
    },
    {
      "epoch": 0.5718957795191271,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00026621521335807045,
      "loss": 1.0347,
      "step": 1235
    },
    {
      "epoch": 0.5723588530248106,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00026618738404452684,
      "loss": 1.0458,
      "step": 1236
    },
    {
      "epoch": 0.5728219265304941,
      "grad_norm": 0.625,
      "learning_rate": 0.0002661595547309833,
      "loss": 0.9941,
      "step": 1237
    },
    {
      "epoch": 0.5732850000361777,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002661317254174397,
      "loss": 0.964,
      "step": 1238
    },
    {
      "epoch": 0.5737480735418611,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00026610389610389607,
      "loss": 0.8989,
      "step": 1239
    },
    {
      "epoch": 0.5742111470475446,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00026607606679035246,
      "loss": 0.8338,
      "step": 1240
    },
    {
      "epoch": 0.5746742205532281,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002660482374768089,
      "loss": 0.7747,
      "step": 1241
    },
    {
      "epoch": 0.5751372940589117,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002660204081632653,
      "loss": 0.9834,
      "step": 1242
    },
    {
      "epoch": 0.5756003675645951,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002659925788497217,
      "loss": 0.9974,
      "step": 1243
    },
    {
      "epoch": 0.5760634410702786,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026596474953617807,
      "loss": 0.9412,
      "step": 1244
    },
    {
      "epoch": 0.5765265145759622,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002659369202226345,
      "loss": 0.8008,
      "step": 1245
    },
    {
      "epoch": 0.5769895880816457,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002659090909090909,
      "loss": 0.805,
      "step": 1246
    },
    {
      "epoch": 0.5774526615873291,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002658812615955473,
      "loss": 0.8592,
      "step": 1247
    },
    {
      "epoch": 0.5779157350930126,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002658534322820037,
      "loss": 0.8152,
      "step": 1248
    },
    {
      "epoch": 0.5783788085986962,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026582560296846007,
      "loss": 0.9088,
      "step": 1249
    },
    {
      "epoch": 0.5788418821043797,
      "grad_norm": 1.078125,
      "learning_rate": 0.0002657977736549165,
      "loss": 0.8655,
      "step": 1250
    },
    {
      "epoch": 0.5793049556100631,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00026576994434137285,
      "loss": 0.8292,
      "step": 1251
    },
    {
      "epoch": 0.5797680291157467,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002657421150278293,
      "loss": 0.9771,
      "step": 1252
    },
    {
      "epoch": 0.5802311026214302,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002657142857142857,
      "loss": 1.0019,
      "step": 1253
    },
    {
      "epoch": 0.5806941761271137,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00026568645640074207,
      "loss": 0.9266,
      "step": 1254
    },
    {
      "epoch": 0.5811572496327971,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00026565862708719846,
      "loss": 0.8187,
      "step": 1255
    },
    {
      "epoch": 0.5816203231384807,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002656307977736549,
      "loss": 0.8968,
      "step": 1256
    },
    {
      "epoch": 0.5820833966441642,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002656029684601113,
      "loss": 0.9999,
      "step": 1257
    },
    {
      "epoch": 0.5825464701498477,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002655751391465677,
      "loss": 0.8613,
      "step": 1258
    },
    {
      "epoch": 0.5830095436555311,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002655473098330241,
      "loss": 0.9532,
      "step": 1259
    },
    {
      "epoch": 0.5834726171612147,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002655194805194805,
      "loss": 0.873,
      "step": 1260
    },
    {
      "epoch": 0.5839356906668982,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002654916512059369,
      "loss": 0.8441,
      "step": 1261
    },
    {
      "epoch": 0.5843987641725817,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002654638218923933,
      "loss": 0.8858,
      "step": 1262
    },
    {
      "epoch": 0.5848618376782652,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002654359925788497,
      "loss": 0.7656,
      "step": 1263
    },
    {
      "epoch": 0.5853249111839487,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002654081632653061,
      "loss": 0.7882,
      "step": 1264
    },
    {
      "epoch": 0.5857879846896322,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002653803339517625,
      "loss": 0.8008,
      "step": 1265
    },
    {
      "epoch": 0.5862510581953158,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002653525046382189,
      "loss": 0.8459,
      "step": 1266
    },
    {
      "epoch": 0.5867141317009992,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002653246753246753,
      "loss": 0.776,
      "step": 1267
    },
    {
      "epoch": 0.5871772052066827,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002652968460111317,
      "loss": 0.9366,
      "step": 1268
    },
    {
      "epoch": 0.5876402787123662,
      "grad_norm": 1.140625,
      "learning_rate": 0.00026526901669758813,
      "loss": 0.989,
      "step": 1269
    },
    {
      "epoch": 0.5881033522180498,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00026524118738404447,
      "loss": 0.9462,
      "step": 1270
    },
    {
      "epoch": 0.5885664257237332,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002652133580705009,
      "loss": 0.7907,
      "step": 1271
    },
    {
      "epoch": 0.5890294992294167,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002651855287569573,
      "loss": 1.02,
      "step": 1272
    },
    {
      "epoch": 0.5894925727351003,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002651576994434137,
      "loss": 0.9105,
      "step": 1273
    },
    {
      "epoch": 0.5899556462407838,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002651298701298701,
      "loss": 0.8892,
      "step": 1274
    },
    {
      "epoch": 0.5904187197464672,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002651020408163265,
      "loss": 0.8228,
      "step": 1275
    },
    {
      "epoch": 0.5908817932521507,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002650742115027829,
      "loss": 0.9113,
      "step": 1276
    },
    {
      "epoch": 0.5913448667578343,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002650463821892393,
      "loss": 0.8983,
      "step": 1277
    },
    {
      "epoch": 0.5918079402635178,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002650185528756957,
      "loss": 0.8602,
      "step": 1278
    },
    {
      "epoch": 0.5922710137692012,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0002649907235621521,
      "loss": 0.9597,
      "step": 1279
    },
    {
      "epoch": 0.5927340872748847,
      "grad_norm": 0.5625,
      "learning_rate": 0.00026496289424860853,
      "loss": 0.8838,
      "step": 1280
    },
    {
      "epoch": 0.5931971607805683,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002649350649350649,
      "loss": 1.0761,
      "step": 1281
    },
    {
      "epoch": 0.5936602342862518,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002649072356215213,
      "loss": 0.8952,
      "step": 1282
    },
    {
      "epoch": 0.5941233077919353,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002648794063079777,
      "loss": 0.8743,
      "step": 1283
    },
    {
      "epoch": 0.5945863812976188,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00026485157699443414,
      "loss": 0.9299,
      "step": 1284
    },
    {
      "epoch": 0.5950494548033023,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00026482374768089053,
      "loss": 0.9002,
      "step": 1285
    },
    {
      "epoch": 0.5955125283089858,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002647959183673469,
      "loss": 0.8958,
      "step": 1286
    },
    {
      "epoch": 0.5959756018146694,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002647680890538033,
      "loss": 0.7644,
      "step": 1287
    },
    {
      "epoch": 0.5964386753203528,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00026474025974025975,
      "loss": 0.8041,
      "step": 1288
    },
    {
      "epoch": 0.5969017488260363,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002647124304267161,
      "loss": 0.9636,
      "step": 1289
    },
    {
      "epoch": 0.5973648223317198,
      "grad_norm": 1.28125,
      "learning_rate": 0.00026468460111317253,
      "loss": 1.0315,
      "step": 1290
    },
    {
      "epoch": 0.5978278958374034,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002646567717996289,
      "loss": 0.8392,
      "step": 1291
    },
    {
      "epoch": 0.5982909693430868,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002646289424860853,
      "loss": 0.9525,
      "step": 1292
    },
    {
      "epoch": 0.5987540428487703,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002646011131725417,
      "loss": 0.8135,
      "step": 1293
    },
    {
      "epoch": 0.5992171163544538,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002645732838589981,
      "loss": 0.8692,
      "step": 1294
    },
    {
      "epoch": 0.5996801898601374,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026454545454545453,
      "loss": 0.8134,
      "step": 1295
    },
    {
      "epoch": 0.6001432633658208,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0002645176252319109,
      "loss": 1.031,
      "step": 1296
    },
    {
      "epoch": 0.6006063368715043,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002644897959183673,
      "loss": 0.8305,
      "step": 1297
    },
    {
      "epoch": 0.6010694103771879,
      "grad_norm": 0.625,
      "learning_rate": 0.0002644619666048237,
      "loss": 0.9347,
      "step": 1298
    },
    {
      "epoch": 0.6015324838828714,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026443413729128015,
      "loss": 0.8779,
      "step": 1299
    },
    {
      "epoch": 0.6019955573885548,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00026440630797773654,
      "loss": 0.8085,
      "step": 1300
    },
    {
      "epoch": 0.6024586308942383,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002643784786641929,
      "loss": 1.0542,
      "step": 1301
    },
    {
      "epoch": 0.6029217043999219,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002643506493506493,
      "loss": 0.9535,
      "step": 1302
    },
    {
      "epoch": 0.6033847779056054,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00026432282003710576,
      "loss": 0.7718,
      "step": 1303
    },
    {
      "epoch": 0.6038478514112888,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00026429499072356215,
      "loss": 0.828,
      "step": 1304
    },
    {
      "epoch": 0.6043109249169724,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00026426716141001854,
      "loss": 0.749,
      "step": 1305
    },
    {
      "epoch": 0.6047739984226559,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026423933209647493,
      "loss": 0.8619,
      "step": 1306
    },
    {
      "epoch": 0.6052370719283394,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002642115027829313,
      "loss": 0.8354,
      "step": 1307
    },
    {
      "epoch": 0.6057001454340228,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00026418367346938776,
      "loss": 0.8405,
      "step": 1308
    },
    {
      "epoch": 0.6061632189397064,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002641558441558441,
      "loss": 0.9385,
      "step": 1309
    },
    {
      "epoch": 0.6066262924453899,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00026412801484230054,
      "loss": 0.9215,
      "step": 1310
    },
    {
      "epoch": 0.6070893659510734,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00026410018552875693,
      "loss": 0.8819,
      "step": 1311
    },
    {
      "epoch": 0.6075524394567569,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002640723562152133,
      "loss": 0.7296,
      "step": 1312
    },
    {
      "epoch": 0.6080155129624404,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002640445269016697,
      "loss": 0.9305,
      "step": 1313
    },
    {
      "epoch": 0.6084785864681239,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00026401669758812615,
      "loss": 0.8703,
      "step": 1314
    },
    {
      "epoch": 0.6089416599738074,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00026398886827458254,
      "loss": 0.9139,
      "step": 1315
    },
    {
      "epoch": 0.6094047334794909,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00026396103896103893,
      "loss": 0.788,
      "step": 1316
    },
    {
      "epoch": 0.6098678069851744,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002639332096474953,
      "loss": 0.8747,
      "step": 1317
    },
    {
      "epoch": 0.6103308804908579,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00026390538033395177,
      "loss": 0.8466,
      "step": 1318
    },
    {
      "epoch": 0.6107939539965415,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00026387755102040816,
      "loss": 0.8311,
      "step": 1319
    },
    {
      "epoch": 0.6112570275022249,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00026384972170686455,
      "loss": 0.9877,
      "step": 1320
    },
    {
      "epoch": 0.6117201010079084,
      "grad_norm": 1.6796875,
      "learning_rate": 0.00026382189239332094,
      "loss": 0.8802,
      "step": 1321
    },
    {
      "epoch": 0.6121831745135919,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0002637940630797773,
      "loss": 0.8807,
      "step": 1322
    },
    {
      "epoch": 0.6126462480192755,
      "grad_norm": 1.03125,
      "learning_rate": 0.00026376623376623377,
      "loss": 0.909,
      "step": 1323
    },
    {
      "epoch": 0.6131093215249589,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00026373840445269016,
      "loss": 0.8464,
      "step": 1324
    },
    {
      "epoch": 0.6135723950306424,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00026371057513914655,
      "loss": 0.8531,
      "step": 1325
    },
    {
      "epoch": 0.614035468536326,
      "grad_norm": 0.65625,
      "learning_rate": 0.00026368274582560294,
      "loss": 0.917,
      "step": 1326
    },
    {
      "epoch": 0.6144985420420095,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002636549165120594,
      "loss": 0.8822,
      "step": 1327
    },
    {
      "epoch": 0.6149616155476929,
      "grad_norm": 0.625,
      "learning_rate": 0.0002636270871985157,
      "loss": 0.8775,
      "step": 1328
    },
    {
      "epoch": 0.6154246890533764,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00026359925788497216,
      "loss": 0.9726,
      "step": 1329
    },
    {
      "epoch": 0.61588776255906,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00026357142857142855,
      "loss": 0.8892,
      "step": 1330
    },
    {
      "epoch": 0.6163508360647435,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00026354359925788494,
      "loss": 0.9639,
      "step": 1331
    },
    {
      "epoch": 0.6168139095704269,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026351576994434133,
      "loss": 0.9702,
      "step": 1332
    },
    {
      "epoch": 0.6172769830761105,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002634879406307978,
      "loss": 1.0228,
      "step": 1333
    },
    {
      "epoch": 0.617740056581794,
      "grad_norm": 0.625,
      "learning_rate": 0.00026346011131725416,
      "loss": 0.985,
      "step": 1334
    },
    {
      "epoch": 0.6182031300874775,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00026343228200371055,
      "loss": 0.9038,
      "step": 1335
    },
    {
      "epoch": 0.6186662035931609,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00026340445269016694,
      "loss": 0.8881,
      "step": 1336
    },
    {
      "epoch": 0.6191292770988445,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026337662337662333,
      "loss": 0.881,
      "step": 1337
    },
    {
      "epoch": 0.619592350604528,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002633487940630798,
      "loss": 0.7645,
      "step": 1338
    },
    {
      "epoch": 0.6200554241102115,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00026332096474953617,
      "loss": 0.7887,
      "step": 1339
    },
    {
      "epoch": 0.620518497615895,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026329313543599255,
      "loss": 0.9014,
      "step": 1340
    },
    {
      "epoch": 0.6209815711215785,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00026326530612244894,
      "loss": 0.9218,
      "step": 1341
    },
    {
      "epoch": 0.621444644627262,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002632374768089054,
      "loss": 0.8967,
      "step": 1342
    },
    {
      "epoch": 0.6219077181329455,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0002632096474953618,
      "loss": 0.9093,
      "step": 1343
    },
    {
      "epoch": 0.622370791638629,
      "grad_norm": 0.671875,
      "learning_rate": 0.00026318181818181817,
      "loss": 0.9147,
      "step": 1344
    },
    {
      "epoch": 0.6228338651443125,
      "grad_norm": 0.609375,
      "learning_rate": 0.00026315398886827456,
      "loss": 0.8606,
      "step": 1345
    },
    {
      "epoch": 0.623296938649996,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000263126159554731,
      "loss": 0.9061,
      "step": 1346
    },
    {
      "epoch": 0.6237600121556796,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00026309833024118734,
      "loss": 0.8511,
      "step": 1347
    },
    {
      "epoch": 0.624223085661363,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002630705009276437,
      "loss": 0.9685,
      "step": 1348
    },
    {
      "epoch": 0.6246861591670465,
      "grad_norm": 1.53125,
      "learning_rate": 0.00026304267161410017,
      "loss": 0.9647,
      "step": 1349
    },
    {
      "epoch": 0.62514923267273,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00026301484230055656,
      "loss": 0.9315,
      "step": 1350
    },
    {
      "epoch": 0.6256123061784136,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00026298701298701295,
      "loss": 0.9033,
      "step": 1351
    },
    {
      "epoch": 0.626075379684097,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026295918367346934,
      "loss": 0.8638,
      "step": 1352
    },
    {
      "epoch": 0.6265384531897805,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002629313543599258,
      "loss": 0.9212,
      "step": 1353
    },
    {
      "epoch": 0.627001526695464,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00026290352504638217,
      "loss": 0.8988,
      "step": 1354
    },
    {
      "epoch": 0.6274646002011476,
      "grad_norm": 0.65625,
      "learning_rate": 0.00026287569573283856,
      "loss": 0.8053,
      "step": 1355
    },
    {
      "epoch": 0.6279276737068311,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00026284786641929495,
      "loss": 0.8465,
      "step": 1356
    },
    {
      "epoch": 0.6283907472125145,
      "grad_norm": 0.78125,
      "learning_rate": 0.0002628200371057514,
      "loss": 1.0047,
      "step": 1357
    },
    {
      "epoch": 0.6288538207181981,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002627922077922078,
      "loss": 0.8536,
      "step": 1358
    },
    {
      "epoch": 0.6293168942238816,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002627643784786642,
      "loss": 0.8718,
      "step": 1359
    },
    {
      "epoch": 0.6297799677295651,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00026273654916512056,
      "loss": 0.7972,
      "step": 1360
    },
    {
      "epoch": 0.6302430412352485,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000262708719851577,
      "loss": 0.7155,
      "step": 1361
    },
    {
      "epoch": 0.6307061147409321,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002626808905380334,
      "loss": 0.8335,
      "step": 1362
    },
    {
      "epoch": 0.6311691882466156,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026265306122448973,
      "loss": 0.907,
      "step": 1363
    },
    {
      "epoch": 0.6316322617522991,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002626252319109462,
      "loss": 0.8408,
      "step": 1364
    },
    {
      "epoch": 0.6320953352579826,
      "grad_norm": 0.546875,
      "learning_rate": 0.00026259740259740257,
      "loss": 0.7875,
      "step": 1365
    },
    {
      "epoch": 0.6325584087636661,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00026256957328385896,
      "loss": 0.9398,
      "step": 1366
    },
    {
      "epoch": 0.6330214822693496,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00026254174397031535,
      "loss": 1.0002,
      "step": 1367
    },
    {
      "epoch": 0.6334845557750332,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002625139146567718,
      "loss": 0.8826,
      "step": 1368
    },
    {
      "epoch": 0.6339476292807166,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002624860853432282,
      "loss": 0.8203,
      "step": 1369
    },
    {
      "epoch": 0.6344107027864001,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00026245825602968457,
      "loss": 0.9216,
      "step": 1370
    },
    {
      "epoch": 0.6348737762920836,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026243042671614096,
      "loss": 0.8992,
      "step": 1371
    },
    {
      "epoch": 0.6353368497977672,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002624025974025974,
      "loss": 0.8629,
      "step": 1372
    },
    {
      "epoch": 0.6357999233034506,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002623747680890538,
      "loss": 0.7965,
      "step": 1373
    },
    {
      "epoch": 0.6362629968091341,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002623469387755102,
      "loss": 0.9126,
      "step": 1374
    },
    {
      "epoch": 0.6367260703148176,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00026231910946196657,
      "loss": 1.0145,
      "step": 1375
    },
    {
      "epoch": 0.6371891438205012,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00026229128014842296,
      "loss": 0.9314,
      "step": 1376
    },
    {
      "epoch": 0.6376522173261846,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002622634508348794,
      "loss": 0.7539,
      "step": 1377
    },
    {
      "epoch": 0.6381152908318681,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002622356215213358,
      "loss": 0.971,
      "step": 1378
    },
    {
      "epoch": 0.6385783643375517,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002622077922077922,
      "loss": 0.8529,
      "step": 1379
    },
    {
      "epoch": 0.6390414378432352,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00026217996289424857,
      "loss": 0.8042,
      "step": 1380
    },
    {
      "epoch": 0.6395045113489186,
      "grad_norm": 0.5625,
      "learning_rate": 0.000262152133580705,
      "loss": 0.8685,
      "step": 1381
    },
    {
      "epoch": 0.6399675848546021,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00026212430426716135,
      "loss": 0.9016,
      "step": 1382
    },
    {
      "epoch": 0.6404306583602857,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002620964749536178,
      "loss": 0.8272,
      "step": 1383
    },
    {
      "epoch": 0.6408937318659692,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002620686456400742,
      "loss": 0.9215,
      "step": 1384
    },
    {
      "epoch": 0.6413568053716526,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002620408163265306,
      "loss": 0.9652,
      "step": 1385
    },
    {
      "epoch": 0.6418198788773362,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00026201298701298696,
      "loss": 0.9963,
      "step": 1386
    },
    {
      "epoch": 0.6422829523830197,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002619851576994434,
      "loss": 0.9621,
      "step": 1387
    },
    {
      "epoch": 0.6427460258887032,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002619573283858998,
      "loss": 0.8098,
      "step": 1388
    },
    {
      "epoch": 0.6432090993943866,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002619294990723562,
      "loss": 0.8525,
      "step": 1389
    },
    {
      "epoch": 0.6436721729000702,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002619016697588126,
      "loss": 0.8474,
      "step": 1390
    },
    {
      "epoch": 0.6441352464057537,
      "grad_norm": 1.1875,
      "learning_rate": 0.00026187384044526897,
      "loss": 0.9698,
      "step": 1391
    },
    {
      "epoch": 0.6445983199114372,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002618460111317254,
      "loss": 0.9209,
      "step": 1392
    },
    {
      "epoch": 0.6450613934171207,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002618181818181818,
      "loss": 0.8242,
      "step": 1393
    },
    {
      "epoch": 0.6455244669228042,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002617903525046382,
      "loss": 0.8534,
      "step": 1394
    },
    {
      "epoch": 0.6459875404284877,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002617625231910946,
      "loss": 0.8354,
      "step": 1395
    },
    {
      "epoch": 0.6464506139341712,
      "grad_norm": 0.53125,
      "learning_rate": 0.000261734693877551,
      "loss": 0.7933,
      "step": 1396
    },
    {
      "epoch": 0.6469136874398547,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002617068645640074,
      "loss": 0.8756,
      "step": 1397
    },
    {
      "epoch": 0.6473767609455382,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002616790352504638,
      "loss": 0.8091,
      "step": 1398
    },
    {
      "epoch": 0.6478398344512217,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002616512059369202,
      "loss": 0.9426,
      "step": 1399
    },
    {
      "epoch": 0.6483029079569053,
      "grad_norm": 0.625,
      "learning_rate": 0.00026162337662337664,
      "loss": 0.9281,
      "step": 1400
    },
    {
      "epoch": 0.6487659814625887,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00026159554730983297,
      "loss": 0.9131,
      "step": 1401
    },
    {
      "epoch": 0.6492290549682722,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002615677179962894,
      "loss": 0.8946,
      "step": 1402
    },
    {
      "epoch": 0.6496921284739557,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002615398886827458,
      "loss": 0.8964,
      "step": 1403
    },
    {
      "epoch": 0.6501552019796393,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002615120593692022,
      "loss": 0.8688,
      "step": 1404
    },
    {
      "epoch": 0.6506182754853227,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002614842300556586,
      "loss": 0.7792,
      "step": 1405
    },
    {
      "epoch": 0.6510813489910062,
      "grad_norm": 0.59375,
      "learning_rate": 0.000261456400742115,
      "loss": 0.874,
      "step": 1406
    },
    {
      "epoch": 0.6515444224966898,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002614285714285714,
      "loss": 0.7748,
      "step": 1407
    },
    {
      "epoch": 0.6520074960023733,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002614007421150278,
      "loss": 0.8609,
      "step": 1408
    },
    {
      "epoch": 0.6524705695080567,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002613729128014842,
      "loss": 0.7392,
      "step": 1409
    },
    {
      "epoch": 0.6529336430137402,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002613450834879406,
      "loss": 0.8822,
      "step": 1410
    },
    {
      "epoch": 0.6533967165194238,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00026131725417439703,
      "loss": 0.8385,
      "step": 1411
    },
    {
      "epoch": 0.6538597900251073,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002612894248608534,
      "loss": 0.8374,
      "step": 1412
    },
    {
      "epoch": 0.6543228635307907,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002612615955473098,
      "loss": 0.933,
      "step": 1413
    },
    {
      "epoch": 0.6547859370364743,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0002612337662337662,
      "loss": 0.9374,
      "step": 1414
    },
    {
      "epoch": 0.6552490105421578,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00026120593692022264,
      "loss": 0.9972,
      "step": 1415
    },
    {
      "epoch": 0.6557120840478413,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00026117810760667903,
      "loss": 0.9029,
      "step": 1416
    },
    {
      "epoch": 0.6561751575535247,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002611502782931354,
      "loss": 0.8758,
      "step": 1417
    },
    {
      "epoch": 0.6566382310592083,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002611224489795918,
      "loss": 0.8232,
      "step": 1418
    },
    {
      "epoch": 0.6571013045648918,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002610946196660482,
      "loss": 0.9029,
      "step": 1419
    },
    {
      "epoch": 0.6575643780705753,
      "grad_norm": 0.75,
      "learning_rate": 0.00026106679035250464,
      "loss": 0.9385,
      "step": 1420
    },
    {
      "epoch": 0.6580274515762587,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000261038961038961,
      "loss": 0.8126,
      "step": 1421
    },
    {
      "epoch": 0.6584905250819423,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002610111317254174,
      "loss": 0.9225,
      "step": 1422
    },
    {
      "epoch": 0.6589535985876258,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002609833024118738,
      "loss": 0.792,
      "step": 1423
    },
    {
      "epoch": 0.6594166720933093,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002609554730983302,
      "loss": 0.8695,
      "step": 1424
    },
    {
      "epoch": 0.6598797455989928,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002609276437847866,
      "loss": 0.9468,
      "step": 1425
    },
    {
      "epoch": 0.6603428191046763,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00026089981447124304,
      "loss": 0.8754,
      "step": 1426
    },
    {
      "epoch": 0.6608058926103598,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002608719851576994,
      "loss": 0.7659,
      "step": 1427
    },
    {
      "epoch": 0.6612689661160434,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002608441558441558,
      "loss": 0.8711,
      "step": 1428
    },
    {
      "epoch": 0.6617320396217268,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002608163265306122,
      "loss": 0.9755,
      "step": 1429
    },
    {
      "epoch": 0.6621951131274103,
      "grad_norm": 6.40625,
      "learning_rate": 0.00026078849721706865,
      "loss": 0.7893,
      "step": 1430
    },
    {
      "epoch": 0.6626581866330938,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00026076066790352504,
      "loss": 0.8878,
      "step": 1431
    },
    {
      "epoch": 0.6631212601387774,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00026073283858998143,
      "loss": 0.8218,
      "step": 1432
    },
    {
      "epoch": 0.6635843336444609,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002607050092764378,
      "loss": 0.9405,
      "step": 1433
    },
    {
      "epoch": 0.6640474071501443,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002606771799628942,
      "loss": 0.9792,
      "step": 1434
    },
    {
      "epoch": 0.6645104806558279,
      "grad_norm": 1.078125,
      "learning_rate": 0.00026064935064935065,
      "loss": 1.0674,
      "step": 1435
    },
    {
      "epoch": 0.6649735541615114,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00026062152133580704,
      "loss": 0.7499,
      "step": 1436
    },
    {
      "epoch": 0.6654366276671949,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00026059369202226343,
      "loss": 0.8283,
      "step": 1437
    },
    {
      "epoch": 0.6658997011728783,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002605658627087198,
      "loss": 0.9039,
      "step": 1438
    },
    {
      "epoch": 0.6663627746785619,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00026053803339517626,
      "loss": 0.858,
      "step": 1439
    },
    {
      "epoch": 0.6668258481842454,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002605102040816326,
      "loss": 0.8492,
      "step": 1440
    },
    {
      "epoch": 0.6672889216899289,
      "grad_norm": 0.578125,
      "learning_rate": 0.00026048237476808904,
      "loss": 0.8456,
      "step": 1441
    },
    {
      "epoch": 0.6677519951956123,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00026045454545454543,
      "loss": 0.9084,
      "step": 1442
    },
    {
      "epoch": 0.6682150687012959,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002604267161410018,
      "loss": 0.6895,
      "step": 1443
    },
    {
      "epoch": 0.6686781422069794,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002603988868274582,
      "loss": 0.8594,
      "step": 1444
    },
    {
      "epoch": 0.6691412157126629,
      "grad_norm": 0.59375,
      "learning_rate": 0.00026037105751391466,
      "loss": 0.8989,
      "step": 1445
    },
    {
      "epoch": 0.6696042892183464,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00026034322820037105,
      "loss": 0.8595,
      "step": 1446
    },
    {
      "epoch": 0.6700673627240299,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00026031539888682744,
      "loss": 0.8742,
      "step": 1447
    },
    {
      "epoch": 0.6705304362297134,
      "grad_norm": 1.3203125,
      "learning_rate": 0.0002602875695732838,
      "loss": 0.953,
      "step": 1448
    },
    {
      "epoch": 0.670993509735397,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002602597402597402,
      "loss": 0.8841,
      "step": 1449
    },
    {
      "epoch": 0.6714565832410804,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00026023191094619666,
      "loss": 0.7876,
      "step": 1450
    },
    {
      "epoch": 0.6719196567467639,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00026020408163265305,
      "loss": 0.8835,
      "step": 1451
    },
    {
      "epoch": 0.6723827302524474,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00026017625231910944,
      "loss": 0.9297,
      "step": 1452
    },
    {
      "epoch": 0.672845803758131,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002601484230055658,
      "loss": 0.8533,
      "step": 1453
    },
    {
      "epoch": 0.6733088772638144,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00026012059369202227,
      "loss": 0.8869,
      "step": 1454
    },
    {
      "epoch": 0.6737719507694979,
      "grad_norm": 1.09375,
      "learning_rate": 0.00026009276437847866,
      "loss": 1.1361,
      "step": 1455
    },
    {
      "epoch": 0.6742350242751814,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00026006493506493505,
      "loss": 0.866,
      "step": 1456
    },
    {
      "epoch": 0.674698097780865,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00026003710575139144,
      "loss": 0.8907,
      "step": 1457
    },
    {
      "epoch": 0.6751611712865484,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002600092764378479,
      "loss": 0.9581,
      "step": 1458
    },
    {
      "epoch": 0.6756242447922319,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002599814471243042,
      "loss": 0.7573,
      "step": 1459
    },
    {
      "epoch": 0.6760873182979155,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00025995361781076066,
      "loss": 0.8404,
      "step": 1460
    },
    {
      "epoch": 0.676550391803599,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00025992578849721705,
      "loss": 0.7955,
      "step": 1461
    },
    {
      "epoch": 0.6770134653092824,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00025989795918367344,
      "loss": 0.867,
      "step": 1462
    },
    {
      "epoch": 0.677476538814966,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00025987012987012983,
      "loss": 0.8999,
      "step": 1463
    },
    {
      "epoch": 0.6779396123206495,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002598423005565862,
      "loss": 0.9178,
      "step": 1464
    },
    {
      "epoch": 0.678402685826333,
      "grad_norm": 0.53125,
      "learning_rate": 0.00025981447124304266,
      "loss": 0.7907,
      "step": 1465
    },
    {
      "epoch": 0.6788657593320164,
      "grad_norm": 0.546875,
      "learning_rate": 0.00025978664192949905,
      "loss": 0.7304,
      "step": 1466
    },
    {
      "epoch": 0.6793288328377,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00025975881261595544,
      "loss": 0.8568,
      "step": 1467
    },
    {
      "epoch": 0.6797919063433835,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00025973098330241183,
      "loss": 1.081,
      "step": 1468
    },
    {
      "epoch": 0.680254979849067,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002597031539888683,
      "loss": 0.9232,
      "step": 1469
    },
    {
      "epoch": 0.6807180533547504,
      "grad_norm": 0.59375,
      "learning_rate": 0.00025967532467532467,
      "loss": 0.8587,
      "step": 1470
    },
    {
      "epoch": 0.681181126860434,
      "grad_norm": 0.609375,
      "learning_rate": 0.00025964749536178106,
      "loss": 0.9309,
      "step": 1471
    },
    {
      "epoch": 0.6816442003661175,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025961966604823745,
      "loss": 0.9061,
      "step": 1472
    },
    {
      "epoch": 0.682107273871801,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002595918367346939,
      "loss": 0.7384,
      "step": 1473
    },
    {
      "epoch": 0.6825703473774845,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002595640074211503,
      "loss": 0.6741,
      "step": 1474
    },
    {
      "epoch": 0.683033420883168,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002595361781076066,
      "loss": 0.8964,
      "step": 1475
    },
    {
      "epoch": 0.6834964943888515,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00025950834879406306,
      "loss": 0.8234,
      "step": 1476
    },
    {
      "epoch": 0.683959567894535,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025948051948051945,
      "loss": 0.9452,
      "step": 1477
    },
    {
      "epoch": 0.6844226414002185,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025945269016697584,
      "loss": 0.9042,
      "step": 1478
    },
    {
      "epoch": 0.684885714905902,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025942486085343223,
      "loss": 0.9157,
      "step": 1479
    },
    {
      "epoch": 0.6853487884115855,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025939703153988867,
      "loss": 0.9162,
      "step": 1480
    },
    {
      "epoch": 0.6858118619172691,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00025936920222634506,
      "loss": 0.7866,
      "step": 1481
    },
    {
      "epoch": 0.6862749354229525,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025934137291280145,
      "loss": 0.8192,
      "step": 1482
    },
    {
      "epoch": 0.686738008928636,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00025931354359925784,
      "loss": 0.814,
      "step": 1483
    },
    {
      "epoch": 0.6872010824343195,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002592857142857143,
      "loss": 0.913,
      "step": 1484
    },
    {
      "epoch": 0.6876641559400031,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002592578849721707,
      "loss": 0.8783,
      "step": 1485
    },
    {
      "epoch": 0.6881272294456865,
      "grad_norm": 0.671875,
      "learning_rate": 0.00025923005565862706,
      "loss": 0.9419,
      "step": 1486
    },
    {
      "epoch": 0.68859030295137,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00025920222634508345,
      "loss": 0.8662,
      "step": 1487
    },
    {
      "epoch": 0.6890533764570536,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002591743970315399,
      "loss": 0.9991,
      "step": 1488
    },
    {
      "epoch": 0.6895164499627371,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002591465677179963,
      "loss": 0.8587,
      "step": 1489
    },
    {
      "epoch": 0.6899795234684205,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002591187384044527,
      "loss": 0.8824,
      "step": 1490
    },
    {
      "epoch": 0.690442596974104,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00025909090909090907,
      "loss": 0.9158,
      "step": 1491
    },
    {
      "epoch": 0.6909056704797876,
      "grad_norm": 0.515625,
      "learning_rate": 0.00025906307977736546,
      "loss": 0.7862,
      "step": 1492
    },
    {
      "epoch": 0.6913687439854711,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002590352504638219,
      "loss": 0.8379,
      "step": 1493
    },
    {
      "epoch": 0.6918318174911545,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00025900742115027823,
      "loss": 0.8378,
      "step": 1494
    },
    {
      "epoch": 0.692294890996838,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002589795918367347,
      "loss": 0.8727,
      "step": 1495
    },
    {
      "epoch": 0.6927579645025216,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025895176252319107,
      "loss": 0.8358,
      "step": 1496
    },
    {
      "epoch": 0.6932210380082051,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00025892393320964746,
      "loss": 0.8399,
      "step": 1497
    },
    {
      "epoch": 0.6936841115138885,
      "grad_norm": 0.6875,
      "learning_rate": 0.00025889610389610385,
      "loss": 0.914,
      "step": 1498
    },
    {
      "epoch": 0.6941471850195721,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002588682745825603,
      "loss": 0.8289,
      "step": 1499
    },
    {
      "epoch": 0.6946102585252556,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002588404452690167,
      "loss": 1.0021,
      "step": 1500
    },
    {
      "epoch": 0.6950733320309391,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00025881261595547307,
      "loss": 0.82,
      "step": 1501
    },
    {
      "epoch": 0.6955364055366225,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00025878478664192946,
      "loss": 0.9725,
      "step": 1502
    },
    {
      "epoch": 0.6959994790423061,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00025875695732838585,
      "loss": 0.8992,
      "step": 1503
    },
    {
      "epoch": 0.6964625525479896,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002587291280148423,
      "loss": 0.8039,
      "step": 1504
    },
    {
      "epoch": 0.6969256260536731,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002587012987012987,
      "loss": 0.8126,
      "step": 1505
    },
    {
      "epoch": 0.6973886995593566,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00025867346938775507,
      "loss": 0.7719,
      "step": 1506
    },
    {
      "epoch": 0.6978517730650401,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00025864564007421146,
      "loss": 0.8162,
      "step": 1507
    },
    {
      "epoch": 0.6983148465707236,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002586178107606679,
      "loss": 0.9996,
      "step": 1508
    },
    {
      "epoch": 0.6987779200764072,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0002585899814471243,
      "loss": 1.0744,
      "step": 1509
    },
    {
      "epoch": 0.6992409935820907,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002585621521335807,
      "loss": 0.9112,
      "step": 1510
    },
    {
      "epoch": 0.6997040670877741,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002585343228200371,
      "loss": 0.7969,
      "step": 1511
    },
    {
      "epoch": 0.7001671405934576,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002585064935064935,
      "loss": 0.7795,
      "step": 1512
    },
    {
      "epoch": 0.7006302140991412,
      "grad_norm": 0.53125,
      "learning_rate": 0.00025847866419294985,
      "loss": 0.7798,
      "step": 1513
    },
    {
      "epoch": 0.7010932876048247,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002584508348794063,
      "loss": 0.7594,
      "step": 1514
    },
    {
      "epoch": 0.7015563611105081,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002584230055658627,
      "loss": 0.9216,
      "step": 1515
    },
    {
      "epoch": 0.7020194346161917,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002583951762523191,
      "loss": 0.8256,
      "step": 1516
    },
    {
      "epoch": 0.7024825081218752,
      "grad_norm": 0.546875,
      "learning_rate": 0.00025836734693877547,
      "loss": 0.8778,
      "step": 1517
    },
    {
      "epoch": 0.7029455816275587,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00025833951762523186,
      "loss": 0.9196,
      "step": 1518
    },
    {
      "epoch": 0.7034086551332421,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002583116883116883,
      "loss": 0.8595,
      "step": 1519
    },
    {
      "epoch": 0.7038717286389257,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002582838589981447,
      "loss": 0.8158,
      "step": 1520
    },
    {
      "epoch": 0.7043348021446092,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002582560296846011,
      "loss": 0.8182,
      "step": 1521
    },
    {
      "epoch": 0.7047978756502927,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00025822820037105747,
      "loss": 0.8244,
      "step": 1522
    },
    {
      "epoch": 0.7052609491559761,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002582003710575139,
      "loss": 0.7858,
      "step": 1523
    },
    {
      "epoch": 0.7057240226616597,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002581725417439703,
      "loss": 0.9587,
      "step": 1524
    },
    {
      "epoch": 0.7061870961673432,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0002581447124304267,
      "loss": 0.8657,
      "step": 1525
    },
    {
      "epoch": 0.7066501696730267,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002581168831168831,
      "loss": 1.0426,
      "step": 1526
    },
    {
      "epoch": 0.7071132431787102,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002580890538033395,
      "loss": 0.8818,
      "step": 1527
    },
    {
      "epoch": 0.7075763166843937,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002580612244897959,
      "loss": 0.7283,
      "step": 1528
    },
    {
      "epoch": 0.7080393901900772,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002580333951762523,
      "loss": 0.8208,
      "step": 1529
    },
    {
      "epoch": 0.7085024636957608,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002580055658627087,
      "loss": 0.955,
      "step": 1530
    },
    {
      "epoch": 0.7089655372014442,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002579777365491651,
      "loss": 0.906,
      "step": 1531
    },
    {
      "epoch": 0.7094286107071277,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002579499072356215,
      "loss": 0.8614,
      "step": 1532
    },
    {
      "epoch": 0.7098916842128112,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025792207792207786,
      "loss": 0.9034,
      "step": 1533
    },
    {
      "epoch": 0.7103547577184948,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002578942486085343,
      "loss": 0.8382,
      "step": 1534
    },
    {
      "epoch": 0.7108178312241782,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002578664192949907,
      "loss": 0.9494,
      "step": 1535
    },
    {
      "epoch": 0.7112809047298617,
      "grad_norm": 0.6875,
      "learning_rate": 0.0002578385899814471,
      "loss": 0.9025,
      "step": 1536
    },
    {
      "epoch": 0.7117439782355452,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002578107606679035,
      "loss": 0.8431,
      "step": 1537
    },
    {
      "epoch": 0.7122070517412288,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002577829313543599,
      "loss": 0.8226,
      "step": 1538
    },
    {
      "epoch": 0.7126701252469122,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002577551020408163,
      "loss": 0.8298,
      "step": 1539
    },
    {
      "epoch": 0.7131331987525957,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002577272727272727,
      "loss": 0.9219,
      "step": 1540
    },
    {
      "epoch": 0.7135962722582793,
      "grad_norm": 1.15625,
      "learning_rate": 0.0002576994434137291,
      "loss": 0.9643,
      "step": 1541
    },
    {
      "epoch": 0.7140593457639628,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00025767161410018553,
      "loss": 0.8703,
      "step": 1542
    },
    {
      "epoch": 0.7145224192696462,
      "grad_norm": 3.328125,
      "learning_rate": 0.0002576437847866419,
      "loss": 0.8238,
      "step": 1543
    },
    {
      "epoch": 0.7149854927753297,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002576159554730983,
      "loss": 0.8336,
      "step": 1544
    },
    {
      "epoch": 0.7154485662810133,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002575881261595547,
      "loss": 0.8762,
      "step": 1545
    },
    {
      "epoch": 0.7159116397866968,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002575602968460111,
      "loss": 0.8183,
      "step": 1546
    },
    {
      "epoch": 0.7163747132923802,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025753246753246753,
      "loss": 0.8894,
      "step": 1547
    },
    {
      "epoch": 0.7168377867980638,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0002575046382189239,
      "loss": 0.8629,
      "step": 1548
    },
    {
      "epoch": 0.7173008603037473,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002574768089053803,
      "loss": 0.7824,
      "step": 1549
    },
    {
      "epoch": 0.7177639338094308,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002574489795918367,
      "loss": 0.7493,
      "step": 1550
    },
    {
      "epoch": 0.7182270073151142,
      "grad_norm": 0.671875,
      "learning_rate": 0.00025742115027829315,
      "loss": 0.9844,
      "step": 1551
    },
    {
      "epoch": 0.7186900808207978,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002573933209647495,
      "loss": 0.9954,
      "step": 1552
    },
    {
      "epoch": 0.7191531543264813,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002573654916512059,
      "loss": 0.8811,
      "step": 1553
    },
    {
      "epoch": 0.7196162278321648,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002573376623376623,
      "loss": 0.909,
      "step": 1554
    },
    {
      "epoch": 0.7200793013378483,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002573098330241187,
      "loss": 0.8314,
      "step": 1555
    },
    {
      "epoch": 0.7205423748435318,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002572820037105751,
      "loss": 0.9314,
      "step": 1556
    },
    {
      "epoch": 0.7210054483492153,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00025725417439703154,
      "loss": 0.8974,
      "step": 1557
    },
    {
      "epoch": 0.7214685218548988,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00025722634508348793,
      "loss": 0.9605,
      "step": 1558
    },
    {
      "epoch": 0.7219315953605823,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002571985157699443,
      "loss": 0.8746,
      "step": 1559
    },
    {
      "epoch": 0.7223946688662658,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002571706864564007,
      "loss": 0.8834,
      "step": 1560
    },
    {
      "epoch": 0.7228577423719493,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002571428571428571,
      "loss": 0.7613,
      "step": 1561
    },
    {
      "epoch": 0.7233208158776329,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00025711502782931354,
      "loss": 0.8695,
      "step": 1562
    },
    {
      "epoch": 0.7237838893833163,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00025708719851576993,
      "loss": 0.7255,
      "step": 1563
    },
    {
      "epoch": 0.7242469628889998,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002570593692022263,
      "loss": 0.9195,
      "step": 1564
    },
    {
      "epoch": 0.7247100363946833,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002570315398886827,
      "loss": 1.0252,
      "step": 1565
    },
    {
      "epoch": 0.7251731099003669,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00025700371057513915,
      "loss": 0.8106,
      "step": 1566
    },
    {
      "epoch": 0.7256361834060503,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025697588126159554,
      "loss": 0.9165,
      "step": 1567
    },
    {
      "epoch": 0.7260992569117338,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00025694805194805193,
      "loss": 0.8172,
      "step": 1568
    },
    {
      "epoch": 0.7265623304174174,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002569202226345083,
      "loss": 0.9048,
      "step": 1569
    },
    {
      "epoch": 0.7270254039231009,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00025689239332096477,
      "loss": 0.8839,
      "step": 1570
    },
    {
      "epoch": 0.7274884774287843,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002568645640074211,
      "loss": 0.7497,
      "step": 1571
    },
    {
      "epoch": 0.7279515509344678,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00025683673469387755,
      "loss": 0.8297,
      "step": 1572
    },
    {
      "epoch": 0.7284146244401514,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00025680890538033393,
      "loss": 0.8998,
      "step": 1573
    },
    {
      "epoch": 0.7288776979458349,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002567810760667903,
      "loss": 1.2039,
      "step": 1574
    },
    {
      "epoch": 0.7293407714515183,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002567532467532467,
      "loss": 0.8085,
      "step": 1575
    },
    {
      "epoch": 0.7298038449572019,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002567254174397031,
      "loss": 0.8606,
      "step": 1576
    },
    {
      "epoch": 0.7302669184628854,
      "grad_norm": 0.625,
      "learning_rate": 0.00025669758812615955,
      "loss": 0.8853,
      "step": 1577
    },
    {
      "epoch": 0.7307299919685689,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025666975881261594,
      "loss": 0.8748,
      "step": 1578
    },
    {
      "epoch": 0.7311930654742523,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002566419294990723,
      "loss": 0.8406,
      "step": 1579
    },
    {
      "epoch": 0.7316561389799359,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002566141001855287,
      "loss": 0.7489,
      "step": 1580
    },
    {
      "epoch": 0.7321192124856194,
      "grad_norm": 0.609375,
      "learning_rate": 0.00025658627087198516,
      "loss": 0.9411,
      "step": 1581
    },
    {
      "epoch": 0.7325822859913029,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00025655844155844155,
      "loss": 0.7678,
      "step": 1582
    },
    {
      "epoch": 0.7330453594969865,
      "grad_norm": 0.59375,
      "learning_rate": 0.00025653061224489794,
      "loss": 0.9736,
      "step": 1583
    },
    {
      "epoch": 0.7335084330026699,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00025650278293135433,
      "loss": 0.864,
      "step": 1584
    },
    {
      "epoch": 0.7339715065083534,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002564749536178108,
      "loss": 0.7846,
      "step": 1585
    },
    {
      "epoch": 0.7344345800140369,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025644712430426716,
      "loss": 0.838,
      "step": 1586
    },
    {
      "epoch": 0.7348976535197205,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00025641929499072355,
      "loss": 0.9401,
      "step": 1587
    },
    {
      "epoch": 0.7353607270254039,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025639146567717994,
      "loss": 0.7826,
      "step": 1588
    },
    {
      "epoch": 0.7358238005310874,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025636363636363633,
      "loss": 0.8914,
      "step": 1589
    },
    {
      "epoch": 0.736286874036771,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002563358070500927,
      "loss": 0.8053,
      "step": 1590
    },
    {
      "epoch": 0.7367499475424545,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002563079777365491,
      "loss": 0.7948,
      "step": 1591
    },
    {
      "epoch": 0.7372130210481379,
      "grad_norm": 0.65625,
      "learning_rate": 0.00025628014842300555,
      "loss": 0.9217,
      "step": 1592
    },
    {
      "epoch": 0.7376760945538214,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00025625231910946194,
      "loss": 0.8044,
      "step": 1593
    },
    {
      "epoch": 0.738139168059505,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00025622448979591833,
      "loss": 0.8634,
      "step": 1594
    },
    {
      "epoch": 0.7386022415651885,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002561966604823747,
      "loss": 0.99,
      "step": 1595
    },
    {
      "epoch": 0.7390653150708719,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00025616883116883117,
      "loss": 0.9891,
      "step": 1596
    },
    {
      "epoch": 0.7395283885765555,
      "grad_norm": 0.84375,
      "learning_rate": 0.00025614100185528756,
      "loss": 0.9304,
      "step": 1597
    },
    {
      "epoch": 0.739991462082239,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00025611317254174395,
      "loss": 0.8599,
      "step": 1598
    },
    {
      "epoch": 0.7404545355879225,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00025608534322820034,
      "loss": 0.8261,
      "step": 1599
    },
    {
      "epoch": 0.7409176090936059,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002560575139146568,
      "loss": 0.8354,
      "step": 1600
    },
    {
      "epoch": 0.7413806825992895,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025602968460111317,
      "loss": 0.7754,
      "step": 1601
    },
    {
      "epoch": 0.741843756104973,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00025600185528756956,
      "loss": 1.0017,
      "step": 1602
    },
    {
      "epoch": 0.7423068296106565,
      "grad_norm": 0.609375,
      "learning_rate": 0.00025597402597402595,
      "loss": 0.9015,
      "step": 1603
    },
    {
      "epoch": 0.74276990311634,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00025594619666048234,
      "loss": 0.9481,
      "step": 1604
    },
    {
      "epoch": 0.7432329766220235,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002559183673469388,
      "loss": 0.9063,
      "step": 1605
    },
    {
      "epoch": 0.743696050127707,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002558905380333951,
      "loss": 0.8734,
      "step": 1606
    },
    {
      "epoch": 0.7441591236333905,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00025586270871985156,
      "loss": 0.9801,
      "step": 1607
    },
    {
      "epoch": 0.744622197139074,
      "grad_norm": 1.21875,
      "learning_rate": 0.00025583487940630795,
      "loss": 0.8254,
      "step": 1608
    },
    {
      "epoch": 0.7450852706447575,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00025580705009276434,
      "loss": 0.8574,
      "step": 1609
    },
    {
      "epoch": 0.745548344150441,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025577922077922073,
      "loss": 0.8263,
      "step": 1610
    },
    {
      "epoch": 0.7460114176561246,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002557513914656772,
      "loss": 0.7887,
      "step": 1611
    },
    {
      "epoch": 0.746474491161808,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025572356215213356,
      "loss": 0.864,
      "step": 1612
    },
    {
      "epoch": 0.7469375646674915,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00025569573283858995,
      "loss": 0.8657,
      "step": 1613
    },
    {
      "epoch": 0.747400638173175,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00025566790352504634,
      "loss": 0.8826,
      "step": 1614
    },
    {
      "epoch": 0.7478637116788586,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002556400742115028,
      "loss": 0.8773,
      "step": 1615
    },
    {
      "epoch": 0.748326785184542,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002556122448979592,
      "loss": 0.9293,
      "step": 1616
    },
    {
      "epoch": 0.7487898586902255,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025558441558441557,
      "loss": 0.8633,
      "step": 1617
    },
    {
      "epoch": 0.749252932195909,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00025555658627087196,
      "loss": 0.955,
      "step": 1618
    },
    {
      "epoch": 0.7497160057015926,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025552875695732834,
      "loss": 0.8746,
      "step": 1619
    },
    {
      "epoch": 0.750179079207276,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002555009276437848,
      "loss": 0.821,
      "step": 1620
    },
    {
      "epoch": 0.7506421527129595,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002554730983302412,
      "loss": 0.9416,
      "step": 1621
    },
    {
      "epoch": 0.7511052262186431,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025544526901669757,
      "loss": 0.8895,
      "step": 1622
    },
    {
      "epoch": 0.7515682997243266,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025541743970315396,
      "loss": 0.7849,
      "step": 1623
    },
    {
      "epoch": 0.75203137323001,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002553896103896104,
      "loss": 0.7984,
      "step": 1624
    },
    {
      "epoch": 0.7524944467356935,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00025536178107606674,
      "loss": 0.8594,
      "step": 1625
    },
    {
      "epoch": 0.7529575202413771,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002553339517625232,
      "loss": 0.822,
      "step": 1626
    },
    {
      "epoch": 0.7534205937470606,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00025530612244897957,
      "loss": 0.7884,
      "step": 1627
    },
    {
      "epoch": 0.753883667252744,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00025527829313543596,
      "loss": 1.0453,
      "step": 1628
    },
    {
      "epoch": 0.7543467407584276,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00025525046382189235,
      "loss": 0.8754,
      "step": 1629
    },
    {
      "epoch": 0.7548098142641111,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025522263450834874,
      "loss": 0.8175,
      "step": 1630
    },
    {
      "epoch": 0.7552728877697946,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002551948051948052,
      "loss": 0.8825,
      "step": 1631
    },
    {
      "epoch": 0.755735961275478,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00025516697588126157,
      "loss": 0.7433,
      "step": 1632
    },
    {
      "epoch": 0.7561990347811616,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00025513914656771796,
      "loss": 0.8685,
      "step": 1633
    },
    {
      "epoch": 0.7566621082868451,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00025511131725417435,
      "loss": 0.8834,
      "step": 1634
    },
    {
      "epoch": 0.7571251817925286,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002550834879406308,
      "loss": 0.8167,
      "step": 1635
    },
    {
      "epoch": 0.757588255298212,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002550556586270872,
      "loss": 0.915,
      "step": 1636
    },
    {
      "epoch": 0.7580513288038956,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002550278293135436,
      "loss": 0.819,
      "step": 1637
    },
    {
      "epoch": 0.7585144023095791,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025499999999999996,
      "loss": 0.8113,
      "step": 1638
    },
    {
      "epoch": 0.7589774758152626,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002549721706864564,
      "loss": 0.9031,
      "step": 1639
    },
    {
      "epoch": 0.7594405493209461,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002549443413729128,
      "loss": 0.9563,
      "step": 1640
    },
    {
      "epoch": 0.7599036228266296,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002549165120593692,
      "loss": 0.8569,
      "step": 1641
    },
    {
      "epoch": 0.7603666963323131,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002548886827458256,
      "loss": 1.1309,
      "step": 1642
    },
    {
      "epoch": 0.7608297698379967,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000254860853432282,
      "loss": 0.9179,
      "step": 1643
    },
    {
      "epoch": 0.7612928433436801,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025483302411873836,
      "loss": 0.8198,
      "step": 1644
    },
    {
      "epoch": 0.7617559168493636,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00025480519480519475,
      "loss": 0.9387,
      "step": 1645
    },
    {
      "epoch": 0.7622189903550471,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002547773654916512,
      "loss": 0.813,
      "step": 1646
    },
    {
      "epoch": 0.7626820638607307,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002547495361781076,
      "loss": 0.9537,
      "step": 1647
    },
    {
      "epoch": 0.7631451373664141,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00025472170686456397,
      "loss": 0.9608,
      "step": 1648
    },
    {
      "epoch": 0.7636082108720976,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00025469387755102036,
      "loss": 0.7673,
      "step": 1649
    },
    {
      "epoch": 0.7640712843777812,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002546660482374768,
      "loss": 0.8193,
      "step": 1650
    },
    {
      "epoch": 0.7645343578834647,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002546382189239332,
      "loss": 0.7805,
      "step": 1651
    },
    {
      "epoch": 0.7649974313891481,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002546103896103896,
      "loss": 0.8573,
      "step": 1652
    },
    {
      "epoch": 0.7654605048948316,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00025458256029684597,
      "loss": 0.9886,
      "step": 1653
    },
    {
      "epoch": 0.7659235784005152,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002545547309833024,
      "loss": 0.8265,
      "step": 1654
    },
    {
      "epoch": 0.7663866519061987,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002545269016697588,
      "loss": 1.0147,
      "step": 1655
    },
    {
      "epoch": 0.7668497254118821,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002544990723562152,
      "loss": 0.7845,
      "step": 1656
    },
    {
      "epoch": 0.7673127989175657,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002544712430426716,
      "loss": 0.8167,
      "step": 1657
    },
    {
      "epoch": 0.7677758724232492,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000254443413729128,
      "loss": 0.944,
      "step": 1658
    },
    {
      "epoch": 0.7682389459289327,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002544155844155844,
      "loss": 0.7891,
      "step": 1659
    },
    {
      "epoch": 0.7687020194346162,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002543877551020408,
      "loss": 0.8751,
      "step": 1660
    },
    {
      "epoch": 0.7691650929402997,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002543599257884972,
      "loss": 0.8981,
      "step": 1661
    },
    {
      "epoch": 0.7696281664459832,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002543320964749536,
      "loss": 0.9047,
      "step": 1662
    },
    {
      "epoch": 0.7700912399516667,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00025430426716141003,
      "loss": 0.9015,
      "step": 1663
    },
    {
      "epoch": 0.7705543134573503,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00025427643784786636,
      "loss": 0.8219,
      "step": 1664
    },
    {
      "epoch": 0.7710173869630337,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002542486085343228,
      "loss": 0.7408,
      "step": 1665
    },
    {
      "epoch": 0.7714804604687172,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002542207792207792,
      "loss": 0.919,
      "step": 1666
    },
    {
      "epoch": 0.7719435339744007,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002541929499072356,
      "loss": 0.7603,
      "step": 1667
    },
    {
      "epoch": 0.7724066074800843,
      "grad_norm": 0.5625,
      "learning_rate": 0.000254165120593692,
      "loss": 0.8651,
      "step": 1668
    },
    {
      "epoch": 0.7728696809857677,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002541372912801484,
      "loss": 0.9742,
      "step": 1669
    },
    {
      "epoch": 0.7733327544914512,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002541094619666048,
      "loss": 0.8017,
      "step": 1670
    },
    {
      "epoch": 0.7737958279971348,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002540816326530612,
      "loss": 0.8191,
      "step": 1671
    },
    {
      "epoch": 0.7742589015028183,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002540538033395176,
      "loss": 0.9767,
      "step": 1672
    },
    {
      "epoch": 0.7747219750085017,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000254025974025974,
      "loss": 0.8087,
      "step": 1673
    },
    {
      "epoch": 0.7751850485141852,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002539981447124304,
      "loss": 0.8149,
      "step": 1674
    },
    {
      "epoch": 0.7756481220198688,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002539703153988868,
      "loss": 0.7885,
      "step": 1675
    },
    {
      "epoch": 0.7761111955255523,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002539424860853432,
      "loss": 0.8405,
      "step": 1676
    },
    {
      "epoch": 0.7765742690312357,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002539146567717996,
      "loss": 0.8946,
      "step": 1677
    },
    {
      "epoch": 0.7770373425369193,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00025388682745825604,
      "loss": 0.7078,
      "step": 1678
    },
    {
      "epoch": 0.7775004160426028,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002538589981447124,
      "loss": 0.8456,
      "step": 1679
    },
    {
      "epoch": 0.7779634895482863,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002538311688311688,
      "loss": 0.8939,
      "step": 1680
    },
    {
      "epoch": 0.7784265630539697,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002538033395176252,
      "loss": 0.9326,
      "step": 1681
    },
    {
      "epoch": 0.7788896365596533,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00025377551020408165,
      "loss": 0.8489,
      "step": 1682
    },
    {
      "epoch": 0.7793527100653368,
      "grad_norm": 0.5,
      "learning_rate": 0.000253747680890538,
      "loss": 0.7323,
      "step": 1683
    },
    {
      "epoch": 0.7798157835710203,
      "grad_norm": 0.609375,
      "learning_rate": 0.00025371985157699443,
      "loss": 0.9022,
      "step": 1684
    },
    {
      "epoch": 0.7802788570767037,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002536920222634508,
      "loss": 0.9364,
      "step": 1685
    },
    {
      "epoch": 0.7807419305823873,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002536641929499072,
      "loss": 0.9108,
      "step": 1686
    },
    {
      "epoch": 0.7812050040880708,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002536363636363636,
      "loss": 0.8005,
      "step": 1687
    },
    {
      "epoch": 0.7816680775937543,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00025360853432282,
      "loss": 0.879,
      "step": 1688
    },
    {
      "epoch": 0.7821311510994378,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00025358070500927643,
      "loss": 0.9824,
      "step": 1689
    },
    {
      "epoch": 0.7825942246051213,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0002535528756957328,
      "loss": 0.7777,
      "step": 1690
    },
    {
      "epoch": 0.7830572981108048,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002535250463821892,
      "loss": 0.8407,
      "step": 1691
    },
    {
      "epoch": 0.7835203716164884,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002534972170686456,
      "loss": 0.7966,
      "step": 1692
    },
    {
      "epoch": 0.7839834451221718,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00025346938775510204,
      "loss": 0.8428,
      "step": 1693
    },
    {
      "epoch": 0.7844465186278553,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00025344155844155843,
      "loss": 0.8281,
      "step": 1694
    },
    {
      "epoch": 0.7849095921335388,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002534137291280148,
      "loss": 0.779,
      "step": 1695
    },
    {
      "epoch": 0.7853726656392224,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002533858998144712,
      "loss": 0.7916,
      "step": 1696
    },
    {
      "epoch": 0.7858357391449058,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00025335807050092766,
      "loss": 0.7778,
      "step": 1697
    },
    {
      "epoch": 0.7862988126505893,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025333024118738405,
      "loss": 0.8164,
      "step": 1698
    },
    {
      "epoch": 0.7867618861562728,
      "grad_norm": 0.578125,
      "learning_rate": 0.00025330241187384043,
      "loss": 0.8987,
      "step": 1699
    },
    {
      "epoch": 0.7872249596619564,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002532745825602968,
      "loss": 0.8634,
      "step": 1700
    },
    {
      "epoch": 0.7876880331676398,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002532467532467532,
      "loss": 0.8003,
      "step": 1701
    },
    {
      "epoch": 0.7881511066733233,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002532189239332096,
      "loss": 1.0721,
      "step": 1702
    },
    {
      "epoch": 0.7886141801790069,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000253191094619666,
      "loss": 0.8415,
      "step": 1703
    },
    {
      "epoch": 0.7890772536846904,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00025316326530612244,
      "loss": 0.8332,
      "step": 1704
    },
    {
      "epoch": 0.7895403271903738,
      "grad_norm": 0.625,
      "learning_rate": 0.0002531354359925788,
      "loss": 0.9169,
      "step": 1705
    },
    {
      "epoch": 0.7900034006960573,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002531076066790352,
      "loss": 0.8374,
      "step": 1706
    },
    {
      "epoch": 0.7904664742017409,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002530797773654916,
      "loss": 0.8114,
      "step": 1707
    },
    {
      "epoch": 0.7909295477074244,
      "grad_norm": 1.140625,
      "learning_rate": 0.00025305194805194805,
      "loss": 0.8528,
      "step": 1708
    },
    {
      "epoch": 0.7913926212131078,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00025302411873840444,
      "loss": 1.078,
      "step": 1709
    },
    {
      "epoch": 0.7918556947187914,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00025299628942486083,
      "loss": 1.0071,
      "step": 1710
    },
    {
      "epoch": 0.7923187682244749,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002529684601113172,
      "loss": 0.8642,
      "step": 1711
    },
    {
      "epoch": 0.7927818417301584,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00025294063079777366,
      "loss": 0.8678,
      "step": 1712
    },
    {
      "epoch": 0.7932449152358418,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00025291280148423005,
      "loss": 1.0113,
      "step": 1713
    },
    {
      "epoch": 0.7937079887415254,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00025288497217068644,
      "loss": 0.9662,
      "step": 1714
    },
    {
      "epoch": 0.7941710622472089,
      "grad_norm": 0.546875,
      "learning_rate": 0.00025285714285714283,
      "loss": 0.8392,
      "step": 1715
    },
    {
      "epoch": 0.7946341357528924,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002528293135435992,
      "loss": 0.817,
      "step": 1716
    },
    {
      "epoch": 0.7950972092585759,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00025280148423005566,
      "loss": 0.7525,
      "step": 1717
    },
    {
      "epoch": 0.7955602827642594,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000252773654916512,
      "loss": 0.8087,
      "step": 1718
    },
    {
      "epoch": 0.7960233562699429,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00025274582560296844,
      "loss": 0.9377,
      "step": 1719
    },
    {
      "epoch": 0.7964864297756264,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00025271799628942483,
      "loss": 0.9923,
      "step": 1720
    },
    {
      "epoch": 0.7969495032813099,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002526901669758812,
      "loss": 0.8361,
      "step": 1721
    },
    {
      "epoch": 0.7974125767869934,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002526623376623376,
      "loss": 0.9399,
      "step": 1722
    },
    {
      "epoch": 0.7978756502926769,
      "grad_norm": 0.53125,
      "learning_rate": 0.00025263450834879406,
      "loss": 0.8404,
      "step": 1723
    },
    {
      "epoch": 0.7983387237983605,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00025260667903525045,
      "loss": 0.9379,
      "step": 1724
    },
    {
      "epoch": 0.7988017973040439,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00025257884972170684,
      "loss": 0.8804,
      "step": 1725
    },
    {
      "epoch": 0.7992648708097274,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002525510204081632,
      "loss": 0.91,
      "step": 1726
    },
    {
      "epoch": 0.7997279443154109,
      "grad_norm": 0.53125,
      "learning_rate": 0.00025252319109461967,
      "loss": 0.7573,
      "step": 1727
    },
    {
      "epoch": 0.8001910178210945,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00025249536178107606,
      "loss": 0.9237,
      "step": 1728
    },
    {
      "epoch": 0.8006540913267779,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00025246753246753245,
      "loss": 0.8414,
      "step": 1729
    },
    {
      "epoch": 0.8011171648324614,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00025243970315398884,
      "loss": 0.8147,
      "step": 1730
    },
    {
      "epoch": 0.801580238338145,
      "grad_norm": 0.5,
      "learning_rate": 0.00025241187384044523,
      "loss": 0.8264,
      "step": 1731
    },
    {
      "epoch": 0.8020433118438285,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025238404452690167,
      "loss": 0.8563,
      "step": 1732
    },
    {
      "epoch": 0.802506385349512,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00025235621521335806,
      "loss": 0.7645,
      "step": 1733
    },
    {
      "epoch": 0.8029694588551954,
      "grad_norm": 0.4375,
      "learning_rate": 0.00025232838589981445,
      "loss": 0.6514,
      "step": 1734
    },
    {
      "epoch": 0.803432532360879,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025230055658627084,
      "loss": 0.789,
      "step": 1735
    },
    {
      "epoch": 0.8038956058665625,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002522727272727273,
      "loss": 0.9284,
      "step": 1736
    },
    {
      "epoch": 0.804358679372246,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002522448979591836,
      "loss": 0.9037,
      "step": 1737
    },
    {
      "epoch": 0.8048217528779295,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00025221706864564006,
      "loss": 0.8508,
      "step": 1738
    },
    {
      "epoch": 0.805284826383613,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00025218923933209645,
      "loss": 0.8295,
      "step": 1739
    },
    {
      "epoch": 0.8057478998892965,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00025216141001855284,
      "loss": 0.7097,
      "step": 1740
    },
    {
      "epoch": 0.80621097339498,
      "grad_norm": 0.5625,
      "learning_rate": 0.00025213358070500923,
      "loss": 0.8146,
      "step": 1741
    },
    {
      "epoch": 0.8066740469006635,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002521057513914657,
      "loss": 0.7955,
      "step": 1742
    },
    {
      "epoch": 0.807137120406347,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00025207792207792207,
      "loss": 0.7622,
      "step": 1743
    },
    {
      "epoch": 0.8076001939120305,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00025205009276437845,
      "loss": 0.9907,
      "step": 1744
    },
    {
      "epoch": 0.8080632674177141,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025202226345083484,
      "loss": 0.8027,
      "step": 1745
    },
    {
      "epoch": 0.8085263409233975,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00025199443413729123,
      "loss": 0.7769,
      "step": 1746
    },
    {
      "epoch": 0.808989414429081,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002519666048237477,
      "loss": 0.8991,
      "step": 1747
    },
    {
      "epoch": 0.8094524879347645,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00025193877551020407,
      "loss": 0.7633,
      "step": 1748
    },
    {
      "epoch": 0.8099155614404481,
      "grad_norm": 0.546875,
      "learning_rate": 0.00025191094619666046,
      "loss": 0.875,
      "step": 1749
    },
    {
      "epoch": 0.8103786349461315,
      "grad_norm": 0.453125,
      "learning_rate": 0.00025188311688311685,
      "loss": 0.8679,
      "step": 1750
    },
    {
      "epoch": 0.810841708451815,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002518552875695733,
      "loss": 0.8627,
      "step": 1751
    },
    {
      "epoch": 0.8113047819574986,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002518274582560297,
      "loss": 0.8365,
      "step": 1752
    },
    {
      "epoch": 0.8117678554631821,
      "grad_norm": 5.96875,
      "learning_rate": 0.00025179962894248607,
      "loss": 0.8473,
      "step": 1753
    },
    {
      "epoch": 0.8122309289688655,
      "grad_norm": 0.5,
      "learning_rate": 0.00025177179962894246,
      "loss": 0.8218,
      "step": 1754
    },
    {
      "epoch": 0.812694002474549,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002517439703153989,
      "loss": 0.8533,
      "step": 1755
    },
    {
      "epoch": 0.8131570759802326,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00025171614100185524,
      "loss": 0.9165,
      "step": 1756
    },
    {
      "epoch": 0.8136201494859161,
      "grad_norm": 0.59375,
      "learning_rate": 0.00025168831168831163,
      "loss": 0.9022,
      "step": 1757
    },
    {
      "epoch": 0.8140832229915995,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00025166048237476807,
      "loss": 0.9393,
      "step": 1758
    },
    {
      "epoch": 0.814546296497283,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00025163265306122446,
      "loss": 0.7751,
      "step": 1759
    },
    {
      "epoch": 0.8150093700029666,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00025160482374768085,
      "loss": 0.8633,
      "step": 1760
    },
    {
      "epoch": 0.8154724435086501,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00025157699443413724,
      "loss": 0.8952,
      "step": 1761
    },
    {
      "epoch": 0.8159355170143335,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002515491651205937,
      "loss": 1.0209,
      "step": 1762
    },
    {
      "epoch": 0.8163985905200171,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002515213358070501,
      "loss": 0.8051,
      "step": 1763
    },
    {
      "epoch": 0.8168616640257006,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025149350649350646,
      "loss": 0.8046,
      "step": 1764
    },
    {
      "epoch": 0.8173247375313841,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00025146567717996285,
      "loss": 0.8615,
      "step": 1765
    },
    {
      "epoch": 0.8177878110370675,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002514378478664193,
      "loss": 0.8041,
      "step": 1766
    },
    {
      "epoch": 0.8182508845427511,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002514100185528757,
      "loss": 0.8659,
      "step": 1767
    },
    {
      "epoch": 0.8187139580484346,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002513821892393321,
      "loss": 0.7774,
      "step": 1768
    },
    {
      "epoch": 0.8191770315541181,
      "grad_norm": 0.5,
      "learning_rate": 0.00025135435992578847,
      "loss": 0.8532,
      "step": 1769
    },
    {
      "epoch": 0.8196401050598016,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002513265306122449,
      "loss": 0.8786,
      "step": 1770
    },
    {
      "epoch": 0.8201031785654851,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002512987012987013,
      "loss": 0.8825,
      "step": 1771
    },
    {
      "epoch": 0.8205662520711686,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002512708719851577,
      "loss": 0.8266,
      "step": 1772
    },
    {
      "epoch": 0.8210293255768522,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002512430426716141,
      "loss": 0.8954,
      "step": 1773
    },
    {
      "epoch": 0.8214923990825356,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00025121521335807047,
      "loss": 0.9003,
      "step": 1774
    },
    {
      "epoch": 0.8219554725882191,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00025118738404452686,
      "loss": 0.8211,
      "step": 1775
    },
    {
      "epoch": 0.8224185460939026,
      "grad_norm": 0.734375,
      "learning_rate": 0.00025115955473098325,
      "loss": 0.9774,
      "step": 1776
    },
    {
      "epoch": 0.8228816195995862,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002511317254174397,
      "loss": 0.8579,
      "step": 1777
    },
    {
      "epoch": 0.8233446931052696,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002511038961038961,
      "loss": 0.9433,
      "step": 1778
    },
    {
      "epoch": 0.8238077666109531,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00025107606679035247,
      "loss": 0.7768,
      "step": 1779
    },
    {
      "epoch": 0.8242708401166366,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00025104823747680886,
      "loss": 0.7367,
      "step": 1780
    },
    {
      "epoch": 0.8247339136223202,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002510204081632653,
      "loss": 0.9276,
      "step": 1781
    },
    {
      "epoch": 0.8251969871280036,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002509925788497217,
      "loss": 0.7486,
      "step": 1782
    },
    {
      "epoch": 0.8256600606336871,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002509647495361781,
      "loss": 0.7698,
      "step": 1783
    },
    {
      "epoch": 0.8261231341393707,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00025093692022263447,
      "loss": 0.8062,
      "step": 1784
    },
    {
      "epoch": 0.8265862076450542,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00025090909090909086,
      "loss": 0.9139,
      "step": 1785
    },
    {
      "epoch": 0.8270492811507376,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002508812615955473,
      "loss": 0.734,
      "step": 1786
    },
    {
      "epoch": 0.8275123546564211,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002508534322820037,
      "loss": 0.8229,
      "step": 1787
    },
    {
      "epoch": 0.8279754281621047,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0002508256029684601,
      "loss": 0.8629,
      "step": 1788
    },
    {
      "epoch": 0.8284385016677882,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002507977736549165,
      "loss": 0.9134,
      "step": 1789
    },
    {
      "epoch": 0.8289015751734716,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002507699443413729,
      "loss": 0.9444,
      "step": 1790
    },
    {
      "epoch": 0.8293646486791552,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002507421150278293,
      "loss": 0.9725,
      "step": 1791
    },
    {
      "epoch": 0.8298277221848387,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0002507142857142857,
      "loss": 0.9718,
      "step": 1792
    },
    {
      "epoch": 0.8302907956905222,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002506864564007421,
      "loss": 0.728,
      "step": 1793
    },
    {
      "epoch": 0.8307538691962056,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00025065862708719853,
      "loss": 0.7855,
      "step": 1794
    },
    {
      "epoch": 0.8312169427018892,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00025063079777365487,
      "loss": 0.9041,
      "step": 1795
    },
    {
      "epoch": 0.8316800162075727,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002506029684601113,
      "loss": 0.7336,
      "step": 1796
    },
    {
      "epoch": 0.8321430897132562,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002505751391465677,
      "loss": 0.9851,
      "step": 1797
    },
    {
      "epoch": 0.8326061632189397,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0002505473098330241,
      "loss": 0.936,
      "step": 1798
    },
    {
      "epoch": 0.8330692367246232,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002505194805194805,
      "loss": 0.7052,
      "step": 1799
    },
    {
      "epoch": 0.8335323102303067,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00025049165120593687,
      "loss": 0.8567,
      "step": 1800
    },
    {
      "epoch": 0.8339953837359902,
      "grad_norm": 5.90625,
      "learning_rate": 0.0002504638218923933,
      "loss": 0.7968,
      "step": 1801
    },
    {
      "epoch": 0.8344584572416737,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002504359925788497,
      "loss": 0.7762,
      "step": 1802
    },
    {
      "epoch": 0.8349215307473572,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002504081632653061,
      "loss": 0.7701,
      "step": 1803
    },
    {
      "epoch": 0.8353846042530407,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002503803339517625,
      "loss": 1.159,
      "step": 1804
    },
    {
      "epoch": 0.8358476777587243,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002503525046382189,
      "loss": 0.7097,
      "step": 1805
    },
    {
      "epoch": 0.8363107512644077,
      "grad_norm": 1.09375,
      "learning_rate": 0.0002503246753246753,
      "loss": 0.8703,
      "step": 1806
    },
    {
      "epoch": 0.8367738247700912,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002502968460111317,
      "loss": 0.9288,
      "step": 1807
    },
    {
      "epoch": 0.8372368982757747,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002502690166975881,
      "loss": 0.9316,
      "step": 1808
    },
    {
      "epoch": 0.8376999717814583,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00025024118738404454,
      "loss": 0.8389,
      "step": 1809
    },
    {
      "epoch": 0.8381630452871418,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00025021335807050093,
      "loss": 0.855,
      "step": 1810
    },
    {
      "epoch": 0.8386261187928252,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002501855287569573,
      "loss": 0.8831,
      "step": 1811
    },
    {
      "epoch": 0.8390891922985088,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002501576994434137,
      "loss": 0.8376,
      "step": 1812
    },
    {
      "epoch": 0.8395522658041923,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002501298701298701,
      "loss": 0.7542,
      "step": 1813
    },
    {
      "epoch": 0.8400153393098758,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002501020408163265,
      "loss": 0.9306,
      "step": 1814
    },
    {
      "epoch": 0.8404784128155592,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002500742115027829,
      "loss": 0.802,
      "step": 1815
    },
    {
      "epoch": 0.8409414863212428,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002500463821892393,
      "loss": 0.8309,
      "step": 1816
    },
    {
      "epoch": 0.8414045598269263,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002500185528756957,
      "loss": 0.8365,
      "step": 1817
    },
    {
      "epoch": 0.8418676333326098,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002499907235621521,
      "loss": 0.9381,
      "step": 1818
    },
    {
      "epoch": 0.8423307068382933,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002499628942486085,
      "loss": 0.8688,
      "step": 1819
    },
    {
      "epoch": 0.8427937803439768,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00024993506493506493,
      "loss": 0.9335,
      "step": 1820
    },
    {
      "epoch": 0.8432568538496603,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002499072356215213,
      "loss": 0.8733,
      "step": 1821
    },
    {
      "epoch": 0.8437199273553438,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002498794063079777,
      "loss": 0.812,
      "step": 1822
    },
    {
      "epoch": 0.8441830008610273,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0002498515769944341,
      "loss": 0.6975,
      "step": 1823
    },
    {
      "epoch": 0.8446460743667108,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024982374768089054,
      "loss": 0.887,
      "step": 1824
    },
    {
      "epoch": 0.8451091478723943,
      "grad_norm": 0.46875,
      "learning_rate": 0.00024979591836734693,
      "loss": 0.7353,
      "step": 1825
    },
    {
      "epoch": 0.8455722213780779,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002497680890538033,
      "loss": 0.7747,
      "step": 1826
    },
    {
      "epoch": 0.8460352948837613,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0002497402597402597,
      "loss": 0.9034,
      "step": 1827
    },
    {
      "epoch": 0.8464983683894448,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002497124304267161,
      "loss": 0.9112,
      "step": 1828
    },
    {
      "epoch": 0.8469614418951283,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00024968460111317255,
      "loss": 0.8674,
      "step": 1829
    },
    {
      "epoch": 0.8474245154008119,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002496567717996289,
      "loss": 0.8876,
      "step": 1830
    },
    {
      "epoch": 0.8478875889064953,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002496289424860853,
      "loss": 0.8521,
      "step": 1831
    },
    {
      "epoch": 0.8483506624121788,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002496011131725417,
      "loss": 0.7446,
      "step": 1832
    },
    {
      "epoch": 0.8488137359178624,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0002495732838589981,
      "loss": 0.798,
      "step": 1833
    },
    {
      "epoch": 0.8492768094235459,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002495454545454545,
      "loss": 0.9958,
      "step": 1834
    },
    {
      "epoch": 0.8497398829292293,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00024951762523191094,
      "loss": 0.902,
      "step": 1835
    },
    {
      "epoch": 0.8502029564349128,
      "grad_norm": 0.640625,
      "learning_rate": 0.00024948979591836733,
      "loss": 0.9785,
      "step": 1836
    },
    {
      "epoch": 0.8506660299405964,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002494619666048237,
      "loss": 0.9046,
      "step": 1837
    },
    {
      "epoch": 0.8511291034462799,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002494341372912801,
      "loss": 0.8919,
      "step": 1838
    },
    {
      "epoch": 0.8515921769519633,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00024940630797773655,
      "loss": 0.7424,
      "step": 1839
    },
    {
      "epoch": 0.8520552504576469,
      "grad_norm": 0.671875,
      "learning_rate": 0.00024937847866419294,
      "loss": 0.878,
      "step": 1840
    },
    {
      "epoch": 0.8525183239633304,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00024935064935064933,
      "loss": 0.8587,
      "step": 1841
    },
    {
      "epoch": 0.8529813974690139,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002493228200371057,
      "loss": 0.7539,
      "step": 1842
    },
    {
      "epoch": 0.8534444709746973,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002492949907235621,
      "loss": 0.7684,
      "step": 1843
    },
    {
      "epoch": 0.8539075444803809,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00024926716141001855,
      "loss": 0.8366,
      "step": 1844
    },
    {
      "epoch": 0.8543706179860644,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00024923933209647494,
      "loss": 0.8912,
      "step": 1845
    },
    {
      "epoch": 0.8548336914917479,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00024921150278293133,
      "loss": 0.8161,
      "step": 1846
    },
    {
      "epoch": 0.8552967649974313,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002491836734693877,
      "loss": 0.9206,
      "step": 1847
    },
    {
      "epoch": 0.8557598385031149,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00024915584415584417,
      "loss": 0.8187,
      "step": 1848
    },
    {
      "epoch": 0.8562229120087984,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002491280148423005,
      "loss": 0.9097,
      "step": 1849
    },
    {
      "epoch": 0.8566859855144819,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024910018552875695,
      "loss": 0.8517,
      "step": 1850
    },
    {
      "epoch": 0.8571490590201654,
      "grad_norm": 0.59375,
      "learning_rate": 0.00024907235621521334,
      "loss": 0.9318,
      "step": 1851
    },
    {
      "epoch": 0.8576121325258489,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002490445269016697,
      "loss": 0.8492,
      "step": 1852
    },
    {
      "epoch": 0.8580752060315324,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002490166975881261,
      "loss": 0.851,
      "step": 1853
    },
    {
      "epoch": 0.858538279537216,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00024898886827458256,
      "loss": 0.8359,
      "step": 1854
    },
    {
      "epoch": 0.8590013530428994,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024896103896103895,
      "loss": 0.7401,
      "step": 1855
    },
    {
      "epoch": 0.8594644265485829,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00024893320964749534,
      "loss": 0.7377,
      "step": 1856
    },
    {
      "epoch": 0.8599275000542664,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024890538033395173,
      "loss": 0.9029,
      "step": 1857
    },
    {
      "epoch": 0.86039057355995,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002488775510204081,
      "loss": 0.876,
      "step": 1858
    },
    {
      "epoch": 0.8608536470656334,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00024884972170686456,
      "loss": 0.8661,
      "step": 1859
    },
    {
      "epoch": 0.8613167205713169,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00024882189239332095,
      "loss": 0.7283,
      "step": 1860
    },
    {
      "epoch": 0.8617797940770004,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024879406307977734,
      "loss": 0.7947,
      "step": 1861
    },
    {
      "epoch": 0.862242867582684,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024876623376623373,
      "loss": 0.7198,
      "step": 1862
    },
    {
      "epoch": 0.8627059410883674,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002487384044526902,
      "loss": 0.8827,
      "step": 1863
    },
    {
      "epoch": 0.8631690145940509,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00024871057513914656,
      "loss": 0.9961,
      "step": 1864
    },
    {
      "epoch": 0.8636320880997345,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024868274582560295,
      "loss": 0.8847,
      "step": 1865
    },
    {
      "epoch": 0.864095161605418,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00024865491651205934,
      "loss": 0.8456,
      "step": 1866
    },
    {
      "epoch": 0.8645582351111014,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002486270871985158,
      "loss": 0.8461,
      "step": 1867
    },
    {
      "epoch": 0.865021308616785,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002485992578849721,
      "loss": 0.8968,
      "step": 1868
    },
    {
      "epoch": 0.8654843821224685,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00024857142857142857,
      "loss": 0.8263,
      "step": 1869
    },
    {
      "epoch": 0.865947455628152,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00024854359925788495,
      "loss": 0.803,
      "step": 1870
    },
    {
      "epoch": 0.8664105291338354,
      "grad_norm": 0.5,
      "learning_rate": 0.00024851576994434134,
      "loss": 0.7964,
      "step": 1871
    },
    {
      "epoch": 0.866873602639519,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00024848794063079773,
      "loss": 0.9314,
      "step": 1872
    },
    {
      "epoch": 0.8673366761452025,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002484601113172541,
      "loss": 0.7932,
      "step": 1873
    },
    {
      "epoch": 0.867799749650886,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024843228200371057,
      "loss": 0.838,
      "step": 1874
    },
    {
      "epoch": 0.8682628231565694,
      "grad_norm": 0.578125,
      "learning_rate": 0.00024840445269016696,
      "loss": 0.7958,
      "step": 1875
    },
    {
      "epoch": 0.868725896662253,
      "grad_norm": 0.578125,
      "learning_rate": 0.00024837662337662335,
      "loss": 0.8928,
      "step": 1876
    },
    {
      "epoch": 0.8691889701679365,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024834879406307974,
      "loss": 0.8535,
      "step": 1877
    },
    {
      "epoch": 0.86965204367362,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002483209647495362,
      "loss": 0.682,
      "step": 1878
    },
    {
      "epoch": 0.8701151171793035,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00024829313543599257,
      "loss": 0.715,
      "step": 1879
    },
    {
      "epoch": 0.870578190684987,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024826530612244896,
      "loss": 0.8882,
      "step": 1880
    },
    {
      "epoch": 0.8710412641906705,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00024823747680890535,
      "loss": 0.8089,
      "step": 1881
    },
    {
      "epoch": 0.871504337696354,
      "grad_norm": 0.5,
      "learning_rate": 0.0002482096474953618,
      "loss": 0.8594,
      "step": 1882
    },
    {
      "epoch": 0.8719674112020376,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0002481818181818182,
      "loss": 0.9891,
      "step": 1883
    },
    {
      "epoch": 0.872430484707721,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002481539888682745,
      "loss": 0.7523,
      "step": 1884
    },
    {
      "epoch": 0.8728935582134045,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024812615955473096,
      "loss": 0.8578,
      "step": 1885
    },
    {
      "epoch": 0.8733566317190881,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024809833024118735,
      "loss": 0.8325,
      "step": 1886
    },
    {
      "epoch": 0.8738197052247716,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00024807050092764374,
      "loss": 0.7844,
      "step": 1887
    },
    {
      "epoch": 0.874282778730455,
      "grad_norm": 0.546875,
      "learning_rate": 0.00024804267161410013,
      "loss": 0.8134,
      "step": 1888
    },
    {
      "epoch": 0.8747458522361385,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002480148423005566,
      "loss": 0.9248,
      "step": 1889
    },
    {
      "epoch": 0.8752089257418221,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024798701298701296,
      "loss": 0.7429,
      "step": 1890
    },
    {
      "epoch": 0.8756719992475056,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00024795918367346935,
      "loss": 0.8491,
      "step": 1891
    },
    {
      "epoch": 0.876135072753189,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00024793135435992574,
      "loss": 0.828,
      "step": 1892
    },
    {
      "epoch": 0.8765981462588726,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002479035250463822,
      "loss": 0.8029,
      "step": 1893
    },
    {
      "epoch": 0.8770612197645561,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002478756957328386,
      "loss": 0.9102,
      "step": 1894
    },
    {
      "epoch": 0.8775242932702396,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024784786641929497,
      "loss": 0.8507,
      "step": 1895
    },
    {
      "epoch": 0.877987366775923,
      "grad_norm": 0.484375,
      "learning_rate": 0.00024782003710575136,
      "loss": 0.793,
      "step": 1896
    },
    {
      "epoch": 0.8784504402816066,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002477922077922078,
      "loss": 0.8565,
      "step": 1897
    },
    {
      "epoch": 0.8789135137872901,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002477643784786642,
      "loss": 0.8922,
      "step": 1898
    },
    {
      "epoch": 0.8793765872929736,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002477365491651206,
      "loss": 0.7956,
      "step": 1899
    },
    {
      "epoch": 0.879839660798657,
      "grad_norm": 0.625,
      "learning_rate": 0.00024770871985157697,
      "loss": 0.8822,
      "step": 1900
    },
    {
      "epoch": 0.8803027343043406,
      "grad_norm": 0.5,
      "learning_rate": 0.00024768089053803336,
      "loss": 0.6939,
      "step": 1901
    },
    {
      "epoch": 0.8807658078100241,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002476530612244898,
      "loss": 0.746,
      "step": 1902
    },
    {
      "epoch": 0.8812288813157076,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002476252319109462,
      "loss": 0.9145,
      "step": 1903
    },
    {
      "epoch": 0.8816919548213911,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002475974025974026,
      "loss": 0.9255,
      "step": 1904
    },
    {
      "epoch": 0.8821550283270746,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00024756957328385897,
      "loss": 0.7665,
      "step": 1905
    },
    {
      "epoch": 0.8826181018327581,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00024754174397031536,
      "loss": 0.7415,
      "step": 1906
    },
    {
      "epoch": 0.8830811753384417,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00024751391465677175,
      "loss": 0.8625,
      "step": 1907
    },
    {
      "epoch": 0.8835442488441251,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002474860853432282,
      "loss": 0.8845,
      "step": 1908
    },
    {
      "epoch": 0.8840073223498086,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002474582560296846,
      "loss": 0.7544,
      "step": 1909
    },
    {
      "epoch": 0.8844703958554921,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024743042671614097,
      "loss": 0.8238,
      "step": 1910
    },
    {
      "epoch": 0.8849334693611757,
      "grad_norm": 0.5625,
      "learning_rate": 0.00024740259740259736,
      "loss": 0.7807,
      "step": 1911
    },
    {
      "epoch": 0.8853965428668591,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00024737476808905375,
      "loss": 0.7802,
      "step": 1912
    },
    {
      "epoch": 0.8858596163725426,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002473469387755102,
      "loss": 0.8179,
      "step": 1913
    },
    {
      "epoch": 0.8863226898782262,
      "grad_norm": 7.5625,
      "learning_rate": 0.0002473191094619666,
      "loss": 0.7953,
      "step": 1914
    },
    {
      "epoch": 0.8867857633839097,
      "grad_norm": 0.58984375,
      "learning_rate": 0.000247291280148423,
      "loss": 0.8545,
      "step": 1915
    },
    {
      "epoch": 0.8872488368895931,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00024726345083487936,
      "loss": 0.9135,
      "step": 1916
    },
    {
      "epoch": 0.8877119103952766,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002472356215213358,
      "loss": 0.7514,
      "step": 1917
    },
    {
      "epoch": 0.8881749839009602,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002472077922077922,
      "loss": 0.8557,
      "step": 1918
    },
    {
      "epoch": 0.8886380574066437,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002471799628942486,
      "loss": 0.884,
      "step": 1919
    },
    {
      "epoch": 0.8891011309123271,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000247152133580705,
      "loss": 0.8969,
      "step": 1920
    },
    {
      "epoch": 0.8895642044180107,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002471243042671614,
      "loss": 0.812,
      "step": 1921
    },
    {
      "epoch": 0.8900272779236942,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002470964749536178,
      "loss": 0.9283,
      "step": 1922
    },
    {
      "epoch": 0.8904903514293777,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002470686456400742,
      "loss": 0.8372,
      "step": 1923
    },
    {
      "epoch": 0.8909534249350611,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002470408163265306,
      "loss": 0.8957,
      "step": 1924
    },
    {
      "epoch": 0.8914164984407447,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024701298701298703,
      "loss": 0.758,
      "step": 1925
    },
    {
      "epoch": 0.8918795719464282,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00024698515769944337,
      "loss": 0.7618,
      "step": 1926
    },
    {
      "epoch": 0.8923426454521117,
      "grad_norm": 0.59375,
      "learning_rate": 0.00024695732838589976,
      "loss": 0.9159,
      "step": 1927
    },
    {
      "epoch": 0.8928057189577951,
      "grad_norm": 0.5,
      "learning_rate": 0.0002469294990723562,
      "loss": 0.8651,
      "step": 1928
    },
    {
      "epoch": 0.8932687924634787,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002469016697588126,
      "loss": 0.9551,
      "step": 1929
    },
    {
      "epoch": 0.8937318659691622,
      "grad_norm": 0.578125,
      "learning_rate": 0.000246873840445269,
      "loss": 0.8245,
      "step": 1930
    },
    {
      "epoch": 0.8941949394748457,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00024684601113172537,
      "loss": 0.8745,
      "step": 1931
    },
    {
      "epoch": 0.8946580129805292,
      "grad_norm": 0.71875,
      "learning_rate": 0.0002468181818181818,
      "loss": 0.8578,
      "step": 1932
    },
    {
      "epoch": 0.8951210864862127,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002467903525046382,
      "loss": 0.7682,
      "step": 1933
    },
    {
      "epoch": 0.8955841599918962,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002467625231910946,
      "loss": 0.8488,
      "step": 1934
    },
    {
      "epoch": 0.8960472334975798,
      "grad_norm": 0.546875,
      "learning_rate": 0.000246734693877551,
      "loss": 0.872,
      "step": 1935
    },
    {
      "epoch": 0.8965103070032632,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00024670686456400743,
      "loss": 0.8956,
      "step": 1936
    },
    {
      "epoch": 0.8969733805089467,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0002466790352504638,
      "loss": 0.6979,
      "step": 1937
    },
    {
      "epoch": 0.8974364540146302,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002466512059369202,
      "loss": 0.8148,
      "step": 1938
    },
    {
      "epoch": 0.8978995275203138,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002466233766233766,
      "loss": 0.8048,
      "step": 1939
    },
    {
      "epoch": 0.8983626010259972,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000246595547309833,
      "loss": 0.8737,
      "step": 1940
    },
    {
      "epoch": 0.8988256745316807,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024656771799628943,
      "loss": 0.8971,
      "step": 1941
    },
    {
      "epoch": 0.8992887480373642,
      "grad_norm": 0.734375,
      "learning_rate": 0.00024653988868274577,
      "loss": 0.8218,
      "step": 1942
    },
    {
      "epoch": 0.8997518215430478,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002465120593692022,
      "loss": 0.8235,
      "step": 1943
    },
    {
      "epoch": 0.9002148950487312,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002464842300556586,
      "loss": 0.8247,
      "step": 1944
    },
    {
      "epoch": 0.9006779685544147,
      "grad_norm": 0.703125,
      "learning_rate": 0.000246456400742115,
      "loss": 0.9577,
      "step": 1945
    },
    {
      "epoch": 0.9011410420600983,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002464285714285714,
      "loss": 0.8495,
      "step": 1946
    },
    {
      "epoch": 0.9016041155657818,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0002464007421150278,
      "loss": 0.7802,
      "step": 1947
    },
    {
      "epoch": 0.9020671890714652,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002463729128014842,
      "loss": 0.8174,
      "step": 1948
    },
    {
      "epoch": 0.9025302625771487,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002463450834879406,
      "loss": 0.8633,
      "step": 1949
    },
    {
      "epoch": 0.9029933360828323,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000246317254174397,
      "loss": 0.7577,
      "step": 1950
    },
    {
      "epoch": 0.9034564095885158,
      "grad_norm": 0.546875,
      "learning_rate": 0.00024628942486085343,
      "loss": 0.8108,
      "step": 1951
    },
    {
      "epoch": 0.9039194830941992,
      "grad_norm": 0.5,
      "learning_rate": 0.0002462615955473098,
      "loss": 0.8214,
      "step": 1952
    },
    {
      "epoch": 0.9043825565998828,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002462337662337662,
      "loss": 0.7596,
      "step": 1953
    },
    {
      "epoch": 0.9048456301055663,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002462059369202226,
      "loss": 0.8061,
      "step": 1954
    },
    {
      "epoch": 0.9053087036112498,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000246178107606679,
      "loss": 0.8278,
      "step": 1955
    },
    {
      "epoch": 0.9057717771169332,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00024615027829313544,
      "loss": 0.7886,
      "step": 1956
    },
    {
      "epoch": 0.9062348506226168,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002461224489795918,
      "loss": 0.831,
      "step": 1957
    },
    {
      "epoch": 0.9066979241283003,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002460946196660482,
      "loss": 0.8542,
      "step": 1958
    },
    {
      "epoch": 0.9071609976339838,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002460667903525046,
      "loss": 0.8056,
      "step": 1959
    },
    {
      "epoch": 0.9076240711396674,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024603896103896105,
      "loss": 0.8399,
      "step": 1960
    },
    {
      "epoch": 0.9080871446453508,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002460111317254174,
      "loss": 0.754,
      "step": 1961
    },
    {
      "epoch": 0.9085502181510343,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00024598330241187383,
      "loss": 0.764,
      "step": 1962
    },
    {
      "epoch": 0.9090132916567178,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002459554730983302,
      "loss": 0.9406,
      "step": 1963
    },
    {
      "epoch": 0.9094763651624014,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002459276437847866,
      "loss": 0.8723,
      "step": 1964
    },
    {
      "epoch": 0.9099394386680848,
      "grad_norm": 0.5,
      "learning_rate": 0.000245899814471243,
      "loss": 0.7485,
      "step": 1965
    },
    {
      "epoch": 0.9104025121737683,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00024587198515769944,
      "loss": 0.8554,
      "step": 1966
    },
    {
      "epoch": 0.9108655856794519,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024584415584415583,
      "loss": 0.7909,
      "step": 1967
    },
    {
      "epoch": 0.9113286591851354,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002458163265306122,
      "loss": 0.949,
      "step": 1968
    },
    {
      "epoch": 0.9117917326908188,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002457884972170686,
      "loss": 0.9127,
      "step": 1969
    },
    {
      "epoch": 0.9122548061965023,
      "grad_norm": 0.470703125,
      "learning_rate": 0.000245760667903525,
      "loss": 0.8202,
      "step": 1970
    },
    {
      "epoch": 0.9127178797021859,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00024573283858998144,
      "loss": 0.7509,
      "step": 1971
    },
    {
      "epoch": 0.9131809532078694,
      "grad_norm": 0.625,
      "learning_rate": 0.00024570500927643783,
      "loss": 0.8968,
      "step": 1972
    },
    {
      "epoch": 0.9136440267135528,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002456771799628942,
      "loss": 0.8115,
      "step": 1973
    },
    {
      "epoch": 0.9141071002192364,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002456493506493506,
      "loss": 0.8523,
      "step": 1974
    },
    {
      "epoch": 0.9145701737249199,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024562152133580706,
      "loss": 0.9258,
      "step": 1975
    },
    {
      "epoch": 0.9150332472306034,
      "grad_norm": 1.15625,
      "learning_rate": 0.00024559369202226345,
      "loss": 0.8546,
      "step": 1976
    },
    {
      "epoch": 0.9154963207362868,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00024556586270871984,
      "loss": 0.9176,
      "step": 1977
    },
    {
      "epoch": 0.9159593942419704,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002455380333951762,
      "loss": 0.9163,
      "step": 1978
    },
    {
      "epoch": 0.9164224677476539,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00024551020408163267,
      "loss": 0.884,
      "step": 1979
    },
    {
      "epoch": 0.9168855412533374,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000245482374768089,
      "loss": 0.9166,
      "step": 1980
    },
    {
      "epoch": 0.9173486147590209,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00024545454545454545,
      "loss": 0.8312,
      "step": 1981
    },
    {
      "epoch": 0.9178116882647044,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00024542671614100184,
      "loss": 0.8485,
      "step": 1982
    },
    {
      "epoch": 0.9182747617703879,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002453988868274582,
      "loss": 0.857,
      "step": 1983
    },
    {
      "epoch": 0.9187378352760714,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002453710575139146,
      "loss": 0.81,
      "step": 1984
    },
    {
      "epoch": 0.9192009087817549,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000245343228200371,
      "loss": 0.8645,
      "step": 1985
    },
    {
      "epoch": 0.9196639822874384,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024531539888682745,
      "loss": 0.8181,
      "step": 1986
    },
    {
      "epoch": 0.9201270557931219,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00024528756957328384,
      "loss": 0.8605,
      "step": 1987
    },
    {
      "epoch": 0.9205901292988055,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00024525974025974023,
      "loss": 0.8153,
      "step": 1988
    },
    {
      "epoch": 0.9210532028044889,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002452319109461966,
      "loss": 0.7599,
      "step": 1989
    },
    {
      "epoch": 0.9215162763101724,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024520408163265306,
      "loss": 0.8158,
      "step": 1990
    },
    {
      "epoch": 0.9219793498158559,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024517625231910945,
      "loss": 0.8086,
      "step": 1991
    },
    {
      "epoch": 0.9224424233215395,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024514842300556584,
      "loss": 0.8873,
      "step": 1992
    },
    {
      "epoch": 0.9229054968272229,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024512059369202223,
      "loss": 0.8617,
      "step": 1993
    },
    {
      "epoch": 0.9233685703329064,
      "grad_norm": 0.625,
      "learning_rate": 0.0002450927643784787,
      "loss": 1.0305,
      "step": 1994
    },
    {
      "epoch": 0.92383164383859,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024506493506493506,
      "loss": 0.8039,
      "step": 1995
    },
    {
      "epoch": 0.9242947173442735,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024503710575139145,
      "loss": 0.9436,
      "step": 1996
    },
    {
      "epoch": 0.9247577908499569,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024500927643784784,
      "loss": 0.8096,
      "step": 1997
    },
    {
      "epoch": 0.9252208643556404,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00024498144712430423,
      "loss": 0.9188,
      "step": 1998
    },
    {
      "epoch": 0.925683937861324,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002449536178107606,
      "loss": 0.994,
      "step": 1999
    },
    {
      "epoch": 0.9261470113670075,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000244925788497217,
      "loss": 0.9665,
      "step": 2000
    },
    {
      "epoch": 0.9266100848726909,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024489795918367346,
      "loss": 0.7587,
      "step": 2001
    },
    {
      "epoch": 0.9270731583783745,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00024487012987012985,
      "loss": 0.8028,
      "step": 2002
    },
    {
      "epoch": 0.927536231884058,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024484230055658624,
      "loss": 0.909,
      "step": 2003
    },
    {
      "epoch": 0.9279993053897415,
      "grad_norm": 0.703125,
      "learning_rate": 0.0002448144712430426,
      "loss": 1.0948,
      "step": 2004
    },
    {
      "epoch": 0.9284623788954249,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024478664192949907,
      "loss": 0.8408,
      "step": 2005
    },
    {
      "epoch": 0.9289254524011085,
      "grad_norm": 0.578125,
      "learning_rate": 0.00024475881261595546,
      "loss": 0.955,
      "step": 2006
    },
    {
      "epoch": 0.929388525906792,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00024473098330241185,
      "loss": 0.6625,
      "step": 2007
    },
    {
      "epoch": 0.9298515994124755,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00024470315398886824,
      "loss": 0.8451,
      "step": 2008
    },
    {
      "epoch": 0.930314672918159,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002446753246753247,
      "loss": 0.8184,
      "step": 2009
    },
    {
      "epoch": 0.9307777464238425,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00024464749536178107,
      "loss": 0.7364,
      "step": 2010
    },
    {
      "epoch": 0.931240819929526,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00024461966604823746,
      "loss": 0.8415,
      "step": 2011
    },
    {
      "epoch": 0.9317038934352095,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00024459183673469385,
      "loss": 0.9009,
      "step": 2012
    },
    {
      "epoch": 0.932166966940893,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024456400742115024,
      "loss": 0.7836,
      "step": 2013
    },
    {
      "epoch": 0.9326300404465765,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002445361781076067,
      "loss": 0.8775,
      "step": 2014
    },
    {
      "epoch": 0.93309311395226,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002445083487940631,
      "loss": 0.8954,
      "step": 2015
    },
    {
      "epoch": 0.9335561874579436,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00024448051948051946,
      "loss": 0.8091,
      "step": 2016
    },
    {
      "epoch": 0.934019260963627,
      "grad_norm": 0.5625,
      "learning_rate": 0.00024445269016697585,
      "loss": 0.7475,
      "step": 2017
    },
    {
      "epoch": 0.9344823344693105,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024442486085343224,
      "loss": 0.9319,
      "step": 2018
    },
    {
      "epoch": 0.934945407974994,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024439703153988863,
      "loss": 0.8186,
      "step": 2019
    },
    {
      "epoch": 0.9354084814806776,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002443692022263451,
      "loss": 0.8638,
      "step": 2020
    },
    {
      "epoch": 0.935871554986361,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00024434137291280147,
      "loss": 0.7802,
      "step": 2021
    },
    {
      "epoch": 0.9363346284920445,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024431354359925786,
      "loss": 0.7807,
      "step": 2022
    },
    {
      "epoch": 0.936797701997728,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00024428571428571424,
      "loss": 0.9708,
      "step": 2023
    },
    {
      "epoch": 0.9372607755034116,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002442578849721707,
      "loss": 0.8867,
      "step": 2024
    },
    {
      "epoch": 0.937723849009095,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002442300556586271,
      "loss": 0.7754,
      "step": 2025
    },
    {
      "epoch": 0.9381869225147785,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00024420222634508347,
      "loss": 0.8024,
      "step": 2026
    },
    {
      "epoch": 0.9386499960204621,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00024417439703153986,
      "loss": 0.9335,
      "step": 2027
    },
    {
      "epoch": 0.9391130695261456,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024414656771799625,
      "loss": 0.8093,
      "step": 2028
    },
    {
      "epoch": 0.939576143031829,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00024411873840445266,
      "loss": 0.8575,
      "step": 2029
    },
    {
      "epoch": 0.9400392165375125,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00024409090909090905,
      "loss": 0.8139,
      "step": 2030
    },
    {
      "epoch": 0.9405022900431961,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00024406307977736547,
      "loss": 0.9052,
      "step": 2031
    },
    {
      "epoch": 0.9409653635488796,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00024403525046382186,
      "loss": 0.7992,
      "step": 2032
    },
    {
      "epoch": 0.941428437054563,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00024400742115027828,
      "loss": 0.866,
      "step": 2033
    },
    {
      "epoch": 0.9418915105602466,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00024397959183673467,
      "loss": 0.8916,
      "step": 2034
    },
    {
      "epoch": 0.9423545840659301,
      "grad_norm": 0.625,
      "learning_rate": 0.00024395176252319108,
      "loss": 0.8329,
      "step": 2035
    },
    {
      "epoch": 0.9428176575716136,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00024392393320964747,
      "loss": 0.9425,
      "step": 2036
    },
    {
      "epoch": 0.9432807310772972,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002438961038961039,
      "loss": 0.9196,
      "step": 2037
    },
    {
      "epoch": 0.9437438045829806,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024386827458256028,
      "loss": 0.8525,
      "step": 2038
    },
    {
      "epoch": 0.9442068780886641,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00024384044526901667,
      "loss": 0.7367,
      "step": 2039
    },
    {
      "epoch": 0.9446699515943476,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00024381261595547309,
      "loss": 0.9251,
      "step": 2040
    },
    {
      "epoch": 0.9451330251000312,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00024378478664192947,
      "loss": 0.7933,
      "step": 2041
    },
    {
      "epoch": 0.9455960986057146,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002437569573283859,
      "loss": 0.7653,
      "step": 2042
    },
    {
      "epoch": 0.9460591721113981,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00024372912801484228,
      "loss": 0.8447,
      "step": 2043
    },
    {
      "epoch": 0.9465222456170816,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002437012987012987,
      "loss": 0.7609,
      "step": 2044
    },
    {
      "epoch": 0.9469853191227652,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024367346938775506,
      "loss": 0.9459,
      "step": 2045
    },
    {
      "epoch": 0.9474483926284486,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00024364564007421148,
      "loss": 0.8319,
      "step": 2046
    },
    {
      "epoch": 0.9479114661341321,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00024361781076066787,
      "loss": 0.7669,
      "step": 2047
    },
    {
      "epoch": 0.9483745396398157,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024358998144712428,
      "loss": 0.9108,
      "step": 2048
    },
    {
      "epoch": 0.9488376131454992,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00024356215213358067,
      "loss": 0.8588,
      "step": 2049
    },
    {
      "epoch": 0.9493006866511826,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002435343228200371,
      "loss": 0.8637,
      "step": 2050
    },
    {
      "epoch": 0.9497637601568661,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00024350649350649348,
      "loss": 0.867,
      "step": 2051
    },
    {
      "epoch": 0.9502268336625497,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002434786641929499,
      "loss": 0.8485,
      "step": 2052
    },
    {
      "epoch": 0.9506899071682332,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00024345083487940629,
      "loss": 0.799,
      "step": 2053
    },
    {
      "epoch": 0.9511529806739166,
      "grad_norm": 0.578125,
      "learning_rate": 0.00024342300556586268,
      "loss": 0.8549,
      "step": 2054
    },
    {
      "epoch": 0.9516160541796002,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002433951762523191,
      "loss": 0.8596,
      "step": 2055
    },
    {
      "epoch": 0.9520791276852837,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00024336734693877548,
      "loss": 0.7591,
      "step": 2056
    },
    {
      "epoch": 0.9525422011909672,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002433395176252319,
      "loss": 0.8968,
      "step": 2057
    },
    {
      "epoch": 0.9530052746966506,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002433116883116883,
      "loss": 0.846,
      "step": 2058
    },
    {
      "epoch": 0.9534683482023342,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002432838589981447,
      "loss": 0.8078,
      "step": 2059
    },
    {
      "epoch": 0.9539314217080177,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002432560296846011,
      "loss": 0.7033,
      "step": 2060
    },
    {
      "epoch": 0.9543944952137012,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002432282003710575,
      "loss": 0.8401,
      "step": 2061
    },
    {
      "epoch": 0.9548575687193847,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002432003710575139,
      "loss": 0.9039,
      "step": 2062
    },
    {
      "epoch": 0.9553206422250682,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00024317254174397032,
      "loss": 0.9587,
      "step": 2063
    },
    {
      "epoch": 0.9557837157307517,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024314471243042668,
      "loss": 0.7871,
      "step": 2064
    },
    {
      "epoch": 0.9562467892364352,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024311688311688312,
      "loss": 0.8622,
      "step": 2065
    },
    {
      "epoch": 0.9567098627421187,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00024308905380333949,
      "loss": 0.8428,
      "step": 2066
    },
    {
      "epoch": 0.9571729362478022,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00024306122448979588,
      "loss": 0.8255,
      "step": 2067
    },
    {
      "epoch": 0.9576360097534857,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002430333951762523,
      "loss": 0.7916,
      "step": 2068
    },
    {
      "epoch": 0.9580990832591693,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024300556586270868,
      "loss": 0.8357,
      "step": 2069
    },
    {
      "epoch": 0.9585621567648527,
      "grad_norm": 0.5,
      "learning_rate": 0.0002429777365491651,
      "loss": 0.913,
      "step": 2070
    },
    {
      "epoch": 0.9590252302705362,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002429499072356215,
      "loss": 0.8535,
      "step": 2071
    },
    {
      "epoch": 0.9594883037762197,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002429220779220779,
      "loss": 0.9153,
      "step": 2072
    },
    {
      "epoch": 0.9599513772819033,
      "grad_norm": 0.453125,
      "learning_rate": 0.0002428942486085343,
      "loss": 0.8274,
      "step": 2073
    },
    {
      "epoch": 0.9604144507875867,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002428664192949907,
      "loss": 0.8737,
      "step": 2074
    },
    {
      "epoch": 0.9608775242932702,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002428385899814471,
      "loss": 0.7338,
      "step": 2075
    },
    {
      "epoch": 0.9613405977989538,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00024281076066790352,
      "loss": 0.7781,
      "step": 2076
    },
    {
      "epoch": 0.9618036713046373,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002427829313543599,
      "loss": 0.9182,
      "step": 2077
    },
    {
      "epoch": 0.9622667448103207,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024275510204081632,
      "loss": 0.8739,
      "step": 2078
    },
    {
      "epoch": 0.9627298183160042,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002427272727272727,
      "loss": 0.8997,
      "step": 2079
    },
    {
      "epoch": 0.9631928918216878,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00024269944341372913,
      "loss": 0.7476,
      "step": 2080
    },
    {
      "epoch": 0.9636559653273713,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024267161410018552,
      "loss": 0.8889,
      "step": 2081
    },
    {
      "epoch": 0.9641190388330547,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024264378478664188,
      "loss": 0.8396,
      "step": 2082
    },
    {
      "epoch": 0.9645821123387383,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002426159554730983,
      "loss": 0.8157,
      "step": 2083
    },
    {
      "epoch": 0.9650451858444218,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002425881261595547,
      "loss": 0.8577,
      "step": 2084
    },
    {
      "epoch": 0.9655082593501053,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002425602968460111,
      "loss": 0.8182,
      "step": 2085
    },
    {
      "epoch": 0.9659713328557887,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002425324675324675,
      "loss": 0.9226,
      "step": 2086
    },
    {
      "epoch": 0.9664344063614723,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002425046382189239,
      "loss": 0.7775,
      "step": 2087
    },
    {
      "epoch": 0.9668974798671558,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002424768089053803,
      "loss": 0.8404,
      "step": 2088
    },
    {
      "epoch": 0.9673605533728393,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00024244897959183672,
      "loss": 0.831,
      "step": 2089
    },
    {
      "epoch": 0.9678236268785227,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002424211502782931,
      "loss": 0.837,
      "step": 2090
    },
    {
      "epoch": 0.9682867003842063,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00024239332096474952,
      "loss": 0.778,
      "step": 2091
    },
    {
      "epoch": 0.9687497738898898,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00024236549165120591,
      "loss": 0.8023,
      "step": 2092
    },
    {
      "epoch": 0.9692128473955733,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00024233766233766233,
      "loss": 1.0125,
      "step": 2093
    },
    {
      "epoch": 0.9696759209012568,
      "grad_norm": 0.546875,
      "learning_rate": 0.00024230983302411872,
      "loss": 0.8877,
      "step": 2094
    },
    {
      "epoch": 0.9701389944069403,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00024228200371057514,
      "loss": 0.7819,
      "step": 2095
    },
    {
      "epoch": 0.9706020679126238,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00024225417439703153,
      "loss": 0.7977,
      "step": 2096
    },
    {
      "epoch": 0.9710651414183074,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00024222634508348792,
      "loss": 0.8592,
      "step": 2097
    },
    {
      "epoch": 0.9715282149239908,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024219851576994433,
      "loss": 0.8785,
      "step": 2098
    },
    {
      "epoch": 0.9719912884296743,
      "grad_norm": 0.5,
      "learning_rate": 0.0002421706864564007,
      "loss": 0.7436,
      "step": 2099
    },
    {
      "epoch": 0.9724543619353578,
      "grad_norm": 0.625,
      "learning_rate": 0.00024214285714285714,
      "loss": 0.8856,
      "step": 2100
    },
    {
      "epoch": 0.9729174354410414,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002421150278293135,
      "loss": 0.7535,
      "step": 2101
    },
    {
      "epoch": 0.9733805089467248,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00024208719851576992,
      "loss": 0.8362,
      "step": 2102
    },
    {
      "epoch": 0.9738435824524083,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002420593692022263,
      "loss": 0.7986,
      "step": 2103
    },
    {
      "epoch": 0.9743066559580918,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00024203153988868272,
      "loss": 0.7623,
      "step": 2104
    },
    {
      "epoch": 0.9747697294637754,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00024200371057513911,
      "loss": 0.9426,
      "step": 2105
    },
    {
      "epoch": 0.9752328029694588,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024197588126159553,
      "loss": 0.8172,
      "step": 2106
    },
    {
      "epoch": 0.9756958764751423,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00024194805194805192,
      "loss": 0.8633,
      "step": 2107
    },
    {
      "epoch": 0.9761589499808259,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00024192022263450834,
      "loss": 0.8713,
      "step": 2108
    },
    {
      "epoch": 0.9766220234865094,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024189239332096473,
      "loss": 0.8905,
      "step": 2109
    },
    {
      "epoch": 0.9770850969921929,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00024186456400742112,
      "loss": 0.9472,
      "step": 2110
    },
    {
      "epoch": 0.9775481704978763,
      "grad_norm": 0.640625,
      "learning_rate": 0.00024183673469387753,
      "loss": 0.9765,
      "step": 2111
    },
    {
      "epoch": 0.9780112440035599,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024180890538033392,
      "loss": 0.8508,
      "step": 2112
    },
    {
      "epoch": 0.9784743175092434,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00024178107606679034,
      "loss": 0.7155,
      "step": 2113
    },
    {
      "epoch": 0.9789373910149269,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024175324675324673,
      "loss": 0.8332,
      "step": 2114
    },
    {
      "epoch": 0.9794004645206104,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00024172541743970315,
      "loss": 0.8035,
      "step": 2115
    },
    {
      "epoch": 0.9798635380262939,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024169758812615954,
      "loss": 0.8432,
      "step": 2116
    },
    {
      "epoch": 0.9803266115319774,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024166975881261595,
      "loss": 0.8063,
      "step": 2117
    },
    {
      "epoch": 0.980789685037661,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00024164192949907234,
      "loss": 0.9772,
      "step": 2118
    },
    {
      "epoch": 0.9812527585433444,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00024161410018552876,
      "loss": 0.8397,
      "step": 2119
    },
    {
      "epoch": 0.9817158320490279,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00024158627087198512,
      "loss": 0.8449,
      "step": 2120
    },
    {
      "epoch": 0.9821789055547114,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024155844155844154,
      "loss": 0.81,
      "step": 2121
    },
    {
      "epoch": 0.982641979060395,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00024153061224489793,
      "loss": 0.9417,
      "step": 2122
    },
    {
      "epoch": 0.9831050525660784,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00024150278293135434,
      "loss": 0.8585,
      "step": 2123
    },
    {
      "epoch": 0.9835681260717619,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00024147495361781073,
      "loss": 0.6877,
      "step": 2124
    },
    {
      "epoch": 0.9840311995774454,
      "grad_norm": 0.5,
      "learning_rate": 0.00024144712430426712,
      "loss": 0.8328,
      "step": 2125
    },
    {
      "epoch": 0.984494273083129,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024141929499072354,
      "loss": 0.8481,
      "step": 2126
    },
    {
      "epoch": 0.9849573465888124,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00024139146567717993,
      "loss": 0.946,
      "step": 2127
    },
    {
      "epoch": 0.9854204200944959,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00024136363636363635,
      "loss": 0.8376,
      "step": 2128
    },
    {
      "epoch": 0.9858834936001795,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00024133580705009274,
      "loss": 0.8751,
      "step": 2129
    },
    {
      "epoch": 0.986346567105863,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00024130797773654915,
      "loss": 1.0059,
      "step": 2130
    },
    {
      "epoch": 0.9868096406115464,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00024128014842300554,
      "loss": 0.8728,
      "step": 2131
    },
    {
      "epoch": 0.9872727141172299,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024125231910946196,
      "loss": 0.8425,
      "step": 2132
    },
    {
      "epoch": 0.9877357876229135,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00024122448979591835,
      "loss": 0.7283,
      "step": 2133
    },
    {
      "epoch": 0.988198861128597,
      "grad_norm": 0.546875,
      "learning_rate": 0.00024119666048237477,
      "loss": 0.7693,
      "step": 2134
    },
    {
      "epoch": 0.9886619346342804,
      "grad_norm": 0.5625,
      "learning_rate": 0.00024116883116883115,
      "loss": 0.8913,
      "step": 2135
    },
    {
      "epoch": 0.989125008139964,
      "grad_norm": 0.703125,
      "learning_rate": 0.00024114100185528757,
      "loss": 0.9975,
      "step": 2136
    },
    {
      "epoch": 0.9895880816456475,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00024111317254174396,
      "loss": 0.6965,
      "step": 2137
    },
    {
      "epoch": 0.990051155151331,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00024108534322820032,
      "loss": 0.8324,
      "step": 2138
    },
    {
      "epoch": 0.9905142286570144,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00024105751391465674,
      "loss": 0.8573,
      "step": 2139
    },
    {
      "epoch": 0.990977302162698,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024102968460111313,
      "loss": 0.8944,
      "step": 2140
    },
    {
      "epoch": 0.9914403756683815,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00024100185528756955,
      "loss": 0.7238,
      "step": 2141
    },
    {
      "epoch": 0.991903449174065,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00024097402597402594,
      "loss": 0.7295,
      "step": 2142
    },
    {
      "epoch": 0.9923665226797485,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00024094619666048235,
      "loss": 0.8734,
      "step": 2143
    },
    {
      "epoch": 0.992829596185432,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00024091836734693874,
      "loss": 0.8239,
      "step": 2144
    },
    {
      "epoch": 0.9932926696911155,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00024089053803339516,
      "loss": 0.7265,
      "step": 2145
    },
    {
      "epoch": 0.993755743196799,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00024086270871985155,
      "loss": 0.7787,
      "step": 2146
    },
    {
      "epoch": 0.9942188167024825,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00024083487940630797,
      "loss": 0.901,
      "step": 2147
    },
    {
      "epoch": 0.994681890208166,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024080705009276436,
      "loss": 0.9772,
      "step": 2148
    },
    {
      "epoch": 0.9951449637138495,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00024077922077922077,
      "loss": 0.8175,
      "step": 2149
    },
    {
      "epoch": 0.9956080372195331,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00024075139146567716,
      "loss": 0.7976,
      "step": 2150
    },
    {
      "epoch": 0.9960711107252165,
      "grad_norm": 0.515625,
      "learning_rate": 0.00024072356215213358,
      "loss": 0.879,
      "step": 2151
    },
    {
      "epoch": 0.9965341842309,
      "grad_norm": 0.53125,
      "learning_rate": 0.00024069573283858997,
      "loss": 0.8372,
      "step": 2152
    },
    {
      "epoch": 0.9969972577365835,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024066790352504636,
      "loss": 0.9528,
      "step": 2153
    },
    {
      "epoch": 0.9974603312422671,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00024064007421150277,
      "loss": 0.9923,
      "step": 2154
    },
    {
      "epoch": 0.9979234047479505,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00024061224489795914,
      "loss": 0.9207,
      "step": 2155
    },
    {
      "epoch": 0.998386478253634,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00024058441558441558,
      "loss": 0.8082,
      "step": 2156
    },
    {
      "epoch": 0.9988495517593176,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00024055658627087194,
      "loss": 0.8479,
      "step": 2157
    },
    {
      "epoch": 0.9993126252650011,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00024052875695732836,
      "loss": 0.7881,
      "step": 2158
    },
    {
      "epoch": 0.9997756987706845,
      "grad_norm": 0.59375,
      "learning_rate": 0.00024050092764378475,
      "loss": 0.9055,
      "step": 2159
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00024047309833024117,
      "loss": 0.7562,
      "step": 2160
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.6753485202789307,
      "eval_runtime": 180.0299,
      "eval_samples_per_second": 34.117,
      "eval_steps_per_second": 34.117,
      "step": 2160
    },
    {
      "epoch": 1.0,
      "eval_title2sid_loss": 1.5337311029434204,
      "eval_title2sid_runtime": 107.9576,
      "eval_title2sid_samples_per_second": 35.255,
      "eval_title2sid_steps_per_second": 35.255,
      "step": 2160
    },
    {
      "epoch": 1.0,
      "eval_title2sid_loss": 1.5337311029434204,
      "eval_title2sid_runtime": 107.9576,
      "eval_title2sid_samples_per_second": 35.255,
      "eval_title2sid_steps_per_second": 35.255,
      "step": 2160
    },
    {
      "epoch": 1.0,
      "eval_sid2title_loss": 0.5844643712043762,
      "eval_sid2title_runtime": 107.4921,
      "eval_sid2title_samples_per_second": 35.603,
      "eval_sid2title_steps_per_second": 35.603,
      "step": 2160
    },
    {
      "epoch": 1.0,
      "eval_sid2title_loss": 0.5844643712043762,
      "eval_sid2title_runtime": 107.4921,
      "eval_sid2title_samples_per_second": 35.603,
      "eval_sid2title_steps_per_second": 35.603,
      "step": 2160
    },
    {
      "epoch": 1.0004630735056834,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00024044526901669756,
      "loss": 0.5653,
      "step": 2161
    },
    {
      "epoch": 1.000926147011367,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00024041743970315397,
      "loss": 0.6191,
      "step": 2162
    },
    {
      "epoch": 1.0013892205170505,
      "grad_norm": 0.609375,
      "learning_rate": 0.00024038961038961036,
      "loss": 0.7397,
      "step": 2163
    },
    {
      "epoch": 1.0018522940227341,
      "grad_norm": 0.640625,
      "learning_rate": 0.00024036178107606678,
      "loss": 0.6938,
      "step": 2164
    },
    {
      "epoch": 1.0023153675284175,
      "grad_norm": 0.59375,
      "learning_rate": 0.00024033395176252317,
      "loss": 0.6702,
      "step": 2165
    },
    {
      "epoch": 1.002778441034101,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00024030612244897956,
      "loss": 0.6391,
      "step": 2166
    },
    {
      "epoch": 1.0032415145397846,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00024027829313543597,
      "loss": 0.7259,
      "step": 2167
    },
    {
      "epoch": 1.003704588045468,
      "grad_norm": 1.65625,
      "learning_rate": 0.00024025046382189236,
      "loss": 0.6848,
      "step": 2168
    },
    {
      "epoch": 1.0041676615511514,
      "grad_norm": 0.5625,
      "learning_rate": 0.00024022263450834878,
      "loss": 0.581,
      "step": 2169
    },
    {
      "epoch": 1.004630735056835,
      "grad_norm": 0.640625,
      "learning_rate": 0.00024019480519480517,
      "loss": 0.6916,
      "step": 2170
    },
    {
      "epoch": 1.0050938085625185,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0002401669758812616,
      "loss": 0.6696,
      "step": 2171
    },
    {
      "epoch": 1.0055568820682022,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00024013914656771798,
      "loss": 0.7004,
      "step": 2172
    },
    {
      "epoch": 1.0060199555738856,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002401113172541744,
      "loss": 0.6231,
      "step": 2173
    },
    {
      "epoch": 1.006483029079569,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00024008348794063078,
      "loss": 0.6975,
      "step": 2174
    },
    {
      "epoch": 1.0069461025852526,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002400556586270872,
      "loss": 0.6055,
      "step": 2175
    },
    {
      "epoch": 1.007409176090936,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00024002782931354356,
      "loss": 0.5967,
      "step": 2176
    },
    {
      "epoch": 1.0078722495966195,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.6509,
      "step": 2177
    },
    {
      "epoch": 1.0083353231023031,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00023997217068645637,
      "loss": 0.8414,
      "step": 2178
    },
    {
      "epoch": 1.0087983966079865,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023994434137291279,
      "loss": 0.6252,
      "step": 2179
    },
    {
      "epoch": 1.0092614701136702,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023991651205936917,
      "loss": 0.6412,
      "step": 2180
    },
    {
      "epoch": 1.0097245436193536,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023988868274582556,
      "loss": 0.6501,
      "step": 2181
    },
    {
      "epoch": 1.010187617125037,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00023986085343228198,
      "loss": 0.5269,
      "step": 2182
    },
    {
      "epoch": 1.0106506906307207,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00023983302411873837,
      "loss": 0.6955,
      "step": 2183
    },
    {
      "epoch": 1.011113764136404,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002398051948051948,
      "loss": 0.713,
      "step": 2184
    },
    {
      "epoch": 1.0115768376420875,
      "grad_norm": 0.625,
      "learning_rate": 0.00023977736549165118,
      "loss": 0.6391,
      "step": 2185
    },
    {
      "epoch": 1.0120399111477711,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002397495361781076,
      "loss": 0.6428,
      "step": 2186
    },
    {
      "epoch": 1.0125029846534546,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023972170686456398,
      "loss": 0.6522,
      "step": 2187
    },
    {
      "epoch": 1.0129660581591382,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002396938775510204,
      "loss": 0.6793,
      "step": 2188
    },
    {
      "epoch": 1.0134291316648216,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002396660482374768,
      "loss": 0.5743,
      "step": 2189
    },
    {
      "epoch": 1.013892205170505,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0002396382189239332,
      "loss": 0.6911,
      "step": 2190
    },
    {
      "epoch": 1.0143552786761887,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002396103896103896,
      "loss": 0.7126,
      "step": 2191
    },
    {
      "epoch": 1.014818352181872,
      "grad_norm": 0.89453125,
      "learning_rate": 0.000239582560296846,
      "loss": 0.7062,
      "step": 2192
    },
    {
      "epoch": 1.0152814256875555,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002395547309833024,
      "loss": 0.6266,
      "step": 2193
    },
    {
      "epoch": 1.0157444991932392,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00023952690166975876,
      "loss": 0.6735,
      "step": 2194
    },
    {
      "epoch": 1.0162075726989226,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00023949907235621518,
      "loss": 0.6751,
      "step": 2195
    },
    {
      "epoch": 1.0166706462046062,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00023947124304267157,
      "loss": 0.659,
      "step": 2196
    },
    {
      "epoch": 1.0171337197102897,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000239443413729128,
      "loss": 0.681,
      "step": 2197
    },
    {
      "epoch": 1.017596793215973,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023941558441558438,
      "loss": 0.6024,
      "step": 2198
    },
    {
      "epoch": 1.0180598667216567,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0002393877551020408,
      "loss": 0.6042,
      "step": 2199
    },
    {
      "epoch": 1.0185229402273401,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023935992578849718,
      "loss": 0.6247,
      "step": 2200
    },
    {
      "epoch": 1.0189860137330236,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002393320964749536,
      "loss": 0.6231,
      "step": 2201
    },
    {
      "epoch": 1.0194490872387072,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023930426716141,
      "loss": 0.6052,
      "step": 2202
    },
    {
      "epoch": 1.0199121607443906,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002392764378478664,
      "loss": 0.6729,
      "step": 2203
    },
    {
      "epoch": 1.0203752342500743,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002392486085343228,
      "loss": 0.6807,
      "step": 2204
    },
    {
      "epoch": 1.0208383077557577,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002392207792207792,
      "loss": 0.6362,
      "step": 2205
    },
    {
      "epoch": 1.021301381261441,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002391929499072356,
      "loss": 0.7175,
      "step": 2206
    },
    {
      "epoch": 1.0217644547671247,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023916512059369202,
      "loss": 0.6218,
      "step": 2207
    },
    {
      "epoch": 1.0222275282728082,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002391372912801484,
      "loss": 0.5554,
      "step": 2208
    },
    {
      "epoch": 1.0226906017784916,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002391094619666048,
      "loss": 0.6482,
      "step": 2209
    },
    {
      "epoch": 1.0231536752841752,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023908163265306122,
      "loss": 0.5413,
      "step": 2210
    },
    {
      "epoch": 1.0236167487898586,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00023905380333951758,
      "loss": 0.7093,
      "step": 2211
    },
    {
      "epoch": 1.0240798222955423,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023902597402597402,
      "loss": 0.6898,
      "step": 2212
    },
    {
      "epoch": 1.0245428958012257,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00023899814471243038,
      "loss": 0.6053,
      "step": 2213
    },
    {
      "epoch": 1.0250059693069091,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002389703153988868,
      "loss": 0.6452,
      "step": 2214
    },
    {
      "epoch": 1.0254690428125928,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002389424860853432,
      "loss": 0.5607,
      "step": 2215
    },
    {
      "epoch": 1.0259321163182762,
      "grad_norm": 0.625,
      "learning_rate": 0.0002389146567717996,
      "loss": 0.7208,
      "step": 2216
    },
    {
      "epoch": 1.0263951898239596,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000238886827458256,
      "loss": 0.6605,
      "step": 2217
    },
    {
      "epoch": 1.0268582633296433,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00023885899814471241,
      "loss": 0.7162,
      "step": 2218
    },
    {
      "epoch": 1.0273213368353267,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002388311688311688,
      "loss": 0.6101,
      "step": 2219
    },
    {
      "epoch": 1.0277844103410103,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00023880333951762522,
      "loss": 0.5924,
      "step": 2220
    },
    {
      "epoch": 1.0282474838466937,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0002387755102040816,
      "loss": 0.6852,
      "step": 2221
    },
    {
      "epoch": 1.0287105573523772,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023874768089053803,
      "loss": 0.603,
      "step": 2222
    },
    {
      "epoch": 1.0291736308580608,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00023871985157699442,
      "loss": 0.6861,
      "step": 2223
    },
    {
      "epoch": 1.0296367043637442,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002386920222634508,
      "loss": 0.5923,
      "step": 2224
    },
    {
      "epoch": 1.0300997778694276,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00023866419294990722,
      "loss": 0.6456,
      "step": 2225
    },
    {
      "epoch": 1.0305628513751113,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002386363636363636,
      "loss": 0.6786,
      "step": 2226
    },
    {
      "epoch": 1.0310259248807947,
      "grad_norm": 0.625,
      "learning_rate": 0.00023860853432282003,
      "loss": 0.7037,
      "step": 2227
    },
    {
      "epoch": 1.0314889983864783,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00023858070500927642,
      "loss": 0.6035,
      "step": 2228
    },
    {
      "epoch": 1.0319520718921618,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023855287569573283,
      "loss": 0.6895,
      "step": 2229
    },
    {
      "epoch": 1.0324151453978452,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00023852504638218922,
      "loss": 0.6534,
      "step": 2230
    },
    {
      "epoch": 1.0328782189035288,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00023849721706864564,
      "loss": 0.7046,
      "step": 2231
    },
    {
      "epoch": 1.0333412924092122,
      "grad_norm": 0.578125,
      "learning_rate": 0.000238469387755102,
      "loss": 0.6563,
      "step": 2232
    },
    {
      "epoch": 1.0338043659148957,
      "grad_norm": 0.625,
      "learning_rate": 0.00023844155844155842,
      "loss": 0.5748,
      "step": 2233
    },
    {
      "epoch": 1.0342674394205793,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002384137291280148,
      "loss": 0.6698,
      "step": 2234
    },
    {
      "epoch": 1.0347305129262627,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023838589981447123,
      "loss": 0.5995,
      "step": 2235
    },
    {
      "epoch": 1.0351935864319464,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023835807050092762,
      "loss": 0.5987,
      "step": 2236
    },
    {
      "epoch": 1.0356566599376298,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000238330241187384,
      "loss": 0.6668,
      "step": 2237
    },
    {
      "epoch": 1.0361197334433132,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023830241187384042,
      "loss": 0.6144,
      "step": 2238
    },
    {
      "epoch": 1.0365828069489968,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002382745825602968,
      "loss": 0.6229,
      "step": 2239
    },
    {
      "epoch": 1.0370458804546803,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023824675324675323,
      "loss": 0.5975,
      "step": 2240
    },
    {
      "epoch": 1.037508953960364,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023821892393320962,
      "loss": 0.5886,
      "step": 2241
    },
    {
      "epoch": 1.0379720274660473,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00023819109461966604,
      "loss": 0.6711,
      "step": 2242
    },
    {
      "epoch": 1.0384351009717308,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00023816326530612242,
      "loss": 0.5822,
      "step": 2243
    },
    {
      "epoch": 1.0388981744774144,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00023813543599257884,
      "loss": 0.6032,
      "step": 2244
    },
    {
      "epoch": 1.0393612479830978,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00023810760667903523,
      "loss": 0.6089,
      "step": 2245
    },
    {
      "epoch": 1.0398243214887812,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023807977736549165,
      "loss": 0.6045,
      "step": 2246
    },
    {
      "epoch": 1.0402873949944649,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023805194805194804,
      "loss": 0.5861,
      "step": 2247
    },
    {
      "epoch": 1.0407504685001483,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00023802411873840445,
      "loss": 0.6656,
      "step": 2248
    },
    {
      "epoch": 1.041213542005832,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00023799628942486084,
      "loss": 0.6142,
      "step": 2249
    },
    {
      "epoch": 1.0416766155115154,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023796846011131726,
      "loss": 0.6536,
      "step": 2250
    },
    {
      "epoch": 1.0421396890171988,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00023794063079777362,
      "loss": 0.6961,
      "step": 2251
    },
    {
      "epoch": 1.0426027625228824,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00023791280148423,
      "loss": 0.6367,
      "step": 2252
    },
    {
      "epoch": 1.0430658360285658,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00023788497217068643,
      "loss": 0.7632,
      "step": 2253
    },
    {
      "epoch": 1.0435289095342493,
      "grad_norm": 0.5,
      "learning_rate": 0.00023785714285714282,
      "loss": 0.5795,
      "step": 2254
    },
    {
      "epoch": 1.043991983039933,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00023782931354359924,
      "loss": 0.6145,
      "step": 2255
    },
    {
      "epoch": 1.0444550565456163,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023780148423005563,
      "loss": 0.6429,
      "step": 2256
    },
    {
      "epoch": 1.0449181300513,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00023777365491651204,
      "loss": 0.6018,
      "step": 2257
    },
    {
      "epoch": 1.0453812035569834,
      "grad_norm": 0.59375,
      "learning_rate": 0.00023774582560296843,
      "loss": 0.6475,
      "step": 2258
    },
    {
      "epoch": 1.0458442770626668,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00023771799628942485,
      "loss": 0.7988,
      "step": 2259
    },
    {
      "epoch": 1.0463073505683504,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023769016697588124,
      "loss": 0.603,
      "step": 2260
    },
    {
      "epoch": 1.0467704240740339,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00023766233766233765,
      "loss": 0.6898,
      "step": 2261
    },
    {
      "epoch": 1.0472334975797173,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023763450834879404,
      "loss": 0.6162,
      "step": 2262
    },
    {
      "epoch": 1.047696571085401,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00023760667903525046,
      "loss": 0.6629,
      "step": 2263
    },
    {
      "epoch": 1.0481596445910843,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00023757884972170685,
      "loss": 0.5778,
      "step": 2264
    },
    {
      "epoch": 1.048622718096768,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00023755102040816324,
      "loss": 0.6192,
      "step": 2265
    },
    {
      "epoch": 1.0490857916024514,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00023752319109461966,
      "loss": 0.6649,
      "step": 2266
    },
    {
      "epoch": 1.0495488651081348,
      "grad_norm": 1.609375,
      "learning_rate": 0.00023749536178107602,
      "loss": 0.5444,
      "step": 2267
    },
    {
      "epoch": 1.0500119386138185,
      "grad_norm": 0.5,
      "learning_rate": 0.00023746753246753246,
      "loss": 0.6078,
      "step": 2268
    },
    {
      "epoch": 1.050475012119502,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00023743970315398883,
      "loss": 0.6426,
      "step": 2269
    },
    {
      "epoch": 1.0509380856251853,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023741187384044524,
      "loss": 0.6544,
      "step": 2270
    },
    {
      "epoch": 1.051401159130869,
      "grad_norm": 0.671875,
      "learning_rate": 0.00023738404452690163,
      "loss": 0.6525,
      "step": 2271
    },
    {
      "epoch": 1.0518642326365524,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00023735621521335805,
      "loss": 0.6787,
      "step": 2272
    },
    {
      "epoch": 1.052327306142236,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023732838589981444,
      "loss": 0.6237,
      "step": 2273
    },
    {
      "epoch": 1.0527903796479194,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023730055658627085,
      "loss": 0.6578,
      "step": 2274
    },
    {
      "epoch": 1.0532534531536029,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00023727272727272724,
      "loss": 0.7005,
      "step": 2275
    },
    {
      "epoch": 1.0537165266592865,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00023724489795918366,
      "loss": 0.6538,
      "step": 2276
    },
    {
      "epoch": 1.05417960016497,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023721706864564005,
      "loss": 0.6474,
      "step": 2277
    },
    {
      "epoch": 1.0546426736706533,
      "grad_norm": 0.65625,
      "learning_rate": 0.00023718923933209647,
      "loss": 0.6345,
      "step": 2278
    },
    {
      "epoch": 1.055105747176337,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00023716141001855286,
      "loss": 0.6494,
      "step": 2279
    },
    {
      "epoch": 1.0555688206820204,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023713358070500925,
      "loss": 0.5711,
      "step": 2280
    },
    {
      "epoch": 1.056031894187704,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023710575139146566,
      "loss": 0.6625,
      "step": 2281
    },
    {
      "epoch": 1.0564949676933875,
      "grad_norm": 0.65625,
      "learning_rate": 0.00023707792207792205,
      "loss": 0.7278,
      "step": 2282
    },
    {
      "epoch": 1.0569580411990709,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00023705009276437847,
      "loss": 0.6002,
      "step": 2283
    },
    {
      "epoch": 1.0574211147047545,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023702226345083486,
      "loss": 0.6708,
      "step": 2284
    },
    {
      "epoch": 1.057884188210438,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00023699443413729128,
      "loss": 0.6049,
      "step": 2285
    },
    {
      "epoch": 1.0583472617161214,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00023696660482374767,
      "loss": 0.6859,
      "step": 2286
    },
    {
      "epoch": 1.058810335221805,
      "grad_norm": 1.015625,
      "learning_rate": 0.00023693877551020408,
      "loss": 0.765,
      "step": 2287
    },
    {
      "epoch": 1.0592734087274884,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023691094619666044,
      "loss": 0.6459,
      "step": 2288
    },
    {
      "epoch": 1.059736482233172,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00023688311688311686,
      "loss": 0.6433,
      "step": 2289
    },
    {
      "epoch": 1.0601995557388555,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023685528756957325,
      "loss": 0.6211,
      "step": 2290
    },
    {
      "epoch": 1.060662629244539,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00023682745825602967,
      "loss": 0.6702,
      "step": 2291
    },
    {
      "epoch": 1.0611257027502226,
      "grad_norm": 0.5,
      "learning_rate": 0.00023679962894248606,
      "loss": 0.5366,
      "step": 2292
    },
    {
      "epoch": 1.061588776255906,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00023677179962894245,
      "loss": 0.6126,
      "step": 2293
    },
    {
      "epoch": 1.0620518497615894,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023674397031539886,
      "loss": 0.6606,
      "step": 2294
    },
    {
      "epoch": 1.062514923267273,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00023671614100185525,
      "loss": 0.6408,
      "step": 2295
    },
    {
      "epoch": 1.0629779967729565,
      "grad_norm": 1.8203125,
      "learning_rate": 0.00023668831168831167,
      "loss": 0.6684,
      "step": 2296
    },
    {
      "epoch": 1.06344107027864,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00023666048237476806,
      "loss": 0.6646,
      "step": 2297
    },
    {
      "epoch": 1.0639041437843235,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00023663265306122448,
      "loss": 0.6224,
      "step": 2298
    },
    {
      "epoch": 1.064367217290007,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00023660482374768087,
      "loss": 0.6061,
      "step": 2299
    },
    {
      "epoch": 1.0648302907956906,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023657699443413728,
      "loss": 0.6993,
      "step": 2300
    },
    {
      "epoch": 1.065293364301374,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00023654916512059367,
      "loss": 0.5595,
      "step": 2301
    },
    {
      "epoch": 1.0657564378070576,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002365213358070501,
      "loss": 0.5988,
      "step": 2302
    },
    {
      "epoch": 1.066219511312741,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023649350649350648,
      "loss": 0.6398,
      "step": 2303
    },
    {
      "epoch": 1.0666825848184245,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002364656771799629,
      "loss": 0.6611,
      "step": 2304
    },
    {
      "epoch": 1.0671456583241081,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00023643784786641929,
      "loss": 0.6078,
      "step": 2305
    },
    {
      "epoch": 1.0676087318297915,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002364100185528757,
      "loss": 0.6404,
      "step": 2306
    },
    {
      "epoch": 1.068071805335475,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023638218923933206,
      "loss": 0.6715,
      "step": 2307
    },
    {
      "epoch": 1.0685348788411586,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023635435992578845,
      "loss": 0.713,
      "step": 2308
    },
    {
      "epoch": 1.068997952346842,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00023632653061224487,
      "loss": 0.6154,
      "step": 2309
    },
    {
      "epoch": 1.0694610258525254,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023629870129870126,
      "loss": 0.6847,
      "step": 2310
    },
    {
      "epoch": 1.069924099358209,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00023627087198515768,
      "loss": 0.7044,
      "step": 2311
    },
    {
      "epoch": 1.0703871728638925,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00023624304267161407,
      "loss": 0.7262,
      "step": 2312
    },
    {
      "epoch": 1.0708502463695762,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00023621521335807048,
      "loss": 0.5701,
      "step": 2313
    },
    {
      "epoch": 1.0713133198752596,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00023618738404452687,
      "loss": 0.6745,
      "step": 2314
    },
    {
      "epoch": 1.071776393380943,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002361595547309833,
      "loss": 0.5994,
      "step": 2315
    },
    {
      "epoch": 1.0722394668866266,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023613172541743968,
      "loss": 0.6396,
      "step": 2316
    },
    {
      "epoch": 1.07270254039231,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002361038961038961,
      "loss": 0.6202,
      "step": 2317
    },
    {
      "epoch": 1.0731656138979937,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00023607606679035249,
      "loss": 0.7377,
      "step": 2318
    },
    {
      "epoch": 1.0736286874036771,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002360482374768089,
      "loss": 0.5556,
      "step": 2319
    },
    {
      "epoch": 1.0740917609093605,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002360204081632653,
      "loss": 0.6571,
      "step": 2320
    },
    {
      "epoch": 1.0745548344150442,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023599257884972168,
      "loss": 0.6253,
      "step": 2321
    },
    {
      "epoch": 1.0750179079207276,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002359647495361781,
      "loss": 0.6033,
      "step": 2322
    },
    {
      "epoch": 1.075480981426411,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023593692022263446,
      "loss": 0.5957,
      "step": 2323
    },
    {
      "epoch": 1.0759440549320947,
      "grad_norm": 1.359375,
      "learning_rate": 0.0002359090909090909,
      "loss": 0.7088,
      "step": 2324
    },
    {
      "epoch": 1.076407128437778,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00023588126159554727,
      "loss": 0.7434,
      "step": 2325
    },
    {
      "epoch": 1.0768702019434615,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00023585343228200368,
      "loss": 0.6627,
      "step": 2326
    },
    {
      "epoch": 1.0773332754491451,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023582560296846007,
      "loss": 0.6606,
      "step": 2327
    },
    {
      "epoch": 1.0777963489548286,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002357977736549165,
      "loss": 0.6201,
      "step": 2328
    },
    {
      "epoch": 1.0782594224605122,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00023576994434137288,
      "loss": 0.6832,
      "step": 2329
    },
    {
      "epoch": 1.0787224959661956,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002357421150278293,
      "loss": 0.6808,
      "step": 2330
    },
    {
      "epoch": 1.079185569471879,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023571428571428569,
      "loss": 0.6189,
      "step": 2331
    },
    {
      "epoch": 1.0796486429775627,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002356864564007421,
      "loss": 0.6299,
      "step": 2332
    },
    {
      "epoch": 1.080111716483246,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002356586270871985,
      "loss": 0.593,
      "step": 2333
    },
    {
      "epoch": 1.0805747899889298,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002356307977736549,
      "loss": 0.5949,
      "step": 2334
    },
    {
      "epoch": 1.0810378634946132,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002356029684601113,
      "loss": 0.6383,
      "step": 2335
    },
    {
      "epoch": 1.0815009370002966,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002355751391465677,
      "loss": 0.6492,
      "step": 2336
    },
    {
      "epoch": 1.0819640105059802,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002355473098330241,
      "loss": 0.6597,
      "step": 2337
    },
    {
      "epoch": 1.0824270840116637,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002355194805194805,
      "loss": 0.5723,
      "step": 2338
    },
    {
      "epoch": 1.082890157517347,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002354916512059369,
      "loss": 0.6582,
      "step": 2339
    },
    {
      "epoch": 1.0833532310230307,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002354638218923933,
      "loss": 0.6263,
      "step": 2340
    },
    {
      "epoch": 1.0838163045287141,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023543599257884972,
      "loss": 0.6175,
      "step": 2341
    },
    {
      "epoch": 1.0842793780343978,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00023540816326530608,
      "loss": 0.6572,
      "step": 2342
    },
    {
      "epoch": 1.0847424515400812,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023538033395176252,
      "loss": 0.6298,
      "step": 2343
    },
    {
      "epoch": 1.0852055250457646,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00023535250463821889,
      "loss": 0.6373,
      "step": 2344
    },
    {
      "epoch": 1.0856685985514483,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002353246753246753,
      "loss": 0.7245,
      "step": 2345
    },
    {
      "epoch": 1.0861316720571317,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002352968460111317,
      "loss": 0.6433,
      "step": 2346
    },
    {
      "epoch": 1.086594745562815,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0002352690166975881,
      "loss": 0.7281,
      "step": 2347
    },
    {
      "epoch": 1.0870578190684987,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002352411873840445,
      "loss": 0.6342,
      "step": 2348
    },
    {
      "epoch": 1.0875208925741822,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00023521335807050092,
      "loss": 0.5722,
      "step": 2349
    },
    {
      "epoch": 1.0879839660798658,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002351855287569573,
      "loss": 0.6424,
      "step": 2350
    },
    {
      "epoch": 1.0884470395855492,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002351576994434137,
      "loss": 0.6511,
      "step": 2351
    },
    {
      "epoch": 1.0889101130912326,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002351298701298701,
      "loss": 0.6694,
      "step": 2352
    },
    {
      "epoch": 1.0893731865969163,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002351020408163265,
      "loss": 0.665,
      "step": 2353
    },
    {
      "epoch": 1.0898362601025997,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023507421150278292,
      "loss": 0.6004,
      "step": 2354
    },
    {
      "epoch": 1.0902993336082831,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002350463821892393,
      "loss": 0.6425,
      "step": 2355
    },
    {
      "epoch": 1.0907624071139668,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023501855287569572,
      "loss": 0.6101,
      "step": 2356
    },
    {
      "epoch": 1.0912254806196502,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00023499072356215211,
      "loss": 0.7244,
      "step": 2357
    },
    {
      "epoch": 1.0916885541253338,
      "grad_norm": 0.625,
      "learning_rate": 0.00023496289424860853,
      "loss": 0.7493,
      "step": 2358
    },
    {
      "epoch": 1.0921516276310173,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00023493506493506492,
      "loss": 0.5897,
      "step": 2359
    },
    {
      "epoch": 1.0926147011367007,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00023490723562152134,
      "loss": 0.6397,
      "step": 2360
    },
    {
      "epoch": 1.0930777746423843,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00023487940630797773,
      "loss": 0.6521,
      "step": 2361
    },
    {
      "epoch": 1.0935408481480677,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023485157699443414,
      "loss": 0.5778,
      "step": 2362
    },
    {
      "epoch": 1.0940039216537512,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002348237476808905,
      "loss": 0.6911,
      "step": 2363
    },
    {
      "epoch": 1.0944669951594348,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002347959183673469,
      "loss": 0.6506,
      "step": 2364
    },
    {
      "epoch": 1.0949300686651182,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002347680890538033,
      "loss": 0.6091,
      "step": 2365
    },
    {
      "epoch": 1.0953931421708019,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0002347402597402597,
      "loss": 0.6921,
      "step": 2366
    },
    {
      "epoch": 1.0958562156764853,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023471243042671612,
      "loss": 0.605,
      "step": 2367
    },
    {
      "epoch": 1.0963192891821687,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002346846011131725,
      "loss": 0.5907,
      "step": 2368
    },
    {
      "epoch": 1.0967823626878523,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023465677179962892,
      "loss": 0.6085,
      "step": 2369
    },
    {
      "epoch": 1.0972454361935358,
      "grad_norm": 0.8125,
      "learning_rate": 0.00023462894248608531,
      "loss": 0.7325,
      "step": 2370
    },
    {
      "epoch": 1.0977085096992192,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023460111317254173,
      "loss": 0.6316,
      "step": 2371
    },
    {
      "epoch": 1.0981715832049028,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00023457328385899812,
      "loss": 0.6101,
      "step": 2372
    },
    {
      "epoch": 1.0986346567105862,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00023454545454545454,
      "loss": 0.581,
      "step": 2373
    },
    {
      "epoch": 1.0990977302162699,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00023451762523191093,
      "loss": 0.6184,
      "step": 2374
    },
    {
      "epoch": 1.0995608037219533,
      "grad_norm": 0.6875,
      "learning_rate": 0.00023448979591836734,
      "loss": 0.7495,
      "step": 2375
    },
    {
      "epoch": 1.1000238772276367,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00023446196660482373,
      "loss": 0.7333,
      "step": 2376
    },
    {
      "epoch": 1.1004869507333204,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023443413729128015,
      "loss": 0.6467,
      "step": 2377
    },
    {
      "epoch": 1.1009500242390038,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023440630797773654,
      "loss": 0.6287,
      "step": 2378
    },
    {
      "epoch": 1.1014130977446874,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002343784786641929,
      "loss": 0.6725,
      "step": 2379
    },
    {
      "epoch": 1.1018761712503709,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023435064935064935,
      "loss": 0.6362,
      "step": 2380
    },
    {
      "epoch": 1.1023392447560543,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002343228200371057,
      "loss": 0.6173,
      "step": 2381
    },
    {
      "epoch": 1.102802318261738,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023429499072356212,
      "loss": 0.6962,
      "step": 2382
    },
    {
      "epoch": 1.1032653917674213,
      "grad_norm": 0.46875,
      "learning_rate": 0.00023426716141001851,
      "loss": 0.5504,
      "step": 2383
    },
    {
      "epoch": 1.1037284652731048,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023423933209647493,
      "loss": 0.5972,
      "step": 2384
    },
    {
      "epoch": 1.1041915387787884,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00023421150278293132,
      "loss": 0.5586,
      "step": 2385
    },
    {
      "epoch": 1.1046546122844718,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00023418367346938774,
      "loss": 0.5663,
      "step": 2386
    },
    {
      "epoch": 1.1051176857901552,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023415584415584413,
      "loss": 0.6369,
      "step": 2387
    },
    {
      "epoch": 1.1055807592958389,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00023412801484230054,
      "loss": 0.6602,
      "step": 2388
    },
    {
      "epoch": 1.1060438328015223,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023410018552875693,
      "loss": 0.655,
      "step": 2389
    },
    {
      "epoch": 1.106506906307206,
      "grad_norm": 0.625,
      "learning_rate": 0.00023407235621521335,
      "loss": 0.6804,
      "step": 2390
    },
    {
      "epoch": 1.1069699798128894,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023404452690166974,
      "loss": 0.709,
      "step": 2391
    },
    {
      "epoch": 1.1074330533185728,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023401669758812613,
      "loss": 0.6918,
      "step": 2392
    },
    {
      "epoch": 1.1078961268242564,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00023398886827458255,
      "loss": 0.7032,
      "step": 2393
    },
    {
      "epoch": 1.1083592003299398,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023396103896103894,
      "loss": 0.6863,
      "step": 2394
    },
    {
      "epoch": 1.1088222738356235,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023393320964749535,
      "loss": 0.617,
      "step": 2395
    },
    {
      "epoch": 1.109285347341307,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023390538033395174,
      "loss": 0.6813,
      "step": 2396
    },
    {
      "epoch": 1.1097484208469903,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023387755102040816,
      "loss": 0.6725,
      "step": 2397
    },
    {
      "epoch": 1.110211494352674,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023384972170686452,
      "loss": 0.6589,
      "step": 2398
    },
    {
      "epoch": 1.1106745678583574,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023382189239332097,
      "loss": 0.6356,
      "step": 2399
    },
    {
      "epoch": 1.1111376413640408,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00023379406307977733,
      "loss": 0.598,
      "step": 2400
    },
    {
      "epoch": 1.1116007148697244,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00023376623376623374,
      "loss": 0.5292,
      "step": 2401
    },
    {
      "epoch": 1.1120637883754079,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00023373840445269013,
      "loss": 0.6182,
      "step": 2402
    },
    {
      "epoch": 1.1125268618810913,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023371057513914655,
      "loss": 0.6331,
      "step": 2403
    },
    {
      "epoch": 1.112989935386775,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023368274582560294,
      "loss": 0.6285,
      "step": 2404
    },
    {
      "epoch": 1.1134530088924584,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00023365491651205936,
      "loss": 0.655,
      "step": 2405
    },
    {
      "epoch": 1.113916082398142,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023362708719851575,
      "loss": 0.6364,
      "step": 2406
    },
    {
      "epoch": 1.1143791559038254,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023359925788497214,
      "loss": 0.7084,
      "step": 2407
    },
    {
      "epoch": 1.1148422294095088,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00023357142857142855,
      "loss": 0.6253,
      "step": 2408
    },
    {
      "epoch": 1.1153053029151925,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023354359925788494,
      "loss": 0.6503,
      "step": 2409
    },
    {
      "epoch": 1.115768376420876,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00023351576994434136,
      "loss": 0.6144,
      "step": 2410
    },
    {
      "epoch": 1.1162314499265595,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023348794063079775,
      "loss": 0.6,
      "step": 2411
    },
    {
      "epoch": 1.116694523432243,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00023346011131725417,
      "loss": 0.6162,
      "step": 2412
    },
    {
      "epoch": 1.1171575969379264,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00023343228200371056,
      "loss": 0.6553,
      "step": 2413
    },
    {
      "epoch": 1.11762067044361,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023340445269016697,
      "loss": 0.7232,
      "step": 2414
    },
    {
      "epoch": 1.1180837439492934,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023337662337662336,
      "loss": 0.6941,
      "step": 2415
    },
    {
      "epoch": 1.1185468174549769,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00023334879406307978,
      "loss": 0.6176,
      "step": 2416
    },
    {
      "epoch": 1.1190098909606605,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00023332096474953617,
      "loss": 0.6663,
      "step": 2417
    },
    {
      "epoch": 1.119472964466344,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023329313543599258,
      "loss": 0.6558,
      "step": 2418
    },
    {
      "epoch": 1.1199360379720276,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023326530612244895,
      "loss": 0.7055,
      "step": 2419
    },
    {
      "epoch": 1.120399111477711,
      "grad_norm": 0.5,
      "learning_rate": 0.00023323747680890534,
      "loss": 0.6332,
      "step": 2420
    },
    {
      "epoch": 1.1208621849833944,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00023320964749536175,
      "loss": 0.6213,
      "step": 2421
    },
    {
      "epoch": 1.121325258489078,
      "grad_norm": 0.765625,
      "learning_rate": 0.00023318181818181814,
      "loss": 0.7209,
      "step": 2422
    },
    {
      "epoch": 1.1217883319947615,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00023315398886827456,
      "loss": 0.54,
      "step": 2423
    },
    {
      "epoch": 1.1222514055004449,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00023312615955473095,
      "loss": 0.644,
      "step": 2424
    },
    {
      "epoch": 1.1227144790061285,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00023309833024118737,
      "loss": 0.6967,
      "step": 2425
    },
    {
      "epoch": 1.123177552511812,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023307050092764376,
      "loss": 0.6448,
      "step": 2426
    },
    {
      "epoch": 1.1236406260174956,
      "grad_norm": 0.671875,
      "learning_rate": 0.00023304267161410017,
      "loss": 0.6892,
      "step": 2427
    },
    {
      "epoch": 1.124103699523179,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00023301484230055656,
      "loss": 0.5988,
      "step": 2428
    },
    {
      "epoch": 1.1245667730288624,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00023298701298701298,
      "loss": 0.6368,
      "step": 2429
    },
    {
      "epoch": 1.125029846534546,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00023295918367346937,
      "loss": 0.627,
      "step": 2430
    },
    {
      "epoch": 1.1254929200402295,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00023293135435992578,
      "loss": 0.5887,
      "step": 2431
    },
    {
      "epoch": 1.125955993545913,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00023290352504638217,
      "loss": 0.6496,
      "step": 2432
    },
    {
      "epoch": 1.1264190670515966,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002328756957328386,
      "loss": 0.6713,
      "step": 2433
    },
    {
      "epoch": 1.12688214055728,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00023284786641929498,
      "loss": 0.6265,
      "step": 2434
    },
    {
      "epoch": 1.1273452140629636,
      "grad_norm": 0.5,
      "learning_rate": 0.00023282003710575134,
      "loss": 0.6363,
      "step": 2435
    },
    {
      "epoch": 1.127808287568647,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002327922077922078,
      "loss": 0.6265,
      "step": 2436
    },
    {
      "epoch": 1.1282713610743305,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00023276437847866415,
      "loss": 0.6999,
      "step": 2437
    },
    {
      "epoch": 1.128734434580014,
      "grad_norm": 0.53125,
      "learning_rate": 0.00023273654916512057,
      "loss": 0.6524,
      "step": 2438
    },
    {
      "epoch": 1.1291975080856975,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00023270871985157696,
      "loss": 0.6828,
      "step": 2439
    },
    {
      "epoch": 1.1296605815913812,
      "grad_norm": 0.671875,
      "learning_rate": 0.00023268089053803337,
      "loss": 0.7518,
      "step": 2440
    },
    {
      "epoch": 1.1301236550970646,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00023265306122448976,
      "loss": 0.6956,
      "step": 2441
    },
    {
      "epoch": 1.130586728602748,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00023262523191094618,
      "loss": 0.7565,
      "step": 2442
    },
    {
      "epoch": 1.1310498021084316,
      "grad_norm": 0.65625,
      "learning_rate": 0.00023259740259740257,
      "loss": 0.658,
      "step": 2443
    },
    {
      "epoch": 1.131512875614115,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00023256957328385899,
      "loss": 0.6077,
      "step": 2444
    },
    {
      "epoch": 1.1319759491197985,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023254174397031537,
      "loss": 0.6098,
      "step": 2445
    },
    {
      "epoch": 1.1324390226254821,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002325139146567718,
      "loss": 0.5681,
      "step": 2446
    },
    {
      "epoch": 1.1329020961311655,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00023248608534322818,
      "loss": 0.5889,
      "step": 2447
    },
    {
      "epoch": 1.133365169636849,
      "grad_norm": 0.484375,
      "learning_rate": 0.00023245825602968457,
      "loss": 0.5666,
      "step": 2448
    },
    {
      "epoch": 1.1338282431425326,
      "grad_norm": 0.734375,
      "learning_rate": 0.000232430426716141,
      "loss": 0.7673,
      "step": 2449
    },
    {
      "epoch": 1.134291316648216,
      "grad_norm": 0.5625,
      "learning_rate": 0.00023240259740259738,
      "loss": 0.6189,
      "step": 2450
    },
    {
      "epoch": 1.1347543901538997,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002323747680890538,
      "loss": 0.6403,
      "step": 2451
    },
    {
      "epoch": 1.135217463659583,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00023234693877551018,
      "loss": 0.6814,
      "step": 2452
    },
    {
      "epoch": 1.1356805371652665,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0002323191094619666,
      "loss": 0.6328,
      "step": 2453
    },
    {
      "epoch": 1.1361436106709502,
      "grad_norm": 0.796875,
      "learning_rate": 0.00023229128014842296,
      "loss": 0.6886,
      "step": 2454
    },
    {
      "epoch": 1.1366066841766336,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002322634508348794,
      "loss": 0.6378,
      "step": 2455
    },
    {
      "epoch": 1.1370697576823172,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023223562152133577,
      "loss": 0.6182,
      "step": 2456
    },
    {
      "epoch": 1.1375328311880006,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00023220779220779219,
      "loss": 0.5982,
      "step": 2457
    },
    {
      "epoch": 1.137995904693684,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023217996289424858,
      "loss": 0.618,
      "step": 2458
    },
    {
      "epoch": 1.1384589781993677,
      "grad_norm": 0.474609375,
      "learning_rate": 0.000232152133580705,
      "loss": 0.6125,
      "step": 2459
    },
    {
      "epoch": 1.1389220517050511,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00023212430426716138,
      "loss": 0.8044,
      "step": 2460
    },
    {
      "epoch": 1.1393851252107345,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002320964749536178,
      "loss": 0.6478,
      "step": 2461
    },
    {
      "epoch": 1.1398481987164182,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002320686456400742,
      "loss": 0.6084,
      "step": 2462
    },
    {
      "epoch": 1.1403112722221016,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00023204081632653058,
      "loss": 0.5731,
      "step": 2463
    },
    {
      "epoch": 1.140774345727785,
      "grad_norm": 0.546875,
      "learning_rate": 0.000232012987012987,
      "loss": 0.6401,
      "step": 2464
    },
    {
      "epoch": 1.1412374192334687,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023198515769944338,
      "loss": 0.6496,
      "step": 2465
    },
    {
      "epoch": 1.141700492739152,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002319573283858998,
      "loss": 0.6104,
      "step": 2466
    },
    {
      "epoch": 1.1421635662448357,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0002319294990723562,
      "loss": 0.6973,
      "step": 2467
    },
    {
      "epoch": 1.1426266397505191,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002319016697588126,
      "loss": 0.6718,
      "step": 2468
    },
    {
      "epoch": 1.1430897132562026,
      "grad_norm": 0.703125,
      "learning_rate": 0.000231873840445269,
      "loss": 0.707,
      "step": 2469
    },
    {
      "epoch": 1.1435527867618862,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002318460111317254,
      "loss": 0.5533,
      "step": 2470
    },
    {
      "epoch": 1.1440158602675696,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002318181818181818,
      "loss": 0.7254,
      "step": 2471
    },
    {
      "epoch": 1.1444789337732533,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00023179035250463822,
      "loss": 0.6361,
      "step": 2472
    },
    {
      "epoch": 1.1449420072789367,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002317625231910946,
      "loss": 0.7218,
      "step": 2473
    },
    {
      "epoch": 1.14540508078462,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00023173469387755103,
      "loss": 0.644,
      "step": 2474
    },
    {
      "epoch": 1.1458681542903038,
      "grad_norm": 1.5390625,
      "learning_rate": 0.0002317068645640074,
      "loss": 0.6579,
      "step": 2475
    },
    {
      "epoch": 1.1463312277959872,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002316790352504638,
      "loss": 0.7016,
      "step": 2476
    },
    {
      "epoch": 1.1467943013016706,
      "grad_norm": 0.828125,
      "learning_rate": 0.0002316512059369202,
      "loss": 0.6591,
      "step": 2477
    },
    {
      "epoch": 1.1472573748073542,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00023162337662337658,
      "loss": 0.6506,
      "step": 2478
    },
    {
      "epoch": 1.1477204483130377,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000231595547309833,
      "loss": 0.8195,
      "step": 2479
    },
    {
      "epoch": 1.148183521818721,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002315677179962894,
      "loss": 0.6731,
      "step": 2480
    },
    {
      "epoch": 1.1486465953244047,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002315398886827458,
      "loss": 0.613,
      "step": 2481
    },
    {
      "epoch": 1.1491096688300881,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002315120593692022,
      "loss": 0.6407,
      "step": 2482
    },
    {
      "epoch": 1.1495727423357718,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023148423005565861,
      "loss": 0.6452,
      "step": 2483
    },
    {
      "epoch": 1.1500358158414552,
      "grad_norm": 0.52734375,
      "learning_rate": 0.000231456400742115,
      "loss": 0.5634,
      "step": 2484
    },
    {
      "epoch": 1.1504988893471386,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023142857142857142,
      "loss": 0.7197,
      "step": 2485
    },
    {
      "epoch": 1.1509619628528223,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002314007421150278,
      "loss": 0.5996,
      "step": 2486
    },
    {
      "epoch": 1.1514250363585057,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023137291280148423,
      "loss": 0.7292,
      "step": 2487
    },
    {
      "epoch": 1.1518881098641893,
      "grad_norm": 0.546875,
      "learning_rate": 0.00023134508348794062,
      "loss": 0.6045,
      "step": 2488
    },
    {
      "epoch": 1.1523511833698727,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00023131725417439703,
      "loss": 0.5981,
      "step": 2489
    },
    {
      "epoch": 1.1528142568755562,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023128942486085342,
      "loss": 0.6376,
      "step": 2490
    },
    {
      "epoch": 1.1532773303812398,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00023126159554730978,
      "loss": 0.6948,
      "step": 2491
    },
    {
      "epoch": 1.1537404038869232,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00023123376623376623,
      "loss": 0.6546,
      "step": 2492
    },
    {
      "epoch": 1.1542034773926066,
      "grad_norm": 15.375,
      "learning_rate": 0.0002312059369202226,
      "loss": 0.9813,
      "step": 2493
    },
    {
      "epoch": 1.1546665508982903,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000231178107606679,
      "loss": 0.6319,
      "step": 2494
    },
    {
      "epoch": 1.1551296244039737,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002311502782931354,
      "loss": 0.5795,
      "step": 2495
    },
    {
      "epoch": 1.1555926979096571,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00023112244897959181,
      "loss": 0.6941,
      "step": 2496
    },
    {
      "epoch": 1.1560557714153408,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002310946196660482,
      "loss": 0.656,
      "step": 2497
    },
    {
      "epoch": 1.1565188449210242,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00023106679035250462,
      "loss": 0.7502,
      "step": 2498
    },
    {
      "epoch": 1.1569819184267078,
      "grad_norm": 0.59375,
      "learning_rate": 0.000231038961038961,
      "loss": 0.7161,
      "step": 2499
    },
    {
      "epoch": 1.1574449919323913,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00023101113172541743,
      "loss": 0.6185,
      "step": 2500
    },
    {
      "epoch": 1.157908065438075,
      "grad_norm": 0.65625,
      "learning_rate": 0.00023098330241187382,
      "loss": 0.6358,
      "step": 2501
    },
    {
      "epoch": 1.1583711389437583,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00023095547309833023,
      "loss": 0.8049,
      "step": 2502
    },
    {
      "epoch": 1.1588342124494417,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023092764378478662,
      "loss": 0.6943,
      "step": 2503
    },
    {
      "epoch": 1.1592972859551254,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023089981447124304,
      "loss": 0.6642,
      "step": 2504
    },
    {
      "epoch": 1.1597603594608088,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00023087198515769943,
      "loss": 0.7772,
      "step": 2505
    },
    {
      "epoch": 1.1602234329664922,
      "grad_norm": 0.625,
      "learning_rate": 0.00023084415584415582,
      "loss": 0.7358,
      "step": 2506
    },
    {
      "epoch": 1.1606865064721759,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00023081632653061224,
      "loss": 0.6085,
      "step": 2507
    },
    {
      "epoch": 1.1611495799778593,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00023078849721706862,
      "loss": 0.6101,
      "step": 2508
    },
    {
      "epoch": 1.1616126534835427,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00023076066790352504,
      "loss": 0.6437,
      "step": 2509
    },
    {
      "epoch": 1.1620757269892263,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002307328385899814,
      "loss": 0.6439,
      "step": 2510
    },
    {
      "epoch": 1.1625388004949098,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00023070500927643785,
      "loss": 0.5868,
      "step": 2511
    },
    {
      "epoch": 1.1630018740005934,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0002306771799628942,
      "loss": 0.7183,
      "step": 2512
    },
    {
      "epoch": 1.1634649475062768,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00023064935064935063,
      "loss": 0.627,
      "step": 2513
    },
    {
      "epoch": 1.1639280210119602,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00023062152133580702,
      "loss": 0.5675,
      "step": 2514
    },
    {
      "epoch": 1.1643910945176439,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00023059369202226343,
      "loss": 0.7182,
      "step": 2515
    },
    {
      "epoch": 1.1648541680233273,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023056586270871982,
      "loss": 0.6807,
      "step": 2516
    },
    {
      "epoch": 1.165317241529011,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00023053803339517624,
      "loss": 0.6597,
      "step": 2517
    },
    {
      "epoch": 1.1657803150346944,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00023051020408163263,
      "loss": 0.7486,
      "step": 2518
    },
    {
      "epoch": 1.1662433885403778,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00023048237476808902,
      "loss": 0.6906,
      "step": 2519
    },
    {
      "epoch": 1.1667064620460614,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00023045454545454544,
      "loss": 0.645,
      "step": 2520
    },
    {
      "epoch": 1.1671695355517449,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023042671614100183,
      "loss": 0.6568,
      "step": 2521
    },
    {
      "epoch": 1.1676326090574283,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00023039888682745824,
      "loss": 0.6348,
      "step": 2522
    },
    {
      "epoch": 1.168095682563112,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00023037105751391463,
      "loss": 0.5922,
      "step": 2523
    },
    {
      "epoch": 1.1685587560687953,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023034322820037105,
      "loss": 0.6452,
      "step": 2524
    },
    {
      "epoch": 1.1690218295744788,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00023031539888682744,
      "loss": 0.599,
      "step": 2525
    },
    {
      "epoch": 1.1694849030801624,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00023028756957328385,
      "loss": 0.5834,
      "step": 2526
    },
    {
      "epoch": 1.1699479765858458,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00023025974025974024,
      "loss": 0.655,
      "step": 2527
    },
    {
      "epoch": 1.1704110500915295,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00023023191094619666,
      "loss": 0.5974,
      "step": 2528
    },
    {
      "epoch": 1.1708741235972129,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00023020408163265302,
      "loss": 0.7616,
      "step": 2529
    },
    {
      "epoch": 1.1713371971028963,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023017625231910947,
      "loss": 0.6348,
      "step": 2530
    },
    {
      "epoch": 1.17180027060858,
      "grad_norm": 0.578125,
      "learning_rate": 0.00023014842300556583,
      "loss": 0.6585,
      "step": 2531
    },
    {
      "epoch": 1.1722633441142634,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00023012059369202225,
      "loss": 0.6408,
      "step": 2532
    },
    {
      "epoch": 1.172726417619947,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00023009276437847864,
      "loss": 0.6057,
      "step": 2533
    },
    {
      "epoch": 1.1731894911256304,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00023006493506493503,
      "loss": 0.6826,
      "step": 2534
    },
    {
      "epoch": 1.1736525646313138,
      "grad_norm": 0.46875,
      "learning_rate": 0.00023003710575139144,
      "loss": 0.6534,
      "step": 2535
    },
    {
      "epoch": 1.1741156381369975,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00023000927643784783,
      "loss": 0.5844,
      "step": 2536
    },
    {
      "epoch": 1.174578711642681,
      "grad_norm": 0.578125,
      "learning_rate": 0.00022998144712430425,
      "loss": 0.6904,
      "step": 2537
    },
    {
      "epoch": 1.1750417851483643,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022995361781076064,
      "loss": 0.6452,
      "step": 2538
    },
    {
      "epoch": 1.175504858654048,
      "grad_norm": 0.640625,
      "learning_rate": 0.00022992578849721705,
      "loss": 0.6982,
      "step": 2539
    },
    {
      "epoch": 1.1759679321597314,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022989795918367344,
      "loss": 0.674,
      "step": 2540
    },
    {
      "epoch": 1.1764310056654148,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00022987012987012986,
      "loss": 0.5834,
      "step": 2541
    },
    {
      "epoch": 1.1768940791710985,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00022984230055658625,
      "loss": 0.5946,
      "step": 2542
    },
    {
      "epoch": 1.1773571526767819,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00022981447124304267,
      "loss": 0.7032,
      "step": 2543
    },
    {
      "epoch": 1.1778202261824655,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00022978664192949906,
      "loss": 0.7176,
      "step": 2544
    },
    {
      "epoch": 1.178283299688149,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022975881261595547,
      "loss": 0.6387,
      "step": 2545
    },
    {
      "epoch": 1.1787463731938324,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00022973098330241186,
      "loss": 0.6866,
      "step": 2546
    },
    {
      "epoch": 1.179209446699516,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00022970315398886823,
      "loss": 0.7346,
      "step": 2547
    },
    {
      "epoch": 1.1796725202051994,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022967532467532467,
      "loss": 0.5822,
      "step": 2548
    },
    {
      "epoch": 1.180135593710883,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022964749536178103,
      "loss": 0.6699,
      "step": 2549
    },
    {
      "epoch": 1.1805986672165665,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00022961966604823745,
      "loss": 0.7442,
      "step": 2550
    },
    {
      "epoch": 1.18106174072225,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022959183673469384,
      "loss": 0.5844,
      "step": 2551
    },
    {
      "epoch": 1.1815248142279335,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022956400742115026,
      "loss": 0.5882,
      "step": 2552
    },
    {
      "epoch": 1.181987887733617,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00022953617810760664,
      "loss": 0.6611,
      "step": 2553
    },
    {
      "epoch": 1.1824509612393004,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022950834879406306,
      "loss": 0.6073,
      "step": 2554
    },
    {
      "epoch": 1.182914034744984,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00022948051948051945,
      "loss": 0.6867,
      "step": 2555
    },
    {
      "epoch": 1.1833771082506674,
      "grad_norm": 0.578125,
      "learning_rate": 0.00022945269016697587,
      "loss": 0.6968,
      "step": 2556
    },
    {
      "epoch": 1.1838401817563509,
      "grad_norm": 0.671875,
      "learning_rate": 0.00022942486085343226,
      "loss": 0.6489,
      "step": 2557
    },
    {
      "epoch": 1.1843032552620345,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022939703153988867,
      "loss": 0.6301,
      "step": 2558
    },
    {
      "epoch": 1.184766328767718,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022936920222634506,
      "loss": 0.628,
      "step": 2559
    },
    {
      "epoch": 1.1852294022734016,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022934137291280148,
      "loss": 0.6554,
      "step": 2560
    },
    {
      "epoch": 1.185692475779085,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00022931354359925787,
      "loss": 0.6954,
      "step": 2561
    },
    {
      "epoch": 1.1861555492847684,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00022928571428571426,
      "loss": 0.6111,
      "step": 2562
    },
    {
      "epoch": 1.186618622790452,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00022925788497217068,
      "loss": 0.6694,
      "step": 2563
    },
    {
      "epoch": 1.1870816962961355,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022923005565862707,
      "loss": 0.7073,
      "step": 2564
    },
    {
      "epoch": 1.1875447698018191,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00022920222634508348,
      "loss": 0.6384,
      "step": 2565
    },
    {
      "epoch": 1.1880078433075025,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00022917439703153985,
      "loss": 0.713,
      "step": 2566
    },
    {
      "epoch": 1.188470916813186,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002291465677179963,
      "loss": 0.5974,
      "step": 2567
    },
    {
      "epoch": 1.1889339903188696,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00022911873840445265,
      "loss": 0.6657,
      "step": 2568
    },
    {
      "epoch": 1.189397063824553,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022909090909090907,
      "loss": 0.7183,
      "step": 2569
    },
    {
      "epoch": 1.1898601373302364,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022906307977736546,
      "loss": 0.7305,
      "step": 2570
    },
    {
      "epoch": 1.19032321083592,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022903525046382187,
      "loss": 0.6567,
      "step": 2571
    },
    {
      "epoch": 1.1907862843416035,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022900742115027826,
      "loss": 0.682,
      "step": 2572
    },
    {
      "epoch": 1.191249357847287,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022897959183673468,
      "loss": 0.6281,
      "step": 2573
    },
    {
      "epoch": 1.1917124313529706,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022895176252319107,
      "loss": 0.5935,
      "step": 2574
    },
    {
      "epoch": 1.192175504858654,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00022892393320964746,
      "loss": 0.6428,
      "step": 2575
    },
    {
      "epoch": 1.1926385783643376,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022889610389610388,
      "loss": 0.6381,
      "step": 2576
    },
    {
      "epoch": 1.193101651870021,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00022886827458256027,
      "loss": 0.676,
      "step": 2577
    },
    {
      "epoch": 1.1935647253757047,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00022884044526901668,
      "loss": 0.6448,
      "step": 2578
    },
    {
      "epoch": 1.194027798881388,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022881261595547307,
      "loss": 0.6685,
      "step": 2579
    },
    {
      "epoch": 1.1944908723870715,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002287847866419295,
      "loss": 0.6756,
      "step": 2580
    },
    {
      "epoch": 1.1949539458927552,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00022875695732838588,
      "loss": 0.5887,
      "step": 2581
    },
    {
      "epoch": 1.1954170193984386,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0002287291280148423,
      "loss": 0.7392,
      "step": 2582
    },
    {
      "epoch": 1.195880092904122,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022870129870129869,
      "loss": 0.5797,
      "step": 2583
    },
    {
      "epoch": 1.1963431664098056,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0002286734693877551,
      "loss": 0.5507,
      "step": 2584
    },
    {
      "epoch": 1.196806239915489,
      "grad_norm": 0.578125,
      "learning_rate": 0.00022864564007421146,
      "loss": 0.6389,
      "step": 2585
    },
    {
      "epoch": 1.1972693134211725,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002286178107606679,
      "loss": 0.6478,
      "step": 2586
    },
    {
      "epoch": 1.1977323869268561,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00022858998144712427,
      "loss": 0.7205,
      "step": 2587
    },
    {
      "epoch": 1.1981954604325395,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002285621521335807,
      "loss": 0.6772,
      "step": 2588
    },
    {
      "epoch": 1.1986585339382232,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022853432282003708,
      "loss": 0.6718,
      "step": 2589
    },
    {
      "epoch": 1.1991216074439066,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00022850649350649347,
      "loss": 0.7237,
      "step": 2590
    },
    {
      "epoch": 1.19958468094959,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022847866419294988,
      "loss": 0.6445,
      "step": 2591
    },
    {
      "epoch": 1.2000477544552737,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022845083487940627,
      "loss": 0.6532,
      "step": 2592
    },
    {
      "epoch": 1.200510827960957,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0002284230055658627,
      "loss": 0.5978,
      "step": 2593
    },
    {
      "epoch": 1.2009739014666407,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022839517625231908,
      "loss": 0.7231,
      "step": 2594
    },
    {
      "epoch": 1.2014369749723242,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0002283673469387755,
      "loss": 0.6129,
      "step": 2595
    },
    {
      "epoch": 1.2019000484780076,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00022833951762523189,
      "loss": 0.6586,
      "step": 2596
    },
    {
      "epoch": 1.2023631219836912,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002283116883116883,
      "loss": 0.5974,
      "step": 2597
    },
    {
      "epoch": 1.2028261954893746,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002282838589981447,
      "loss": 0.6397,
      "step": 2598
    },
    {
      "epoch": 1.203289268995058,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002282560296846011,
      "loss": 0.6075,
      "step": 2599
    },
    {
      "epoch": 1.2037523425007417,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002282282003710575,
      "loss": 0.7174,
      "step": 2600
    },
    {
      "epoch": 1.2042154160064251,
      "grad_norm": 0.765625,
      "learning_rate": 0.00022820037105751392,
      "loss": 0.7471,
      "step": 2601
    },
    {
      "epoch": 1.2046784895121085,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0002281725417439703,
      "loss": 0.6347,
      "step": 2602
    },
    {
      "epoch": 1.2051415630177922,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022814471243042672,
      "loss": 0.6232,
      "step": 2603
    },
    {
      "epoch": 1.2056046365234756,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002281168831168831,
      "loss": 0.636,
      "step": 2604
    },
    {
      "epoch": 1.2060677100291592,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00022808905380333947,
      "loss": 0.705,
      "step": 2605
    },
    {
      "epoch": 1.2065307835348427,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002280612244897959,
      "loss": 0.7151,
      "step": 2606
    },
    {
      "epoch": 1.206993857040526,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00022803339517625228,
      "loss": 0.5549,
      "step": 2607
    },
    {
      "epoch": 1.2074569305462097,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002280055658627087,
      "loss": 0.595,
      "step": 2608
    },
    {
      "epoch": 1.2079200040518931,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022797773654916509,
      "loss": 0.6232,
      "step": 2609
    },
    {
      "epoch": 1.2083830775575768,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002279499072356215,
      "loss": 0.6518,
      "step": 2610
    },
    {
      "epoch": 1.2088461510632602,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002279220779220779,
      "loss": 0.6996,
      "step": 2611
    },
    {
      "epoch": 1.2093092245689436,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002278942486085343,
      "loss": 0.639,
      "step": 2612
    },
    {
      "epoch": 1.2097722980746273,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002278664192949907,
      "loss": 0.5817,
      "step": 2613
    },
    {
      "epoch": 1.2102353715803107,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00022783858998144712,
      "loss": 0.5973,
      "step": 2614
    },
    {
      "epoch": 1.2106984450859941,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002278107606679035,
      "loss": 0.6653,
      "step": 2615
    },
    {
      "epoch": 1.2111615185916778,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022778293135435992,
      "loss": 0.6439,
      "step": 2616
    },
    {
      "epoch": 1.2116245920973612,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002277551020408163,
      "loss": 0.6019,
      "step": 2617
    },
    {
      "epoch": 1.2120876656030446,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002277272727272727,
      "loss": 0.646,
      "step": 2618
    },
    {
      "epoch": 1.2125507391087282,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022769944341372912,
      "loss": 0.648,
      "step": 2619
    },
    {
      "epoch": 1.2130138126144117,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002276716141001855,
      "loss": 0.6583,
      "step": 2620
    },
    {
      "epoch": 1.2134768861200953,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00022764378478664192,
      "loss": 0.6288,
      "step": 2621
    },
    {
      "epoch": 1.2139399596257787,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002276159554730983,
      "loss": 0.7446,
      "step": 2622
    },
    {
      "epoch": 1.2144030331314621,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022758812615955473,
      "loss": 0.6923,
      "step": 2623
    },
    {
      "epoch": 1.2148661066371458,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002275602968460111,
      "loss": 0.6907,
      "step": 2624
    },
    {
      "epoch": 1.2153291801428292,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002275324675324675,
      "loss": 0.6031,
      "step": 2625
    },
    {
      "epoch": 1.2157922536485128,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002275046382189239,
      "loss": 0.6027,
      "step": 2626
    },
    {
      "epoch": 1.2162553271541963,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022747680890538032,
      "loss": 0.6257,
      "step": 2627
    },
    {
      "epoch": 1.2167184006598797,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002274489795918367,
      "loss": 0.6486,
      "step": 2628
    },
    {
      "epoch": 1.2171814741655633,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00022742115027829312,
      "loss": 0.5911,
      "step": 2629
    },
    {
      "epoch": 1.2176445476712467,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002273933209647495,
      "loss": 0.6904,
      "step": 2630
    },
    {
      "epoch": 1.2181076211769302,
      "grad_norm": 0.75,
      "learning_rate": 0.00022736549165120593,
      "loss": 0.755,
      "step": 2631
    },
    {
      "epoch": 1.2185706946826138,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00022733766233766232,
      "loss": 0.5628,
      "step": 2632
    },
    {
      "epoch": 1.2190337681882972,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0002273098330241187,
      "loss": 0.5926,
      "step": 2633
    },
    {
      "epoch": 1.2194968416939806,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00022728200371057512,
      "loss": 0.6783,
      "step": 2634
    },
    {
      "epoch": 1.2199599151996643,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022725417439703151,
      "loss": 0.6026,
      "step": 2635
    },
    {
      "epoch": 1.2204229887053477,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00022722634508348793,
      "loss": 0.6773,
      "step": 2636
    },
    {
      "epoch": 1.2208860622110314,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022719851576994432,
      "loss": 0.6808,
      "step": 2637
    },
    {
      "epoch": 1.2213491357167148,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022717068645640074,
      "loss": 0.5856,
      "step": 2638
    },
    {
      "epoch": 1.2218122092223982,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022714285714285713,
      "loss": 0.6147,
      "step": 2639
    },
    {
      "epoch": 1.2222752827280818,
      "grad_norm": 0.640625,
      "learning_rate": 0.00022711502782931354,
      "loss": 0.7292,
      "step": 2640
    },
    {
      "epoch": 1.2227383562337653,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002270871985157699,
      "loss": 0.6691,
      "step": 2641
    },
    {
      "epoch": 1.223201429739449,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00022705936920222635,
      "loss": 0.6889,
      "step": 2642
    },
    {
      "epoch": 1.2236645032451323,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002270315398886827,
      "loss": 0.6019,
      "step": 2643
    },
    {
      "epoch": 1.2241275767508157,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022700371057513913,
      "loss": 0.6951,
      "step": 2644
    },
    {
      "epoch": 1.2245906502564994,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022697588126159552,
      "loss": 0.6906,
      "step": 2645
    },
    {
      "epoch": 1.2250537237621828,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002269480519480519,
      "loss": 0.6852,
      "step": 2646
    },
    {
      "epoch": 1.2255167972678662,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022692022263450832,
      "loss": 0.634,
      "step": 2647
    },
    {
      "epoch": 1.2259798707735499,
      "grad_norm": 0.609375,
      "learning_rate": 0.00022689239332096471,
      "loss": 0.7211,
      "step": 2648
    },
    {
      "epoch": 1.2264429442792333,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022686456400742113,
      "loss": 0.6617,
      "step": 2649
    },
    {
      "epoch": 1.226906017784917,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00022683673469387752,
      "loss": 0.6984,
      "step": 2650
    },
    {
      "epoch": 1.2273690912906003,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00022680890538033394,
      "loss": 0.6108,
      "step": 2651
    },
    {
      "epoch": 1.2278321647962838,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022678107606679033,
      "loss": 0.6598,
      "step": 2652
    },
    {
      "epoch": 1.2282952383019674,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00022675324675324674,
      "loss": 0.6669,
      "step": 2653
    },
    {
      "epoch": 1.2287583118076508,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00022672541743970313,
      "loss": 0.7304,
      "step": 2654
    },
    {
      "epoch": 1.2292213853133345,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00022669758812615955,
      "loss": 0.6176,
      "step": 2655
    },
    {
      "epoch": 1.229684458819018,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00022666975881261594,
      "loss": 0.691,
      "step": 2656
    },
    {
      "epoch": 1.2301475323247013,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022664192949907236,
      "loss": 0.6787,
      "step": 2657
    },
    {
      "epoch": 1.230610605830385,
      "grad_norm": 0.625,
      "learning_rate": 0.00022661410018552875,
      "loss": 0.749,
      "step": 2658
    },
    {
      "epoch": 1.2310736793360684,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022658627087198516,
      "loss": 0.6121,
      "step": 2659
    },
    {
      "epoch": 1.2315367528417518,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022655844155844155,
      "loss": 0.6231,
      "step": 2660
    },
    {
      "epoch": 1.2319998263474354,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022653061224489791,
      "loss": 0.6464,
      "step": 2661
    },
    {
      "epoch": 1.2324628998531189,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022650278293135433,
      "loss": 0.5681,
      "step": 2662
    },
    {
      "epoch": 1.2329259733588023,
      "grad_norm": 0.578125,
      "learning_rate": 0.00022647495361781072,
      "loss": 0.631,
      "step": 2663
    },
    {
      "epoch": 1.233389046864486,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022644712430426714,
      "loss": 0.6199,
      "step": 2664
    },
    {
      "epoch": 1.2338521203701693,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022641929499072353,
      "loss": 0.6034,
      "step": 2665
    },
    {
      "epoch": 1.234315193875853,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022639146567717994,
      "loss": 0.6204,
      "step": 2666
    },
    {
      "epoch": 1.2347782673815364,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022636363636363633,
      "loss": 0.5725,
      "step": 2667
    },
    {
      "epoch": 1.2352413408872198,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022633580705009275,
      "loss": 0.6173,
      "step": 2668
    },
    {
      "epoch": 1.2357044143929035,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022630797773654914,
      "loss": 0.7144,
      "step": 2669
    },
    {
      "epoch": 1.2361674878985869,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00022628014842300556,
      "loss": 0.6678,
      "step": 2670
    },
    {
      "epoch": 1.2366305614042705,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022625231910946195,
      "loss": 0.6638,
      "step": 2671
    },
    {
      "epoch": 1.237093634909954,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022622448979591836,
      "loss": 0.6427,
      "step": 2672
    },
    {
      "epoch": 1.2375567084156374,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022619666048237475,
      "loss": 0.7413,
      "step": 2673
    },
    {
      "epoch": 1.238019781921321,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00022616883116883114,
      "loss": 0.6945,
      "step": 2674
    },
    {
      "epoch": 1.2384828554270044,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022614100185528756,
      "loss": 0.6079,
      "step": 2675
    },
    {
      "epoch": 1.2389459289326878,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00022611317254174395,
      "loss": 0.6051,
      "step": 2676
    },
    {
      "epoch": 1.2394090024383715,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00022608534322820037,
      "loss": 0.573,
      "step": 2677
    },
    {
      "epoch": 1.239872075944055,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00022605751391465673,
      "loss": 0.6236,
      "step": 2678
    },
    {
      "epoch": 1.2403351494497383,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022602968460111317,
      "loss": 0.6511,
      "step": 2679
    },
    {
      "epoch": 1.240798222955422,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00022600185528756953,
      "loss": 0.6544,
      "step": 2680
    },
    {
      "epoch": 1.2412612964611054,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00022597402597402595,
      "loss": 0.6176,
      "step": 2681
    },
    {
      "epoch": 1.241724369966789,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00022594619666048234,
      "loss": 0.663,
      "step": 2682
    },
    {
      "epoch": 1.2421874434724725,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00022591836734693876,
      "loss": 0.5964,
      "step": 2683
    },
    {
      "epoch": 1.2426505169781559,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00022589053803339515,
      "loss": 0.6743,
      "step": 2684
    },
    {
      "epoch": 1.2431135904838395,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022586270871985156,
      "loss": 0.5968,
      "step": 2685
    },
    {
      "epoch": 1.243576663989523,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022583487940630795,
      "loss": 0.6045,
      "step": 2686
    },
    {
      "epoch": 1.2440397374952066,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00022580705009276437,
      "loss": 0.5496,
      "step": 2687
    },
    {
      "epoch": 1.24450281100089,
      "grad_norm": 0.546875,
      "learning_rate": 0.00022577922077922076,
      "loss": 0.6562,
      "step": 2688
    },
    {
      "epoch": 1.2449658845065734,
      "grad_norm": 0.59375,
      "learning_rate": 0.00022575139146567715,
      "loss": 0.6206,
      "step": 2689
    },
    {
      "epoch": 1.245428958012257,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022572356215213357,
      "loss": 0.6616,
      "step": 2690
    },
    {
      "epoch": 1.2458920315179405,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00022569573283858996,
      "loss": 0.7247,
      "step": 2691
    },
    {
      "epoch": 1.246355105023624,
      "grad_norm": 0.71875,
      "learning_rate": 0.00022566790352504637,
      "loss": 0.8028,
      "step": 2692
    },
    {
      "epoch": 1.2468181785293075,
      "grad_norm": 0.65625,
      "learning_rate": 0.00022564007421150276,
      "loss": 0.7288,
      "step": 2693
    },
    {
      "epoch": 1.247281252034991,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00022561224489795918,
      "loss": 0.6219,
      "step": 2694
    },
    {
      "epoch": 1.2477443255406744,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00022558441558441557,
      "loss": 0.7373,
      "step": 2695
    },
    {
      "epoch": 1.248207399046358,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022555658627087198,
      "loss": 0.6238,
      "step": 2696
    },
    {
      "epoch": 1.2486704725520414,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022552875695732835,
      "loss": 0.6451,
      "step": 2697
    },
    {
      "epoch": 1.249133546057725,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002255009276437848,
      "loss": 0.568,
      "step": 2698
    },
    {
      "epoch": 1.2495966195634085,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00022547309833024115,
      "loss": 0.6723,
      "step": 2699
    },
    {
      "epoch": 1.2500596930690921,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00022544526901669757,
      "loss": 0.7191,
      "step": 2700
    },
    {
      "epoch": 1.2505227665747756,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00022541743970315396,
      "loss": 0.6248,
      "step": 2701
    },
    {
      "epoch": 1.250985840080459,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00022538961038961035,
      "loss": 0.6892,
      "step": 2702
    },
    {
      "epoch": 1.2514489135861426,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022536178107606677,
      "loss": 0.5908,
      "step": 2703
    },
    {
      "epoch": 1.251911987091826,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022533395176252316,
      "loss": 0.6713,
      "step": 2704
    },
    {
      "epoch": 1.2523750605975095,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00022530612244897957,
      "loss": 0.6337,
      "step": 2705
    },
    {
      "epoch": 1.2528381341031931,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00022527829313543596,
      "loss": 0.6657,
      "step": 2706
    },
    {
      "epoch": 1.2533012076088765,
      "grad_norm": 0.578125,
      "learning_rate": 0.00022525046382189238,
      "loss": 0.6787,
      "step": 2707
    },
    {
      "epoch": 1.25376428111456,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022522263450834877,
      "loss": 0.6625,
      "step": 2708
    },
    {
      "epoch": 1.2542273546202436,
      "grad_norm": 0.609375,
      "learning_rate": 0.00022519480519480519,
      "loss": 0.5996,
      "step": 2709
    },
    {
      "epoch": 1.254690428125927,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00022516697588126157,
      "loss": 0.6567,
      "step": 2710
    },
    {
      "epoch": 1.2551535016316104,
      "grad_norm": 0.546875,
      "learning_rate": 0.000225139146567718,
      "loss": 0.624,
      "step": 2711
    },
    {
      "epoch": 1.255616575137294,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022511131725417438,
      "loss": 0.6453,
      "step": 2712
    },
    {
      "epoch": 1.2560796486429775,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002250834879406308,
      "loss": 0.6246,
      "step": 2713
    },
    {
      "epoch": 1.2565427221486611,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002250556586270872,
      "loss": 0.6901,
      "step": 2714
    },
    {
      "epoch": 1.2570057956543446,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002250278293135436,
      "loss": 0.6632,
      "step": 2715
    },
    {
      "epoch": 1.2574688691600282,
      "grad_norm": 0.498046875,
      "learning_rate": 0.000225,
      "loss": 0.6156,
      "step": 2716
    },
    {
      "epoch": 1.2579319426657116,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00022497217068645636,
      "loss": 0.6382,
      "step": 2717
    },
    {
      "epoch": 1.258395016171395,
      "grad_norm": 0.59375,
      "learning_rate": 0.00022494434137291277,
      "loss": 0.6328,
      "step": 2718
    },
    {
      "epoch": 1.2588580896770787,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022491651205936916,
      "loss": 0.6854,
      "step": 2719
    },
    {
      "epoch": 1.259321163182762,
      "grad_norm": 0.734375,
      "learning_rate": 0.00022488868274582558,
      "loss": 0.7862,
      "step": 2720
    },
    {
      "epoch": 1.2597842366884455,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00022486085343228197,
      "loss": 0.7071,
      "step": 2721
    },
    {
      "epoch": 1.2602473101941292,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00022483302411873839,
      "loss": 0.62,
      "step": 2722
    },
    {
      "epoch": 1.2607103836998126,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00022480519480519478,
      "loss": 0.6565,
      "step": 2723
    },
    {
      "epoch": 1.261173457205496,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002247773654916512,
      "loss": 0.6835,
      "step": 2724
    },
    {
      "epoch": 1.2616365307111796,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022474953617810758,
      "loss": 0.6052,
      "step": 2725
    },
    {
      "epoch": 1.262099604216863,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000224721706864564,
      "loss": 0.5635,
      "step": 2726
    },
    {
      "epoch": 1.2625626777225465,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002246938775510204,
      "loss": 0.6195,
      "step": 2727
    },
    {
      "epoch": 1.2630257512282301,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002246660482374768,
      "loss": 0.6047,
      "step": 2728
    },
    {
      "epoch": 1.2634888247339136,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002246382189239332,
      "loss": 0.6368,
      "step": 2729
    },
    {
      "epoch": 1.2639518982395972,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002246103896103896,
      "loss": 0.7677,
      "step": 2730
    },
    {
      "epoch": 1.2644149717452806,
      "grad_norm": 0.609375,
      "learning_rate": 0.000224582560296846,
      "loss": 0.7154,
      "step": 2731
    },
    {
      "epoch": 1.2648780452509643,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002245547309833024,
      "loss": 0.6151,
      "step": 2732
    },
    {
      "epoch": 1.2653411187566477,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002245269016697588,
      "loss": 0.5497,
      "step": 2733
    },
    {
      "epoch": 1.265804192262331,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00022449907235621517,
      "loss": 0.599,
      "step": 2734
    },
    {
      "epoch": 1.2662672657680147,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002244712430426716,
      "loss": 0.617,
      "step": 2735
    },
    {
      "epoch": 1.2667303392736982,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022444341372912798,
      "loss": 0.59,
      "step": 2736
    },
    {
      "epoch": 1.2671934127793816,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002244155844155844,
      "loss": 0.624,
      "step": 2737
    },
    {
      "epoch": 1.2676564862850652,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022438775510204078,
      "loss": 0.62,
      "step": 2738
    },
    {
      "epoch": 1.2681195597907486,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002243599257884972,
      "loss": 0.7334,
      "step": 2739
    },
    {
      "epoch": 1.268582633296432,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002243320964749536,
      "loss": 0.6277,
      "step": 2740
    },
    {
      "epoch": 1.2690457068021157,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022430426716141,
      "loss": 0.6138,
      "step": 2741
    },
    {
      "epoch": 1.2695087803077991,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002242764378478664,
      "loss": 0.6999,
      "step": 2742
    },
    {
      "epoch": 1.2699718538134825,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002242486085343228,
      "loss": 0.6225,
      "step": 2743
    },
    {
      "epoch": 1.2704349273191662,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002242207792207792,
      "loss": 0.622,
      "step": 2744
    },
    {
      "epoch": 1.2708980008248496,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002241929499072356,
      "loss": 0.7316,
      "step": 2745
    },
    {
      "epoch": 1.2713610743305332,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000224165120593692,
      "loss": 0.5991,
      "step": 2746
    },
    {
      "epoch": 1.2718241478362167,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002241372912801484,
      "loss": 0.6486,
      "step": 2747
    },
    {
      "epoch": 1.2722872213419003,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022410946196660481,
      "loss": 0.6799,
      "step": 2748
    },
    {
      "epoch": 1.2727502948475837,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002240816326530612,
      "loss": 0.6327,
      "step": 2749
    },
    {
      "epoch": 1.2732133683532671,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00022405380333951762,
      "loss": 0.5848,
      "step": 2750
    },
    {
      "epoch": 1.2736764418589508,
      "grad_norm": 0.484375,
      "learning_rate": 0.000224025974025974,
      "loss": 0.598,
      "step": 2751
    },
    {
      "epoch": 1.2741395153646342,
      "grad_norm": 0.484375,
      "learning_rate": 0.00022399814471243043,
      "loss": 0.5855,
      "step": 2752
    },
    {
      "epoch": 1.2746025888703176,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0002239703153988868,
      "loss": 0.5806,
      "step": 2753
    },
    {
      "epoch": 1.2750656623760013,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00022394248608534323,
      "loss": 0.5634,
      "step": 2754
    },
    {
      "epoch": 1.2755287358816847,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002239146567717996,
      "loss": 0.6337,
      "step": 2755
    },
    {
      "epoch": 1.2759918093873681,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000223886827458256,
      "loss": 0.5839,
      "step": 2756
    },
    {
      "epoch": 1.2764548828930518,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002238589981447124,
      "loss": 0.6318,
      "step": 2757
    },
    {
      "epoch": 1.2769179563987352,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00022383116883116882,
      "loss": 0.6545,
      "step": 2758
    },
    {
      "epoch": 1.2773810299044186,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002238033395176252,
      "loss": 0.6112,
      "step": 2759
    },
    {
      "epoch": 1.2778441034101022,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0002237755102040816,
      "loss": 0.6702,
      "step": 2760
    },
    {
      "epoch": 1.2783071769157857,
      "grad_norm": 0.65625,
      "learning_rate": 0.00022374768089053801,
      "loss": 0.6479,
      "step": 2761
    },
    {
      "epoch": 1.2787702504214693,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0002237198515769944,
      "loss": 0.5886,
      "step": 2762
    },
    {
      "epoch": 1.2792333239271527,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00022369202226345082,
      "loss": 0.5805,
      "step": 2763
    },
    {
      "epoch": 1.2796963974328364,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002236641929499072,
      "loss": 0.6649,
      "step": 2764
    },
    {
      "epoch": 1.2801594709385198,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022363636363636363,
      "loss": 0.6384,
      "step": 2765
    },
    {
      "epoch": 1.2806225444442032,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022360853432282002,
      "loss": 0.5933,
      "step": 2766
    },
    {
      "epoch": 1.2810856179498868,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00022358070500927643,
      "loss": 0.6692,
      "step": 2767
    },
    {
      "epoch": 1.2815486914555703,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00022355287569573282,
      "loss": 0.7186,
      "step": 2768
    },
    {
      "epoch": 1.2820117649612537,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00022352504638218924,
      "loss": 0.5612,
      "step": 2769
    },
    {
      "epoch": 1.2824748384669373,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022349721706864563,
      "loss": 0.6516,
      "step": 2770
    },
    {
      "epoch": 1.2829379119726207,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022346938775510205,
      "loss": 0.5615,
      "step": 2771
    },
    {
      "epoch": 1.2834009854783042,
      "grad_norm": 0.5,
      "learning_rate": 0.0002234415584415584,
      "loss": 0.6711,
      "step": 2772
    },
    {
      "epoch": 1.2838640589839878,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002234137291280148,
      "loss": 0.5928,
      "step": 2773
    },
    {
      "epoch": 1.2843271324896712,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022338589981447121,
      "loss": 0.6317,
      "step": 2774
    },
    {
      "epoch": 1.2847902059953549,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002233580705009276,
      "loss": 0.6435,
      "step": 2775
    },
    {
      "epoch": 1.2852532795010383,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022333024118738402,
      "loss": 0.6846,
      "step": 2776
    },
    {
      "epoch": 1.285716353006722,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002233024118738404,
      "loss": 0.6098,
      "step": 2777
    },
    {
      "epoch": 1.2861794265124054,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022327458256029683,
      "loss": 0.6457,
      "step": 2778
    },
    {
      "epoch": 1.2866425000180888,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00022324675324675322,
      "loss": 0.6238,
      "step": 2779
    },
    {
      "epoch": 1.2871055735237724,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00022321892393320963,
      "loss": 0.663,
      "step": 2780
    },
    {
      "epoch": 1.2875686470294558,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022319109461966602,
      "loss": 0.6306,
      "step": 2781
    },
    {
      "epoch": 1.2880317205351393,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00022316326530612244,
      "loss": 0.5854,
      "step": 2782
    },
    {
      "epoch": 1.288494794040823,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00022313543599257883,
      "loss": 0.5156,
      "step": 2783
    },
    {
      "epoch": 1.2889578675465063,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00022310760667903525,
      "loss": 0.5904,
      "step": 2784
    },
    {
      "epoch": 1.2894209410521897,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00022307977736549164,
      "loss": 0.5775,
      "step": 2785
    },
    {
      "epoch": 1.2898840145578734,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022305194805194805,
      "loss": 0.6579,
      "step": 2786
    },
    {
      "epoch": 1.2903470880635568,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022302411873840444,
      "loss": 0.575,
      "step": 2787
    },
    {
      "epoch": 1.2908101615692402,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022299628942486083,
      "loss": 0.6055,
      "step": 2788
    },
    {
      "epoch": 1.2912732350749239,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022296846011131725,
      "loss": 0.6111,
      "step": 2789
    },
    {
      "epoch": 1.2917363085806073,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002229406307977736,
      "loss": 0.6112,
      "step": 2790
    },
    {
      "epoch": 1.292199382086291,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022291280148423005,
      "loss": 0.6784,
      "step": 2791
    },
    {
      "epoch": 1.2926624555919743,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022288497217068642,
      "loss": 0.7125,
      "step": 2792
    },
    {
      "epoch": 1.293125529097658,
      "grad_norm": 0.65625,
      "learning_rate": 0.00022285714285714283,
      "loss": 0.6277,
      "step": 2793
    },
    {
      "epoch": 1.2935886026033414,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00022282931354359922,
      "loss": 0.6958,
      "step": 2794
    },
    {
      "epoch": 1.2940516761090248,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00022280148423005564,
      "loss": 0.6081,
      "step": 2795
    },
    {
      "epoch": 1.2945147496147085,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00022277365491651203,
      "loss": 0.7775,
      "step": 2796
    },
    {
      "epoch": 1.294977823120392,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00022274582560296845,
      "loss": 0.582,
      "step": 2797
    },
    {
      "epoch": 1.2954408966260753,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00022271799628942484,
      "loss": 0.6885,
      "step": 2798
    },
    {
      "epoch": 1.295903970131759,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022269016697588125,
      "loss": 0.6122,
      "step": 2799
    },
    {
      "epoch": 1.2963670436374424,
      "grad_norm": 0.640625,
      "learning_rate": 0.00022266233766233764,
      "loss": 0.6848,
      "step": 2800
    },
    {
      "epoch": 1.2968301171431258,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00022263450834879403,
      "loss": 0.7175,
      "step": 2801
    },
    {
      "epoch": 1.2972931906488094,
      "grad_norm": 0.640625,
      "learning_rate": 0.00022260667903525045,
      "loss": 0.7522,
      "step": 2802
    },
    {
      "epoch": 1.2977562641544929,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00022257884972170684,
      "loss": 0.5907,
      "step": 2803
    },
    {
      "epoch": 1.2982193376601763,
      "grad_norm": 0.546875,
      "learning_rate": 0.00022255102040816325,
      "loss": 0.7021,
      "step": 2804
    },
    {
      "epoch": 1.29868241116586,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022252319109461964,
      "loss": 0.6993,
      "step": 2805
    },
    {
      "epoch": 1.2991454846715433,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022249536178107606,
      "loss": 0.5551,
      "step": 2806
    },
    {
      "epoch": 1.299608558177227,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00022246753246753245,
      "loss": 0.6145,
      "step": 2807
    },
    {
      "epoch": 1.3000716316829104,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00022243970315398887,
      "loss": 0.5902,
      "step": 2808
    },
    {
      "epoch": 1.300534705188594,
      "grad_norm": 0.671875,
      "learning_rate": 0.00022241187384044523,
      "loss": 0.7985,
      "step": 2809
    },
    {
      "epoch": 1.3009977786942775,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00022238404452690167,
      "loss": 0.6723,
      "step": 2810
    },
    {
      "epoch": 1.3014608521999609,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022235621521335804,
      "loss": 0.6216,
      "step": 2811
    },
    {
      "epoch": 1.3019239257056445,
      "grad_norm": 0.59375,
      "learning_rate": 0.00022232838589981445,
      "loss": 0.6315,
      "step": 2812
    },
    {
      "epoch": 1.302386999211328,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022230055658627084,
      "loss": 0.658,
      "step": 2813
    },
    {
      "epoch": 1.3028500727170114,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00022227272727272726,
      "loss": 0.6652,
      "step": 2814
    },
    {
      "epoch": 1.303313146222695,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022224489795918365,
      "loss": 0.6646,
      "step": 2815
    },
    {
      "epoch": 1.3037762197283784,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022221706864564004,
      "loss": 0.6352,
      "step": 2816
    },
    {
      "epoch": 1.3042392932340618,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022218923933209646,
      "loss": 0.6761,
      "step": 2817
    },
    {
      "epoch": 1.3047023667397455,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00022216141001855284,
      "loss": 0.6868,
      "step": 2818
    },
    {
      "epoch": 1.305165440245429,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00022213358070500926,
      "loss": 0.6903,
      "step": 2819
    },
    {
      "epoch": 1.3056285137511123,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022210575139146565,
      "loss": 0.605,
      "step": 2820
    },
    {
      "epoch": 1.306091587256796,
      "grad_norm": 0.53125,
      "learning_rate": 0.00022207792207792207,
      "loss": 0.5948,
      "step": 2821
    },
    {
      "epoch": 1.3065546607624794,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00022205009276437846,
      "loss": 0.6392,
      "step": 2822
    },
    {
      "epoch": 1.307017734268163,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00022202226345083487,
      "loss": 0.5557,
      "step": 2823
    },
    {
      "epoch": 1.3074808077738465,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00022199443413729126,
      "loss": 0.6234,
      "step": 2824
    },
    {
      "epoch": 1.30794388127953,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00022196660482374768,
      "loss": 0.6616,
      "step": 2825
    },
    {
      "epoch": 1.3084069547852135,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022193877551020407,
      "loss": 0.6483,
      "step": 2826
    },
    {
      "epoch": 1.308870028290897,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0002219109461966605,
      "loss": 0.7555,
      "step": 2827
    },
    {
      "epoch": 1.3093331017965806,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00022188311688311685,
      "loss": 0.6919,
      "step": 2828
    },
    {
      "epoch": 1.309796175302264,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00022185528756957324,
      "loss": 0.591,
      "step": 2829
    },
    {
      "epoch": 1.3102592488079474,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022182745825602966,
      "loss": 0.6542,
      "step": 2830
    },
    {
      "epoch": 1.310722322313631,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00022179962894248605,
      "loss": 0.6244,
      "step": 2831
    },
    {
      "epoch": 1.3111853958193145,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00022177179962894246,
      "loss": 0.667,
      "step": 2832
    },
    {
      "epoch": 1.311648469324998,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00022174397031539885,
      "loss": 0.6883,
      "step": 2833
    },
    {
      "epoch": 1.3121115428306815,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00022171614100185527,
      "loss": 0.6432,
      "step": 2834
    },
    {
      "epoch": 1.312574616336365,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00022168831168831166,
      "loss": 0.6173,
      "step": 2835
    },
    {
      "epoch": 1.3130376898420484,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022166048237476807,
      "loss": 0.6305,
      "step": 2836
    },
    {
      "epoch": 1.313500763347732,
      "grad_norm": 0.546875,
      "learning_rate": 0.00022163265306122446,
      "loss": 0.5525,
      "step": 2837
    },
    {
      "epoch": 1.3139638368534157,
      "grad_norm": 0.515625,
      "learning_rate": 0.00022160482374768088,
      "loss": 0.5708,
      "step": 2838
    },
    {
      "epoch": 1.314426910359099,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022157699443413727,
      "loss": 0.6531,
      "step": 2839
    },
    {
      "epoch": 1.3148899838647825,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002215491651205937,
      "loss": 0.6667,
      "step": 2840
    },
    {
      "epoch": 1.3153530573704661,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022152133580705008,
      "loss": 0.5811,
      "step": 2841
    },
    {
      "epoch": 1.3158161308761496,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002214935064935065,
      "loss": 0.6048,
      "step": 2842
    },
    {
      "epoch": 1.316279204381833,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00022146567717996288,
      "loss": 0.5847,
      "step": 2843
    },
    {
      "epoch": 1.3167422778875166,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00022143784786641927,
      "loss": 0.5916,
      "step": 2844
    },
    {
      "epoch": 1.3172053513932,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002214100185528757,
      "loss": 0.6265,
      "step": 2845
    },
    {
      "epoch": 1.3176684248988835,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00022138218923933205,
      "loss": 0.5747,
      "step": 2846
    },
    {
      "epoch": 1.3181314984045671,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002213543599257885,
      "loss": 0.6875,
      "step": 2847
    },
    {
      "epoch": 1.3185945719102505,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00022132653061224486,
      "loss": 0.678,
      "step": 2848
    },
    {
      "epoch": 1.319057645415934,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00022129870129870128,
      "loss": 0.6612,
      "step": 2849
    },
    {
      "epoch": 1.3195207189216176,
      "grad_norm": 0.75,
      "learning_rate": 0.00022127087198515766,
      "loss": 0.6933,
      "step": 2850
    },
    {
      "epoch": 1.319983792427301,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00022124304267161408,
      "loss": 0.5477,
      "step": 2851
    },
    {
      "epoch": 1.3204468659329847,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00022121521335807047,
      "loss": 0.5806,
      "step": 2852
    },
    {
      "epoch": 1.320909939438668,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002211873840445269,
      "loss": 0.6145,
      "step": 2853
    },
    {
      "epoch": 1.3213730129443517,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022115955473098328,
      "loss": 0.621,
      "step": 2854
    },
    {
      "epoch": 1.3218360864500351,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002211317254174397,
      "loss": 0.6804,
      "step": 2855
    },
    {
      "epoch": 1.3222991599557186,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00022110389610389608,
      "loss": 0.634,
      "step": 2856
    },
    {
      "epoch": 1.3227622334614022,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002210760667903525,
      "loss": 0.6192,
      "step": 2857
    },
    {
      "epoch": 1.3232253069670856,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002210482374768089,
      "loss": 0.6401,
      "step": 2858
    },
    {
      "epoch": 1.323688380472769,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00022102040816326528,
      "loss": 0.6471,
      "step": 2859
    },
    {
      "epoch": 1.3241514539784527,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002209925788497217,
      "loss": 0.6444,
      "step": 2860
    },
    {
      "epoch": 1.324614527484136,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00022096474953617809,
      "loss": 0.6508,
      "step": 2861
    },
    {
      "epoch": 1.3250776009898195,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002209369202226345,
      "loss": 0.5601,
      "step": 2862
    },
    {
      "epoch": 1.3255406744955032,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0002209090909090909,
      "loss": 0.6023,
      "step": 2863
    },
    {
      "epoch": 1.3260037480011866,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002208812615955473,
      "loss": 0.6215,
      "step": 2864
    },
    {
      "epoch": 1.32646682150687,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00022085343228200367,
      "loss": 0.6123,
      "step": 2865
    },
    {
      "epoch": 1.3269298950125537,
      "grad_norm": 0.75,
      "learning_rate": 0.00022082560296846012,
      "loss": 0.7006,
      "step": 2866
    },
    {
      "epoch": 1.327392968518237,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00022079777365491648,
      "loss": 0.5961,
      "step": 2867
    },
    {
      "epoch": 1.3278560420239207,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002207699443413729,
      "loss": 0.5763,
      "step": 2868
    },
    {
      "epoch": 1.3283191155296041,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00022074211502782928,
      "loss": 0.6019,
      "step": 2869
    },
    {
      "epoch": 1.3287821890352878,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002207142857142857,
      "loss": 0.6035,
      "step": 2870
    },
    {
      "epoch": 1.3292452625409712,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002206864564007421,
      "loss": 0.6159,
      "step": 2871
    },
    {
      "epoch": 1.3297083360466546,
      "grad_norm": 0.5625,
      "learning_rate": 0.00022065862708719848,
      "loss": 0.6875,
      "step": 2872
    },
    {
      "epoch": 1.3301714095523383,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002206307977736549,
      "loss": 0.6101,
      "step": 2873
    },
    {
      "epoch": 1.3306344830580217,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00022060296846011129,
      "loss": 0.6372,
      "step": 2874
    },
    {
      "epoch": 1.331097556563705,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002205751391465677,
      "loss": 0.6275,
      "step": 2875
    },
    {
      "epoch": 1.3315606300693887,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002205473098330241,
      "loss": 0.7247,
      "step": 2876
    },
    {
      "epoch": 1.3320237035750722,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002205194805194805,
      "loss": 0.6682,
      "step": 2877
    },
    {
      "epoch": 1.3324867770807556,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002204916512059369,
      "loss": 0.6566,
      "step": 2878
    },
    {
      "epoch": 1.3329498505864392,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00022046382189239332,
      "loss": 0.6401,
      "step": 2879
    },
    {
      "epoch": 1.3334129240921226,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002204359925788497,
      "loss": 0.6979,
      "step": 2880
    },
    {
      "epoch": 1.333875997597806,
      "grad_norm": 0.703125,
      "learning_rate": 0.00022040816326530612,
      "loss": 0.7495,
      "step": 2881
    },
    {
      "epoch": 1.3343390711034897,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002203803339517625,
      "loss": 0.6768,
      "step": 2882
    },
    {
      "epoch": 1.3348021446091731,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00022035250463821893,
      "loss": 0.7113,
      "step": 2883
    },
    {
      "epoch": 1.3352652181148568,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002203246753246753,
      "loss": 0.6438,
      "step": 2884
    },
    {
      "epoch": 1.3357282916205402,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00022029684601113173,
      "loss": 0.6381,
      "step": 2885
    },
    {
      "epoch": 1.3361913651262238,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0002202690166975881,
      "loss": 0.7164,
      "step": 2886
    },
    {
      "epoch": 1.3366544386319072,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002202411873840445,
      "loss": 0.6225,
      "step": 2887
    },
    {
      "epoch": 1.3371175121375907,
      "grad_norm": 0.765625,
      "learning_rate": 0.0002202133580705009,
      "loss": 0.7454,
      "step": 2888
    },
    {
      "epoch": 1.3375805856432743,
      "grad_norm": 0.625,
      "learning_rate": 0.0002201855287569573,
      "loss": 0.6449,
      "step": 2889
    },
    {
      "epoch": 1.3380436591489577,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002201576994434137,
      "loss": 0.7389,
      "step": 2890
    },
    {
      "epoch": 1.3385067326546412,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0002201298701298701,
      "loss": 0.5975,
      "step": 2891
    },
    {
      "epoch": 1.3389698061603248,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00022010204081632652,
      "loss": 0.6262,
      "step": 2892
    },
    {
      "epoch": 1.3394328796660082,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0002200742115027829,
      "loss": 0.7741,
      "step": 2893
    },
    {
      "epoch": 1.3398959531716916,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00022004638218923932,
      "loss": 0.6575,
      "step": 2894
    },
    {
      "epoch": 1.3403590266773753,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002200185528756957,
      "loss": 0.7062,
      "step": 2895
    },
    {
      "epoch": 1.3408221001830587,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021999072356215213,
      "loss": 0.6381,
      "step": 2896
    },
    {
      "epoch": 1.3412851736887421,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00021996289424860852,
      "loss": 0.7297,
      "step": 2897
    },
    {
      "epoch": 1.3417482471944258,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021993506493506493,
      "loss": 0.6876,
      "step": 2898
    },
    {
      "epoch": 1.3422113207001092,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00021990723562152132,
      "loss": 0.7347,
      "step": 2899
    },
    {
      "epoch": 1.3426743942057928,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00021987940630797771,
      "loss": 0.6904,
      "step": 2900
    },
    {
      "epoch": 1.3431374677114762,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00021985157699443413,
      "loss": 0.6379,
      "step": 2901
    },
    {
      "epoch": 1.3436005412171599,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002198237476808905,
      "loss": 0.6363,
      "step": 2902
    },
    {
      "epoch": 1.3440636147228433,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021979591836734694,
      "loss": 0.7277,
      "step": 2903
    },
    {
      "epoch": 1.3445266882285267,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0002197680890538033,
      "loss": 0.7454,
      "step": 2904
    },
    {
      "epoch": 1.3449897617342104,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021974025974025972,
      "loss": 0.7232,
      "step": 2905
    },
    {
      "epoch": 1.3454528352398938,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002197124304267161,
      "loss": 0.6054,
      "step": 2906
    },
    {
      "epoch": 1.3459159087455772,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00021968460111317252,
      "loss": 0.574,
      "step": 2907
    },
    {
      "epoch": 1.3463789822512608,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0002196567717996289,
      "loss": 0.7347,
      "step": 2908
    },
    {
      "epoch": 1.3468420557569443,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021962894248608533,
      "loss": 0.6002,
      "step": 2909
    },
    {
      "epoch": 1.3473051292626277,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00021960111317254172,
      "loss": 0.6598,
      "step": 2910
    },
    {
      "epoch": 1.3477682027683113,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021957328385899814,
      "loss": 0.6073,
      "step": 2911
    },
    {
      "epoch": 1.3482312762739947,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021954545454545452,
      "loss": 0.6764,
      "step": 2912
    },
    {
      "epoch": 1.3486943497796782,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00021951762523191094,
      "loss": 0.5938,
      "step": 2913
    },
    {
      "epoch": 1.3491574232853618,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021948979591836733,
      "loss": 0.5505,
      "step": 2914
    },
    {
      "epoch": 1.3496204967910455,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021946196660482372,
      "loss": 0.7256,
      "step": 2915
    },
    {
      "epoch": 1.3500835702967289,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021943413729128014,
      "loss": 0.746,
      "step": 2916
    },
    {
      "epoch": 1.3505466438024123,
      "grad_norm": 0.46875,
      "learning_rate": 0.00021940630797773653,
      "loss": 0.6218,
      "step": 2917
    },
    {
      "epoch": 1.351009717308096,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00021937847866419294,
      "loss": 0.6512,
      "step": 2918
    },
    {
      "epoch": 1.3514727908137794,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021935064935064933,
      "loss": 0.599,
      "step": 2919
    },
    {
      "epoch": 1.3519358643194628,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00021932282003710575,
      "loss": 0.6501,
      "step": 2920
    },
    {
      "epoch": 1.3523989378251464,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002192949907235621,
      "loss": 0.6211,
      "step": 2921
    },
    {
      "epoch": 1.3528620113308298,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021926716141001856,
      "loss": 0.6122,
      "step": 2922
    },
    {
      "epoch": 1.3533250848365133,
      "grad_norm": 0.625,
      "learning_rate": 0.00021923933209647492,
      "loss": 0.702,
      "step": 2923
    },
    {
      "epoch": 1.353788158342197,
      "grad_norm": 0.5,
      "learning_rate": 0.00021921150278293134,
      "loss": 0.6781,
      "step": 2924
    },
    {
      "epoch": 1.3542512318478803,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00021918367346938773,
      "loss": 0.6459,
      "step": 2925
    },
    {
      "epoch": 1.3547143053535637,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00021915584415584414,
      "loss": 0.7431,
      "step": 2926
    },
    {
      "epoch": 1.3551773788592474,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00021912801484230053,
      "loss": 0.5857,
      "step": 2927
    },
    {
      "epoch": 1.3556404523649308,
      "grad_norm": 0.84375,
      "learning_rate": 0.00021910018552875692,
      "loss": 0.8356,
      "step": 2928
    },
    {
      "epoch": 1.3561035258706144,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00021907235621521334,
      "loss": 0.6958,
      "step": 2929
    },
    {
      "epoch": 1.3565665993762979,
      "grad_norm": 0.59375,
      "learning_rate": 0.00021904452690166973,
      "loss": 0.6998,
      "step": 2930
    },
    {
      "epoch": 1.3570296728819815,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021901669758812614,
      "loss": 0.6323,
      "step": 2931
    },
    {
      "epoch": 1.357492746387665,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00021898886827458253,
      "loss": 0.6799,
      "step": 2932
    },
    {
      "epoch": 1.3579558198933483,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00021896103896103895,
      "loss": 0.5734,
      "step": 2933
    },
    {
      "epoch": 1.358418893399032,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021893320964749534,
      "loss": 0.6492,
      "step": 2934
    },
    {
      "epoch": 1.3588819669047154,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00021890538033395176,
      "loss": 0.6165,
      "step": 2935
    },
    {
      "epoch": 1.3593450404103988,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021887755102040815,
      "loss": 0.6533,
      "step": 2936
    },
    {
      "epoch": 1.3598081139160825,
      "grad_norm": 0.515625,
      "learning_rate": 0.00021884972170686456,
      "loss": 0.5591,
      "step": 2937
    },
    {
      "epoch": 1.360271187421766,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00021882189239332095,
      "loss": 0.5792,
      "step": 2938
    },
    {
      "epoch": 1.3607342609274493,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021879406307977737,
      "loss": 0.6516,
      "step": 2939
    },
    {
      "epoch": 1.361197334433133,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00021876623376623373,
      "loss": 0.5849,
      "step": 2940
    },
    {
      "epoch": 1.3616604079388164,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021873840445269018,
      "loss": 0.6787,
      "step": 2941
    },
    {
      "epoch": 1.3621234814444998,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021871057513914654,
      "loss": 0.6301,
      "step": 2942
    },
    {
      "epoch": 1.3625865549501834,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021868274582560293,
      "loss": 0.5754,
      "step": 2943
    },
    {
      "epoch": 1.3630496284558669,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00021865491651205934,
      "loss": 0.608,
      "step": 2944
    },
    {
      "epoch": 1.3635127019615505,
      "grad_norm": 0.515625,
      "learning_rate": 0.00021862708719851573,
      "loss": 0.655,
      "step": 2945
    },
    {
      "epoch": 1.363975775467234,
      "grad_norm": 0.640625,
      "learning_rate": 0.00021859925788497215,
      "loss": 0.654,
      "step": 2946
    },
    {
      "epoch": 1.3644388489729176,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021857142857142854,
      "loss": 0.6415,
      "step": 2947
    },
    {
      "epoch": 1.364901922478601,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00021854359925788496,
      "loss": 0.6369,
      "step": 2948
    },
    {
      "epoch": 1.3653649959842844,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021851576994434135,
      "loss": 0.7391,
      "step": 2949
    },
    {
      "epoch": 1.365828069489968,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00021848794063079776,
      "loss": 0.6716,
      "step": 2950
    },
    {
      "epoch": 1.3662911429956515,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021846011131725415,
      "loss": 0.5661,
      "step": 2951
    },
    {
      "epoch": 1.3667542165013349,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021843228200371057,
      "loss": 0.6299,
      "step": 2952
    },
    {
      "epoch": 1.3672172900070185,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00021840445269016696,
      "loss": 0.7577,
      "step": 2953
    },
    {
      "epoch": 1.367680363512702,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021837662337662338,
      "loss": 0.5984,
      "step": 2954
    },
    {
      "epoch": 1.3681434370183854,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00021834879406307977,
      "loss": 0.6762,
      "step": 2955
    },
    {
      "epoch": 1.368606510524069,
      "grad_norm": 0.625,
      "learning_rate": 0.00021832096474953616,
      "loss": 0.6798,
      "step": 2956
    },
    {
      "epoch": 1.3690695840297524,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021829313543599257,
      "loss": 0.6998,
      "step": 2957
    },
    {
      "epoch": 1.3695326575354358,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021826530612244893,
      "loss": 0.6463,
      "step": 2958
    },
    {
      "epoch": 1.3699957310411195,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00021823747680890535,
      "loss": 0.7245,
      "step": 2959
    },
    {
      "epoch": 1.370458804546803,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021820964749536174,
      "loss": 0.5662,
      "step": 2960
    },
    {
      "epoch": 1.3709218780524866,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021818181818181816,
      "loss": 0.7007,
      "step": 2961
    },
    {
      "epoch": 1.37138495155817,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021815398886827455,
      "loss": 0.7028,
      "step": 2962
    },
    {
      "epoch": 1.3718480250638536,
      "grad_norm": 0.59375,
      "learning_rate": 0.00021812615955473096,
      "loss": 0.6496,
      "step": 2963
    },
    {
      "epoch": 1.372311098569537,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021809833024118735,
      "loss": 0.7241,
      "step": 2964
    },
    {
      "epoch": 1.3727741720752205,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00021807050092764377,
      "loss": 0.6337,
      "step": 2965
    },
    {
      "epoch": 1.373237245580904,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021804267161410016,
      "loss": 0.6206,
      "step": 2966
    },
    {
      "epoch": 1.3737003190865875,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00021801484230055658,
      "loss": 0.6452,
      "step": 2967
    },
    {
      "epoch": 1.374163392592271,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021798701298701297,
      "loss": 0.5999,
      "step": 2968
    },
    {
      "epoch": 1.3746264660979546,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021795918367346938,
      "loss": 0.619,
      "step": 2969
    },
    {
      "epoch": 1.375089539603638,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021793135435992577,
      "loss": 0.7311,
      "step": 2970
    },
    {
      "epoch": 1.3755526131093214,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00021790352504638216,
      "loss": 0.6857,
      "step": 2971
    },
    {
      "epoch": 1.376015686615005,
      "grad_norm": 19.0,
      "learning_rate": 0.00021787569573283858,
      "loss": 0.8533,
      "step": 2972
    },
    {
      "epoch": 1.3764787601206885,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021784786641929497,
      "loss": 0.6261,
      "step": 2973
    },
    {
      "epoch": 1.376941833626372,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021782003710575139,
      "loss": 0.6463,
      "step": 2974
    },
    {
      "epoch": 1.3774049071320555,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021779220779220777,
      "loss": 0.6873,
      "step": 2975
    },
    {
      "epoch": 1.377867980637739,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002177643784786642,
      "loss": 0.6422,
      "step": 2976
    },
    {
      "epoch": 1.3783310541434226,
      "grad_norm": 0.625,
      "learning_rate": 0.00021773654916512055,
      "loss": 0.7178,
      "step": 2977
    },
    {
      "epoch": 1.378794127649106,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000217708719851577,
      "loss": 0.6563,
      "step": 2978
    },
    {
      "epoch": 1.3792572011547897,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021768089053803336,
      "loss": 0.7092,
      "step": 2979
    },
    {
      "epoch": 1.379720274660473,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021765306122448978,
      "loss": 0.6486,
      "step": 2980
    },
    {
      "epoch": 1.3801833481661565,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021762523191094617,
      "loss": 0.5508,
      "step": 2981
    },
    {
      "epoch": 1.3806464216718402,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021759740259740258,
      "loss": 0.6745,
      "step": 2982
    },
    {
      "epoch": 1.3811094951775236,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021756957328385897,
      "loss": 0.614,
      "step": 2983
    },
    {
      "epoch": 1.381572568683207,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002175417439703154,
      "loss": 0.6696,
      "step": 2984
    },
    {
      "epoch": 1.3820356421888906,
      "grad_norm": 0.515625,
      "learning_rate": 0.00021751391465677178,
      "loss": 0.626,
      "step": 2985
    },
    {
      "epoch": 1.382498715694574,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00021748608534322817,
      "loss": 0.6128,
      "step": 2986
    },
    {
      "epoch": 1.3829617892002575,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021745825602968459,
      "loss": 0.7003,
      "step": 2987
    },
    {
      "epoch": 1.3834248627059411,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00021743042671614098,
      "loss": 0.5541,
      "step": 2988
    },
    {
      "epoch": 1.3838879362116245,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002174025974025974,
      "loss": 0.6469,
      "step": 2989
    },
    {
      "epoch": 1.3843510097173082,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021737476808905378,
      "loss": 0.5485,
      "step": 2990
    },
    {
      "epoch": 1.3848140832229916,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0002173469387755102,
      "loss": 0.5982,
      "step": 2991
    },
    {
      "epoch": 1.3852771567286752,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002173191094619666,
      "loss": 0.7153,
      "step": 2992
    },
    {
      "epoch": 1.3857402302343587,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000217291280148423,
      "loss": 0.7553,
      "step": 2993
    },
    {
      "epoch": 1.386203303740042,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002172634508348794,
      "loss": 0.6885,
      "step": 2994
    },
    {
      "epoch": 1.3866663772457257,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0002172356215213358,
      "loss": 0.5694,
      "step": 2995
    },
    {
      "epoch": 1.3871294507514091,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00021720779220779217,
      "loss": 0.6713,
      "step": 2996
    },
    {
      "epoch": 1.3875925242570926,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021717996289424862,
      "loss": 0.6426,
      "step": 2997
    },
    {
      "epoch": 1.3880555977627762,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00021715213358070498,
      "loss": 0.7021,
      "step": 2998
    },
    {
      "epoch": 1.3885186712684596,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021712430426716137,
      "loss": 0.6323,
      "step": 2999
    },
    {
      "epoch": 1.388981744774143,
      "grad_norm": 0.6875,
      "learning_rate": 0.00021709647495361779,
      "loss": 0.6574,
      "step": 3000
    },
    {
      "epoch": 1.3894448182798267,
      "grad_norm": 0.515625,
      "learning_rate": 0.00021706864564007418,
      "loss": 0.5276,
      "step": 3001
    },
    {
      "epoch": 1.38990789178551,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002170408163265306,
      "loss": 0.6404,
      "step": 3002
    },
    {
      "epoch": 1.3903709652911935,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00021701298701298698,
      "loss": 0.5937,
      "step": 3003
    },
    {
      "epoch": 1.3908340387968772,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002169851576994434,
      "loss": 0.6081,
      "step": 3004
    },
    {
      "epoch": 1.3912971123025606,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002169573283858998,
      "loss": 0.6965,
      "step": 3005
    },
    {
      "epoch": 1.3917601858082442,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002169294990723562,
      "loss": 0.6291,
      "step": 3006
    },
    {
      "epoch": 1.3922232593139277,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002169016697588126,
      "loss": 0.6367,
      "step": 3007
    },
    {
      "epoch": 1.3926863328196113,
      "grad_norm": 0.50390625,
      "learning_rate": 0.000216873840445269,
      "loss": 0.69,
      "step": 3008
    },
    {
      "epoch": 1.3931494063252947,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002168460111317254,
      "loss": 0.7163,
      "step": 3009
    },
    {
      "epoch": 1.3936124798309781,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021681818181818182,
      "loss": 0.5948,
      "step": 3010
    },
    {
      "epoch": 1.3940755533366618,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0002167903525046382,
      "loss": 0.6484,
      "step": 3011
    },
    {
      "epoch": 1.3945386268423452,
      "grad_norm": 0.5,
      "learning_rate": 0.00021676252319109462,
      "loss": 0.6148,
      "step": 3012
    },
    {
      "epoch": 1.3950017003480286,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00021673469387755101,
      "loss": 0.6578,
      "step": 3013
    },
    {
      "epoch": 1.3954647738537123,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021670686456400738,
      "loss": 0.6078,
      "step": 3014
    },
    {
      "epoch": 1.3959278473593957,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0002166790352504638,
      "loss": 0.5524,
      "step": 3015
    },
    {
      "epoch": 1.396390920865079,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00021665120593692018,
      "loss": 0.572,
      "step": 3016
    },
    {
      "epoch": 1.3968539943707627,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002166233766233766,
      "loss": 0.6052,
      "step": 3017
    },
    {
      "epoch": 1.3973170678764462,
      "grad_norm": 0.609375,
      "learning_rate": 0.000216595547309833,
      "loss": 0.7835,
      "step": 3018
    },
    {
      "epoch": 1.3977801413821296,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002165677179962894,
      "loss": 0.6264,
      "step": 3019
    },
    {
      "epoch": 1.3982432148878132,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002165398886827458,
      "loss": 0.6459,
      "step": 3020
    },
    {
      "epoch": 1.3987062883934966,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002165120593692022,
      "loss": 0.6884,
      "step": 3021
    },
    {
      "epoch": 1.3991693618991803,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0002164842300556586,
      "loss": 0.6189,
      "step": 3022
    },
    {
      "epoch": 1.3996324354048637,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00021645640074211502,
      "loss": 0.7036,
      "step": 3023
    },
    {
      "epoch": 1.4000955089105473,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002164285714285714,
      "loss": 0.677,
      "step": 3024
    },
    {
      "epoch": 1.4005585824162308,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00021640074211502782,
      "loss": 0.6088,
      "step": 3025
    },
    {
      "epoch": 1.4010216559219142,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00021637291280148421,
      "loss": 0.6548,
      "step": 3026
    },
    {
      "epoch": 1.4014847294275978,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002163450834879406,
      "loss": 0.5755,
      "step": 3027
    },
    {
      "epoch": 1.4019478029332813,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021631725417439702,
      "loss": 0.7098,
      "step": 3028
    },
    {
      "epoch": 1.4024108764389647,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002162894248608534,
      "loss": 0.6158,
      "step": 3029
    },
    {
      "epoch": 1.4028739499446483,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021626159554730983,
      "loss": 0.5999,
      "step": 3030
    },
    {
      "epoch": 1.4033370234503317,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021623376623376622,
      "loss": 0.6303,
      "step": 3031
    },
    {
      "epoch": 1.4038000969560152,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00021620593692022263,
      "loss": 0.7138,
      "step": 3032
    },
    {
      "epoch": 1.4042631704616988,
      "grad_norm": 0.7421875,
      "learning_rate": 0.000216178107606679,
      "loss": 0.6306,
      "step": 3033
    },
    {
      "epoch": 1.4047262439673822,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00021615027829313544,
      "loss": 0.5639,
      "step": 3034
    },
    {
      "epoch": 1.4051893174730656,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002161224489795918,
      "loss": 0.6645,
      "step": 3035
    },
    {
      "epoch": 1.4056523909787493,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021609461966604822,
      "loss": 0.6232,
      "step": 3036
    },
    {
      "epoch": 1.4061154644844327,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002160667903525046,
      "loss": 0.6706,
      "step": 3037
    },
    {
      "epoch": 1.4065785379901163,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00021603896103896102,
      "loss": 0.6137,
      "step": 3038
    },
    {
      "epoch": 1.4070416114957998,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021601113172541741,
      "loss": 0.6476,
      "step": 3039
    },
    {
      "epoch": 1.4075046850014834,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021598330241187383,
      "loss": 0.6997,
      "step": 3040
    },
    {
      "epoch": 1.4079677585071668,
      "grad_norm": 0.671875,
      "learning_rate": 0.00021595547309833022,
      "loss": 0.6762,
      "step": 3041
    },
    {
      "epoch": 1.4084308320128502,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002159276437847866,
      "loss": 0.6789,
      "step": 3042
    },
    {
      "epoch": 1.4088939055185339,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00021589981447124303,
      "loss": 0.6391,
      "step": 3043
    },
    {
      "epoch": 1.4093569790242173,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021587198515769942,
      "loss": 0.7256,
      "step": 3044
    },
    {
      "epoch": 1.4098200525299007,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021584415584415583,
      "loss": 0.6666,
      "step": 3045
    },
    {
      "epoch": 1.4102831260355844,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021581632653061222,
      "loss": 0.644,
      "step": 3046
    },
    {
      "epoch": 1.4107461995412678,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021578849721706864,
      "loss": 0.6054,
      "step": 3047
    },
    {
      "epoch": 1.4112092730469512,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00021576066790352503,
      "loss": 0.594,
      "step": 3048
    },
    {
      "epoch": 1.4116723465526348,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00021573283858998145,
      "loss": 0.5228,
      "step": 3049
    },
    {
      "epoch": 1.4121354200583183,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00021570500927643784,
      "loss": 0.6613,
      "step": 3050
    },
    {
      "epoch": 1.4125984935640017,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00021567717996289425,
      "loss": 0.6998,
      "step": 3051
    },
    {
      "epoch": 1.4130615670696853,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00021564935064935061,
      "loss": 0.7386,
      "step": 3052
    },
    {
      "epoch": 1.4135246405753688,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021562152133580706,
      "loss": 0.6803,
      "step": 3053
    },
    {
      "epoch": 1.4139877140810524,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021559369202226342,
      "loss": 0.6159,
      "step": 3054
    },
    {
      "epoch": 1.4144507875867358,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002155658627087198,
      "loss": 0.6565,
      "step": 3055
    },
    {
      "epoch": 1.4149138610924195,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021553803339517623,
      "loss": 0.7317,
      "step": 3056
    },
    {
      "epoch": 1.4153769345981029,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00021551020408163262,
      "loss": 0.6564,
      "step": 3057
    },
    {
      "epoch": 1.4158400081037863,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00021548237476808903,
      "loss": 0.6083,
      "step": 3058
    },
    {
      "epoch": 1.41630308160947,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021545454545454542,
      "loss": 0.6696,
      "step": 3059
    },
    {
      "epoch": 1.4167661551151534,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021542671614100184,
      "loss": 0.6827,
      "step": 3060
    },
    {
      "epoch": 1.4172292286208368,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00021539888682745823,
      "loss": 0.6045,
      "step": 3061
    },
    {
      "epoch": 1.4176923021265204,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021537105751391465,
      "loss": 0.5991,
      "step": 3062
    },
    {
      "epoch": 1.4181553756322038,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00021534322820037104,
      "loss": 0.5985,
      "step": 3063
    },
    {
      "epoch": 1.4186184491378873,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021531539888682745,
      "loss": 0.6902,
      "step": 3064
    },
    {
      "epoch": 1.419081522643571,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021528756957328384,
      "loss": 0.6804,
      "step": 3065
    },
    {
      "epoch": 1.4195445961492543,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021525974025974026,
      "loss": 0.6515,
      "step": 3066
    },
    {
      "epoch": 1.420007669654938,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00021523191094619665,
      "loss": 0.656,
      "step": 3067
    },
    {
      "epoch": 1.4204707431606214,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00021520408163265307,
      "loss": 0.6567,
      "step": 3068
    },
    {
      "epoch": 1.420933816666305,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00021517625231910945,
      "loss": 0.7145,
      "step": 3069
    },
    {
      "epoch": 1.4213968901719884,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00021514842300556582,
      "loss": 0.5923,
      "step": 3070
    },
    {
      "epoch": 1.4218599636776719,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021512059369202223,
      "loss": 0.6212,
      "step": 3071
    },
    {
      "epoch": 1.4223230371833555,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021509276437847862,
      "loss": 0.6443,
      "step": 3072
    },
    {
      "epoch": 1.422786110689039,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021506493506493504,
      "loss": 0.7017,
      "step": 3073
    },
    {
      "epoch": 1.4232491841947223,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00021503710575139143,
      "loss": 0.6494,
      "step": 3074
    },
    {
      "epoch": 1.423712257700406,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00021500927643784785,
      "loss": 0.6481,
      "step": 3075
    },
    {
      "epoch": 1.4241753312060894,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021498144712430424,
      "loss": 0.5267,
      "step": 3076
    },
    {
      "epoch": 1.4246384047117728,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021495361781076065,
      "loss": 0.6053,
      "step": 3077
    },
    {
      "epoch": 1.4251014782174565,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021492578849721704,
      "loss": 0.6244,
      "step": 3078
    },
    {
      "epoch": 1.42556455172314,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021489795918367346,
      "loss": 0.6586,
      "step": 3079
    },
    {
      "epoch": 1.4260276252288233,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00021487012987012985,
      "loss": 0.6675,
      "step": 3080
    },
    {
      "epoch": 1.426490698734507,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00021484230055658627,
      "loss": 0.5777,
      "step": 3081
    },
    {
      "epoch": 1.4269537722401904,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00021481447124304266,
      "loss": 0.624,
      "step": 3082
    },
    {
      "epoch": 1.427416845745874,
      "grad_norm": 0.5,
      "learning_rate": 0.00021478664192949904,
      "loss": 0.6846,
      "step": 3083
    },
    {
      "epoch": 1.4278799192515574,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00021475881261595546,
      "loss": 0.6397,
      "step": 3084
    },
    {
      "epoch": 1.428342992757241,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021473098330241185,
      "loss": 0.6494,
      "step": 3085
    },
    {
      "epoch": 1.4288060662629245,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021470315398886827,
      "loss": 0.6862,
      "step": 3086
    },
    {
      "epoch": 1.429269139768608,
      "grad_norm": 0.484375,
      "learning_rate": 0.00021467532467532466,
      "loss": 0.6156,
      "step": 3087
    },
    {
      "epoch": 1.4297322132742916,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00021464749536178107,
      "loss": 0.5683,
      "step": 3088
    },
    {
      "epoch": 1.430195286779975,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021461966604823744,
      "loss": 0.645,
      "step": 3089
    },
    {
      "epoch": 1.4306583602856584,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00021459183673469388,
      "loss": 0.641,
      "step": 3090
    },
    {
      "epoch": 1.431121433791342,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021456400742115024,
      "loss": 0.6302,
      "step": 3091
    },
    {
      "epoch": 1.4315845072970255,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021453617810760666,
      "loss": 0.607,
      "step": 3092
    },
    {
      "epoch": 1.4320475808027089,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021450834879406305,
      "loss": 0.5727,
      "step": 3093
    },
    {
      "epoch": 1.4325106543083925,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021448051948051947,
      "loss": 0.6772,
      "step": 3094
    },
    {
      "epoch": 1.432973727814076,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00021445269016697586,
      "loss": 0.6124,
      "step": 3095
    },
    {
      "epoch": 1.4334368013197594,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021442486085343227,
      "loss": 0.6541,
      "step": 3096
    },
    {
      "epoch": 1.433899874825443,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00021439703153988866,
      "loss": 0.6097,
      "step": 3097
    },
    {
      "epoch": 1.4343629483311264,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00021436920222634505,
      "loss": 0.67,
      "step": 3098
    },
    {
      "epoch": 1.43482602183681,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00021434137291280147,
      "loss": 0.5581,
      "step": 3099
    },
    {
      "epoch": 1.4352890953424935,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00021431354359925786,
      "loss": 0.7495,
      "step": 3100
    },
    {
      "epoch": 1.4357521688481771,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021428571428571427,
      "loss": 0.6258,
      "step": 3101
    },
    {
      "epoch": 1.4362152423538606,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021425788497217066,
      "loss": 0.6191,
      "step": 3102
    },
    {
      "epoch": 1.436678315859544,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00021423005565862708,
      "loss": 0.6555,
      "step": 3103
    },
    {
      "epoch": 1.4371413893652276,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021420222634508347,
      "loss": 0.7017,
      "step": 3104
    },
    {
      "epoch": 1.437604462870911,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002141743970315399,
      "loss": 0.6868,
      "step": 3105
    },
    {
      "epoch": 1.4380675363765945,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00021414656771799628,
      "loss": 0.6287,
      "step": 3106
    },
    {
      "epoch": 1.438530609882278,
      "grad_norm": 0.5,
      "learning_rate": 0.0002141187384044527,
      "loss": 0.5994,
      "step": 3107
    },
    {
      "epoch": 1.4389936833879615,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021409090909090906,
      "loss": 0.5888,
      "step": 3108
    },
    {
      "epoch": 1.439456756893645,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002140630797773655,
      "loss": 0.7341,
      "step": 3109
    },
    {
      "epoch": 1.4399198303993286,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00021403525046382186,
      "loss": 0.5982,
      "step": 3110
    },
    {
      "epoch": 1.440382903905012,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021400742115027828,
      "loss": 0.7199,
      "step": 3111
    },
    {
      "epoch": 1.4408459774106954,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00021397959183673467,
      "loss": 0.6035,
      "step": 3112
    },
    {
      "epoch": 1.441309050916379,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00021395176252319106,
      "loss": 0.6532,
      "step": 3113
    },
    {
      "epoch": 1.4417721244220625,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00021392393320964748,
      "loss": 0.6622,
      "step": 3114
    },
    {
      "epoch": 1.4422351979277461,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00021389610389610386,
      "loss": 0.6979,
      "step": 3115
    },
    {
      "epoch": 1.4426982714334295,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021386827458256028,
      "loss": 0.7038,
      "step": 3116
    },
    {
      "epoch": 1.4431613449391132,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00021384044526901667,
      "loss": 0.607,
      "step": 3117
    },
    {
      "epoch": 1.4436244184447966,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0002138126159554731,
      "loss": 0.723,
      "step": 3118
    },
    {
      "epoch": 1.44408749195048,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00021378478664192948,
      "loss": 0.6242,
      "step": 3119
    },
    {
      "epoch": 1.4445505654561637,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0002137569573283859,
      "loss": 0.6345,
      "step": 3120
    },
    {
      "epoch": 1.445013638961847,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021372912801484228,
      "loss": 0.5838,
      "step": 3121
    },
    {
      "epoch": 1.4454767124675305,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002137012987012987,
      "loss": 0.5933,
      "step": 3122
    },
    {
      "epoch": 1.4459397859732142,
      "grad_norm": 47.0,
      "learning_rate": 0.0002136734693877551,
      "loss": 1.2843,
      "step": 3123
    },
    {
      "epoch": 1.4464028594788976,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002136456400742115,
      "loss": 0.6293,
      "step": 3124
    },
    {
      "epoch": 1.446865932984581,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002136178107606679,
      "loss": 0.673,
      "step": 3125
    },
    {
      "epoch": 1.4473290064902646,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021358998144712426,
      "loss": 0.6886,
      "step": 3126
    },
    {
      "epoch": 1.447792079995948,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021356215213358068,
      "loss": 0.6925,
      "step": 3127
    },
    {
      "epoch": 1.4482551535016315,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021353432282003707,
      "loss": 0.645,
      "step": 3128
    },
    {
      "epoch": 1.4487182270073151,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00021350649350649348,
      "loss": 0.6682,
      "step": 3129
    },
    {
      "epoch": 1.4491813005129985,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00021347866419294987,
      "loss": 0.5594,
      "step": 3130
    },
    {
      "epoch": 1.4496443740186822,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002134508348794063,
      "loss": 0.6479,
      "step": 3131
    },
    {
      "epoch": 1.4501074475243656,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021342300556586268,
      "loss": 0.6448,
      "step": 3132
    },
    {
      "epoch": 1.4505705210300492,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002133951762523191,
      "loss": 0.6281,
      "step": 3133
    },
    {
      "epoch": 1.4510335945357327,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021336734693877548,
      "loss": 0.6951,
      "step": 3134
    },
    {
      "epoch": 1.451496668041416,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002133395176252319,
      "loss": 0.6337,
      "step": 3135
    },
    {
      "epoch": 1.4519597415470997,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002133116883116883,
      "loss": 0.6492,
      "step": 3136
    },
    {
      "epoch": 1.4524228150527831,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002132838589981447,
      "loss": 0.6148,
      "step": 3137
    },
    {
      "epoch": 1.4528858885584666,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002132560296846011,
      "loss": 0.6603,
      "step": 3138
    },
    {
      "epoch": 1.4533489620641502,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002132282003710575,
      "loss": 0.6872,
      "step": 3139
    },
    {
      "epoch": 1.4538120355698336,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002132003710575139,
      "loss": 0.7027,
      "step": 3140
    },
    {
      "epoch": 1.454275109075517,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002131725417439703,
      "loss": 0.5906,
      "step": 3141
    },
    {
      "epoch": 1.4547381825812007,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002131447124304267,
      "loss": 0.742,
      "step": 3142
    },
    {
      "epoch": 1.455201256086884,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002131168831168831,
      "loss": 0.6299,
      "step": 3143
    },
    {
      "epoch": 1.4556643295925678,
      "grad_norm": 0.625,
      "learning_rate": 0.00021308905380333952,
      "loss": 0.6846,
      "step": 3144
    },
    {
      "epoch": 1.4561274030982512,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021306122448979588,
      "loss": 0.6859,
      "step": 3145
    },
    {
      "epoch": 1.4565904766039348,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002130333951762523,
      "loss": 0.5799,
      "step": 3146
    },
    {
      "epoch": 1.4570535501096182,
      "grad_norm": 0.53125,
      "learning_rate": 0.00021300556586270868,
      "loss": 0.6958,
      "step": 3147
    },
    {
      "epoch": 1.4575166236153017,
      "grad_norm": 9.5,
      "learning_rate": 0.0002129777365491651,
      "loss": 0.6003,
      "step": 3148
    },
    {
      "epoch": 1.4579796971209853,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0002129499072356215,
      "loss": 0.6464,
      "step": 3149
    },
    {
      "epoch": 1.4584427706266687,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0002129220779220779,
      "loss": 0.5767,
      "step": 3150
    },
    {
      "epoch": 1.4589058441323521,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002128942486085343,
      "loss": 0.5786,
      "step": 3151
    },
    {
      "epoch": 1.4593689176380358,
      "grad_norm": 0.53125,
      "learning_rate": 0.00021286641929499071,
      "loss": 0.6749,
      "step": 3152
    },
    {
      "epoch": 1.4598319911437192,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002128385899814471,
      "loss": 0.6897,
      "step": 3153
    },
    {
      "epoch": 1.4602950646494026,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002128107606679035,
      "loss": 0.647,
      "step": 3154
    },
    {
      "epoch": 1.4607581381550863,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0002127829313543599,
      "loss": 0.6769,
      "step": 3155
    },
    {
      "epoch": 1.4612212116607697,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0002127551020408163,
      "loss": 0.6119,
      "step": 3156
    },
    {
      "epoch": 1.461684285166453,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00021272727272727272,
      "loss": 0.6025,
      "step": 3157
    },
    {
      "epoch": 1.4621473586721367,
      "grad_norm": 0.640625,
      "learning_rate": 0.0002126994434137291,
      "loss": 0.6798,
      "step": 3158
    },
    {
      "epoch": 1.4626104321778202,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021267161410018552,
      "loss": 0.5966,
      "step": 3159
    },
    {
      "epoch": 1.4630735056835038,
      "grad_norm": 1.578125,
      "learning_rate": 0.0002126437847866419,
      "loss": 0.72,
      "step": 3160
    },
    {
      "epoch": 1.4635365791891872,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00021261595547309833,
      "loss": 0.6651,
      "step": 3161
    },
    {
      "epoch": 1.4639996526948709,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00021258812615955472,
      "loss": 0.5967,
      "step": 3162
    },
    {
      "epoch": 1.4644627262005543,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00021256029684601113,
      "loss": 0.6221,
      "step": 3163
    },
    {
      "epoch": 1.4649257997062377,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0002125324675324675,
      "loss": 0.6457,
      "step": 3164
    },
    {
      "epoch": 1.4653888732119214,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021250463821892394,
      "loss": 0.6509,
      "step": 3165
    },
    {
      "epoch": 1.4658519467176048,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002124768089053803,
      "loss": 0.5723,
      "step": 3166
    },
    {
      "epoch": 1.4663150202232882,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021244897959183672,
      "loss": 0.6049,
      "step": 3167
    },
    {
      "epoch": 1.4667780937289718,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002124211502782931,
      "loss": 0.5976,
      "step": 3168
    },
    {
      "epoch": 1.4672411672346553,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002123933209647495,
      "loss": 0.587,
      "step": 3169
    },
    {
      "epoch": 1.4677042407403387,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00021236549165120592,
      "loss": 0.6751,
      "step": 3170
    },
    {
      "epoch": 1.4681673142460223,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002123376623376623,
      "loss": 0.6081,
      "step": 3171
    },
    {
      "epoch": 1.4686303877517057,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021230983302411872,
      "loss": 0.6473,
      "step": 3172
    },
    {
      "epoch": 1.4690934612573892,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002122820037105751,
      "loss": 0.6902,
      "step": 3173
    },
    {
      "epoch": 1.4695565347630728,
      "grad_norm": 0.5,
      "learning_rate": 0.00021225417439703153,
      "loss": 0.6583,
      "step": 3174
    },
    {
      "epoch": 1.4700196082687562,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00021222634508348792,
      "loss": 0.5939,
      "step": 3175
    },
    {
      "epoch": 1.4704826817744399,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00021219851576994434,
      "loss": 0.5568,
      "step": 3176
    },
    {
      "epoch": 1.4709457552801233,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021217068645640072,
      "loss": 0.654,
      "step": 3177
    },
    {
      "epoch": 1.471408828785807,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021214285714285714,
      "loss": 0.7085,
      "step": 3178
    },
    {
      "epoch": 1.4718719022914903,
      "grad_norm": 0.671875,
      "learning_rate": 0.00021211502782931353,
      "loss": 0.7162,
      "step": 3179
    },
    {
      "epoch": 1.4723349757971738,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00021208719851576995,
      "loss": 0.614,
      "step": 3180
    },
    {
      "epoch": 1.4727980493028574,
      "grad_norm": 0.640625,
      "learning_rate": 0.00021205936920222634,
      "loss": 0.6057,
      "step": 3181
    },
    {
      "epoch": 1.4732611228085408,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0002120315398886827,
      "loss": 0.5749,
      "step": 3182
    },
    {
      "epoch": 1.4737241963142242,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021200371057513912,
      "loss": 0.6849,
      "step": 3183
    },
    {
      "epoch": 1.4741872698199079,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002119758812615955,
      "loss": 0.6294,
      "step": 3184
    },
    {
      "epoch": 1.4746503433255913,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021194805194805192,
      "loss": 0.6319,
      "step": 3185
    },
    {
      "epoch": 1.4751134168312747,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002119202226345083,
      "loss": 0.6084,
      "step": 3186
    },
    {
      "epoch": 1.4755764903369584,
      "grad_norm": 0.515625,
      "learning_rate": 0.00021189239332096473,
      "loss": 0.6083,
      "step": 3187
    },
    {
      "epoch": 1.4760395638426418,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021186456400742112,
      "loss": 0.6773,
      "step": 3188
    },
    {
      "epoch": 1.4765026373483252,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00021183673469387754,
      "loss": 0.6947,
      "step": 3189
    },
    {
      "epoch": 1.4769657108540089,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021180890538033393,
      "loss": 0.618,
      "step": 3190
    },
    {
      "epoch": 1.4774287843596923,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021178107606679034,
      "loss": 0.6571,
      "step": 3191
    },
    {
      "epoch": 1.477891857865376,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00021175324675324673,
      "loss": 0.7138,
      "step": 3192
    },
    {
      "epoch": 1.4783549313710593,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00021172541743970315,
      "loss": 0.6432,
      "step": 3193
    },
    {
      "epoch": 1.478818004876743,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021169758812615954,
      "loss": 0.5223,
      "step": 3194
    },
    {
      "epoch": 1.4792810783824264,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021166975881261595,
      "loss": 0.6668,
      "step": 3195
    },
    {
      "epoch": 1.4797441518881098,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00021164192949907234,
      "loss": 0.6265,
      "step": 3196
    },
    {
      "epoch": 1.4802072253937935,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00021161410018552873,
      "loss": 0.5899,
      "step": 3197
    },
    {
      "epoch": 1.4806702988994769,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021158627087198515,
      "loss": 0.6121,
      "step": 3198
    },
    {
      "epoch": 1.4811333724051603,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021155844155844154,
      "loss": 0.6552,
      "step": 3199
    },
    {
      "epoch": 1.481596445910844,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021153061224489796,
      "loss": 0.6267,
      "step": 3200
    },
    {
      "epoch": 1.4820595194165274,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021150278293135432,
      "loss": 0.6207,
      "step": 3201
    },
    {
      "epoch": 1.4825225929222108,
      "grad_norm": 0.53125,
      "learning_rate": 0.00021147495361781074,
      "loss": 0.6245,
      "step": 3202
    },
    {
      "epoch": 1.4829856664278944,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021144712430426713,
      "loss": 0.5944,
      "step": 3203
    },
    {
      "epoch": 1.4834487399335778,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00021141929499072354,
      "loss": 0.6958,
      "step": 3204
    },
    {
      "epoch": 1.4839118134392613,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00021139146567717993,
      "loss": 0.6735,
      "step": 3205
    },
    {
      "epoch": 1.484374886944945,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021136363636363635,
      "loss": 0.6339,
      "step": 3206
    },
    {
      "epoch": 1.4848379604506283,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021133580705009274,
      "loss": 0.6449,
      "step": 3207
    },
    {
      "epoch": 1.485301033956312,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00021130797773654916,
      "loss": 0.6569,
      "step": 3208
    },
    {
      "epoch": 1.4857641074619954,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00021128014842300554,
      "loss": 0.5649,
      "step": 3209
    },
    {
      "epoch": 1.486227180967679,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00021125231910946193,
      "loss": 0.7114,
      "step": 3210
    },
    {
      "epoch": 1.4866902544733624,
      "grad_norm": 0.6875,
      "learning_rate": 0.00021122448979591835,
      "loss": 0.6622,
      "step": 3211
    },
    {
      "epoch": 1.4871533279790459,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00021119666048237474,
      "loss": 0.6571,
      "step": 3212
    },
    {
      "epoch": 1.4876164014847295,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021116883116883116,
      "loss": 0.605,
      "step": 3213
    },
    {
      "epoch": 1.488079474990413,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00021114100185528755,
      "loss": 0.6664,
      "step": 3214
    },
    {
      "epoch": 1.4885425484960964,
      "grad_norm": 0.578125,
      "learning_rate": 0.00021111317254174396,
      "loss": 0.6804,
      "step": 3215
    },
    {
      "epoch": 1.48900562200178,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00021108534322820035,
      "loss": 0.6238,
      "step": 3216
    },
    {
      "epoch": 1.4894686955074634,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00021105751391465677,
      "loss": 0.6476,
      "step": 3217
    },
    {
      "epoch": 1.4899317690131468,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00021102968460111316,
      "loss": 0.5998,
      "step": 3218
    },
    {
      "epoch": 1.4903948425188305,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00021100185528756958,
      "loss": 0.6537,
      "step": 3219
    },
    {
      "epoch": 1.490857916024514,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00021097402597402594,
      "loss": 0.5639,
      "step": 3220
    },
    {
      "epoch": 1.4913209895301975,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021094619666048238,
      "loss": 0.6216,
      "step": 3221
    },
    {
      "epoch": 1.491784063035881,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00021091836734693875,
      "loss": 0.6071,
      "step": 3222
    },
    {
      "epoch": 1.4922471365415646,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021089053803339516,
      "loss": 0.6944,
      "step": 3223
    },
    {
      "epoch": 1.492710210047248,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00021086270871985155,
      "loss": 0.6516,
      "step": 3224
    },
    {
      "epoch": 1.4931732835529314,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00021083487940630794,
      "loss": 0.5684,
      "step": 3225
    },
    {
      "epoch": 1.493636357058615,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00021080705009276436,
      "loss": 0.7135,
      "step": 3226
    },
    {
      "epoch": 1.4940994305642985,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00021077922077922075,
      "loss": 0.6942,
      "step": 3227
    },
    {
      "epoch": 1.494562504069982,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021075139146567716,
      "loss": 0.5853,
      "step": 3228
    },
    {
      "epoch": 1.4950255775756656,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00021072356215213355,
      "loss": 0.7341,
      "step": 3229
    },
    {
      "epoch": 1.495488651081349,
      "grad_norm": 0.65625,
      "learning_rate": 0.00021069573283858997,
      "loss": 0.6245,
      "step": 3230
    },
    {
      "epoch": 1.4959517245870324,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00021066790352504636,
      "loss": 0.6869,
      "step": 3231
    },
    {
      "epoch": 1.496414798092716,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00021064007421150278,
      "loss": 0.5617,
      "step": 3232
    },
    {
      "epoch": 1.4968778715983995,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00021061224489795917,
      "loss": 0.6279,
      "step": 3233
    },
    {
      "epoch": 1.4973409451040829,
      "grad_norm": 0.59375,
      "learning_rate": 0.00021058441558441558,
      "loss": 0.6595,
      "step": 3234
    },
    {
      "epoch": 1.4978040186097665,
      "grad_norm": 0.546875,
      "learning_rate": 0.00021055658627087197,
      "loss": 0.613,
      "step": 3235
    },
    {
      "epoch": 1.49826709211545,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002105287569573284,
      "loss": 0.6354,
      "step": 3236
    },
    {
      "epoch": 1.4987301656211336,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00021050092764378478,
      "loss": 0.647,
      "step": 3237
    },
    {
      "epoch": 1.499193239126817,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002104730983302412,
      "loss": 0.6454,
      "step": 3238
    },
    {
      "epoch": 1.4996563126325007,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021044526901669756,
      "loss": 0.5886,
      "step": 3239
    },
    {
      "epoch": 1.500119386138184,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00021041743970315395,
      "loss": 0.6319,
      "step": 3240
    },
    {
      "epoch": 1.5005824596438675,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021038961038961036,
      "loss": 0.6403,
      "step": 3241
    },
    {
      "epoch": 1.5010455331495511,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00021036178107606675,
      "loss": 0.6283,
      "step": 3242
    },
    {
      "epoch": 1.5015086066552346,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00021033395176252317,
      "loss": 0.5988,
      "step": 3243
    },
    {
      "epoch": 1.501971680160918,
      "grad_norm": 0.609375,
      "learning_rate": 0.00021030612244897956,
      "loss": 0.6308,
      "step": 3244
    },
    {
      "epoch": 1.5024347536666016,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00021027829313543598,
      "loss": 0.61,
      "step": 3245
    },
    {
      "epoch": 1.502897827172285,
      "grad_norm": 0.5625,
      "learning_rate": 0.00021025046382189237,
      "loss": 0.6668,
      "step": 3246
    },
    {
      "epoch": 1.5033609006779685,
      "grad_norm": 0.625,
      "learning_rate": 0.00021022263450834878,
      "loss": 0.7216,
      "step": 3247
    },
    {
      "epoch": 1.503823974183652,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00021019480519480517,
      "loss": 0.6299,
      "step": 3248
    },
    {
      "epoch": 1.5042870476893355,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002101669758812616,
      "loss": 0.6862,
      "step": 3249
    },
    {
      "epoch": 1.504750121195019,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00021013914656771798,
      "loss": 0.6268,
      "step": 3250
    },
    {
      "epoch": 1.5052131947007026,
      "grad_norm": 0.625,
      "learning_rate": 0.0002101113172541744,
      "loss": 0.7265,
      "step": 3251
    },
    {
      "epoch": 1.5056762682063862,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00021008348794063079,
      "loss": 0.639,
      "step": 3252
    },
    {
      "epoch": 1.5061393417120694,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00021005565862708718,
      "loss": 0.5682,
      "step": 3253
    },
    {
      "epoch": 1.506602415217753,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0002100278293135436,
      "loss": 0.5736,
      "step": 3254
    },
    {
      "epoch": 1.5070654887234367,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00020999999999999998,
      "loss": 0.7222,
      "step": 3255
    },
    {
      "epoch": 1.5075285622291201,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0002099721706864564,
      "loss": 0.6112,
      "step": 3256
    },
    {
      "epoch": 1.5079916357348035,
      "grad_norm": 0.5,
      "learning_rate": 0.00020994434137291276,
      "loss": 0.5942,
      "step": 3257
    },
    {
      "epoch": 1.5084547092404872,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020991651205936918,
      "loss": 0.6695,
      "step": 3258
    },
    {
      "epoch": 1.5089177827461706,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00020988868274582557,
      "loss": 0.6841,
      "step": 3259
    },
    {
      "epoch": 1.509380856251854,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00020986085343228198,
      "loss": 0.7673,
      "step": 3260
    },
    {
      "epoch": 1.5098439297575377,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00020983302411873837,
      "loss": 0.6622,
      "step": 3261
    },
    {
      "epoch": 1.510307003263221,
      "grad_norm": 0.484375,
      "learning_rate": 0.0002098051948051948,
      "loss": 0.6736,
      "step": 3262
    },
    {
      "epoch": 1.5107700767689045,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00020977736549165118,
      "loss": 0.6343,
      "step": 3263
    },
    {
      "epoch": 1.5112331502745882,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002097495361781076,
      "loss": 0.709,
      "step": 3264
    },
    {
      "epoch": 1.5116962237802716,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020972170686456399,
      "loss": 0.6656,
      "step": 3265
    },
    {
      "epoch": 1.512159297285955,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0002096938775510204,
      "loss": 0.7229,
      "step": 3266
    },
    {
      "epoch": 1.5126223707916386,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002096660482374768,
      "loss": 0.5958,
      "step": 3267
    },
    {
      "epoch": 1.5130854442973223,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020963821892393318,
      "loss": 0.6021,
      "step": 3268
    },
    {
      "epoch": 1.5135485178030055,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002096103896103896,
      "loss": 0.6381,
      "step": 3269
    },
    {
      "epoch": 1.5140115913086891,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000209582560296846,
      "loss": 0.6519,
      "step": 3270
    },
    {
      "epoch": 1.5144746648143728,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002095547309833024,
      "loss": 0.6608,
      "step": 3271
    },
    {
      "epoch": 1.5149377383200562,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002095269016697588,
      "loss": 0.6189,
      "step": 3272
    },
    {
      "epoch": 1.5154008118257396,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002094990723562152,
      "loss": 0.6479,
      "step": 3273
    },
    {
      "epoch": 1.5158638853314232,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002094712430426716,
      "loss": 0.6098,
      "step": 3274
    },
    {
      "epoch": 1.5163269588371067,
      "grad_norm": 0.734375,
      "learning_rate": 0.00020944341372912802,
      "loss": 0.7435,
      "step": 3275
    },
    {
      "epoch": 1.51679003234279,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00020941558441558438,
      "loss": 0.6234,
      "step": 3276
    },
    {
      "epoch": 1.5172531058484737,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020938775510204082,
      "loss": 0.6563,
      "step": 3277
    },
    {
      "epoch": 1.5177161793541571,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002093599257884972,
      "loss": 0.6415,
      "step": 3278
    },
    {
      "epoch": 1.5181792528598406,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0002093320964749536,
      "loss": 0.719,
      "step": 3279
    },
    {
      "epoch": 1.5186423263655242,
      "grad_norm": 0.515625,
      "learning_rate": 0.00020930426716141,
      "loss": 0.6798,
      "step": 3280
    },
    {
      "epoch": 1.5191053998712076,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00020927643784786638,
      "loss": 0.7101,
      "step": 3281
    },
    {
      "epoch": 1.519568473376891,
      "grad_norm": 0.59375,
      "learning_rate": 0.0002092486085343228,
      "loss": 0.648,
      "step": 3282
    },
    {
      "epoch": 1.5200315468825747,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002092207792207792,
      "loss": 0.7513,
      "step": 3283
    },
    {
      "epoch": 1.5204946203882583,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002091929499072356,
      "loss": 0.5871,
      "step": 3284
    },
    {
      "epoch": 1.5209576938939415,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000209165120593692,
      "loss": 0.6899,
      "step": 3285
    },
    {
      "epoch": 1.5214207673996252,
      "grad_norm": 0.5,
      "learning_rate": 0.0002091372912801484,
      "loss": 0.5281,
      "step": 3286
    },
    {
      "epoch": 1.5218838409053088,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0002091094619666048,
      "loss": 0.6311,
      "step": 3287
    },
    {
      "epoch": 1.5223469144109922,
      "grad_norm": 0.671875,
      "learning_rate": 0.00020908163265306122,
      "loss": 0.7081,
      "step": 3288
    },
    {
      "epoch": 1.5228099879166757,
      "grad_norm": 0.5,
      "learning_rate": 0.0002090538033395176,
      "loss": 0.6409,
      "step": 3289
    },
    {
      "epoch": 1.5232730614223593,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00020902597402597402,
      "loss": 0.673,
      "step": 3290
    },
    {
      "epoch": 1.5237361349280427,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00020899814471243041,
      "loss": 0.7358,
      "step": 3291
    },
    {
      "epoch": 1.5241992084337261,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00020897031539888683,
      "loss": 0.6112,
      "step": 3292
    },
    {
      "epoch": 1.5246622819394098,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00020894248608534322,
      "loss": 0.603,
      "step": 3293
    },
    {
      "epoch": 1.5251253554450932,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00020891465677179964,
      "loss": 0.7663,
      "step": 3294
    },
    {
      "epoch": 1.5255884289507766,
      "grad_norm": 0.53125,
      "learning_rate": 0.000208886827458256,
      "loss": 0.5618,
      "step": 3295
    },
    {
      "epoch": 1.5260515024564603,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002088589981447124,
      "loss": 0.6699,
      "step": 3296
    },
    {
      "epoch": 1.526514575962144,
      "grad_norm": 0.5625,
      "learning_rate": 0.0002088311688311688,
      "loss": 0.6599,
      "step": 3297
    },
    {
      "epoch": 1.526977649467827,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0002088033395176252,
      "loss": 0.5978,
      "step": 3298
    },
    {
      "epoch": 1.5274407229735107,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002087755102040816,
      "loss": 0.6333,
      "step": 3299
    },
    {
      "epoch": 1.5279037964791944,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000208747680890538,
      "loss": 0.5978,
      "step": 3300
    },
    {
      "epoch": 1.5283668699848778,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00020871985157699442,
      "loss": 0.6659,
      "step": 3301
    },
    {
      "epoch": 1.5288299434905612,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002086920222634508,
      "loss": 0.5728,
      "step": 3302
    },
    {
      "epoch": 1.5292930169962449,
      "grad_norm": 0.703125,
      "learning_rate": 0.00020866419294990722,
      "loss": 0.6989,
      "step": 3303
    },
    {
      "epoch": 1.5297560905019283,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020863636363636361,
      "loss": 0.6708,
      "step": 3304
    },
    {
      "epoch": 1.5302191640076117,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020860853432282003,
      "loss": 0.6973,
      "step": 3305
    },
    {
      "epoch": 1.5306822375132954,
      "grad_norm": 0.515625,
      "learning_rate": 0.00020858070500927642,
      "loss": 0.6313,
      "step": 3306
    },
    {
      "epoch": 1.5311453110189788,
      "grad_norm": 0.625,
      "learning_rate": 0.00020855287569573284,
      "loss": 0.603,
      "step": 3307
    },
    {
      "epoch": 1.5316083845246622,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020852504638218923,
      "loss": 0.6496,
      "step": 3308
    },
    {
      "epoch": 1.5320714580303458,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00020849721706864562,
      "loss": 0.6664,
      "step": 3309
    },
    {
      "epoch": 1.5325345315360293,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00020846938775510203,
      "loss": 0.5943,
      "step": 3310
    },
    {
      "epoch": 1.5329976050417127,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00020844155844155842,
      "loss": 0.689,
      "step": 3311
    },
    {
      "epoch": 1.5334606785473963,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00020841372912801484,
      "loss": 0.6356,
      "step": 3312
    },
    {
      "epoch": 1.53392375205308,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0002083858998144712,
      "loss": 0.7015,
      "step": 3313
    },
    {
      "epoch": 1.5343868255587632,
      "grad_norm": 0.5,
      "learning_rate": 0.00020835807050092762,
      "loss": 0.6373,
      "step": 3314
    },
    {
      "epoch": 1.5348498990644468,
      "grad_norm": 0.498046875,
      "learning_rate": 0.000208330241187384,
      "loss": 0.6085,
      "step": 3315
    },
    {
      "epoch": 1.5353129725701304,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020830241187384043,
      "loss": 0.6286,
      "step": 3316
    },
    {
      "epoch": 1.5357760460758139,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020827458256029681,
      "loss": 0.6758,
      "step": 3317
    },
    {
      "epoch": 1.5362391195814973,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00020824675324675323,
      "loss": 0.6826,
      "step": 3318
    },
    {
      "epoch": 1.536702193087181,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00020821892393320962,
      "loss": 0.6537,
      "step": 3319
    },
    {
      "epoch": 1.5371652665928643,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00020819109461966604,
      "loss": 0.7016,
      "step": 3320
    },
    {
      "epoch": 1.5376283400985478,
      "grad_norm": 0.59375,
      "learning_rate": 0.00020816326530612243,
      "loss": 0.6519,
      "step": 3321
    },
    {
      "epoch": 1.5380914136042314,
      "grad_norm": 0.515625,
      "learning_rate": 0.00020813543599257884,
      "loss": 0.56,
      "step": 3322
    },
    {
      "epoch": 1.5385544871099148,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00020810760667903523,
      "loss": 0.7186,
      "step": 3323
    },
    {
      "epoch": 1.5390175606155982,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00020807977736549162,
      "loss": 0.6222,
      "step": 3324
    },
    {
      "epoch": 1.5394806341212819,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00020805194805194804,
      "loss": 0.5578,
      "step": 3325
    },
    {
      "epoch": 1.5399437076269653,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020802411873840443,
      "loss": 0.6676,
      "step": 3326
    },
    {
      "epoch": 1.5404067811326487,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020799628942486085,
      "loss": 0.5923,
      "step": 3327
    },
    {
      "epoch": 1.5408698546383324,
      "grad_norm": 0.484375,
      "learning_rate": 0.00020796846011131724,
      "loss": 0.6296,
      "step": 3328
    },
    {
      "epoch": 1.541332928144016,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020794063079777365,
      "loss": 0.6323,
      "step": 3329
    },
    {
      "epoch": 1.5417960016496992,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020791280148423004,
      "loss": 0.6642,
      "step": 3330
    },
    {
      "epoch": 1.5422590751553829,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020788497217068646,
      "loss": 0.6345,
      "step": 3331
    },
    {
      "epoch": 1.5427221486610665,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020785714285714282,
      "loss": 0.7449,
      "step": 3332
    },
    {
      "epoch": 1.54318522216675,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020782931354359927,
      "loss": 0.7538,
      "step": 3333
    },
    {
      "epoch": 1.5436482956724333,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020780148423005563,
      "loss": 0.7027,
      "step": 3334
    },
    {
      "epoch": 1.544111369178117,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00020777365491651204,
      "loss": 0.6555,
      "step": 3335
    },
    {
      "epoch": 1.5445744426838004,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020774582560296843,
      "loss": 0.6607,
      "step": 3336
    },
    {
      "epoch": 1.5450375161894838,
      "grad_norm": 0.5,
      "learning_rate": 0.00020771799628942482,
      "loss": 0.6299,
      "step": 3337
    },
    {
      "epoch": 1.5455005896951675,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020769016697588124,
      "loss": 0.6371,
      "step": 3338
    },
    {
      "epoch": 1.5459636632008509,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00020766233766233763,
      "loss": 0.6607,
      "step": 3339
    },
    {
      "epoch": 1.5464267367065343,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020763450834879405,
      "loss": 0.5813,
      "step": 3340
    },
    {
      "epoch": 1.546889810212218,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00020760667903525044,
      "loss": 0.729,
      "step": 3341
    },
    {
      "epoch": 1.5473528837179014,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00020757884972170685,
      "loss": 0.6313,
      "step": 3342
    },
    {
      "epoch": 1.5478159572235848,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00020755102040816324,
      "loss": 0.6115,
      "step": 3343
    },
    {
      "epoch": 1.5482790307292684,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00020752319109461966,
      "loss": 0.6086,
      "step": 3344
    },
    {
      "epoch": 1.548742104234952,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00020749536178107605,
      "loss": 0.6197,
      "step": 3345
    },
    {
      "epoch": 1.5492051777406353,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020746753246753247,
      "loss": 0.5685,
      "step": 3346
    },
    {
      "epoch": 1.549668251246319,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020743970315398886,
      "loss": 0.6329,
      "step": 3347
    },
    {
      "epoch": 1.5501313247520025,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00020741187384044527,
      "loss": 0.6489,
      "step": 3348
    },
    {
      "epoch": 1.550594398257686,
      "grad_norm": 0.859375,
      "learning_rate": 0.00020738404452690166,
      "loss": 0.625,
      "step": 3349
    },
    {
      "epoch": 1.5510574717633694,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00020735621521335808,
      "loss": 0.651,
      "step": 3350
    },
    {
      "epoch": 1.551520545269053,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020732838589981444,
      "loss": 0.7075,
      "step": 3351
    },
    {
      "epoch": 1.5519836187747365,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020730055658627083,
      "loss": 0.6403,
      "step": 3352
    },
    {
      "epoch": 1.5524466922804199,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00020727272727272725,
      "loss": 0.7677,
      "step": 3353
    },
    {
      "epoch": 1.5529097657861035,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020724489795918364,
      "loss": 0.6414,
      "step": 3354
    },
    {
      "epoch": 1.553372839291787,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00020721706864564005,
      "loss": 0.6306,
      "step": 3355
    },
    {
      "epoch": 1.5538359127974704,
      "grad_norm": 0.65625,
      "learning_rate": 0.00020718923933209644,
      "loss": 0.7515,
      "step": 3356
    },
    {
      "epoch": 1.554298986303154,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00020716141001855286,
      "loss": 0.7507,
      "step": 3357
    },
    {
      "epoch": 1.5547620598088374,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020713358070500925,
      "loss": 0.6428,
      "step": 3358
    },
    {
      "epoch": 1.5552251333145208,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020710575139146567,
      "loss": 0.562,
      "step": 3359
    },
    {
      "epoch": 1.5556882068202045,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00020707792207792206,
      "loss": 0.58,
      "step": 3360
    },
    {
      "epoch": 1.5561512803258881,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00020705009276437847,
      "loss": 0.5896,
      "step": 3361
    },
    {
      "epoch": 1.5566143538315713,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020702226345083486,
      "loss": 0.6493,
      "step": 3362
    },
    {
      "epoch": 1.557077427337255,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00020699443413729128,
      "loss": 0.6961,
      "step": 3363
    },
    {
      "epoch": 1.5575405008429386,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00020696660482374767,
      "loss": 0.6902,
      "step": 3364
    },
    {
      "epoch": 1.558003574348622,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00020693877551020409,
      "loss": 0.6806,
      "step": 3365
    },
    {
      "epoch": 1.5584666478543054,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020691094619666047,
      "loss": 0.6768,
      "step": 3366
    },
    {
      "epoch": 1.558929721359989,
      "grad_norm": 0.5625,
      "learning_rate": 0.00020688311688311684,
      "loss": 0.6144,
      "step": 3367
    },
    {
      "epoch": 1.5593927948656725,
      "grad_norm": 0.484375,
      "learning_rate": 0.00020685528756957328,
      "loss": 0.6336,
      "step": 3368
    },
    {
      "epoch": 1.559855868371356,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00020682745825602964,
      "loss": 0.6346,
      "step": 3369
    },
    {
      "epoch": 1.5603189418770396,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00020679962894248606,
      "loss": 0.5966,
      "step": 3370
    },
    {
      "epoch": 1.560782015382723,
      "grad_norm": 0.515625,
      "learning_rate": 0.00020677179962894245,
      "loss": 0.6347,
      "step": 3371
    },
    {
      "epoch": 1.5612450888884064,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020674397031539887,
      "loss": 0.5948,
      "step": 3372
    },
    {
      "epoch": 1.56170816239409,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020671614100185526,
      "loss": 0.5831,
      "step": 3373
    },
    {
      "epoch": 1.5621712358997737,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020668831168831167,
      "loss": 0.6316,
      "step": 3374
    },
    {
      "epoch": 1.562634309405457,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00020666048237476806,
      "loss": 0.7556,
      "step": 3375
    },
    {
      "epoch": 1.5630973829111405,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00020663265306122448,
      "loss": 0.6432,
      "step": 3376
    },
    {
      "epoch": 1.5635604564168242,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00020660482374768087,
      "loss": 0.725,
      "step": 3377
    },
    {
      "epoch": 1.5640235299225076,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00020657699443413729,
      "loss": 0.5988,
      "step": 3378
    },
    {
      "epoch": 1.564486603428191,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020654916512059368,
      "loss": 0.65,
      "step": 3379
    },
    {
      "epoch": 1.5649496769338747,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00020652133580705006,
      "loss": 0.5716,
      "step": 3380
    },
    {
      "epoch": 1.565412750439558,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020649350649350648,
      "loss": 0.6365,
      "step": 3381
    },
    {
      "epoch": 1.5658758239452415,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020646567717996287,
      "loss": 0.7052,
      "step": 3382
    },
    {
      "epoch": 1.5663388974509251,
      "grad_norm": 0.5,
      "learning_rate": 0.0002064378478664193,
      "loss": 0.536,
      "step": 3383
    },
    {
      "epoch": 1.5668019709566086,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020641001855287568,
      "loss": 0.5917,
      "step": 3384
    },
    {
      "epoch": 1.567265044462292,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002063821892393321,
      "loss": 0.6412,
      "step": 3385
    },
    {
      "epoch": 1.5677281179679756,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00020635435992578848,
      "loss": 0.7057,
      "step": 3386
    },
    {
      "epoch": 1.568191191473659,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0002063265306122449,
      "loss": 0.629,
      "step": 3387
    },
    {
      "epoch": 1.5686542649793425,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00020629870129870126,
      "loss": 0.6385,
      "step": 3388
    },
    {
      "epoch": 1.569117338485026,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020627087198515768,
      "loss": 0.6319,
      "step": 3389
    },
    {
      "epoch": 1.5695804119907097,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00020624304267161407,
      "loss": 0.6155,
      "step": 3390
    },
    {
      "epoch": 1.570043485496393,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020621521335807049,
      "loss": 0.6128,
      "step": 3391
    },
    {
      "epoch": 1.5705065590020766,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00020618738404452688,
      "loss": 0.7472,
      "step": 3392
    },
    {
      "epoch": 1.5709696325077602,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002061595547309833,
      "loss": 0.6797,
      "step": 3393
    },
    {
      "epoch": 1.5714327060134436,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00020613172541743968,
      "loss": 0.683,
      "step": 3394
    },
    {
      "epoch": 1.571895779519127,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00020610389610389607,
      "loss": 0.6134,
      "step": 3395
    },
    {
      "epoch": 1.5723588530248107,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0002060760667903525,
      "loss": 0.6169,
      "step": 3396
    },
    {
      "epoch": 1.5728219265304941,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00020604823747680888,
      "loss": 0.6243,
      "step": 3397
    },
    {
      "epoch": 1.5732850000361775,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002060204081632653,
      "loss": 0.6059,
      "step": 3398
    },
    {
      "epoch": 1.5737480735418612,
      "grad_norm": 0.5625,
      "learning_rate": 0.00020599257884972168,
      "loss": 0.6659,
      "step": 3399
    },
    {
      "epoch": 1.5742111470475446,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002059647495361781,
      "loss": 0.6152,
      "step": 3400
    },
    {
      "epoch": 1.574674220553228,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002059369202226345,
      "loss": 0.7059,
      "step": 3401
    },
    {
      "epoch": 1.5751372940589117,
      "grad_norm": 0.96875,
      "learning_rate": 0.0002059090909090909,
      "loss": 0.6544,
      "step": 3402
    },
    {
      "epoch": 1.575600367564595,
      "grad_norm": 0.515625,
      "learning_rate": 0.0002058812615955473,
      "loss": 0.6299,
      "step": 3403
    },
    {
      "epoch": 1.5760634410702785,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002058534322820037,
      "loss": 0.6097,
      "step": 3404
    },
    {
      "epoch": 1.5765265145759622,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002058256029684601,
      "loss": 0.6206,
      "step": 3405
    },
    {
      "epoch": 1.5769895880816458,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020579777365491652,
      "loss": 0.652,
      "step": 3406
    },
    {
      "epoch": 1.577452661587329,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00020576994434137288,
      "loss": 0.652,
      "step": 3407
    },
    {
      "epoch": 1.5779157350930126,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00020574211502782927,
      "loss": 0.5577,
      "step": 3408
    },
    {
      "epoch": 1.5783788085986963,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002057142857142857,
      "loss": 0.6154,
      "step": 3409
    },
    {
      "epoch": 1.5788418821043797,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020568645640074208,
      "loss": 0.6409,
      "step": 3410
    },
    {
      "epoch": 1.5793049556100631,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002056586270871985,
      "loss": 0.6447,
      "step": 3411
    },
    {
      "epoch": 1.5797680291157468,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00020563079777365488,
      "loss": 0.6007,
      "step": 3412
    },
    {
      "epoch": 1.5802311026214302,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002056029684601113,
      "loss": 0.7551,
      "step": 3413
    },
    {
      "epoch": 1.5806941761271136,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0002055751391465677,
      "loss": 0.6618,
      "step": 3414
    },
    {
      "epoch": 1.5811572496327972,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002055473098330241,
      "loss": 0.7051,
      "step": 3415
    },
    {
      "epoch": 1.5816203231384807,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0002055194805194805,
      "loss": 0.6904,
      "step": 3416
    },
    {
      "epoch": 1.582083396644164,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00020549165120593691,
      "loss": 0.6537,
      "step": 3417
    },
    {
      "epoch": 1.5825464701498477,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0002054638218923933,
      "loss": 0.5553,
      "step": 3418
    },
    {
      "epoch": 1.5830095436555311,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00020543599257884972,
      "loss": 0.5856,
      "step": 3419
    },
    {
      "epoch": 1.5834726171612146,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0002054081632653061,
      "loss": 0.6669,
      "step": 3420
    },
    {
      "epoch": 1.5839356906668982,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00020538033395176253,
      "loss": 0.685,
      "step": 3421
    },
    {
      "epoch": 1.5843987641725819,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020535250463821892,
      "loss": 0.6637,
      "step": 3422
    },
    {
      "epoch": 1.584861837678265,
      "grad_norm": 0.625,
      "learning_rate": 0.00020532467532467528,
      "loss": 0.6766,
      "step": 3423
    },
    {
      "epoch": 1.5853249111839487,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020529684601113172,
      "loss": 0.654,
      "step": 3424
    },
    {
      "epoch": 1.5857879846896323,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00020526901669758808,
      "loss": 0.6123,
      "step": 3425
    },
    {
      "epoch": 1.5862510581953158,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0002052411873840445,
      "loss": 0.6412,
      "step": 3426
    },
    {
      "epoch": 1.5867141317009992,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002052133580705009,
      "loss": 0.626,
      "step": 3427
    },
    {
      "epoch": 1.5871772052066828,
      "grad_norm": 0.671875,
      "learning_rate": 0.0002051855287569573,
      "loss": 0.6858,
      "step": 3428
    },
    {
      "epoch": 1.5876402787123662,
      "grad_norm": 0.546875,
      "learning_rate": 0.0002051576994434137,
      "loss": 0.6532,
      "step": 3429
    },
    {
      "epoch": 1.5881033522180497,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00020512987012987011,
      "loss": 0.6549,
      "step": 3430
    },
    {
      "epoch": 1.5885664257237333,
      "grad_norm": 0.609375,
      "learning_rate": 0.0002051020408163265,
      "loss": 0.5845,
      "step": 3431
    },
    {
      "epoch": 1.5890294992294167,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00020507421150278292,
      "loss": 0.6128,
      "step": 3432
    },
    {
      "epoch": 1.5894925727351001,
      "grad_norm": 0.5,
      "learning_rate": 0.0002050463821892393,
      "loss": 0.698,
      "step": 3433
    },
    {
      "epoch": 1.5899556462407838,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020501855287569573,
      "loss": 0.6289,
      "step": 3434
    },
    {
      "epoch": 1.5904187197464672,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020499072356215212,
      "loss": 0.6277,
      "step": 3435
    },
    {
      "epoch": 1.5908817932521506,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002049628942486085,
      "loss": 0.6592,
      "step": 3436
    },
    {
      "epoch": 1.5913448667578343,
      "grad_norm": 4.8125,
      "learning_rate": 0.00020493506493506492,
      "loss": 0.5996,
      "step": 3437
    },
    {
      "epoch": 1.591807940263518,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0002049072356215213,
      "loss": 0.6653,
      "step": 3438
    },
    {
      "epoch": 1.592271013769201,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00020487940630797773,
      "loss": 0.6498,
      "step": 3439
    },
    {
      "epoch": 1.5927340872748847,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00020485157699443412,
      "loss": 0.6722,
      "step": 3440
    },
    {
      "epoch": 1.5931971607805684,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020482374768089054,
      "loss": 0.6557,
      "step": 3441
    },
    {
      "epoch": 1.5936602342862518,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00020479591836734692,
      "loss": 0.6988,
      "step": 3442
    },
    {
      "epoch": 1.5941233077919352,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00020476808905380334,
      "loss": 0.6619,
      "step": 3443
    },
    {
      "epoch": 1.5945863812976189,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002047402597402597,
      "loss": 0.6176,
      "step": 3444
    },
    {
      "epoch": 1.5950494548033023,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020471243042671612,
      "loss": 0.6481,
      "step": 3445
    },
    {
      "epoch": 1.5955125283089857,
      "grad_norm": 0.46875,
      "learning_rate": 0.0002046846011131725,
      "loss": 0.5924,
      "step": 3446
    },
    {
      "epoch": 1.5959756018146694,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00020465677179962893,
      "loss": 0.6711,
      "step": 3447
    },
    {
      "epoch": 1.5964386753203528,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00020462894248608532,
      "loss": 0.689,
      "step": 3448
    },
    {
      "epoch": 1.5969017488260362,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020460111317254173,
      "loss": 0.6307,
      "step": 3449
    },
    {
      "epoch": 1.5973648223317198,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020457328385899812,
      "loss": 0.649,
      "step": 3450
    },
    {
      "epoch": 1.5978278958374035,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002045454545454545,
      "loss": 0.5925,
      "step": 3451
    },
    {
      "epoch": 1.5982909693430867,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00020451762523191093,
      "loss": 0.6807,
      "step": 3452
    },
    {
      "epoch": 1.5987540428487703,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00020448979591836732,
      "loss": 0.6433,
      "step": 3453
    },
    {
      "epoch": 1.599217116354454,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00020446196660482374,
      "loss": 0.6831,
      "step": 3454
    },
    {
      "epoch": 1.5996801898601374,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00020443413729128013,
      "loss": 0.6096,
      "step": 3455
    },
    {
      "epoch": 1.6001432633658208,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020440630797773654,
      "loss": 0.6501,
      "step": 3456
    },
    {
      "epoch": 1.6006063368715044,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00020437847866419293,
      "loss": 0.7571,
      "step": 3457
    },
    {
      "epoch": 1.6010694103771879,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00020435064935064935,
      "loss": 0.6278,
      "step": 3458
    },
    {
      "epoch": 1.6015324838828713,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020432282003710574,
      "loss": 0.6066,
      "step": 3459
    },
    {
      "epoch": 1.601995557388555,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020429499072356215,
      "loss": 0.5956,
      "step": 3460
    },
    {
      "epoch": 1.6024586308942383,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00020426716141001854,
      "loss": 0.6799,
      "step": 3461
    },
    {
      "epoch": 1.6029217043999218,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00020423933209647496,
      "loss": 0.6697,
      "step": 3462
    },
    {
      "epoch": 1.6033847779056054,
      "grad_norm": 0.5625,
      "learning_rate": 0.00020421150278293132,
      "loss": 0.6945,
      "step": 3463
    },
    {
      "epoch": 1.6038478514112888,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0002041836734693877,
      "loss": 0.7591,
      "step": 3464
    },
    {
      "epoch": 1.6043109249169722,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00020415584415584413,
      "loss": 0.6028,
      "step": 3465
    },
    {
      "epoch": 1.604773998422656,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020412801484230052,
      "loss": 0.6385,
      "step": 3466
    },
    {
      "epoch": 1.6052370719283395,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020410018552875694,
      "loss": 0.6012,
      "step": 3467
    },
    {
      "epoch": 1.6057001454340227,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020407235621521333,
      "loss": 0.6383,
      "step": 3468
    },
    {
      "epoch": 1.6061632189397064,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020404452690166974,
      "loss": 0.692,
      "step": 3469
    },
    {
      "epoch": 1.60662629244539,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00020401669758812613,
      "loss": 0.5473,
      "step": 3470
    },
    {
      "epoch": 1.6070893659510734,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00020398886827458255,
      "loss": 0.5861,
      "step": 3471
    },
    {
      "epoch": 1.6075524394567569,
      "grad_norm": 0.640625,
      "learning_rate": 0.00020396103896103894,
      "loss": 0.6295,
      "step": 3472
    },
    {
      "epoch": 1.6080155129624405,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020393320964749536,
      "loss": 0.5842,
      "step": 3473
    },
    {
      "epoch": 1.608478586468124,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00020390538033395174,
      "loss": 0.5926,
      "step": 3474
    },
    {
      "epoch": 1.6089416599738073,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020387755102040816,
      "loss": 0.6363,
      "step": 3475
    },
    {
      "epoch": 1.609404733479491,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00020384972170686455,
      "loss": 0.6652,
      "step": 3476
    },
    {
      "epoch": 1.6098678069851744,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020382189239332097,
      "loss": 0.6106,
      "step": 3477
    },
    {
      "epoch": 1.6103308804908578,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00020379406307977736,
      "loss": 0.6103,
      "step": 3478
    },
    {
      "epoch": 1.6107939539965415,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00020376623376623372,
      "loss": 0.681,
      "step": 3479
    },
    {
      "epoch": 1.6112570275022249,
      "grad_norm": 0.515625,
      "learning_rate": 0.00020373840445269016,
      "loss": 0.6037,
      "step": 3480
    },
    {
      "epoch": 1.6117201010079083,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020371057513914653,
      "loss": 0.6709,
      "step": 3481
    },
    {
      "epoch": 1.612183174513592,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00020368274582560294,
      "loss": 0.626,
      "step": 3482
    },
    {
      "epoch": 1.6126462480192756,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00020365491651205933,
      "loss": 0.7097,
      "step": 3483
    },
    {
      "epoch": 1.6131093215249588,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020362708719851575,
      "loss": 0.7473,
      "step": 3484
    },
    {
      "epoch": 1.6135723950306424,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00020359925788497214,
      "loss": 0.5589,
      "step": 3485
    },
    {
      "epoch": 1.614035468536326,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020357142857142856,
      "loss": 0.6436,
      "step": 3486
    },
    {
      "epoch": 1.6144985420420095,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00020354359925788495,
      "loss": 0.6439,
      "step": 3487
    },
    {
      "epoch": 1.614961615547693,
      "grad_norm": 0.59375,
      "learning_rate": 0.00020351576994434136,
      "loss": 0.698,
      "step": 3488
    },
    {
      "epoch": 1.6154246890533766,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00020348794063079775,
      "loss": 0.5938,
      "step": 3489
    },
    {
      "epoch": 1.61588776255906,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00020346011131725417,
      "loss": 0.7028,
      "step": 3490
    },
    {
      "epoch": 1.6163508360647434,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020343228200371056,
      "loss": 0.6619,
      "step": 3491
    },
    {
      "epoch": 1.616813909570427,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00020340445269016697,
      "loss": 0.5937,
      "step": 3492
    },
    {
      "epoch": 1.6172769830761105,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00020337662337662336,
      "loss": 0.64,
      "step": 3493
    },
    {
      "epoch": 1.6177400565817939,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00020334879406307975,
      "loss": 0.5641,
      "step": 3494
    },
    {
      "epoch": 1.6182031300874775,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00020332096474953617,
      "loss": 0.7172,
      "step": 3495
    },
    {
      "epoch": 1.618666203593161,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00020329313543599256,
      "loss": 0.5808,
      "step": 3496
    },
    {
      "epoch": 1.6191292770988444,
      "grad_norm": 0.59375,
      "learning_rate": 0.00020326530612244898,
      "loss": 0.6808,
      "step": 3497
    },
    {
      "epoch": 1.619592350604528,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00020323747680890537,
      "loss": 0.6597,
      "step": 3498
    },
    {
      "epoch": 1.6200554241102116,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00020320964749536178,
      "loss": 0.6581,
      "step": 3499
    },
    {
      "epoch": 1.6205184976158948,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020318181818181815,
      "loss": 0.5939,
      "step": 3500
    },
    {
      "epoch": 1.6209815711215785,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00020315398886827456,
      "loss": 0.6153,
      "step": 3501
    },
    {
      "epoch": 1.6214446446272621,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020312615955473095,
      "loss": 0.6498,
      "step": 3502
    },
    {
      "epoch": 1.6219077181329455,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00020309833024118737,
      "loss": 0.5695,
      "step": 3503
    },
    {
      "epoch": 1.622370791638629,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00020307050092764376,
      "loss": 0.6607,
      "step": 3504
    },
    {
      "epoch": 1.6228338651443126,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020304267161410017,
      "loss": 0.6141,
      "step": 3505
    },
    {
      "epoch": 1.623296938649996,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00020301484230055656,
      "loss": 0.6017,
      "step": 3506
    },
    {
      "epoch": 1.6237600121556794,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00020298701298701295,
      "loss": 0.6751,
      "step": 3507
    },
    {
      "epoch": 1.624223085661363,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020295918367346937,
      "loss": 0.6869,
      "step": 3508
    },
    {
      "epoch": 1.6246861591670465,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020293135435992576,
      "loss": 0.6345,
      "step": 3509
    },
    {
      "epoch": 1.62514923267273,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020290352504638218,
      "loss": 0.6943,
      "step": 3510
    },
    {
      "epoch": 1.6256123061784136,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020287569573283857,
      "loss": 0.6468,
      "step": 3511
    },
    {
      "epoch": 1.626075379684097,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020284786641929498,
      "loss": 0.5882,
      "step": 3512
    },
    {
      "epoch": 1.6265384531897804,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020282003710575137,
      "loss": 0.7097,
      "step": 3513
    },
    {
      "epoch": 1.627001526695464,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0002027922077922078,
      "loss": 0.6098,
      "step": 3514
    },
    {
      "epoch": 1.6274646002011477,
      "grad_norm": 0.5,
      "learning_rate": 0.00020276437847866418,
      "loss": 0.6734,
      "step": 3515
    },
    {
      "epoch": 1.6279276737068311,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0002027365491651206,
      "loss": 0.6113,
      "step": 3516
    },
    {
      "epoch": 1.6283907472125145,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00020270871985157699,
      "loss": 0.6411,
      "step": 3517
    },
    {
      "epoch": 1.6288538207181982,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002026808905380334,
      "loss": 0.5726,
      "step": 3518
    },
    {
      "epoch": 1.6293168942238816,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00020265306122448976,
      "loss": 0.6663,
      "step": 3519
    },
    {
      "epoch": 1.629779967729565,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0002026252319109462,
      "loss": 0.6801,
      "step": 3520
    },
    {
      "epoch": 1.6302430412352487,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020259740259740257,
      "loss": 0.7388,
      "step": 3521
    },
    {
      "epoch": 1.630706114740932,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00020256957328385896,
      "loss": 0.6572,
      "step": 3522
    },
    {
      "epoch": 1.6311691882466155,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00020254174397031538,
      "loss": 0.6408,
      "step": 3523
    },
    {
      "epoch": 1.6316322617522991,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00020251391465677177,
      "loss": 0.6846,
      "step": 3524
    },
    {
      "epoch": 1.6320953352579826,
      "grad_norm": 0.453125,
      "learning_rate": 0.00020248608534322818,
      "loss": 0.6427,
      "step": 3525
    },
    {
      "epoch": 1.632558408763666,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020245825602968457,
      "loss": 0.6475,
      "step": 3526
    },
    {
      "epoch": 1.6330214822693496,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000202430426716141,
      "loss": 0.6616,
      "step": 3527
    },
    {
      "epoch": 1.6334845557750333,
      "grad_norm": 0.59375,
      "learning_rate": 0.00020240259740259738,
      "loss": 0.625,
      "step": 3528
    },
    {
      "epoch": 1.6339476292807165,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0002023747680890538,
      "loss": 0.7644,
      "step": 3529
    },
    {
      "epoch": 1.6344107027864,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020234693877551019,
      "loss": 0.6469,
      "step": 3530
    },
    {
      "epoch": 1.6348737762920837,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0002023191094619666,
      "loss": 0.6402,
      "step": 3531
    },
    {
      "epoch": 1.6353368497977672,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000202291280148423,
      "loss": 0.748,
      "step": 3532
    },
    {
      "epoch": 1.6357999233034506,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0002022634508348794,
      "loss": 0.7248,
      "step": 3533
    },
    {
      "epoch": 1.6362629968091342,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002022356215213358,
      "loss": 0.5886,
      "step": 3534
    },
    {
      "epoch": 1.6367260703148176,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00020220779220779216,
      "loss": 0.6643,
      "step": 3535
    },
    {
      "epoch": 1.637189143820501,
      "grad_norm": 0.578125,
      "learning_rate": 0.0002021799628942486,
      "loss": 0.6821,
      "step": 3536
    },
    {
      "epoch": 1.6376522173261847,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00020215213358070497,
      "loss": 0.5984,
      "step": 3537
    },
    {
      "epoch": 1.6381152908318681,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00020212430426716138,
      "loss": 0.5992,
      "step": 3538
    },
    {
      "epoch": 1.6385783643375516,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020209647495361777,
      "loss": 0.5884,
      "step": 3539
    },
    {
      "epoch": 1.6390414378432352,
      "grad_norm": 0.4375,
      "learning_rate": 0.0002020686456400742,
      "loss": 0.5935,
      "step": 3540
    },
    {
      "epoch": 1.6395045113489186,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00020204081632653058,
      "loss": 0.5794,
      "step": 3541
    },
    {
      "epoch": 1.639967584854602,
      "grad_norm": 0.546875,
      "learning_rate": 0.000202012987012987,
      "loss": 0.63,
      "step": 3542
    },
    {
      "epoch": 1.6404306583602857,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002019851576994434,
      "loss": 0.6326,
      "step": 3543
    },
    {
      "epoch": 1.6408937318659693,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0002019573283858998,
      "loss": 0.6208,
      "step": 3544
    },
    {
      "epoch": 1.6413568053716525,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0002019294990723562,
      "loss": 0.7453,
      "step": 3545
    },
    {
      "epoch": 1.6418198788773362,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0002019016697588126,
      "loss": 0.6636,
      "step": 3546
    },
    {
      "epoch": 1.6422829523830198,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000201873840445269,
      "loss": 0.6327,
      "step": 3547
    },
    {
      "epoch": 1.6427460258887032,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00020184601113172542,
      "loss": 0.6078,
      "step": 3548
    },
    {
      "epoch": 1.6432090993943866,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0002018181818181818,
      "loss": 0.5861,
      "step": 3549
    },
    {
      "epoch": 1.6436721729000703,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0002017903525046382,
      "loss": 0.6679,
      "step": 3550
    },
    {
      "epoch": 1.6441352464057537,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0002017625231910946,
      "loss": 0.6094,
      "step": 3551
    },
    {
      "epoch": 1.6445983199114371,
      "grad_norm": 0.61328125,
      "learning_rate": 0.000201734693877551,
      "loss": 0.7065,
      "step": 3552
    },
    {
      "epoch": 1.6450613934171208,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020170686456400742,
      "loss": 0.6421,
      "step": 3553
    },
    {
      "epoch": 1.6455244669228042,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020167903525046378,
      "loss": 0.6393,
      "step": 3554
    },
    {
      "epoch": 1.6459875404284876,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00020165120593692022,
      "loss": 0.6536,
      "step": 3555
    },
    {
      "epoch": 1.6464506139341712,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0002016233766233766,
      "loss": 0.546,
      "step": 3556
    },
    {
      "epoch": 1.6469136874398547,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000201595547309833,
      "loss": 0.5751,
      "step": 3557
    },
    {
      "epoch": 1.647376760945538,
      "grad_norm": 0.734375,
      "learning_rate": 0.0002015677179962894,
      "loss": 0.7147,
      "step": 3558
    },
    {
      "epoch": 1.6478398344512217,
      "grad_norm": 0.53125,
      "learning_rate": 0.0002015398886827458,
      "loss": 0.5784,
      "step": 3559
    },
    {
      "epoch": 1.6483029079569054,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0002015120593692022,
      "loss": 0.6596,
      "step": 3560
    },
    {
      "epoch": 1.6487659814625886,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00020148423005565862,
      "loss": 0.5789,
      "step": 3561
    },
    {
      "epoch": 1.6492290549682722,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000201456400742115,
      "loss": 0.6565,
      "step": 3562
    },
    {
      "epoch": 1.6496921284739559,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0002014285714285714,
      "loss": 0.5659,
      "step": 3563
    },
    {
      "epoch": 1.6501552019796393,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0002014007421150278,
      "loss": 0.6817,
      "step": 3564
    },
    {
      "epoch": 1.6506182754853227,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0002013729128014842,
      "loss": 0.6263,
      "step": 3565
    },
    {
      "epoch": 1.6510813489910063,
      "grad_norm": 0.65625,
      "learning_rate": 0.00020134508348794062,
      "loss": 0.7427,
      "step": 3566
    },
    {
      "epoch": 1.6515444224966898,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000201317254174397,
      "loss": 0.6242,
      "step": 3567
    },
    {
      "epoch": 1.6520074960023732,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00020128942486085342,
      "loss": 0.5671,
      "step": 3568
    },
    {
      "epoch": 1.6524705695080568,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00020126159554730981,
      "loss": 0.6039,
      "step": 3569
    },
    {
      "epoch": 1.6529336430137402,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020123376623376623,
      "loss": 0.6152,
      "step": 3570
    },
    {
      "epoch": 1.6533967165194237,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020120593692022262,
      "loss": 0.6299,
      "step": 3571
    },
    {
      "epoch": 1.6538597900251073,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00020117810760667904,
      "loss": 0.6246,
      "step": 3572
    },
    {
      "epoch": 1.6543228635307907,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020115027829313543,
      "loss": 0.6251,
      "step": 3573
    },
    {
      "epoch": 1.6547859370364741,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00020112244897959184,
      "loss": 0.61,
      "step": 3574
    },
    {
      "epoch": 1.6552490105421578,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0002010946196660482,
      "loss": 0.5666,
      "step": 3575
    },
    {
      "epoch": 1.6557120840478414,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00020106679035250462,
      "loss": 0.6891,
      "step": 3576
    },
    {
      "epoch": 1.6561751575535246,
      "grad_norm": 0.453125,
      "learning_rate": 0.000201038961038961,
      "loss": 0.5654,
      "step": 3577
    },
    {
      "epoch": 1.6566382310592083,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002010111317254174,
      "loss": 0.663,
      "step": 3578
    },
    {
      "epoch": 1.657101304564892,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020098330241187382,
      "loss": 0.6554,
      "step": 3579
    },
    {
      "epoch": 1.6575643780705753,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0002009554730983302,
      "loss": 0.6885,
      "step": 3580
    },
    {
      "epoch": 1.6580274515762587,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00020092764378478663,
      "loss": 0.5932,
      "step": 3581
    },
    {
      "epoch": 1.6584905250819424,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020089981447124301,
      "loss": 0.6025,
      "step": 3582
    },
    {
      "epoch": 1.6589535985876258,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020087198515769943,
      "loss": 0.5819,
      "step": 3583
    },
    {
      "epoch": 1.6594166720933092,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00020084415584415582,
      "loss": 0.7399,
      "step": 3584
    },
    {
      "epoch": 1.6598797455989929,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020081632653061224,
      "loss": 0.6769,
      "step": 3585
    },
    {
      "epoch": 1.6603428191046763,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020078849721706863,
      "loss": 0.685,
      "step": 3586
    },
    {
      "epoch": 1.6608058926103597,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020076066790352504,
      "loss": 0.633,
      "step": 3587
    },
    {
      "epoch": 1.6612689661160434,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020073283858998143,
      "loss": 0.574,
      "step": 3588
    },
    {
      "epoch": 1.6617320396217268,
      "grad_norm": 0.609375,
      "learning_rate": 0.00020070500927643785,
      "loss": 0.6799,
      "step": 3589
    },
    {
      "epoch": 1.6621951131274102,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00020067717996289424,
      "loss": 0.6466,
      "step": 3590
    },
    {
      "epoch": 1.6626581866330938,
      "grad_norm": 0.65625,
      "learning_rate": 0.0002006493506493506,
      "loss": 0.6582,
      "step": 3591
    },
    {
      "epoch": 1.6631212601387775,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00020062152133580705,
      "loss": 0.5941,
      "step": 3592
    },
    {
      "epoch": 1.663584333644461,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0002005936920222634,
      "loss": 0.6078,
      "step": 3593
    },
    {
      "epoch": 1.6640474071501443,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00020056586270871983,
      "loss": 0.6646,
      "step": 3594
    },
    {
      "epoch": 1.664510480655828,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020053803339517622,
      "loss": 0.6656,
      "step": 3595
    },
    {
      "epoch": 1.6649735541615114,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00020051020408163263,
      "loss": 0.6325,
      "step": 3596
    },
    {
      "epoch": 1.6654366276671948,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00020048237476808902,
      "loss": 0.7552,
      "step": 3597
    },
    {
      "epoch": 1.6658997011728784,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00020045454545454544,
      "loss": 0.6464,
      "step": 3598
    },
    {
      "epoch": 1.6663627746785619,
      "grad_norm": 0.546875,
      "learning_rate": 0.00020042671614100183,
      "loss": 0.6653,
      "step": 3599
    },
    {
      "epoch": 1.6668258481842453,
      "grad_norm": 0.703125,
      "learning_rate": 0.00020039888682745824,
      "loss": 0.7017,
      "step": 3600
    },
    {
      "epoch": 1.667288921689929,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00020037105751391463,
      "loss": 0.6565,
      "step": 3601
    },
    {
      "epoch": 1.6677519951956123,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00020034322820037105,
      "loss": 0.6562,
      "step": 3602
    },
    {
      "epoch": 1.6682150687012958,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00020031539888682744,
      "loss": 0.6394,
      "step": 3603
    },
    {
      "epoch": 1.6686781422069794,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00020028756957328386,
      "loss": 0.6395,
      "step": 3604
    },
    {
      "epoch": 1.669141215712663,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020025974025974025,
      "loss": 0.668,
      "step": 3605
    },
    {
      "epoch": 1.6696042892183462,
      "grad_norm": 0.53125,
      "learning_rate": 0.00020023191094619664,
      "loss": 0.6352,
      "step": 3606
    },
    {
      "epoch": 1.67006736272403,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00020020408163265305,
      "loss": 0.6937,
      "step": 3607
    },
    {
      "epoch": 1.6705304362297135,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00020017625231910944,
      "loss": 0.6215,
      "step": 3608
    },
    {
      "epoch": 1.670993509735397,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020014842300556586,
      "loss": 0.6515,
      "step": 3609
    },
    {
      "epoch": 1.6714565832410804,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00020012059369202222,
      "loss": 0.679,
      "step": 3610
    },
    {
      "epoch": 1.671919656746764,
      "grad_norm": 0.578125,
      "learning_rate": 0.00020009276437847867,
      "loss": 0.7503,
      "step": 3611
    },
    {
      "epoch": 1.6723827302524474,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00020006493506493503,
      "loss": 0.5898,
      "step": 3612
    },
    {
      "epoch": 1.6728458037581309,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00020003710575139144,
      "loss": 0.6365,
      "step": 3613
    },
    {
      "epoch": 1.6733088772638145,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00020000927643784783,
      "loss": 0.585,
      "step": 3614
    },
    {
      "epoch": 1.673771950769498,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019998144712430425,
      "loss": 0.6338,
      "step": 3615
    },
    {
      "epoch": 1.6742350242751813,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019995361781076064,
      "loss": 0.5893,
      "step": 3616
    },
    {
      "epoch": 1.674698097780865,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019992578849721706,
      "loss": 0.558,
      "step": 3617
    },
    {
      "epoch": 1.6751611712865484,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019989795918367345,
      "loss": 0.6853,
      "step": 3618
    },
    {
      "epoch": 1.6756242447922318,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019987012987012986,
      "loss": 0.6775,
      "step": 3619
    },
    {
      "epoch": 1.6760873182979155,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019984230055658625,
      "loss": 0.6508,
      "step": 3620
    },
    {
      "epoch": 1.676550391803599,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019981447124304264,
      "loss": 0.6043,
      "step": 3621
    },
    {
      "epoch": 1.6770134653092823,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019978664192949906,
      "loss": 0.5504,
      "step": 3622
    },
    {
      "epoch": 1.677476538814966,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019975881261595545,
      "loss": 0.6992,
      "step": 3623
    },
    {
      "epoch": 1.6779396123206496,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019973098330241187,
      "loss": 0.6185,
      "step": 3624
    },
    {
      "epoch": 1.678402685826333,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019970315398886826,
      "loss": 0.6668,
      "step": 3625
    },
    {
      "epoch": 1.6788657593320164,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019967532467532467,
      "loss": 0.5934,
      "step": 3626
    },
    {
      "epoch": 1.6793288328377,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019964749536178106,
      "loss": 0.6425,
      "step": 3627
    },
    {
      "epoch": 1.6797919063433835,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019961966604823748,
      "loss": 0.5316,
      "step": 3628
    },
    {
      "epoch": 1.680254979849067,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019959183673469387,
      "loss": 0.5617,
      "step": 3629
    },
    {
      "epoch": 1.6807180533547506,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019956400742115029,
      "loss": 0.7504,
      "step": 3630
    },
    {
      "epoch": 1.681181126860434,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019953617810760665,
      "loss": 0.6626,
      "step": 3631
    },
    {
      "epoch": 1.6816442003661174,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019950834879406306,
      "loss": 0.6388,
      "step": 3632
    },
    {
      "epoch": 1.682107273871801,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019948051948051945,
      "loss": 0.6035,
      "step": 3633
    },
    {
      "epoch": 1.6825703473774845,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019945269016697584,
      "loss": 0.6523,
      "step": 3634
    },
    {
      "epoch": 1.6830334208831679,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00019942486085343226,
      "loss": 0.6252,
      "step": 3635
    },
    {
      "epoch": 1.6834964943888515,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019939703153988865,
      "loss": 0.5985,
      "step": 3636
    },
    {
      "epoch": 1.6839595678945352,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019936920222634507,
      "loss": 0.7364,
      "step": 3637
    },
    {
      "epoch": 1.6844226414002184,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019934137291280146,
      "loss": 0.6071,
      "step": 3638
    },
    {
      "epoch": 1.684885714905902,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019931354359925787,
      "loss": 0.5329,
      "step": 3639
    },
    {
      "epoch": 1.6853487884115856,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019928571428571426,
      "loss": 0.6273,
      "step": 3640
    },
    {
      "epoch": 1.685811861917269,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019925788497217068,
      "loss": 0.7006,
      "step": 3641
    },
    {
      "epoch": 1.6862749354229525,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019923005565862707,
      "loss": 0.5953,
      "step": 3642
    },
    {
      "epoch": 1.6867380089286361,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019920222634508349,
      "loss": 0.6731,
      "step": 3643
    },
    {
      "epoch": 1.6872010824343195,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019917439703153988,
      "loss": 0.6237,
      "step": 3644
    },
    {
      "epoch": 1.687664155940003,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001991465677179963,
      "loss": 0.6875,
      "step": 3645
    },
    {
      "epoch": 1.6881272294456866,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019911873840445268,
      "loss": 0.6439,
      "step": 3646
    },
    {
      "epoch": 1.68859030295137,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001990909090909091,
      "loss": 0.6917,
      "step": 3647
    },
    {
      "epoch": 1.6890533764570534,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001990630797773655,
      "loss": 0.6083,
      "step": 3648
    },
    {
      "epoch": 1.689516449962737,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00019903525046382185,
      "loss": 0.6487,
      "step": 3649
    },
    {
      "epoch": 1.6899795234684205,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019900742115027827,
      "loss": 0.6802,
      "step": 3650
    },
    {
      "epoch": 1.690442596974104,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019897959183673466,
      "loss": 0.6582,
      "step": 3651
    },
    {
      "epoch": 1.6909056704797876,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019895176252319107,
      "loss": 0.5988,
      "step": 3652
    },
    {
      "epoch": 1.6913687439854712,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019892393320964746,
      "loss": 0.6299,
      "step": 3653
    },
    {
      "epoch": 1.6918318174911544,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019889610389610388,
      "loss": 0.6234,
      "step": 3654
    },
    {
      "epoch": 1.692294890996838,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019886827458256027,
      "loss": 0.6871,
      "step": 3655
    },
    {
      "epoch": 1.6927579645025217,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019884044526901669,
      "loss": 0.682,
      "step": 3656
    },
    {
      "epoch": 1.6932210380082051,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019881261595547308,
      "loss": 0.6832,
      "step": 3657
    },
    {
      "epoch": 1.6936841115138885,
      "grad_norm": 0.5,
      "learning_rate": 0.0001987847866419295,
      "loss": 0.5799,
      "step": 3658
    },
    {
      "epoch": 1.6941471850195722,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019875695732838588,
      "loss": 0.6489,
      "step": 3659
    },
    {
      "epoch": 1.6946102585252556,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001987291280148423,
      "loss": 0.6998,
      "step": 3660
    },
    {
      "epoch": 1.695073332030939,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001987012987012987,
      "loss": 0.6938,
      "step": 3661
    },
    {
      "epoch": 1.6955364055366227,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019867346938775508,
      "loss": 0.6057,
      "step": 3662
    },
    {
      "epoch": 1.695999479042306,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001986456400742115,
      "loss": 0.6897,
      "step": 3663
    },
    {
      "epoch": 1.6964625525479895,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019861781076066788,
      "loss": 0.6743,
      "step": 3664
    },
    {
      "epoch": 1.6969256260536731,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001985899814471243,
      "loss": 0.5825,
      "step": 3665
    },
    {
      "epoch": 1.6973886995593566,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019856215213358066,
      "loss": 0.5859,
      "step": 3666
    },
    {
      "epoch": 1.69785177306504,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001985343228200371,
      "loss": 0.6491,
      "step": 3667
    },
    {
      "epoch": 1.6983148465707236,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019850649350649347,
      "loss": 0.6577,
      "step": 3668
    },
    {
      "epoch": 1.6987779200764073,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019847866419294989,
      "loss": 0.6765,
      "step": 3669
    },
    {
      "epoch": 1.6992409935820907,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019845083487940628,
      "loss": 0.7325,
      "step": 3670
    },
    {
      "epoch": 1.699704067087774,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001984230055658627,
      "loss": 0.6872,
      "step": 3671
    },
    {
      "epoch": 1.7001671405934577,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019839517625231908,
      "loss": 0.6151,
      "step": 3672
    },
    {
      "epoch": 1.7006302140991412,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001983673469387755,
      "loss": 0.6149,
      "step": 3673
    },
    {
      "epoch": 1.7010932876048246,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001983395176252319,
      "loss": 0.6942,
      "step": 3674
    },
    {
      "epoch": 1.7015563611105082,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001983116883116883,
      "loss": 0.6753,
      "step": 3675
    },
    {
      "epoch": 1.7020194346161917,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001982838589981447,
      "loss": 0.6096,
      "step": 3676
    },
    {
      "epoch": 1.702482508121875,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019825602968460108,
      "loss": 0.7082,
      "step": 3677
    },
    {
      "epoch": 1.7029455816275587,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001982282003710575,
      "loss": 0.683,
      "step": 3678
    },
    {
      "epoch": 1.7034086551332421,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001982003710575139,
      "loss": 0.6109,
      "step": 3679
    },
    {
      "epoch": 1.7038717286389256,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001981725417439703,
      "loss": 0.525,
      "step": 3680
    },
    {
      "epoch": 1.7043348021446092,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001981447124304267,
      "loss": 0.5526,
      "step": 3681
    },
    {
      "epoch": 1.7047978756502928,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019811688311688311,
      "loss": 0.6759,
      "step": 3682
    },
    {
      "epoch": 1.705260949155976,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001980890538033395,
      "loss": 0.7044,
      "step": 3683
    },
    {
      "epoch": 1.7057240226616597,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019806122448979592,
      "loss": 0.5499,
      "step": 3684
    },
    {
      "epoch": 1.7061870961673433,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001980333951762523,
      "loss": 0.5935,
      "step": 3685
    },
    {
      "epoch": 1.7066501696730267,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019800556586270873,
      "loss": 0.6243,
      "step": 3686
    },
    {
      "epoch": 1.7071132431787102,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001979777365491651,
      "loss": 0.5917,
      "step": 3687
    },
    {
      "epoch": 1.7075763166843938,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001979499072356215,
      "loss": 0.6004,
      "step": 3688
    },
    {
      "epoch": 1.7080393901900772,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001979220779220779,
      "loss": 0.6805,
      "step": 3689
    },
    {
      "epoch": 1.7085024636957606,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019789424860853428,
      "loss": 0.6651,
      "step": 3690
    },
    {
      "epoch": 1.7089655372014443,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001978664192949907,
      "loss": 0.5995,
      "step": 3691
    },
    {
      "epoch": 1.7094286107071277,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001978385899814471,
      "loss": 0.7575,
      "step": 3692
    },
    {
      "epoch": 1.7098916842128111,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001978107606679035,
      "loss": 0.5844,
      "step": 3693
    },
    {
      "epoch": 1.7103547577184948,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001977829313543599,
      "loss": 0.6378,
      "step": 3694
    },
    {
      "epoch": 1.7108178312241782,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019775510204081631,
      "loss": 0.5879,
      "step": 3695
    },
    {
      "epoch": 1.7112809047298616,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001977272727272727,
      "loss": 0.6824,
      "step": 3696
    },
    {
      "epoch": 1.7117439782355452,
      "grad_norm": 0.5,
      "learning_rate": 0.00019769944341372912,
      "loss": 0.5671,
      "step": 3697
    },
    {
      "epoch": 1.712207051741229,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001976716141001855,
      "loss": 0.6472,
      "step": 3698
    },
    {
      "epoch": 1.712670125246912,
      "grad_norm": 0.625,
      "learning_rate": 0.00019764378478664193,
      "loss": 0.6786,
      "step": 3699
    },
    {
      "epoch": 1.7131331987525957,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019761595547309832,
      "loss": 0.5768,
      "step": 3700
    },
    {
      "epoch": 1.7135962722582794,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019758812615955473,
      "loss": 0.6509,
      "step": 3701
    },
    {
      "epoch": 1.7140593457639628,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019756029684601112,
      "loss": 0.6502,
      "step": 3702
    },
    {
      "epoch": 1.7145224192696462,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019753246753246754,
      "loss": 0.6932,
      "step": 3703
    },
    {
      "epoch": 1.7149854927753299,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019750463821892393,
      "loss": 0.676,
      "step": 3704
    },
    {
      "epoch": 1.7154485662810133,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001974768089053803,
      "loss": 0.6079,
      "step": 3705
    },
    {
      "epoch": 1.7159116397866967,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001974489795918367,
      "loss": 0.6842,
      "step": 3706
    },
    {
      "epoch": 1.7163747132923803,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001974211502782931,
      "loss": 0.6319,
      "step": 3707
    },
    {
      "epoch": 1.7168377867980638,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019739332096474951,
      "loss": 0.6004,
      "step": 3708
    },
    {
      "epoch": 1.7173008603037472,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001973654916512059,
      "loss": 0.6502,
      "step": 3709
    },
    {
      "epoch": 1.7177639338094308,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019733766233766232,
      "loss": 0.657,
      "step": 3710
    },
    {
      "epoch": 1.7182270073151142,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001973098330241187,
      "loss": 0.6718,
      "step": 3711
    },
    {
      "epoch": 1.7186900808207977,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019728200371057513,
      "loss": 0.6343,
      "step": 3712
    },
    {
      "epoch": 1.7191531543264813,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019725417439703152,
      "loss": 0.7229,
      "step": 3713
    },
    {
      "epoch": 1.719616227832165,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019722634508348793,
      "loss": 0.5785,
      "step": 3714
    },
    {
      "epoch": 1.7200793013378481,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019719851576994432,
      "loss": 0.626,
      "step": 3715
    },
    {
      "epoch": 1.7205423748435318,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019717068645640074,
      "loss": 0.617,
      "step": 3716
    },
    {
      "epoch": 1.7210054483492154,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019714285714285713,
      "loss": 0.668,
      "step": 3717
    },
    {
      "epoch": 1.7214685218548988,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019711502782931352,
      "loss": 0.6874,
      "step": 3718
    },
    {
      "epoch": 1.7219315953605823,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019708719851576994,
      "loss": 0.6565,
      "step": 3719
    },
    {
      "epoch": 1.722394668866266,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019705936920222633,
      "loss": 0.5926,
      "step": 3720
    },
    {
      "epoch": 1.7228577423719493,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019703153988868274,
      "loss": 0.6112,
      "step": 3721
    },
    {
      "epoch": 1.7233208158776328,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001970037105751391,
      "loss": 0.6654,
      "step": 3722
    },
    {
      "epoch": 1.7237838893833164,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019697588126159555,
      "loss": 0.6264,
      "step": 3723
    },
    {
      "epoch": 1.7242469628889998,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001969480519480519,
      "loss": 0.7187,
      "step": 3724
    },
    {
      "epoch": 1.7247100363946832,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019692022263450833,
      "loss": 0.6785,
      "step": 3725
    },
    {
      "epoch": 1.7251731099003669,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019689239332096472,
      "loss": 0.6142,
      "step": 3726
    },
    {
      "epoch": 1.7256361834060503,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019686456400742113,
      "loss": 0.5743,
      "step": 3727
    },
    {
      "epoch": 1.7260992569117337,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019683673469387752,
      "loss": 0.632,
      "step": 3728
    },
    {
      "epoch": 1.7265623304174174,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019680890538033394,
      "loss": 0.6252,
      "step": 3729
    },
    {
      "epoch": 1.727025403923101,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019678107606679033,
      "loss": 0.6666,
      "step": 3730
    },
    {
      "epoch": 1.7274884774287842,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019675324675324675,
      "loss": 0.5947,
      "step": 3731
    },
    {
      "epoch": 1.7279515509344678,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019672541743970314,
      "loss": 0.6415,
      "step": 3732
    },
    {
      "epoch": 1.7284146244401515,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019669758812615953,
      "loss": 0.6457,
      "step": 3733
    },
    {
      "epoch": 1.728877697945835,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019666975881261594,
      "loss": 0.6756,
      "step": 3734
    },
    {
      "epoch": 1.7293407714515183,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019664192949907233,
      "loss": 0.7015,
      "step": 3735
    },
    {
      "epoch": 1.729803844957202,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019661410018552875,
      "loss": 0.6711,
      "step": 3736
    },
    {
      "epoch": 1.7302669184628854,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019658627087198514,
      "loss": 0.66,
      "step": 3737
    },
    {
      "epoch": 1.7307299919685688,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019655844155844156,
      "loss": 0.6576,
      "step": 3738
    },
    {
      "epoch": 1.7311930654742524,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019653061224489794,
      "loss": 0.6148,
      "step": 3739
    },
    {
      "epoch": 1.7316561389799359,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019650278293135436,
      "loss": 0.5793,
      "step": 3740
    },
    {
      "epoch": 1.7321192124856193,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019647495361781075,
      "loss": 0.4935,
      "step": 3741
    },
    {
      "epoch": 1.732582285991303,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019644712430426717,
      "loss": 0.61,
      "step": 3742
    },
    {
      "epoch": 1.7330453594969866,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019641929499072353,
      "loss": 0.6543,
      "step": 3743
    },
    {
      "epoch": 1.7335084330026698,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019639146567717995,
      "loss": 0.6431,
      "step": 3744
    },
    {
      "epoch": 1.7339715065083534,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019636363636363634,
      "loss": 0.6564,
      "step": 3745
    },
    {
      "epoch": 1.734434580014037,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019633580705009275,
      "loss": 0.608,
      "step": 3746
    },
    {
      "epoch": 1.7348976535197205,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019630797773654914,
      "loss": 0.6139,
      "step": 3747
    },
    {
      "epoch": 1.735360727025404,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019628014842300553,
      "loss": 0.6233,
      "step": 3748
    },
    {
      "epoch": 1.7358238005310875,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019625231910946195,
      "loss": 0.6196,
      "step": 3749
    },
    {
      "epoch": 1.736286874036771,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019622448979591834,
      "loss": 0.6672,
      "step": 3750
    },
    {
      "epoch": 1.7367499475424544,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019619666048237476,
      "loss": 0.5701,
      "step": 3751
    },
    {
      "epoch": 1.737213021048138,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019616883116883115,
      "loss": 0.585,
      "step": 3752
    },
    {
      "epoch": 1.7376760945538214,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019614100185528756,
      "loss": 0.6305,
      "step": 3753
    },
    {
      "epoch": 1.7381391680595049,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019611317254174395,
      "loss": 0.5834,
      "step": 3754
    },
    {
      "epoch": 1.7386022415651885,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019608534322820037,
      "loss": 0.7076,
      "step": 3755
    },
    {
      "epoch": 1.739065315070872,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019605751391465676,
      "loss": 0.5109,
      "step": 3756
    },
    {
      "epoch": 1.7395283885765553,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019602968460111317,
      "loss": 0.6341,
      "step": 3757
    },
    {
      "epoch": 1.739991462082239,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019600185528756956,
      "loss": 0.6518,
      "step": 3758
    },
    {
      "epoch": 1.7404545355879226,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019597402597402598,
      "loss": 0.5703,
      "step": 3759
    },
    {
      "epoch": 1.7409176090936058,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019594619666048237,
      "loss": 0.592,
      "step": 3760
    },
    {
      "epoch": 1.7413806825992895,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019591836734693873,
      "loss": 0.6027,
      "step": 3761
    },
    {
      "epoch": 1.741843756104973,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019589053803339515,
      "loss": 0.7357,
      "step": 3762
    },
    {
      "epoch": 1.7423068296106565,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019586270871985154,
      "loss": 0.6845,
      "step": 3763
    },
    {
      "epoch": 1.74276990311634,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019583487940630796,
      "loss": 0.608,
      "step": 3764
    },
    {
      "epoch": 1.7432329766220236,
      "grad_norm": 0.5,
      "learning_rate": 0.00019580705009276435,
      "loss": 0.6501,
      "step": 3765
    },
    {
      "epoch": 1.743696050127707,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019577922077922076,
      "loss": 0.6826,
      "step": 3766
    },
    {
      "epoch": 1.7441591236333904,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019575139146567715,
      "loss": 0.6589,
      "step": 3767
    },
    {
      "epoch": 1.744622197139074,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019572356215213357,
      "loss": 0.6587,
      "step": 3768
    },
    {
      "epoch": 1.7450852706447575,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019569573283858996,
      "loss": 0.5305,
      "step": 3769
    },
    {
      "epoch": 1.745548344150441,
      "grad_norm": 0.5,
      "learning_rate": 0.00019566790352504637,
      "loss": 0.6098,
      "step": 3770
    },
    {
      "epoch": 1.7460114176561246,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019564007421150276,
      "loss": 0.5987,
      "step": 3771
    },
    {
      "epoch": 1.746474491161808,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019561224489795918,
      "loss": 0.6538,
      "step": 3772
    },
    {
      "epoch": 1.7469375646674914,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019558441558441557,
      "loss": 0.6491,
      "step": 3773
    },
    {
      "epoch": 1.747400638173175,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000195556586270872,
      "loss": 0.5938,
      "step": 3774
    },
    {
      "epoch": 1.7478637116788587,
      "grad_norm": 3.75,
      "learning_rate": 0.00019552875695732838,
      "loss": 0.6439,
      "step": 3775
    },
    {
      "epoch": 1.7483267851845419,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019550092764378477,
      "loss": 0.5826,
      "step": 3776
    },
    {
      "epoch": 1.7487898586902255,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019547309833024118,
      "loss": 0.6907,
      "step": 3777
    },
    {
      "epoch": 1.7492529321959092,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019544526901669755,
      "loss": 0.6315,
      "step": 3778
    },
    {
      "epoch": 1.7497160057015926,
      "grad_norm": 0.431640625,
      "learning_rate": 0.000195417439703154,
      "loss": 0.6038,
      "step": 3779
    },
    {
      "epoch": 1.750179079207276,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019538961038961035,
      "loss": 0.6336,
      "step": 3780
    },
    {
      "epoch": 1.7506421527129596,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019536178107606677,
      "loss": 0.6924,
      "step": 3781
    },
    {
      "epoch": 1.751105226218643,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019533395176252316,
      "loss": 0.799,
      "step": 3782
    },
    {
      "epoch": 1.7515682997243265,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019530612244897958,
      "loss": 0.649,
      "step": 3783
    },
    {
      "epoch": 1.7520313732300101,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019527829313543596,
      "loss": 0.5584,
      "step": 3784
    },
    {
      "epoch": 1.7524944467356935,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019525046382189238,
      "loss": 0.5858,
      "step": 3785
    },
    {
      "epoch": 1.752957520241377,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019522263450834877,
      "loss": 0.6189,
      "step": 3786
    },
    {
      "epoch": 1.7534205937470606,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001951948051948052,
      "loss": 0.5503,
      "step": 3787
    },
    {
      "epoch": 1.753883667252744,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019516697588126158,
      "loss": 0.5975,
      "step": 3788
    },
    {
      "epoch": 1.7543467407584274,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019513914656771797,
      "loss": 0.7249,
      "step": 3789
    },
    {
      "epoch": 1.754809814264111,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019511131725417438,
      "loss": 0.628,
      "step": 3790
    },
    {
      "epoch": 1.7552728877697947,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019508348794063077,
      "loss": 0.635,
      "step": 3791
    },
    {
      "epoch": 1.755735961275478,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001950556586270872,
      "loss": 0.6776,
      "step": 3792
    },
    {
      "epoch": 1.7561990347811616,
      "grad_norm": 0.5,
      "learning_rate": 0.00019502782931354358,
      "loss": 0.6198,
      "step": 3793
    },
    {
      "epoch": 1.7566621082868452,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000195,
      "loss": 0.6814,
      "step": 3794
    },
    {
      "epoch": 1.7571251817925286,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019497217068645639,
      "loss": 0.6086,
      "step": 3795
    },
    {
      "epoch": 1.757588255298212,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001949443413729128,
      "loss": 0.6687,
      "step": 3796
    },
    {
      "epoch": 1.7580513288038957,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019491651205936917,
      "loss": 0.6214,
      "step": 3797
    },
    {
      "epoch": 1.7585144023095791,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001948886827458256,
      "loss": 0.678,
      "step": 3798
    },
    {
      "epoch": 1.7589774758152625,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00019486085343228197,
      "loss": 0.5438,
      "step": 3799
    },
    {
      "epoch": 1.7594405493209462,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001948330241187384,
      "loss": 0.7087,
      "step": 3800
    },
    {
      "epoch": 1.7599036228266296,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019480519480519478,
      "loss": 0.6896,
      "step": 3801
    },
    {
      "epoch": 1.760366696332313,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001947773654916512,
      "loss": 0.6787,
      "step": 3802
    },
    {
      "epoch": 1.7608297698379967,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019474953617810758,
      "loss": 0.617,
      "step": 3803
    },
    {
      "epoch": 1.76129284334368,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019472170686456397,
      "loss": 0.6345,
      "step": 3804
    },
    {
      "epoch": 1.7617559168493635,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001946938775510204,
      "loss": 0.6492,
      "step": 3805
    },
    {
      "epoch": 1.7622189903550471,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019466604823747678,
      "loss": 0.6251,
      "step": 3806
    },
    {
      "epoch": 1.7626820638607308,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001946382189239332,
      "loss": 0.6245,
      "step": 3807
    },
    {
      "epoch": 1.763145137366414,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001946103896103896,
      "loss": 0.7084,
      "step": 3808
    },
    {
      "epoch": 1.7636082108720976,
      "grad_norm": 0.490234375,
      "learning_rate": 0.000194582560296846,
      "loss": 0.6764,
      "step": 3809
    },
    {
      "epoch": 1.7640712843777813,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001945547309833024,
      "loss": 0.6,
      "step": 3810
    },
    {
      "epoch": 1.7645343578834647,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001945269016697588,
      "loss": 0.6122,
      "step": 3811
    },
    {
      "epoch": 1.764997431389148,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001944990723562152,
      "loss": 0.6529,
      "step": 3812
    },
    {
      "epoch": 1.7654605048948318,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019447124304267162,
      "loss": 0.5764,
      "step": 3813
    },
    {
      "epoch": 1.7659235784005152,
      "grad_norm": 0.56640625,
      "learning_rate": 0.000194443413729128,
      "loss": 0.6681,
      "step": 3814
    },
    {
      "epoch": 1.7663866519061986,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019441558441558442,
      "loss": 0.6039,
      "step": 3815
    },
    {
      "epoch": 1.7668497254118822,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001943877551020408,
      "loss": 0.6141,
      "step": 3816
    },
    {
      "epoch": 1.7673127989175657,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019435992578849717,
      "loss": 0.6482,
      "step": 3817
    },
    {
      "epoch": 1.767775872423249,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001943320964749536,
      "loss": 0.5825,
      "step": 3818
    },
    {
      "epoch": 1.7682389459289327,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019430426716140998,
      "loss": 0.71,
      "step": 3819
    },
    {
      "epoch": 1.7687020194346164,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001942764378478664,
      "loss": 0.6165,
      "step": 3820
    },
    {
      "epoch": 1.7691650929402996,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001942486085343228,
      "loss": 0.6282,
      "step": 3821
    },
    {
      "epoch": 1.7696281664459832,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001942207792207792,
      "loss": 0.6614,
      "step": 3822
    },
    {
      "epoch": 1.7700912399516668,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001941929499072356,
      "loss": 0.6818,
      "step": 3823
    },
    {
      "epoch": 1.7705543134573503,
      "grad_norm": 0.46484375,
      "learning_rate": 0.000194165120593692,
      "loss": 0.5256,
      "step": 3824
    },
    {
      "epoch": 1.7710173869630337,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001941372912801484,
      "loss": 0.7059,
      "step": 3825
    },
    {
      "epoch": 1.7714804604687173,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019410946196660482,
      "loss": 0.7467,
      "step": 3826
    },
    {
      "epoch": 1.7719435339744007,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001940816326530612,
      "loss": 0.6801,
      "step": 3827
    },
    {
      "epoch": 1.7724066074800842,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019405380333951762,
      "loss": 0.6797,
      "step": 3828
    },
    {
      "epoch": 1.7728696809857678,
      "grad_norm": 0.5546875,
      "learning_rate": 0.000194025974025974,
      "loss": 0.6143,
      "step": 3829
    },
    {
      "epoch": 1.7733327544914512,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019399814471243043,
      "loss": 0.5868,
      "step": 3830
    },
    {
      "epoch": 1.7737958279971346,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019397031539888682,
      "loss": 0.5965,
      "step": 3831
    },
    {
      "epoch": 1.7742589015028183,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001939424860853432,
      "loss": 0.7358,
      "step": 3832
    },
    {
      "epoch": 1.7747219750085017,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00019391465677179962,
      "loss": 0.5754,
      "step": 3833
    },
    {
      "epoch": 1.7751850485141851,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000193886827458256,
      "loss": 0.6085,
      "step": 3834
    },
    {
      "epoch": 1.7756481220198688,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019385899814471243,
      "loss": 0.6052,
      "step": 3835
    },
    {
      "epoch": 1.7761111955255524,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001938311688311688,
      "loss": 0.725,
      "step": 3836
    },
    {
      "epoch": 1.7765742690312356,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001938033395176252,
      "loss": 0.6474,
      "step": 3837
    },
    {
      "epoch": 1.7770373425369193,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001937755102040816,
      "loss": 0.5857,
      "step": 3838
    },
    {
      "epoch": 1.777500416042603,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019374768089053802,
      "loss": 0.5954,
      "step": 3839
    },
    {
      "epoch": 1.7779634895482863,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001937198515769944,
      "loss": 0.6008,
      "step": 3840
    },
    {
      "epoch": 1.7784265630539697,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00019369202226345082,
      "loss": 0.6297,
      "step": 3841
    },
    {
      "epoch": 1.7788896365596534,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001936641929499072,
      "loss": 0.6489,
      "step": 3842
    },
    {
      "epoch": 1.7793527100653368,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019363636363636363,
      "loss": 0.6355,
      "step": 3843
    },
    {
      "epoch": 1.7798157835710202,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019360853432282002,
      "loss": 0.5872,
      "step": 3844
    },
    {
      "epoch": 1.7802788570767039,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001935807050092764,
      "loss": 0.5918,
      "step": 3845
    },
    {
      "epoch": 1.7807419305823873,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019355287569573283,
      "loss": 0.6386,
      "step": 3846
    },
    {
      "epoch": 1.7812050040880707,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019352504638218921,
      "loss": 0.6455,
      "step": 3847
    },
    {
      "epoch": 1.7816680775937543,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00019349721706864563,
      "loss": 0.6348,
      "step": 3848
    },
    {
      "epoch": 1.7821311510994378,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019346938775510202,
      "loss": 0.5706,
      "step": 3849
    },
    {
      "epoch": 1.7825942246051212,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019344155844155844,
      "loss": 0.6432,
      "step": 3850
    },
    {
      "epoch": 1.7830572981108048,
      "grad_norm": 0.625,
      "learning_rate": 0.00019341372912801483,
      "loss": 0.7596,
      "step": 3851
    },
    {
      "epoch": 1.7835203716164885,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019338589981447124,
      "loss": 0.7827,
      "step": 3852
    },
    {
      "epoch": 1.7839834451221717,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001933580705009276,
      "loss": 0.6571,
      "step": 3853
    },
    {
      "epoch": 1.7844465186278553,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019333024118738405,
      "loss": 0.5751,
      "step": 3854
    },
    {
      "epoch": 1.784909592133539,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001933024118738404,
      "loss": 0.6885,
      "step": 3855
    },
    {
      "epoch": 1.7853726656392224,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019327458256029683,
      "loss": 0.6439,
      "step": 3856
    },
    {
      "epoch": 1.7858357391449058,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019324675324675322,
      "loss": 0.5678,
      "step": 3857
    },
    {
      "epoch": 1.7862988126505894,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019321892393320964,
      "loss": 0.6549,
      "step": 3858
    },
    {
      "epoch": 1.7867618861562728,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019319109461966603,
      "loss": 0.6845,
      "step": 3859
    },
    {
      "epoch": 1.7872249596619563,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019316326530612242,
      "loss": 0.6404,
      "step": 3860
    },
    {
      "epoch": 1.78768803316764,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019313543599257883,
      "loss": 0.6624,
      "step": 3861
    },
    {
      "epoch": 1.7881511066733233,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019310760667903522,
      "loss": 0.6306,
      "step": 3862
    },
    {
      "epoch": 1.7886141801790068,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00019307977736549164,
      "loss": 0.555,
      "step": 3863
    },
    {
      "epoch": 1.7890772536846904,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019305194805194803,
      "loss": 0.5935,
      "step": 3864
    },
    {
      "epoch": 1.7895403271903738,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019302411873840444,
      "loss": 0.6648,
      "step": 3865
    },
    {
      "epoch": 1.7900034006960572,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019299628942486083,
      "loss": 0.6007,
      "step": 3866
    },
    {
      "epoch": 1.7904664742017409,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00019296846011131725,
      "loss": 0.6032,
      "step": 3867
    },
    {
      "epoch": 1.7909295477074245,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00019294063079777364,
      "loss": 0.5572,
      "step": 3868
    },
    {
      "epoch": 1.7913926212131077,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019291280148423006,
      "loss": 0.6566,
      "step": 3869
    },
    {
      "epoch": 1.7918556947187914,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019288497217068645,
      "loss": 0.6823,
      "step": 3870
    },
    {
      "epoch": 1.792318768224475,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019285714285714286,
      "loss": 0.6801,
      "step": 3871
    },
    {
      "epoch": 1.7927818417301584,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019282931354359925,
      "loss": 0.6659,
      "step": 3872
    },
    {
      "epoch": 1.7932449152358418,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019280148423005567,
      "loss": 0.6181,
      "step": 3873
    },
    {
      "epoch": 1.7937079887415255,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019277365491651203,
      "loss": 0.6937,
      "step": 3874
    },
    {
      "epoch": 1.794171062247209,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019274582560296842,
      "loss": 0.621,
      "step": 3875
    },
    {
      "epoch": 1.7946341357528923,
      "grad_norm": 0.625,
      "learning_rate": 0.00019271799628942484,
      "loss": 0.6751,
      "step": 3876
    },
    {
      "epoch": 1.795097209258576,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00019269016697588123,
      "loss": 0.5711,
      "step": 3877
    },
    {
      "epoch": 1.7955602827642594,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019266233766233764,
      "loss": 0.6272,
      "step": 3878
    },
    {
      "epoch": 1.7960233562699428,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019263450834879403,
      "loss": 0.6642,
      "step": 3879
    },
    {
      "epoch": 1.7964864297756264,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019260667903525045,
      "loss": 0.5815,
      "step": 3880
    },
    {
      "epoch": 1.7969495032813099,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019257884972170684,
      "loss": 0.5584,
      "step": 3881
    },
    {
      "epoch": 1.7974125767869933,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019255102040816326,
      "loss": 0.6019,
      "step": 3882
    },
    {
      "epoch": 1.797875650292677,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019252319109461965,
      "loss": 0.6503,
      "step": 3883
    },
    {
      "epoch": 1.7983387237983606,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019249536178107606,
      "loss": 0.5541,
      "step": 3884
    },
    {
      "epoch": 1.7988017973040438,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019246753246753245,
      "loss": 0.5605,
      "step": 3885
    },
    {
      "epoch": 1.7992648708097274,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019243970315398887,
      "loss": 0.6215,
      "step": 3886
    },
    {
      "epoch": 1.799727944315411,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019241187384044526,
      "loss": 0.6076,
      "step": 3887
    },
    {
      "epoch": 1.8001910178210945,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019238404452690165,
      "loss": 0.7043,
      "step": 3888
    },
    {
      "epoch": 1.800654091326778,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019235621521335807,
      "loss": 0.5938,
      "step": 3889
    },
    {
      "epoch": 1.8011171648324615,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019232838589981443,
      "loss": 0.6632,
      "step": 3890
    },
    {
      "epoch": 1.801580238338145,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019230055658627087,
      "loss": 0.6079,
      "step": 3891
    },
    {
      "epoch": 1.8020433118438284,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019227272727272723,
      "loss": 0.6095,
      "step": 3892
    },
    {
      "epoch": 1.802506385349512,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019224489795918365,
      "loss": 0.6778,
      "step": 3893
    },
    {
      "epoch": 1.8029694588551954,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00019221706864564004,
      "loss": 0.5564,
      "step": 3894
    },
    {
      "epoch": 1.8034325323608789,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019218923933209646,
      "loss": 0.6438,
      "step": 3895
    },
    {
      "epoch": 1.8038956058665625,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019216141001855285,
      "loss": 0.6946,
      "step": 3896
    },
    {
      "epoch": 1.8043586793722461,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019213358070500926,
      "loss": 0.6149,
      "step": 3897
    },
    {
      "epoch": 1.8048217528779293,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019210575139146565,
      "loss": 0.5982,
      "step": 3898
    },
    {
      "epoch": 1.805284826383613,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019207792207792207,
      "loss": 0.6729,
      "step": 3899
    },
    {
      "epoch": 1.8057478998892966,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019205009276437846,
      "loss": 0.5613,
      "step": 3900
    },
    {
      "epoch": 1.80621097339498,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019202226345083488,
      "loss": 0.6734,
      "step": 3901
    },
    {
      "epoch": 1.8066740469006635,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019199443413729127,
      "loss": 0.6141,
      "step": 3902
    },
    {
      "epoch": 1.807137120406347,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019196660482374766,
      "loss": 0.6709,
      "step": 3903
    },
    {
      "epoch": 1.8076001939120305,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019193877551020407,
      "loss": 0.673,
      "step": 3904
    },
    {
      "epoch": 1.808063267417714,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019191094619666046,
      "loss": 0.6731,
      "step": 3905
    },
    {
      "epoch": 1.8085263409233976,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019188311688311688,
      "loss": 0.6868,
      "step": 3906
    },
    {
      "epoch": 1.808989414429081,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019185528756957327,
      "loss": 0.6321,
      "step": 3907
    },
    {
      "epoch": 1.8094524879347644,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019182745825602969,
      "loss": 0.6608,
      "step": 3908
    },
    {
      "epoch": 1.809915561440448,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019179962894248605,
      "loss": 0.6497,
      "step": 3909
    },
    {
      "epoch": 1.8103786349461315,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001917717996289425,
      "loss": 0.642,
      "step": 3910
    },
    {
      "epoch": 1.810841708451815,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019174397031539885,
      "loss": 0.6434,
      "step": 3911
    },
    {
      "epoch": 1.8113047819574986,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019171614100185527,
      "loss": 0.7175,
      "step": 3912
    },
    {
      "epoch": 1.8117678554631822,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019168831168831166,
      "loss": 0.594,
      "step": 3913
    },
    {
      "epoch": 1.8122309289688654,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019166048237476808,
      "loss": 0.6655,
      "step": 3914
    },
    {
      "epoch": 1.812694002474549,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019163265306122447,
      "loss": 0.6049,
      "step": 3915
    },
    {
      "epoch": 1.8131570759802327,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019160482374768086,
      "loss": 0.7101,
      "step": 3916
    },
    {
      "epoch": 1.813620149485916,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019157699443413727,
      "loss": 0.6353,
      "step": 3917
    },
    {
      "epoch": 1.8140832229915995,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019154916512059366,
      "loss": 0.6007,
      "step": 3918
    },
    {
      "epoch": 1.8145462964972832,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019152133580705008,
      "loss": 0.6983,
      "step": 3919
    },
    {
      "epoch": 1.8150093700029666,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019149350649350647,
      "loss": 0.6508,
      "step": 3920
    },
    {
      "epoch": 1.81547244350865,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019146567717996289,
      "loss": 0.6843,
      "step": 3921
    },
    {
      "epoch": 1.8159355170143336,
      "grad_norm": 0.421875,
      "learning_rate": 0.00019143784786641928,
      "loss": 0.5587,
      "step": 3922
    },
    {
      "epoch": 1.816398590520017,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001914100185528757,
      "loss": 0.5861,
      "step": 3923
    },
    {
      "epoch": 1.8168616640257005,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019138218923933208,
      "loss": 0.5726,
      "step": 3924
    },
    {
      "epoch": 1.8173247375313841,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001913543599257885,
      "loss": 0.6466,
      "step": 3925
    },
    {
      "epoch": 1.8177878110370675,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001913265306122449,
      "loss": 0.6667,
      "step": 3926
    },
    {
      "epoch": 1.818250884542751,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001912987012987013,
      "loss": 0.6931,
      "step": 3927
    },
    {
      "epoch": 1.8187139580484346,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001912708719851577,
      "loss": 0.7043,
      "step": 3928
    },
    {
      "epoch": 1.8191770315541183,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001912430426716141,
      "loss": 0.673,
      "step": 3929
    },
    {
      "epoch": 1.8196401050598014,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019121521335807047,
      "loss": 0.6795,
      "step": 3930
    },
    {
      "epoch": 1.820103178565485,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019118738404452686,
      "loss": 0.6559,
      "step": 3931
    },
    {
      "epoch": 1.8205662520711687,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019115955473098328,
      "loss": 0.6006,
      "step": 3932
    },
    {
      "epoch": 1.8210293255768522,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019113172541743967,
      "loss": 0.786,
      "step": 3933
    },
    {
      "epoch": 1.8214923990825356,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019110389610389609,
      "loss": 0.5934,
      "step": 3934
    },
    {
      "epoch": 1.8219554725882192,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019107606679035248,
      "loss": 0.6365,
      "step": 3935
    },
    {
      "epoch": 1.8224185460939026,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001910482374768089,
      "loss": 0.6118,
      "step": 3936
    },
    {
      "epoch": 1.822881619599586,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019102040816326528,
      "loss": 0.5751,
      "step": 3937
    },
    {
      "epoch": 1.8233446931052697,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001909925788497217,
      "loss": 0.582,
      "step": 3938
    },
    {
      "epoch": 1.8238077666109531,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001909647495361781,
      "loss": 0.6292,
      "step": 3939
    },
    {
      "epoch": 1.8242708401166365,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001909369202226345,
      "loss": 0.6704,
      "step": 3940
    },
    {
      "epoch": 1.8247339136223202,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001909090909090909,
      "loss": 0.6393,
      "step": 3941
    },
    {
      "epoch": 1.8251969871280036,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001908812615955473,
      "loss": 0.6204,
      "step": 3942
    },
    {
      "epoch": 1.825660060633687,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001908534322820037,
      "loss": 0.6276,
      "step": 3943
    },
    {
      "epoch": 1.8261231341393707,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001908256029684601,
      "loss": 0.5586,
      "step": 3944
    },
    {
      "epoch": 1.8265862076450543,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001907977736549165,
      "loss": 0.701,
      "step": 3945
    },
    {
      "epoch": 1.8270492811507375,
      "grad_norm": 0.5,
      "learning_rate": 0.00019076994434137287,
      "loss": 0.6343,
      "step": 3946
    },
    {
      "epoch": 1.8275123546564211,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019074211502782931,
      "loss": 0.6018,
      "step": 3947
    },
    {
      "epoch": 1.8279754281621048,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019071428571428568,
      "loss": 0.646,
      "step": 3948
    },
    {
      "epoch": 1.8284385016677882,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001906864564007421,
      "loss": 0.6691,
      "step": 3949
    },
    {
      "epoch": 1.8289015751734716,
      "grad_norm": 0.625,
      "learning_rate": 0.00019065862708719848,
      "loss": 0.7771,
      "step": 3950
    },
    {
      "epoch": 1.8293646486791553,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001906307977736549,
      "loss": 0.6146,
      "step": 3951
    },
    {
      "epoch": 1.8298277221848387,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001906029684601113,
      "loss": 0.6635,
      "step": 3952
    },
    {
      "epoch": 1.830290795690522,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001905751391465677,
      "loss": 0.5574,
      "step": 3953
    },
    {
      "epoch": 1.8307538691962058,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001905473098330241,
      "loss": 0.6144,
      "step": 3954
    },
    {
      "epoch": 1.8312169427018892,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001905194805194805,
      "loss": 0.6563,
      "step": 3955
    },
    {
      "epoch": 1.8316800162075726,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001904916512059369,
      "loss": 0.5828,
      "step": 3956
    },
    {
      "epoch": 1.8321430897132562,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019046382189239332,
      "loss": 0.6516,
      "step": 3957
    },
    {
      "epoch": 1.8326061632189397,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001904359925788497,
      "loss": 0.6886,
      "step": 3958
    },
    {
      "epoch": 1.833069236724623,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001904081632653061,
      "loss": 0.7305,
      "step": 3959
    },
    {
      "epoch": 1.8335323102303067,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019038033395176251,
      "loss": 0.6175,
      "step": 3960
    },
    {
      "epoch": 1.8339953837359904,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001903525046382189,
      "loss": 0.5791,
      "step": 3961
    },
    {
      "epoch": 1.8344584572416736,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00019032467532467532,
      "loss": 0.6395,
      "step": 3962
    },
    {
      "epoch": 1.8349215307473572,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001902968460111317,
      "loss": 0.6003,
      "step": 3963
    },
    {
      "epoch": 1.8353846042530408,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019026901669758813,
      "loss": 0.6251,
      "step": 3964
    },
    {
      "epoch": 1.8358476777587243,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001902411873840445,
      "loss": 0.6267,
      "step": 3965
    },
    {
      "epoch": 1.8363107512644077,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019021335807050093,
      "loss": 0.5903,
      "step": 3966
    },
    {
      "epoch": 1.8367738247700913,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001901855287569573,
      "loss": 0.6428,
      "step": 3967
    },
    {
      "epoch": 1.8372368982757747,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001901576994434137,
      "loss": 0.6176,
      "step": 3968
    },
    {
      "epoch": 1.8376999717814582,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001901298701298701,
      "loss": 0.6032,
      "step": 3969
    },
    {
      "epoch": 1.8381630452871418,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019010204081632652,
      "loss": 0.7418,
      "step": 3970
    },
    {
      "epoch": 1.8386261187928252,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001900742115027829,
      "loss": 0.6502,
      "step": 3971
    },
    {
      "epoch": 1.8390891922985086,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001900463821892393,
      "loss": 0.6439,
      "step": 3972
    },
    {
      "epoch": 1.8395522658041923,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019001855287569571,
      "loss": 0.6395,
      "step": 3973
    },
    {
      "epoch": 1.840015339309876,
      "grad_norm": 0.5,
      "learning_rate": 0.0001899907235621521,
      "loss": 0.6181,
      "step": 3974
    },
    {
      "epoch": 1.8404784128155591,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018996289424860852,
      "loss": 0.6395,
      "step": 3975
    },
    {
      "epoch": 1.8409414863212428,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001899350649350649,
      "loss": 0.6297,
      "step": 3976
    },
    {
      "epoch": 1.8414045598269264,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018990723562152133,
      "loss": 0.6607,
      "step": 3977
    },
    {
      "epoch": 1.8418676333326098,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018987940630797772,
      "loss": 0.554,
      "step": 3978
    },
    {
      "epoch": 1.8423307068382933,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018985157699443413,
      "loss": 0.6224,
      "step": 3979
    },
    {
      "epoch": 1.842793780343977,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018982374768089052,
      "loss": 0.6005,
      "step": 3980
    },
    {
      "epoch": 1.8432568538496603,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018979591836734694,
      "loss": 0.5759,
      "step": 3981
    },
    {
      "epoch": 1.8437199273553437,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00018976808905380333,
      "loss": 0.7513,
      "step": 3982
    },
    {
      "epoch": 1.8441830008610274,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018974025974025975,
      "loss": 0.6756,
      "step": 3983
    },
    {
      "epoch": 1.8446460743667108,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001897124304267161,
      "loss": 0.7102,
      "step": 3984
    },
    {
      "epoch": 1.8451091478723942,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018968460111317255,
      "loss": 0.5709,
      "step": 3985
    },
    {
      "epoch": 1.8455722213780779,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018965677179962891,
      "loss": 0.6556,
      "step": 3986
    },
    {
      "epoch": 1.8460352948837613,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001896289424860853,
      "loss": 0.6702,
      "step": 3987
    },
    {
      "epoch": 1.8464983683894447,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018960111317254172,
      "loss": 0.5853,
      "step": 3988
    },
    {
      "epoch": 1.8469614418951283,
      "grad_norm": 0.5,
      "learning_rate": 0.0001895732838589981,
      "loss": 0.6175,
      "step": 3989
    },
    {
      "epoch": 1.847424515400812,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018954545454545453,
      "loss": 0.6489,
      "step": 3990
    },
    {
      "epoch": 1.8478875889064952,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00018951762523191092,
      "loss": 0.5174,
      "step": 3991
    },
    {
      "epoch": 1.8483506624121788,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018948979591836733,
      "loss": 0.5665,
      "step": 3992
    },
    {
      "epoch": 1.8488137359178625,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00018946196660482372,
      "loss": 0.7781,
      "step": 3993
    },
    {
      "epoch": 1.8492768094235459,
      "grad_norm": 0.5,
      "learning_rate": 0.00018943413729128014,
      "loss": 0.5855,
      "step": 3994
    },
    {
      "epoch": 1.8497398829292293,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018940630797773653,
      "loss": 0.543,
      "step": 3995
    },
    {
      "epoch": 1.850202956434913,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018937847866419295,
      "loss": 0.7005,
      "step": 3996
    },
    {
      "epoch": 1.8506660299405964,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018935064935064934,
      "loss": 0.6204,
      "step": 3997
    },
    {
      "epoch": 1.8511291034462798,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018932282003710575,
      "loss": 0.6174,
      "step": 3998
    },
    {
      "epoch": 1.8515921769519634,
      "grad_norm": 0.46875,
      "learning_rate": 0.00018929499072356214,
      "loss": 0.5972,
      "step": 3999
    },
    {
      "epoch": 1.8520552504576469,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018926716141001856,
      "loss": 0.6472,
      "step": 4000
    },
    {
      "epoch": 1.8525183239633303,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018923933209647495,
      "loss": 0.6023,
      "step": 4001
    },
    {
      "epoch": 1.852981397469014,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001892115027829313,
      "loss": 0.6169,
      "step": 4002
    },
    {
      "epoch": 1.8534444709746973,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018918367346938776,
      "loss": 0.6494,
      "step": 4003
    },
    {
      "epoch": 1.8539075444803808,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018915584415584412,
      "loss": 0.6001,
      "step": 4004
    },
    {
      "epoch": 1.8543706179860644,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018912801484230053,
      "loss": 0.6833,
      "step": 4005
    },
    {
      "epoch": 1.854833691491748,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018910018552875692,
      "loss": 0.6711,
      "step": 4006
    },
    {
      "epoch": 1.8552967649974312,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00018907235621521334,
      "loss": 0.5579,
      "step": 4007
    },
    {
      "epoch": 1.8557598385031149,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018904452690166973,
      "loss": 0.6199,
      "step": 4008
    },
    {
      "epoch": 1.8562229120087985,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018901669758812615,
      "loss": 0.5933,
      "step": 4009
    },
    {
      "epoch": 1.856685985514482,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018898886827458254,
      "loss": 0.6021,
      "step": 4010
    },
    {
      "epoch": 1.8571490590201654,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018896103896103895,
      "loss": 0.6562,
      "step": 4011
    },
    {
      "epoch": 1.857612132525849,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018893320964749534,
      "loss": 0.6385,
      "step": 4012
    },
    {
      "epoch": 1.8580752060315324,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00018890538033395176,
      "loss": 0.5942,
      "step": 4013
    },
    {
      "epoch": 1.8585382795372158,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018887755102040815,
      "loss": 0.5375,
      "step": 4014
    },
    {
      "epoch": 1.8590013530428995,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00018884972170686454,
      "loss": 0.6362,
      "step": 4015
    },
    {
      "epoch": 1.859464426548583,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018882189239332096,
      "loss": 0.6443,
      "step": 4016
    },
    {
      "epoch": 1.8599275000542663,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00018879406307977735,
      "loss": 0.6358,
      "step": 4017
    },
    {
      "epoch": 1.86039057355995,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018876623376623376,
      "loss": 0.6602,
      "step": 4018
    },
    {
      "epoch": 1.8608536470656334,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018873840445269015,
      "loss": 0.6107,
      "step": 4019
    },
    {
      "epoch": 1.8613167205713168,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018871057513914657,
      "loss": 0.6792,
      "step": 4020
    },
    {
      "epoch": 1.8617797940770004,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018868274582560293,
      "loss": 0.6016,
      "step": 4021
    },
    {
      "epoch": 1.862242867582684,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018865491651205937,
      "loss": 0.6759,
      "step": 4022
    },
    {
      "epoch": 1.8627059410883673,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018862708719851574,
      "loss": 0.6214,
      "step": 4023
    },
    {
      "epoch": 1.863169014594051,
      "grad_norm": 0.625,
      "learning_rate": 0.00018859925788497215,
      "loss": 0.7181,
      "step": 4024
    },
    {
      "epoch": 1.8636320880997346,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018857142857142854,
      "loss": 0.5996,
      "step": 4025
    },
    {
      "epoch": 1.864095161605418,
      "grad_norm": 0.609375,
      "learning_rate": 0.00018854359925788496,
      "loss": 0.6539,
      "step": 4026
    },
    {
      "epoch": 1.8645582351111014,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00018851576994434135,
      "loss": 0.4819,
      "step": 4027
    },
    {
      "epoch": 1.865021308616785,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018848794063079777,
      "loss": 0.5975,
      "step": 4028
    },
    {
      "epoch": 1.8654843821224685,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018846011131725416,
      "loss": 0.6117,
      "step": 4029
    },
    {
      "epoch": 1.865947455628152,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018843228200371055,
      "loss": 0.6704,
      "step": 4030
    },
    {
      "epoch": 1.8664105291338355,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018840445269016696,
      "loss": 0.6412,
      "step": 4031
    },
    {
      "epoch": 1.866873602639519,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018837662337662335,
      "loss": 0.5595,
      "step": 4032
    },
    {
      "epoch": 1.8673366761452024,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00018834879406307977,
      "loss": 0.645,
      "step": 4033
    },
    {
      "epoch": 1.867799749650886,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018832096474953616,
      "loss": 0.5891,
      "step": 4034
    },
    {
      "epoch": 1.8682628231565694,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018829313543599257,
      "loss": 0.6938,
      "step": 4035
    },
    {
      "epoch": 1.8687258966622529,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00018826530612244896,
      "loss": 0.6796,
      "step": 4036
    },
    {
      "epoch": 1.8691889701679365,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018823747680890538,
      "loss": 0.6718,
      "step": 4037
    },
    {
      "epoch": 1.8696520436736201,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018820964749536177,
      "loss": 0.5866,
      "step": 4038
    },
    {
      "epoch": 1.8701151171793033,
      "grad_norm": 1.6484375,
      "learning_rate": 0.0001881818181818182,
      "loss": 0.9119,
      "step": 4039
    },
    {
      "epoch": 1.870578190684987,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00018815398886827455,
      "loss": 0.6726,
      "step": 4040
    },
    {
      "epoch": 1.8710412641906706,
      "grad_norm": 0.546875,
      "learning_rate": 0.000188126159554731,
      "loss": 0.6351,
      "step": 4041
    },
    {
      "epoch": 1.871504337696354,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018809833024118736,
      "loss": 0.6865,
      "step": 4042
    },
    {
      "epoch": 1.8719674112020375,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018807050092764375,
      "loss": 0.614,
      "step": 4043
    },
    {
      "epoch": 1.872430484707721,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018804267161410016,
      "loss": 0.5638,
      "step": 4044
    },
    {
      "epoch": 1.8728935582134045,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018801484230055655,
      "loss": 0.5984,
      "step": 4045
    },
    {
      "epoch": 1.873356631719088,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00018798701298701297,
      "loss": 0.5429,
      "step": 4046
    },
    {
      "epoch": 1.8738197052247716,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018795918367346936,
      "loss": 0.6225,
      "step": 4047
    },
    {
      "epoch": 1.874282778730455,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018793135435992578,
      "loss": 0.6218,
      "step": 4048
    },
    {
      "epoch": 1.8747458522361384,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00018790352504638216,
      "loss": 0.5559,
      "step": 4049
    },
    {
      "epoch": 1.875208925741822,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018787569573283858,
      "loss": 0.617,
      "step": 4050
    },
    {
      "epoch": 1.8756719992475057,
      "grad_norm": 0.453125,
      "learning_rate": 0.00018784786641929497,
      "loss": 0.6043,
      "step": 4051
    },
    {
      "epoch": 1.876135072753189,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001878200371057514,
      "loss": 0.6763,
      "step": 4052
    },
    {
      "epoch": 1.8765981462588726,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018779220779220778,
      "loss": 0.6942,
      "step": 4053
    },
    {
      "epoch": 1.8770612197645562,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001877643784786642,
      "loss": 0.6613,
      "step": 4054
    },
    {
      "epoch": 1.8775242932702396,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00018773654916512058,
      "loss": 0.6882,
      "step": 4055
    },
    {
      "epoch": 1.877987366775923,
      "grad_norm": 0.45703125,
      "learning_rate": 0.000187708719851577,
      "loss": 0.5794,
      "step": 4056
    },
    {
      "epoch": 1.8784504402816067,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001876808905380334,
      "loss": 0.6334,
      "step": 4057
    },
    {
      "epoch": 1.87891351378729,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00018765306122448975,
      "loss": 0.613,
      "step": 4058
    },
    {
      "epoch": 1.8793765872929735,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001876252319109462,
      "loss": 0.5968,
      "step": 4059
    },
    {
      "epoch": 1.8798396607986572,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018759740259740256,
      "loss": 0.5499,
      "step": 4060
    },
    {
      "epoch": 1.8803027343043406,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018756957328385898,
      "loss": 0.5606,
      "step": 4061
    },
    {
      "epoch": 1.880765807810024,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018754174397031537,
      "loss": 0.6709,
      "step": 4062
    },
    {
      "epoch": 1.8812288813157076,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018751391465677178,
      "loss": 0.6465,
      "step": 4063
    },
    {
      "epoch": 1.881691954821391,
      "grad_norm": 0.6875,
      "learning_rate": 0.00018748608534322817,
      "loss": 0.7132,
      "step": 4064
    },
    {
      "epoch": 1.8821550283270745,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001874582560296846,
      "loss": 0.6818,
      "step": 4065
    },
    {
      "epoch": 1.8826181018327581,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018743042671614098,
      "loss": 0.6077,
      "step": 4066
    },
    {
      "epoch": 1.8830811753384418,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001874025974025974,
      "loss": 0.6169,
      "step": 4067
    },
    {
      "epoch": 1.883544248844125,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018737476808905378,
      "loss": 0.6818,
      "step": 4068
    },
    {
      "epoch": 1.8840073223498086,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001873469387755102,
      "loss": 0.567,
      "step": 4069
    },
    {
      "epoch": 1.8844703958554923,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001873191094619666,
      "loss": 0.7272,
      "step": 4070
    },
    {
      "epoch": 1.8849334693611757,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018729128014842298,
      "loss": 0.6265,
      "step": 4071
    },
    {
      "epoch": 1.885396542866859,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001872634508348794,
      "loss": 0.6487,
      "step": 4072
    },
    {
      "epoch": 1.8858596163725427,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001872356215213358,
      "loss": 0.6397,
      "step": 4073
    },
    {
      "epoch": 1.8863226898782262,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001872077922077922,
      "loss": 0.6353,
      "step": 4074
    },
    {
      "epoch": 1.8867857633839096,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001871799628942486,
      "loss": 0.716,
      "step": 4075
    },
    {
      "epoch": 1.8872488368895932,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000187152133580705,
      "loss": 0.6066,
      "step": 4076
    },
    {
      "epoch": 1.8877119103952766,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018712430426716137,
      "loss": 0.6768,
      "step": 4077
    },
    {
      "epoch": 1.88817498390096,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018709647495361782,
      "loss": 0.6382,
      "step": 4078
    },
    {
      "epoch": 1.8886380574066437,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018706864564007418,
      "loss": 0.5743,
      "step": 4079
    },
    {
      "epoch": 1.8891011309123271,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001870408163265306,
      "loss": 0.7122,
      "step": 4080
    },
    {
      "epoch": 1.8895642044180105,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018701298701298698,
      "loss": 0.5968,
      "step": 4081
    },
    {
      "epoch": 1.8900272779236942,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001869851576994434,
      "loss": 0.5674,
      "step": 4082
    },
    {
      "epoch": 1.8904903514293778,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001869573283858998,
      "loss": 0.6464,
      "step": 4083
    },
    {
      "epoch": 1.890953424935061,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001869294990723562,
      "loss": 0.6445,
      "step": 4084
    },
    {
      "epoch": 1.8914164984407447,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001869016697588126,
      "loss": 0.6825,
      "step": 4085
    },
    {
      "epoch": 1.8918795719464283,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000186873840445269,
      "loss": 0.5509,
      "step": 4086
    },
    {
      "epoch": 1.8923426454521117,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001868460111317254,
      "loss": 0.6683,
      "step": 4087
    },
    {
      "epoch": 1.8928057189577951,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001868181818181818,
      "loss": 0.5515,
      "step": 4088
    },
    {
      "epoch": 1.8932687924634788,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001867903525046382,
      "loss": 0.617,
      "step": 4089
    },
    {
      "epoch": 1.8937318659691622,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001867625231910946,
      "loss": 0.6095,
      "step": 4090
    },
    {
      "epoch": 1.8941949394748456,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00018673469387755102,
      "loss": 0.6377,
      "step": 4091
    },
    {
      "epoch": 1.8946580129805293,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001867068645640074,
      "loss": 0.6275,
      "step": 4092
    },
    {
      "epoch": 1.8951210864862127,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018667903525046382,
      "loss": 0.5605,
      "step": 4093
    },
    {
      "epoch": 1.8955841599918961,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001866512059369202,
      "loss": 0.6393,
      "step": 4094
    },
    {
      "epoch": 1.8960472334975798,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018662337662337663,
      "loss": 0.6412,
      "step": 4095
    },
    {
      "epoch": 1.8965103070032632,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000186595547309833,
      "loss": 0.6631,
      "step": 4096
    },
    {
      "epoch": 1.8969733805089466,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018656771799628944,
      "loss": 0.6147,
      "step": 4097
    },
    {
      "epoch": 1.8974364540146302,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001865398886827458,
      "loss": 0.6721,
      "step": 4098
    },
    {
      "epoch": 1.8978995275203139,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001865120593692022,
      "loss": 0.6,
      "step": 4099
    },
    {
      "epoch": 1.898362601025997,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001864842300556586,
      "loss": 0.7086,
      "step": 4100
    },
    {
      "epoch": 1.8988256745316807,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000186456400742115,
      "loss": 0.6471,
      "step": 4101
    },
    {
      "epoch": 1.8992887480373644,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001864285714285714,
      "loss": 0.6868,
      "step": 4102
    },
    {
      "epoch": 1.8997518215430478,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001864007421150278,
      "loss": 0.5949,
      "step": 4103
    },
    {
      "epoch": 1.9002148950487312,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00018637291280148422,
      "loss": 0.4809,
      "step": 4104
    },
    {
      "epoch": 1.9006779685544148,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001863450834879406,
      "loss": 0.7149,
      "step": 4105
    },
    {
      "epoch": 1.9011410420600983,
      "grad_norm": 0.671875,
      "learning_rate": 0.00018631725417439702,
      "loss": 0.7433,
      "step": 4106
    },
    {
      "epoch": 1.9016041155657817,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001862894248608534,
      "loss": 0.7007,
      "step": 4107
    },
    {
      "epoch": 1.9020671890714653,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018626159554730983,
      "loss": 0.6218,
      "step": 4108
    },
    {
      "epoch": 1.9025302625771487,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018623376623376622,
      "loss": 0.6047,
      "step": 4109
    },
    {
      "epoch": 1.9029933360828322,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018620593692022264,
      "loss": 0.7181,
      "step": 4110
    },
    {
      "epoch": 1.9034564095885158,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00018617810760667903,
      "loss": 0.6245,
      "step": 4111
    },
    {
      "epoch": 1.9039194830941992,
      "grad_norm": 0.5,
      "learning_rate": 0.00018615027829313544,
      "loss": 0.5711,
      "step": 4112
    },
    {
      "epoch": 1.9043825565998826,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00018612244897959183,
      "loss": 0.7703,
      "step": 4113
    },
    {
      "epoch": 1.9048456301055663,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001860946196660482,
      "loss": 0.6944,
      "step": 4114
    },
    {
      "epoch": 1.90530870361125,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018606679035250464,
      "loss": 0.6814,
      "step": 4115
    },
    {
      "epoch": 1.9057717771169331,
      "grad_norm": 0.423828125,
      "learning_rate": 0.000186038961038961,
      "loss": 0.5795,
      "step": 4116
    },
    {
      "epoch": 1.9062348506226168,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00018601113172541742,
      "loss": 0.6812,
      "step": 4117
    },
    {
      "epoch": 1.9066979241283004,
      "grad_norm": 0.625,
      "learning_rate": 0.0001859833024118738,
      "loss": 0.6589,
      "step": 4118
    },
    {
      "epoch": 1.9071609976339838,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00018595547309833022,
      "loss": 0.5789,
      "step": 4119
    },
    {
      "epoch": 1.9076240711396673,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001859276437847866,
      "loss": 0.6442,
      "step": 4120
    },
    {
      "epoch": 1.908087144645351,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018589981447124303,
      "loss": 0.6318,
      "step": 4121
    },
    {
      "epoch": 1.9085502181510343,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018587198515769942,
      "loss": 0.5535,
      "step": 4122
    },
    {
      "epoch": 1.9090132916567177,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018584415584415584,
      "loss": 0.6157,
      "step": 4123
    },
    {
      "epoch": 1.9094763651624014,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018581632653061223,
      "loss": 0.5914,
      "step": 4124
    },
    {
      "epoch": 1.9099394386680848,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018578849721706864,
      "loss": 0.6787,
      "step": 4125
    },
    {
      "epoch": 1.9104025121737682,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00018576066790352503,
      "loss": 0.5588,
      "step": 4126
    },
    {
      "epoch": 1.9108655856794519,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018573283858998142,
      "loss": 0.6555,
      "step": 4127
    },
    {
      "epoch": 1.9113286591851355,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018570500927643784,
      "loss": 0.6077,
      "step": 4128
    },
    {
      "epoch": 1.9117917326908187,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018567717996289423,
      "loss": 0.6699,
      "step": 4129
    },
    {
      "epoch": 1.9122548061965023,
      "grad_norm": 0.4375,
      "learning_rate": 0.00018564935064935064,
      "loss": 0.5874,
      "step": 4130
    },
    {
      "epoch": 1.912717879702186,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018562152133580703,
      "loss": 0.6168,
      "step": 4131
    },
    {
      "epoch": 1.9131809532078694,
      "grad_norm": 0.625,
      "learning_rate": 0.00018559369202226345,
      "loss": 0.6528,
      "step": 4132
    },
    {
      "epoch": 1.9136440267135528,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001855658627087198,
      "loss": 0.665,
      "step": 4133
    },
    {
      "epoch": 1.9141071002192365,
      "grad_norm": 0.609375,
      "learning_rate": 0.00018553803339517626,
      "loss": 0.6613,
      "step": 4134
    },
    {
      "epoch": 1.9145701737249199,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018551020408163262,
      "loss": 0.588,
      "step": 4135
    },
    {
      "epoch": 1.9150332472306033,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018548237476808904,
      "loss": 0.5993,
      "step": 4136
    },
    {
      "epoch": 1.915496320736287,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018545454545454543,
      "loss": 0.6124,
      "step": 4137
    },
    {
      "epoch": 1.9159593942419704,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018542671614100184,
      "loss": 0.6745,
      "step": 4138
    },
    {
      "epoch": 1.9164224677476538,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018539888682745823,
      "loss": 0.6004,
      "step": 4139
    },
    {
      "epoch": 1.9168855412533374,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018537105751391465,
      "loss": 0.7636,
      "step": 4140
    },
    {
      "epoch": 1.9173486147590209,
      "grad_norm": 0.71875,
      "learning_rate": 0.00018534322820037104,
      "loss": 0.744,
      "step": 4141
    },
    {
      "epoch": 1.9178116882647043,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018531539888682743,
      "loss": 0.5974,
      "step": 4142
    },
    {
      "epoch": 1.918274761770388,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018528756957328384,
      "loss": 0.6397,
      "step": 4143
    },
    {
      "epoch": 1.9187378352760716,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00018525974025974023,
      "loss": 0.5233,
      "step": 4144
    },
    {
      "epoch": 1.9192009087817548,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00018523191094619665,
      "loss": 0.6128,
      "step": 4145
    },
    {
      "epoch": 1.9196639822874384,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018520408163265304,
      "loss": 0.6075,
      "step": 4146
    },
    {
      "epoch": 1.920127055793122,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018517625231910946,
      "loss": 0.6341,
      "step": 4147
    },
    {
      "epoch": 1.9205901292988055,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018514842300556585,
      "loss": 0.6225,
      "step": 4148
    },
    {
      "epoch": 1.9210532028044889,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018512059369202226,
      "loss": 0.6118,
      "step": 4149
    },
    {
      "epoch": 1.9215162763101725,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018509276437847865,
      "loss": 0.6225,
      "step": 4150
    },
    {
      "epoch": 1.921979349815856,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00018506493506493507,
      "loss": 0.5962,
      "step": 4151
    },
    {
      "epoch": 1.9224424233215394,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018503710575139143,
      "loss": 0.6076,
      "step": 4152
    },
    {
      "epoch": 1.922905496827223,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00018500927643784788,
      "loss": 0.7218,
      "step": 4153
    },
    {
      "epoch": 1.9233685703329064,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018498144712430424,
      "loss": 0.5957,
      "step": 4154
    },
    {
      "epoch": 1.9238316438385898,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018495361781076066,
      "loss": 0.6261,
      "step": 4155
    },
    {
      "epoch": 1.9242947173442735,
      "grad_norm": 0.625,
      "learning_rate": 0.00018492578849721705,
      "loss": 0.6658,
      "step": 4156
    },
    {
      "epoch": 1.924757790849957,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018489795918367343,
      "loss": 0.7192,
      "step": 4157
    },
    {
      "epoch": 1.9252208643556403,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018487012987012985,
      "loss": 0.668,
      "step": 4158
    },
    {
      "epoch": 1.925683937861324,
      "grad_norm": 0.5,
      "learning_rate": 0.00018484230055658624,
      "loss": 0.6483,
      "step": 4159
    },
    {
      "epoch": 1.9261470113670076,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018481447124304266,
      "loss": 0.5796,
      "step": 4160
    },
    {
      "epoch": 1.9266100848726908,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018478664192949905,
      "loss": 0.5831,
      "step": 4161
    },
    {
      "epoch": 1.9270731583783745,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018475881261595546,
      "loss": 0.6217,
      "step": 4162
    },
    {
      "epoch": 1.927536231884058,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018473098330241185,
      "loss": 0.6038,
      "step": 4163
    },
    {
      "epoch": 1.9279993053897415,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018470315398886827,
      "loss": 0.6864,
      "step": 4164
    },
    {
      "epoch": 1.928462378895425,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018467532467532466,
      "loss": 0.6261,
      "step": 4165
    },
    {
      "epoch": 1.9289254524011086,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00018464749536178108,
      "loss": 0.5582,
      "step": 4166
    },
    {
      "epoch": 1.929388525906792,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018461966604823747,
      "loss": 0.6048,
      "step": 4167
    },
    {
      "epoch": 1.9298515994124754,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018459183673469388,
      "loss": 0.6707,
      "step": 4168
    },
    {
      "epoch": 1.930314672918159,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00018456400742115027,
      "loss": 0.7206,
      "step": 4169
    },
    {
      "epoch": 1.9307777464238425,
      "grad_norm": 0.5,
      "learning_rate": 0.00018453617810760664,
      "loss": 0.6814,
      "step": 4170
    },
    {
      "epoch": 1.931240819929526,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018450834879406305,
      "loss": 0.6301,
      "step": 4171
    },
    {
      "epoch": 1.9317038934352095,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018448051948051944,
      "loss": 0.6975,
      "step": 4172
    },
    {
      "epoch": 1.932166966940893,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018445269016697586,
      "loss": 0.6513,
      "step": 4173
    },
    {
      "epoch": 1.9326300404465764,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018442486085343225,
      "loss": 0.5709,
      "step": 4174
    },
    {
      "epoch": 1.93309311395226,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018439703153988866,
      "loss": 0.6026,
      "step": 4175
    },
    {
      "epoch": 1.9335561874579437,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018436920222634505,
      "loss": 0.6549,
      "step": 4176
    },
    {
      "epoch": 1.9340192609636269,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018434137291280147,
      "loss": 0.6234,
      "step": 4177
    },
    {
      "epoch": 1.9344823344693105,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018431354359925786,
      "loss": 0.6259,
      "step": 4178
    },
    {
      "epoch": 1.9349454079749941,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018428571428571428,
      "loss": 0.6078,
      "step": 4179
    },
    {
      "epoch": 1.9354084814806776,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018425788497217067,
      "loss": 0.6667,
      "step": 4180
    },
    {
      "epoch": 1.935871554986361,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018423005565862708,
      "loss": 0.658,
      "step": 4181
    },
    {
      "epoch": 1.9363346284920446,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018420222634508347,
      "loss": 0.602,
      "step": 4182
    },
    {
      "epoch": 1.936797701997728,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001841743970315399,
      "loss": 0.6262,
      "step": 4183
    },
    {
      "epoch": 1.9372607755034115,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018414656771799628,
      "loss": 0.6739,
      "step": 4184
    },
    {
      "epoch": 1.9377238490090951,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018411873840445267,
      "loss": 0.6214,
      "step": 4185
    },
    {
      "epoch": 1.9381869225147785,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018409090909090909,
      "loss": 0.5932,
      "step": 4186
    },
    {
      "epoch": 1.938649996020462,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00018406307977736548,
      "loss": 0.6284,
      "step": 4187
    },
    {
      "epoch": 1.9391130695261456,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001840352504638219,
      "loss": 0.6332,
      "step": 4188
    },
    {
      "epoch": 1.939576143031829,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018400742115027825,
      "loss": 0.6842,
      "step": 4189
    },
    {
      "epoch": 1.9400392165375124,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001839795918367347,
      "loss": 0.6348,
      "step": 4190
    },
    {
      "epoch": 1.940502290043196,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00018395176252319106,
      "loss": 0.6957,
      "step": 4191
    },
    {
      "epoch": 1.9409653635488797,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00018392393320964748,
      "loss": 0.6908,
      "step": 4192
    },
    {
      "epoch": 1.941428437054563,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00018389610389610387,
      "loss": 0.683,
      "step": 4193
    },
    {
      "epoch": 1.9418915105602466,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00018386827458256028,
      "loss": 0.5774,
      "step": 4194
    },
    {
      "epoch": 1.9423545840659302,
      "grad_norm": 0.515625,
      "learning_rate": 0.00018384044526901667,
      "loss": 0.6616,
      "step": 4195
    },
    {
      "epoch": 1.9428176575716136,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001838126159554731,
      "loss": 0.6295,
      "step": 4196
    },
    {
      "epoch": 1.943280731077297,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00018378478664192948,
      "loss": 0.6551,
      "step": 4197
    },
    {
      "epoch": 1.9437438045829807,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018375695732838587,
      "loss": 0.5871,
      "step": 4198
    },
    {
      "epoch": 1.944206878088664,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00018372912801484229,
      "loss": 0.5412,
      "step": 4199
    },
    {
      "epoch": 1.9446699515943475,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00018370129870129868,
      "loss": 0.5494,
      "step": 4200
    },
    {
      "epoch": 1.9451330251000312,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001836734693877551,
      "loss": 0.6886,
      "step": 4201
    },
    {
      "epoch": 1.9455960986057146,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018364564007421148,
      "loss": 0.6284,
      "step": 4202
    },
    {
      "epoch": 1.946059172111398,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001836178107606679,
      "loss": 0.6271,
      "step": 4203
    },
    {
      "epoch": 1.9465222456170816,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001835899814471243,
      "loss": 0.5854,
      "step": 4204
    },
    {
      "epoch": 1.9469853191227653,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001835621521335807,
      "loss": 0.6525,
      "step": 4205
    },
    {
      "epoch": 1.9474483926284485,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001835343228200371,
      "loss": 0.591,
      "step": 4206
    },
    {
      "epoch": 1.9479114661341321,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001835064935064935,
      "loss": 0.6211,
      "step": 4207
    },
    {
      "epoch": 1.9483745396398158,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018347866419294987,
      "loss": 0.5925,
      "step": 4208
    },
    {
      "epoch": 1.9488376131454992,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00018345083487940632,
      "loss": 0.5778,
      "step": 4209
    },
    {
      "epoch": 1.9493006866511826,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00018342300556586268,
      "loss": 0.6524,
      "step": 4210
    },
    {
      "epoch": 1.9497637601568663,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001833951762523191,
      "loss": 0.5729,
      "step": 4211
    },
    {
      "epoch": 1.9502268336625497,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001833673469387755,
      "loss": 0.6403,
      "step": 4212
    },
    {
      "epoch": 1.950689907168233,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018333951762523188,
      "loss": 0.611,
      "step": 4213
    },
    {
      "epoch": 1.9511529806739167,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001833116883116883,
      "loss": 0.6227,
      "step": 4214
    },
    {
      "epoch": 1.9516160541796002,
      "grad_norm": 0.625,
      "learning_rate": 0.00018328385899814468,
      "loss": 0.691,
      "step": 4215
    },
    {
      "epoch": 1.9520791276852836,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001832560296846011,
      "loss": 0.6175,
      "step": 4216
    },
    {
      "epoch": 1.9525422011909672,
      "grad_norm": 0.5,
      "learning_rate": 0.0001832282003710575,
      "loss": 0.6609,
      "step": 4217
    },
    {
      "epoch": 1.9530052746966506,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001832003710575139,
      "loss": 0.6605,
      "step": 4218
    },
    {
      "epoch": 1.953468348202334,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001831725417439703,
      "loss": 0.6326,
      "step": 4219
    },
    {
      "epoch": 1.9539314217080177,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001831447124304267,
      "loss": 0.6175,
      "step": 4220
    },
    {
      "epoch": 1.9543944952137013,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001831168831168831,
      "loss": 0.595,
      "step": 4221
    },
    {
      "epoch": 1.9548575687193845,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018308905380333952,
      "loss": 0.6573,
      "step": 4222
    },
    {
      "epoch": 1.9553206422250682,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001830612244897959,
      "loss": 0.6105,
      "step": 4223
    },
    {
      "epoch": 1.9557837157307518,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018303339517625232,
      "loss": 0.6313,
      "step": 4224
    },
    {
      "epoch": 1.9562467892364352,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018300556586270871,
      "loss": 0.6269,
      "step": 4225
    },
    {
      "epoch": 1.9567098627421187,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00018297773654916508,
      "loss": 0.7132,
      "step": 4226
    },
    {
      "epoch": 1.9571729362478023,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001829499072356215,
      "loss": 0.6014,
      "step": 4227
    },
    {
      "epoch": 1.9576360097534857,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018292207792207788,
      "loss": 0.6638,
      "step": 4228
    },
    {
      "epoch": 1.9580990832591691,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001828942486085343,
      "loss": 0.6321,
      "step": 4229
    },
    {
      "epoch": 1.9585621567648528,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001828664192949907,
      "loss": 0.7292,
      "step": 4230
    },
    {
      "epoch": 1.9590252302705362,
      "grad_norm": 0.5,
      "learning_rate": 0.0001828385899814471,
      "loss": 0.5771,
      "step": 4231
    },
    {
      "epoch": 1.9594883037762196,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001828107606679035,
      "loss": 0.5925,
      "step": 4232
    },
    {
      "epoch": 1.9599513772819033,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001827829313543599,
      "loss": 0.6039,
      "step": 4233
    },
    {
      "epoch": 1.9604144507875867,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001827551020408163,
      "loss": 0.6372,
      "step": 4234
    },
    {
      "epoch": 1.9608775242932701,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00018272727272727272,
      "loss": 0.5856,
      "step": 4235
    },
    {
      "epoch": 1.9613405977989538,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001826994434137291,
      "loss": 0.719,
      "step": 4236
    },
    {
      "epoch": 1.9618036713046374,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018267161410018552,
      "loss": 0.6049,
      "step": 4237
    },
    {
      "epoch": 1.9622667448103206,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018264378478664191,
      "loss": 0.6658,
      "step": 4238
    },
    {
      "epoch": 1.9627298183160042,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018261595547309833,
      "loss": 0.7045,
      "step": 4239
    },
    {
      "epoch": 1.9631928918216879,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018258812615955472,
      "loss": 0.6016,
      "step": 4240
    },
    {
      "epoch": 1.9636559653273713,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001825602968460111,
      "loss": 0.5976,
      "step": 4241
    },
    {
      "epoch": 1.9641190388330547,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018253246753246753,
      "loss": 0.7292,
      "step": 4242
    },
    {
      "epoch": 1.9645821123387384,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00018250463821892392,
      "loss": 0.6094,
      "step": 4243
    },
    {
      "epoch": 1.9650451858444218,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018247680890538033,
      "loss": 0.6203,
      "step": 4244
    },
    {
      "epoch": 1.9655082593501052,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001824489795918367,
      "loss": 0.6528,
      "step": 4245
    },
    {
      "epoch": 1.9659713328557888,
      "grad_norm": 0.5,
      "learning_rate": 0.00018242115027829314,
      "loss": 0.6159,
      "step": 4246
    },
    {
      "epoch": 1.9664344063614723,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001823933209647495,
      "loss": 0.6217,
      "step": 4247
    },
    {
      "epoch": 1.9668974798671557,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00018236549165120592,
      "loss": 0.7522,
      "step": 4248
    },
    {
      "epoch": 1.9673605533728393,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001823376623376623,
      "loss": 0.628,
      "step": 4249
    },
    {
      "epoch": 1.9678236268785227,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018230983302411873,
      "loss": 0.6939,
      "step": 4250
    },
    {
      "epoch": 1.9682867003842062,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018228200371057511,
      "loss": 0.6564,
      "step": 4251
    },
    {
      "epoch": 1.9687497738898898,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00018225417439703153,
      "loss": 0.7021,
      "step": 4252
    },
    {
      "epoch": 1.9692128473955735,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018222634508348792,
      "loss": 0.5831,
      "step": 4253
    },
    {
      "epoch": 1.9696759209012566,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001821985157699443,
      "loss": 0.604,
      "step": 4254
    },
    {
      "epoch": 1.9701389944069403,
      "grad_norm": 0.546875,
      "learning_rate": 0.00018217068645640073,
      "loss": 0.6961,
      "step": 4255
    },
    {
      "epoch": 1.970602067912624,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00018214285714285712,
      "loss": 0.6,
      "step": 4256
    },
    {
      "epoch": 1.9710651414183074,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018211502782931353,
      "loss": 0.5995,
      "step": 4257
    },
    {
      "epoch": 1.9715282149239908,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00018208719851576992,
      "loss": 0.5139,
      "step": 4258
    },
    {
      "epoch": 1.9719912884296744,
      "grad_norm": 0.53125,
      "learning_rate": 0.00018205936920222634,
      "loss": 0.5937,
      "step": 4259
    },
    {
      "epoch": 1.9724543619353578,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018203153988868273,
      "loss": 0.5882,
      "step": 4260
    },
    {
      "epoch": 1.9729174354410413,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00018200371057513915,
      "loss": 0.6036,
      "step": 4261
    },
    {
      "epoch": 1.973380508946725,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00018197588126159554,
      "loss": 0.7551,
      "step": 4262
    },
    {
      "epoch": 1.9738435824524083,
      "grad_norm": 0.5,
      "learning_rate": 0.00018194805194805195,
      "loss": 0.6603,
      "step": 4263
    },
    {
      "epoch": 1.9743066559580917,
      "grad_norm": 0.59375,
      "learning_rate": 0.00018192022263450832,
      "loss": 0.7166,
      "step": 4264
    },
    {
      "epoch": 1.9747697294637754,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018189239332096476,
      "loss": 0.6454,
      "step": 4265
    },
    {
      "epoch": 1.9752328029694588,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018186456400742112,
      "loss": 0.6153,
      "step": 4266
    },
    {
      "epoch": 1.9756958764751422,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00018183673469387754,
      "loss": 0.6214,
      "step": 4267
    },
    {
      "epoch": 1.9761589499808259,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018180890538033393,
      "loss": 0.5786,
      "step": 4268
    },
    {
      "epoch": 1.9766220234865095,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00018178107606679032,
      "loss": 0.6788,
      "step": 4269
    },
    {
      "epoch": 1.977085096992193,
      "grad_norm": 0.65625,
      "learning_rate": 0.00018175324675324673,
      "loss": 0.6504,
      "step": 4270
    },
    {
      "epoch": 1.9775481704978763,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00018172541743970312,
      "loss": 0.6046,
      "step": 4271
    },
    {
      "epoch": 1.97801124400356,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018169758812615954,
      "loss": 0.6585,
      "step": 4272
    },
    {
      "epoch": 1.9784743175092434,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018166975881261593,
      "loss": 0.6754,
      "step": 4273
    },
    {
      "epoch": 1.9789373910149268,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018164192949907235,
      "loss": 0.6207,
      "step": 4274
    },
    {
      "epoch": 1.9794004645206105,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018161410018552874,
      "loss": 0.6066,
      "step": 4275
    },
    {
      "epoch": 1.979863538026294,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018158627087198515,
      "loss": 0.658,
      "step": 4276
    },
    {
      "epoch": 1.9803266115319773,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018155844155844154,
      "loss": 0.7402,
      "step": 4277
    },
    {
      "epoch": 1.980789685037661,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00018153061224489796,
      "loss": 0.6026,
      "step": 4278
    },
    {
      "epoch": 1.9812527585433444,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018150278293135435,
      "loss": 0.6853,
      "step": 4279
    },
    {
      "epoch": 1.9817158320490278,
      "grad_norm": 0.578125,
      "learning_rate": 0.00018147495361781077,
      "loss": 0.6425,
      "step": 4280
    },
    {
      "epoch": 1.9821789055547114,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018144712430426716,
      "loss": 0.6247,
      "step": 4281
    },
    {
      "epoch": 1.982641979060395,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00018141929499072357,
      "loss": 0.5581,
      "step": 4282
    },
    {
      "epoch": 1.9831050525660783,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018139146567717993,
      "loss": 0.6107,
      "step": 4283
    },
    {
      "epoch": 1.983568126071762,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00018136363636363632,
      "loss": 0.5626,
      "step": 4284
    },
    {
      "epoch": 1.9840311995774456,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00018133580705009274,
      "loss": 0.6315,
      "step": 4285
    },
    {
      "epoch": 1.984494273083129,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00018130797773654913,
      "loss": 0.5587,
      "step": 4286
    },
    {
      "epoch": 1.9849573465888124,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018128014842300555,
      "loss": 0.6317,
      "step": 4287
    },
    {
      "epoch": 1.985420420094496,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018125231910946194,
      "loss": 0.6225,
      "step": 4288
    },
    {
      "epoch": 1.9858834936001795,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018122448979591835,
      "loss": 0.6085,
      "step": 4289
    },
    {
      "epoch": 1.9863465671058629,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00018119666048237474,
      "loss": 0.6028,
      "step": 4290
    },
    {
      "epoch": 1.9868096406115465,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018116883116883116,
      "loss": 0.62,
      "step": 4291
    },
    {
      "epoch": 1.98727271411723,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00018114100185528755,
      "loss": 0.6675,
      "step": 4292
    },
    {
      "epoch": 1.9877357876229134,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018111317254174397,
      "loss": 0.6536,
      "step": 4293
    },
    {
      "epoch": 1.988198861128597,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00018108534322820036,
      "loss": 0.566,
      "step": 4294
    },
    {
      "epoch": 1.9886619346342804,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00018105751391465677,
      "loss": 0.6102,
      "step": 4295
    },
    {
      "epoch": 1.9891250081399638,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00018102968460111316,
      "loss": 0.6707,
      "step": 4296
    },
    {
      "epoch": 1.9895880816456475,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018100185528756955,
      "loss": 0.7712,
      "step": 4297
    },
    {
      "epoch": 1.9900511551513311,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00018097402597402597,
      "loss": 0.625,
      "step": 4298
    },
    {
      "epoch": 1.9905142286570143,
      "grad_norm": 1.5546875,
      "learning_rate": 0.00018094619666048236,
      "loss": 0.6358,
      "step": 4299
    },
    {
      "epoch": 1.990977302162698,
      "grad_norm": 0.5625,
      "learning_rate": 0.00018091836734693877,
      "loss": 0.6711,
      "step": 4300
    },
    {
      "epoch": 1.9914403756683816,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00018089053803339514,
      "loss": 0.5725,
      "step": 4301
    },
    {
      "epoch": 1.991903449174065,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00018086270871985158,
      "loss": 0.5598,
      "step": 4302
    },
    {
      "epoch": 1.9923665226797485,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00018083487940630794,
      "loss": 0.6016,
      "step": 4303
    },
    {
      "epoch": 1.992829596185432,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00018080705009276436,
      "loss": 0.6496,
      "step": 4304
    },
    {
      "epoch": 1.9932926696911155,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00018077922077922075,
      "loss": 0.62,
      "step": 4305
    },
    {
      "epoch": 1.993755743196799,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00018075139146567717,
      "loss": 0.5834,
      "step": 4306
    },
    {
      "epoch": 1.9942188167024826,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00018072356215213356,
      "loss": 0.5709,
      "step": 4307
    },
    {
      "epoch": 1.994681890208166,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00018069573283858997,
      "loss": 0.6057,
      "step": 4308
    },
    {
      "epoch": 1.9951449637138494,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00018066790352504636,
      "loss": 0.6344,
      "step": 4309
    },
    {
      "epoch": 1.995608037219533,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00018064007421150278,
      "loss": 0.5527,
      "step": 4310
    },
    {
      "epoch": 1.9960711107252165,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018061224489795917,
      "loss": 0.6355,
      "step": 4311
    },
    {
      "epoch": 1.9965341842309,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00018058441558441556,
      "loss": 0.5985,
      "step": 4312
    },
    {
      "epoch": 1.9969972577365835,
      "grad_norm": 0.703125,
      "learning_rate": 0.00018055658627087198,
      "loss": 0.6862,
      "step": 4313
    },
    {
      "epoch": 1.9974603312422672,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00018052875695732836,
      "loss": 0.6085,
      "step": 4314
    },
    {
      "epoch": 1.9979234047479504,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00018050092764378478,
      "loss": 0.6161,
      "step": 4315
    },
    {
      "epoch": 1.998386478253634,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00018047309833024117,
      "loss": 0.6304,
      "step": 4316
    },
    {
      "epoch": 1.9988495517593177,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001804452690166976,
      "loss": 0.5727,
      "step": 4317
    },
    {
      "epoch": 1.999312625265001,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00018041743970315398,
      "loss": 0.5457,
      "step": 4318
    },
    {
      "epoch": 1.9997756987706845,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001803896103896104,
      "loss": 0.6081,
      "step": 4319
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018036178107606676,
      "loss": 0.646,
      "step": 4320
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.488868474960327,
      "eval_runtime": 178.7276,
      "eval_samples_per_second": 34.365,
      "eval_steps_per_second": 34.365,
      "step": 4320
    },
    {
      "epoch": 2.0,
      "eval_title2sid_loss": 0.8923724293708801,
      "eval_title2sid_runtime": 114.7832,
      "eval_title2sid_samples_per_second": 33.158,
      "eval_title2sid_steps_per_second": 33.158,
      "step": 4320
    },
    {
      "epoch": 2.0,
      "eval_title2sid_loss": 0.8923724293708801,
      "eval_title2sid_runtime": 114.7832,
      "eval_title2sid_samples_per_second": 33.158,
      "eval_title2sid_steps_per_second": 33.158,
      "step": 4320
    },
    {
      "epoch": 2.0,
      "eval_sid2title_loss": 0.35291633009910583,
      "eval_sid2title_runtime": 110.853,
      "eval_sid2title_samples_per_second": 34.523,
      "eval_sid2title_steps_per_second": 34.523,
      "step": 4320
    },
    {
      "epoch": 2.0,
      "eval_sid2title_loss": 0.35291633009910583,
      "eval_sid2title_runtime": 110.853,
      "eval_sid2title_samples_per_second": 34.523,
      "eval_sid2title_steps_per_second": 34.523,
      "step": 4320
    }
  ],
  "logging_steps": 1,
  "max_steps": 10800,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 3,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.596441123570483e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}