{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9998660056277636,
  "eval_steps": 500,
  "global_step": 3731,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00026798874447273214,
      "grad_norm": 2.25720278847754,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.906,
      "step": 1
    },
    {
      "epoch": 0.0005359774889454643,
      "grad_norm": 2.39171793296628,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.9075,
      "step": 2
    },
    {
      "epoch": 0.0008039662334181965,
      "grad_norm": 2.117373423926525,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.8754,
      "step": 3
    },
    {
      "epoch": 0.0010719549778909286,
      "grad_norm": 2.2540361976902803,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.8775,
      "step": 4
    },
    {
      "epoch": 0.0013399437223636608,
      "grad_norm": 2.311350136670924,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.9075,
      "step": 5
    },
    {
      "epoch": 0.001607932466836393,
      "grad_norm": 2.269220073575924,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.9361,
      "step": 6
    },
    {
      "epoch": 0.001875921211309125,
      "grad_norm": 2.478025488177133,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.8996,
      "step": 7
    },
    {
      "epoch": 0.002143909955781857,
      "grad_norm": 2.1701895144821557,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.9001,
      "step": 8
    },
    {
      "epoch": 0.002411898700254589,
      "grad_norm": 2.5379858515264013,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 0.9043,
      "step": 9
    },
    {
      "epoch": 0.0026798874447273215,
      "grad_norm": 1.8710167577693437,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.8562,
      "step": 10
    },
    {
      "epoch": 0.0029478761892000535,
      "grad_norm": 2.1576626601473383,
      "learning_rate": 5.5e-07,
      "loss": 0.8915,
      "step": 11
    },
    {
      "epoch": 0.003215864933672786,
      "grad_norm": 1.9468628150051608,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.8941,
      "step": 12
    },
    {
      "epoch": 0.003483853678145518,
      "grad_norm": 1.867368495547589,
      "learning_rate": 6.5e-07,
      "loss": 0.8991,
      "step": 13
    },
    {
      "epoch": 0.00375184242261825,
      "grad_norm": 1.73695267084012,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.8674,
      "step": 14
    },
    {
      "epoch": 0.004019831167090982,
      "grad_norm": 1.6635178694010648,
      "learning_rate": 7.5e-07,
      "loss": 0.8973,
      "step": 15
    },
    {
      "epoch": 0.004287819911563714,
      "grad_norm": 1.742660861100219,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.8976,
      "step": 16
    },
    {
      "epoch": 0.004555808656036446,
      "grad_norm": 1.5209446843040308,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.8926,
      "step": 17
    },
    {
      "epoch": 0.004823797400509178,
      "grad_norm": 1.4715495709622453,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.8769,
      "step": 18
    },
    {
      "epoch": 0.005091786144981911,
      "grad_norm": 1.3354841490040308,
      "learning_rate": 9.500000000000001e-07,
      "loss": 0.8645,
      "step": 19
    },
    {
      "epoch": 0.005359774889454643,
      "grad_norm": 1.3098681863098105,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.867,
      "step": 20
    },
    {
      "epoch": 0.005627763633927375,
      "grad_norm": 1.2683468325882374,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.8804,
      "step": 21
    },
    {
      "epoch": 0.005895752378400107,
      "grad_norm": 1.2109668644560807,
      "learning_rate": 1.1e-06,
      "loss": 0.8709,
      "step": 22
    },
    {
      "epoch": 0.006163741122872839,
      "grad_norm": 1.2034122538951713,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 0.8563,
      "step": 23
    },
    {
      "epoch": 0.006431729867345572,
      "grad_norm": 1.1961400386180532,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.8892,
      "step": 24
    },
    {
      "epoch": 0.006699718611818304,
      "grad_norm": 1.4643071297829495,
      "learning_rate": 1.25e-06,
      "loss": 0.8097,
      "step": 25
    },
    {
      "epoch": 0.006967707356291036,
      "grad_norm": 1.0336188537431046,
      "learning_rate": 1.3e-06,
      "loss": 0.8512,
      "step": 26
    },
    {
      "epoch": 0.007235696100763768,
      "grad_norm": 1.0303702619850084,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 0.8417,
      "step": 27
    },
    {
      "epoch": 0.0075036848452365,
      "grad_norm": 0.963043102812572,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.8023,
      "step": 28
    },
    {
      "epoch": 0.007771673589709233,
      "grad_norm": 0.9036622151252024,
      "learning_rate": 1.45e-06,
      "loss": 0.7845,
      "step": 29
    },
    {
      "epoch": 0.008039662334181965,
      "grad_norm": 0.9435664347842717,
      "learning_rate": 1.5e-06,
      "loss": 0.851,
      "step": 30
    },
    {
      "epoch": 0.008307651078654697,
      "grad_norm": 0.8804630513554519,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 0.8107,
      "step": 31
    },
    {
      "epoch": 0.008575639823127429,
      "grad_norm": 0.8212843559219073,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.7738,
      "step": 32
    },
    {
      "epoch": 0.00884362856760016,
      "grad_norm": 1.0258081854447259,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 0.8492,
      "step": 33
    },
    {
      "epoch": 0.009111617312072893,
      "grad_norm": 0.8291050030270384,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.8123,
      "step": 34
    },
    {
      "epoch": 0.009379606056545624,
      "grad_norm": 0.8796346267577275,
      "learning_rate": 1.75e-06,
      "loss": 0.846,
      "step": 35
    },
    {
      "epoch": 0.009647594801018356,
      "grad_norm": 0.8647596993137628,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.7915,
      "step": 36
    },
    {
      "epoch": 0.00991558354549109,
      "grad_norm": 0.8487256177334335,
      "learning_rate": 1.85e-06,
      "loss": 0.776,
      "step": 37
    },
    {
      "epoch": 0.010183572289963822,
      "grad_norm": 0.7937985373954708,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.7657,
      "step": 38
    },
    {
      "epoch": 0.010451561034436554,
      "grad_norm": 0.8149083575940699,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.8017,
      "step": 39
    },
    {
      "epoch": 0.010719549778909286,
      "grad_norm": 0.7966041265699525,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.8066,
      "step": 40
    },
    {
      "epoch": 0.010987538523382018,
      "grad_norm": 0.7558521661511102,
      "learning_rate": 2.05e-06,
      "loss": 0.7766,
      "step": 41
    },
    {
      "epoch": 0.01125552726785475,
      "grad_norm": 0.7511838136376909,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.7772,
      "step": 42
    },
    {
      "epoch": 0.011523516012327482,
      "grad_norm": 0.740066613225746,
      "learning_rate": 2.15e-06,
      "loss": 0.7721,
      "step": 43
    },
    {
      "epoch": 0.011791504756800214,
      "grad_norm": 0.753644703081159,
      "learning_rate": 2.2e-06,
      "loss": 0.7662,
      "step": 44
    },
    {
      "epoch": 0.012059493501272946,
      "grad_norm": 0.734386240323295,
      "learning_rate": 2.25e-06,
      "loss": 0.7504,
      "step": 45
    },
    {
      "epoch": 0.012327482245745678,
      "grad_norm": 0.7272936315263336,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.736,
      "step": 46
    },
    {
      "epoch": 0.01259547099021841,
      "grad_norm": 0.6994662363407925,
      "learning_rate": 2.35e-06,
      "loss": 0.7563,
      "step": 47
    },
    {
      "epoch": 0.012863459734691144,
      "grad_norm": 0.7063884865165266,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.7903,
      "step": 48
    },
    {
      "epoch": 0.013131448479163876,
      "grad_norm": 0.7074574870594601,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.7489,
      "step": 49
    },
    {
      "epoch": 0.013399437223636608,
      "grad_norm": 0.697130120070113,
      "learning_rate": 2.5e-06,
      "loss": 0.756,
      "step": 50
    },
    {
      "epoch": 0.01366742596810934,
      "grad_norm": 0.7152301556974937,
      "learning_rate": 2.55e-06,
      "loss": 0.7811,
      "step": 51
    },
    {
      "epoch": 0.013935414712582072,
      "grad_norm": 0.7135866297261962,
      "learning_rate": 2.6e-06,
      "loss": 0.7657,
      "step": 52
    },
    {
      "epoch": 0.014203403457054804,
      "grad_norm": 0.6871176808562605,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.7668,
      "step": 53
    },
    {
      "epoch": 0.014471392201527536,
      "grad_norm": 0.6895833904138279,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.768,
      "step": 54
    },
    {
      "epoch": 0.014739380946000268,
      "grad_norm": 0.6798647196037325,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.7775,
      "step": 55
    },
    {
      "epoch": 0.015007369690473,
      "grad_norm": 0.7137733924049838,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.7526,
      "step": 56
    },
    {
      "epoch": 0.015275358434945732,
      "grad_norm": 0.7139801212518347,
      "learning_rate": 2.85e-06,
      "loss": 0.7685,
      "step": 57
    },
    {
      "epoch": 0.015543347179418465,
      "grad_norm": 0.6761038127805458,
      "learning_rate": 2.9e-06,
      "loss": 0.7136,
      "step": 58
    },
    {
      "epoch": 0.015811335923891195,
      "grad_norm": 0.6779110668429545,
      "learning_rate": 2.95e-06,
      "loss": 0.7453,
      "step": 59
    },
    {
      "epoch": 0.01607932466836393,
      "grad_norm": 0.6556368879123587,
      "learning_rate": 3e-06,
      "loss": 0.7804,
      "step": 60
    },
    {
      "epoch": 0.01634731341283666,
      "grad_norm": 0.6719670262451065,
      "learning_rate": 3.05e-06,
      "loss": 0.7142,
      "step": 61
    },
    {
      "epoch": 0.016615302157309393,
      "grad_norm": 0.6646700612645033,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.7989,
      "step": 62
    },
    {
      "epoch": 0.016883290901782127,
      "grad_norm": 0.6381679044186503,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.7317,
      "step": 63
    },
    {
      "epoch": 0.017151279646254857,
      "grad_norm": 0.6474041972163426,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.7331,
      "step": 64
    },
    {
      "epoch": 0.01741926839072759,
      "grad_norm": 0.7016614904648757,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.7447,
      "step": 65
    },
    {
      "epoch": 0.01768725713520032,
      "grad_norm": 0.6625760890168111,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.7321,
      "step": 66
    },
    {
      "epoch": 0.017955245879673055,
      "grad_norm": 0.6512487052628951,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.7238,
      "step": 67
    },
    {
      "epoch": 0.018223234624145785,
      "grad_norm": 0.6715447659597985,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.7401,
      "step": 68
    },
    {
      "epoch": 0.01849122336861852,
      "grad_norm": 0.6300613900503144,
      "learning_rate": 3.45e-06,
      "loss": 0.6822,
      "step": 69
    },
    {
      "epoch": 0.01875921211309125,
      "grad_norm": 0.6621506807701116,
      "learning_rate": 3.5e-06,
      "loss": 0.7552,
      "step": 70
    },
    {
      "epoch": 0.019027200857563983,
      "grad_norm": 0.6902002004071224,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.7329,
      "step": 71
    },
    {
      "epoch": 0.019295189602036713,
      "grad_norm": 0.6462970862893366,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.7317,
      "step": 72
    },
    {
      "epoch": 0.019563178346509447,
      "grad_norm": 0.6577848216155666,
      "learning_rate": 3.65e-06,
      "loss": 0.7363,
      "step": 73
    },
    {
      "epoch": 0.01983116709098218,
      "grad_norm": 0.6898233385816993,
      "learning_rate": 3.7e-06,
      "loss": 0.7719,
      "step": 74
    },
    {
      "epoch": 0.02009915583545491,
      "grad_norm": 0.6473154408340821,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.6984,
      "step": 75
    },
    {
      "epoch": 0.020367144579927644,
      "grad_norm": 0.6639562386303538,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.7295,
      "step": 76
    },
    {
      "epoch": 0.020635133324400375,
      "grad_norm": 0.6550344403989367,
      "learning_rate": 3.85e-06,
      "loss": 0.7268,
      "step": 77
    },
    {
      "epoch": 0.02090312206887311,
      "grad_norm": 0.6307224417427757,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.7343,
      "step": 78
    },
    {
      "epoch": 0.02117111081334584,
      "grad_norm": 0.637206256082345,
      "learning_rate": 3.95e-06,
      "loss": 0.7266,
      "step": 79
    },
    {
      "epoch": 0.021439099557818572,
      "grad_norm": 0.67981113221639,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.7669,
      "step": 80
    },
    {
      "epoch": 0.021707088302291302,
      "grad_norm": 0.6485594579689294,
      "learning_rate": 4.05e-06,
      "loss": 0.7474,
      "step": 81
    },
    {
      "epoch": 0.021975077046764036,
      "grad_norm": 0.6868023930819824,
      "learning_rate": 4.1e-06,
      "loss": 0.7716,
      "step": 82
    },
    {
      "epoch": 0.022243065791236766,
      "grad_norm": 0.6576727714390032,
      "learning_rate": 4.15e-06,
      "loss": 0.7607,
      "step": 83
    },
    {
      "epoch": 0.0225110545357095,
      "grad_norm": 0.6389350119005764,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.7126,
      "step": 84
    },
    {
      "epoch": 0.022779043280182234,
      "grad_norm": 0.6396784633626287,
      "learning_rate": 4.25e-06,
      "loss": 0.7287,
      "step": 85
    },
    {
      "epoch": 0.023047032024654964,
      "grad_norm": 0.627067242244011,
      "learning_rate": 4.3e-06,
      "loss": 0.7106,
      "step": 86
    },
    {
      "epoch": 0.023315020769127698,
      "grad_norm": 0.6538916463200231,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.7158,
      "step": 87
    },
    {
      "epoch": 0.023583009513600428,
      "grad_norm": 0.6448819585144561,
      "learning_rate": 4.4e-06,
      "loss": 0.6979,
      "step": 88
    },
    {
      "epoch": 0.023850998258073162,
      "grad_norm": 0.7177049009596135,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.6938,
      "step": 89
    },
    {
      "epoch": 0.024118987002545892,
      "grad_norm": 0.6480193849427767,
      "learning_rate": 4.5e-06,
      "loss": 0.7213,
      "step": 90
    },
    {
      "epoch": 0.024386975747018626,
      "grad_norm": 0.7319312567301275,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.6911,
      "step": 91
    },
    {
      "epoch": 0.024654964491491356,
      "grad_norm": 0.6319998067147801,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.6965,
      "step": 92
    },
    {
      "epoch": 0.02492295323596409,
      "grad_norm": 0.6619137812674756,
      "learning_rate": 4.65e-06,
      "loss": 0.7333,
      "step": 93
    },
    {
      "epoch": 0.02519094198043682,
      "grad_norm": 0.6688602683830973,
      "learning_rate": 4.7e-06,
      "loss": 0.6673,
      "step": 94
    },
    {
      "epoch": 0.025458930724909554,
      "grad_norm": 0.6760510462724317,
      "learning_rate": 4.75e-06,
      "loss": 0.7249,
      "step": 95
    },
    {
      "epoch": 0.025726919469382287,
      "grad_norm": 0.6541007344866108,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.7502,
      "step": 96
    },
    {
      "epoch": 0.025994908213855018,
      "grad_norm": 0.6598842719707059,
      "learning_rate": 4.85e-06,
      "loss": 0.7334,
      "step": 97
    },
    {
      "epoch": 0.02626289695832775,
      "grad_norm": 0.6710516095785296,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.739,
      "step": 98
    },
    {
      "epoch": 0.02653088570280048,
      "grad_norm": 0.6774821515467728,
      "learning_rate": 4.95e-06,
      "loss": 0.7191,
      "step": 99
    },
    {
      "epoch": 0.026798874447273215,
      "grad_norm": 0.7121812965969331,
      "learning_rate": 5e-06,
      "loss": 0.7603,
      "step": 100
    },
    {
      "epoch": 0.027066863191745946,
      "grad_norm": 0.662503800123449,
      "learning_rate": 4.999999975160348e-06,
      "loss": 0.7174,
      "step": 101
    },
    {
      "epoch": 0.02733485193621868,
      "grad_norm": 0.6406527358376657,
      "learning_rate": 4.999999900641393e-06,
      "loss": 0.7274,
      "step": 102
    },
    {
      "epoch": 0.02760284068069141,
      "grad_norm": 0.6475760309997134,
      "learning_rate": 4.999999776443136e-06,
      "loss": 0.7173,
      "step": 103
    },
    {
      "epoch": 0.027870829425164143,
      "grad_norm": 0.6508222374535301,
      "learning_rate": 4.9999996025655804e-06,
      "loss": 0.7273,
      "step": 104
    },
    {
      "epoch": 0.028138818169636877,
      "grad_norm": 0.6422845025330729,
      "learning_rate": 4.999999379008728e-06,
      "loss": 0.7101,
      "step": 105
    },
    {
      "epoch": 0.028406806914109607,
      "grad_norm": 0.6894148699577247,
      "learning_rate": 4.999999105772584e-06,
      "loss": 0.6883,
      "step": 106
    },
    {
      "epoch": 0.02867479565858234,
      "grad_norm": 0.6786448816107133,
      "learning_rate": 4.999998782857154e-06,
      "loss": 0.7208,
      "step": 107
    },
    {
      "epoch": 0.02894278440305507,
      "grad_norm": 0.6563727203784894,
      "learning_rate": 4.999998410262445e-06,
      "loss": 0.6873,
      "step": 108
    },
    {
      "epoch": 0.029210773147527805,
      "grad_norm": 0.6798363655004392,
      "learning_rate": 4.999997987988463e-06,
      "loss": 0.7399,
      "step": 109
    },
    {
      "epoch": 0.029478761892000535,
      "grad_norm": 0.6687257805055732,
      "learning_rate": 4.999997516035218e-06,
      "loss": 0.7274,
      "step": 110
    },
    {
      "epoch": 0.02974675063647327,
      "grad_norm": 0.683690509332196,
      "learning_rate": 4.999996994402718e-06,
      "loss": 0.7236,
      "step": 111
    },
    {
      "epoch": 0.030014739380946,
      "grad_norm": 0.6617859651724926,
      "learning_rate": 4.9999964230909745e-06,
      "loss": 0.7047,
      "step": 112
    },
    {
      "epoch": 0.030282728125418733,
      "grad_norm": 0.7093346085901996,
      "learning_rate": 4.999995802099998e-06,
      "loss": 0.6947,
      "step": 113
    },
    {
      "epoch": 0.030550716869891463,
      "grad_norm": 0.6842538326100387,
      "learning_rate": 4.999995131429801e-06,
      "loss": 0.692,
      "step": 114
    },
    {
      "epoch": 0.030818705614364197,
      "grad_norm": 0.658526235386219,
      "learning_rate": 4.9999944110803975e-06,
      "loss": 0.7112,
      "step": 115
    },
    {
      "epoch": 0.03108669435883693,
      "grad_norm": 0.7079888961087176,
      "learning_rate": 4.9999936410518e-06,
      "loss": 0.6781,
      "step": 116
    },
    {
      "epoch": 0.03135468310330966,
      "grad_norm": 0.6566184724767354,
      "learning_rate": 4.999992821344026e-06,
      "loss": 0.7253,
      "step": 117
    },
    {
      "epoch": 0.03162267184778239,
      "grad_norm": 0.667044349429523,
      "learning_rate": 4.999991951957091e-06,
      "loss": 0.6844,
      "step": 118
    },
    {
      "epoch": 0.03189066059225513,
      "grad_norm": 0.6981136774306808,
      "learning_rate": 4.999991032891012e-06,
      "loss": 0.7168,
      "step": 119
    },
    {
      "epoch": 0.03215864933672786,
      "grad_norm": 0.6694802209707341,
      "learning_rate": 4.999990064145807e-06,
      "loss": 0.711,
      "step": 120
    },
    {
      "epoch": 0.03242663808120059,
      "grad_norm": 0.6908705309905183,
      "learning_rate": 4.999989045721496e-06,
      "loss": 0.6747,
      "step": 121
    },
    {
      "epoch": 0.03269462682567332,
      "grad_norm": 0.6592775130597462,
      "learning_rate": 4.999987977618099e-06,
      "loss": 0.6752,
      "step": 122
    },
    {
      "epoch": 0.032962615570146056,
      "grad_norm": 0.6710613575749856,
      "learning_rate": 4.999986859835637e-06,
      "loss": 0.665,
      "step": 123
    },
    {
      "epoch": 0.033230604314618786,
      "grad_norm": 0.6756997971686135,
      "learning_rate": 4.999985692374132e-06,
      "loss": 0.7149,
      "step": 124
    },
    {
      "epoch": 0.033498593059091517,
      "grad_norm": 0.720111164104893,
      "learning_rate": 4.9999844752336076e-06,
      "loss": 0.6982,
      "step": 125
    },
    {
      "epoch": 0.033766581803564254,
      "grad_norm": 0.7281942457446365,
      "learning_rate": 4.999983208414089e-06,
      "loss": 0.7293,
      "step": 126
    },
    {
      "epoch": 0.034034570548036984,
      "grad_norm": 0.697618015655422,
      "learning_rate": 4.999981891915599e-06,
      "loss": 0.7556,
      "step": 127
    },
    {
      "epoch": 0.034302559292509714,
      "grad_norm": 0.6670427501733045,
      "learning_rate": 4.9999805257381655e-06,
      "loss": 0.7216,
      "step": 128
    },
    {
      "epoch": 0.034570548036982444,
      "grad_norm": 0.6788323532017639,
      "learning_rate": 4.999979109881815e-06,
      "loss": 0.7187,
      "step": 129
    },
    {
      "epoch": 0.03483853678145518,
      "grad_norm": 0.7198575536196861,
      "learning_rate": 4.9999776443465764e-06,
      "loss": 0.7217,
      "step": 130
    },
    {
      "epoch": 0.03510652552592791,
      "grad_norm": 0.6628844565381232,
      "learning_rate": 4.999976129132478e-06,
      "loss": 0.6864,
      "step": 131
    },
    {
      "epoch": 0.03537451427040064,
      "grad_norm": 0.6659518186659468,
      "learning_rate": 4.9999745642395494e-06,
      "loss": 0.7091,
      "step": 132
    },
    {
      "epoch": 0.03564250301487337,
      "grad_norm": 0.7479769949175036,
      "learning_rate": 4.999972949667823e-06,
      "loss": 0.6698,
      "step": 133
    },
    {
      "epoch": 0.03591049175934611,
      "grad_norm": 0.6875985191349179,
      "learning_rate": 4.99997128541733e-06,
      "loss": 0.7245,
      "step": 134
    },
    {
      "epoch": 0.03617848050381884,
      "grad_norm": 0.642495021266916,
      "learning_rate": 4.999969571488105e-06,
      "loss": 0.6791,
      "step": 135
    },
    {
      "epoch": 0.03644646924829157,
      "grad_norm": 0.6873501316777596,
      "learning_rate": 4.99996780788018e-06,
      "loss": 0.7512,
      "step": 136
    },
    {
      "epoch": 0.03671445799276431,
      "grad_norm": 0.6986437383111996,
      "learning_rate": 4.999965994593591e-06,
      "loss": 0.72,
      "step": 137
    },
    {
      "epoch": 0.03698244673723704,
      "grad_norm": 0.6553595477312628,
      "learning_rate": 4.999964131628374e-06,
      "loss": 0.6659,
      "step": 138
    },
    {
      "epoch": 0.03725043548170977,
      "grad_norm": 0.679109555923408,
      "learning_rate": 4.999962218984566e-06,
      "loss": 0.7055,
      "step": 139
    },
    {
      "epoch": 0.0375184242261825,
      "grad_norm": 0.6510382838909927,
      "learning_rate": 4.9999602566622055e-06,
      "loss": 0.7174,
      "step": 140
    },
    {
      "epoch": 0.037786412970655235,
      "grad_norm": 0.6974225807602472,
      "learning_rate": 4.99995824466133e-06,
      "loss": 0.7159,
      "step": 141
    },
    {
      "epoch": 0.038054401715127965,
      "grad_norm": 0.7278354928779344,
      "learning_rate": 4.999956182981981e-06,
      "loss": 0.6387,
      "step": 142
    },
    {
      "epoch": 0.038322390459600696,
      "grad_norm": 0.675117966369623,
      "learning_rate": 4.999954071624198e-06,
      "loss": 0.7023,
      "step": 143
    },
    {
      "epoch": 0.038590379204073426,
      "grad_norm": 0.6375421251369112,
      "learning_rate": 4.999951910588025e-06,
      "loss": 0.6618,
      "step": 144
    },
    {
      "epoch": 0.03885836794854616,
      "grad_norm": 0.6844552013559212,
      "learning_rate": 4.999949699873504e-06,
      "loss": 0.7008,
      "step": 145
    },
    {
      "epoch": 0.03912635669301889,
      "grad_norm": 0.6978528021660166,
      "learning_rate": 4.999947439480678e-06,
      "loss": 0.717,
      "step": 146
    },
    {
      "epoch": 0.039394345437491624,
      "grad_norm": 0.6690310402203973,
      "learning_rate": 4.999945129409593e-06,
      "loss": 0.7016,
      "step": 147
    },
    {
      "epoch": 0.03966233418196436,
      "grad_norm": 0.6921430209410233,
      "learning_rate": 4.999942769660294e-06,
      "loss": 0.7362,
      "step": 148
    },
    {
      "epoch": 0.03993032292643709,
      "grad_norm": 0.6810326681225496,
      "learning_rate": 4.999940360232829e-06,
      "loss": 0.7046,
      "step": 149
    },
    {
      "epoch": 0.04019831167090982,
      "grad_norm": 0.697699105861109,
      "learning_rate": 4.999937901127246e-06,
      "loss": 0.6969,
      "step": 150
    },
    {
      "epoch": 0.04046630041538255,
      "grad_norm": 0.7220338080748768,
      "learning_rate": 4.999935392343592e-06,
      "loss": 0.7232,
      "step": 151
    },
    {
      "epoch": 0.04073428915985529,
      "grad_norm": 0.6903633630163049,
      "learning_rate": 4.999932833881919e-06,
      "loss": 0.6947,
      "step": 152
    },
    {
      "epoch": 0.04100227790432802,
      "grad_norm": 0.6521501350213141,
      "learning_rate": 4.999930225742276e-06,
      "loss": 0.6866,
      "step": 153
    },
    {
      "epoch": 0.04127026664880075,
      "grad_norm": 0.6741178391234263,
      "learning_rate": 4.999927567924718e-06,
      "loss": 0.6665,
      "step": 154
    },
    {
      "epoch": 0.04153825539327348,
      "grad_norm": 0.6675565205922853,
      "learning_rate": 4.9999248604292926e-06,
      "loss": 0.7163,
      "step": 155
    },
    {
      "epoch": 0.04180624413774622,
      "grad_norm": 0.6867973039593035,
      "learning_rate": 4.999922103256058e-06,
      "loss": 0.7148,
      "step": 156
    },
    {
      "epoch": 0.04207423288221895,
      "grad_norm": 0.7118969463851695,
      "learning_rate": 4.999919296405067e-06,
      "loss": 0.7172,
      "step": 157
    },
    {
      "epoch": 0.04234222162669168,
      "grad_norm": 0.6495962340736448,
      "learning_rate": 4.999916439876377e-06,
      "loss": 0.7034,
      "step": 158
    },
    {
      "epoch": 0.042610210371164414,
      "grad_norm": 0.674279291343455,
      "learning_rate": 4.999913533670043e-06,
      "loss": 0.6736,
      "step": 159
    },
    {
      "epoch": 0.042878199115637144,
      "grad_norm": 0.6611066930729592,
      "learning_rate": 4.9999105777861236e-06,
      "loss": 0.7031,
      "step": 160
    },
    {
      "epoch": 0.043146187860109875,
      "grad_norm": 0.6541531089178088,
      "learning_rate": 4.999907572224677e-06,
      "loss": 0.6953,
      "step": 161
    },
    {
      "epoch": 0.043414176604582605,
      "grad_norm": 0.6479938487970461,
      "learning_rate": 4.9999045169857634e-06,
      "loss": 0.6805,
      "step": 162
    },
    {
      "epoch": 0.04368216534905534,
      "grad_norm": 0.675466945015435,
      "learning_rate": 4.999901412069443e-06,
      "loss": 0.7002,
      "step": 163
    },
    {
      "epoch": 0.04395015409352807,
      "grad_norm": 0.6741522614424119,
      "learning_rate": 4.999898257475779e-06,
      "loss": 0.71,
      "step": 164
    },
    {
      "epoch": 0.0442181428380008,
      "grad_norm": 0.6784809770720124,
      "learning_rate": 4.999895053204833e-06,
      "loss": 0.6974,
      "step": 165
    },
    {
      "epoch": 0.04448613158247353,
      "grad_norm": 0.660076178113711,
      "learning_rate": 4.999891799256668e-06,
      "loss": 0.6956,
      "step": 166
    },
    {
      "epoch": 0.04475412032694627,
      "grad_norm": 0.6688931265898396,
      "learning_rate": 4.999888495631349e-06,
      "loss": 0.6932,
      "step": 167
    },
    {
      "epoch": 0.045022109071419,
      "grad_norm": 0.6852709166298661,
      "learning_rate": 4.9998851423289426e-06,
      "loss": 0.7078,
      "step": 168
    },
    {
      "epoch": 0.04529009781589173,
      "grad_norm": 0.6712167698996772,
      "learning_rate": 4.999881739349515e-06,
      "loss": 0.6925,
      "step": 169
    },
    {
      "epoch": 0.04555808656036447,
      "grad_norm": 0.7838455350835052,
      "learning_rate": 4.999878286693133e-06,
      "loss": 0.7256,
      "step": 170
    },
    {
      "epoch": 0.0458260753048372,
      "grad_norm": 0.6982742105757692,
      "learning_rate": 4.999874784359866e-06,
      "loss": 0.6509,
      "step": 171
    },
    {
      "epoch": 0.04609406404930993,
      "grad_norm": 0.6837866579372184,
      "learning_rate": 4.999871232349785e-06,
      "loss": 0.7065,
      "step": 172
    },
    {
      "epoch": 0.04636205279378266,
      "grad_norm": 0.7171788397937933,
      "learning_rate": 4.9998676306629565e-06,
      "loss": 0.6816,
      "step": 173
    },
    {
      "epoch": 0.046630041538255396,
      "grad_norm": 0.6693609952402656,
      "learning_rate": 4.999863979299457e-06,
      "loss": 0.6495,
      "step": 174
    },
    {
      "epoch": 0.046898030282728126,
      "grad_norm": 0.6775841239302329,
      "learning_rate": 4.999860278259355e-06,
      "loss": 0.7164,
      "step": 175
    },
    {
      "epoch": 0.047166019027200856,
      "grad_norm": 0.7221811255858932,
      "learning_rate": 4.999856527542726e-06,
      "loss": 0.7121,
      "step": 176
    },
    {
      "epoch": 0.047434007771673586,
      "grad_norm": 0.6950519075922282,
      "learning_rate": 4.999852727149645e-06,
      "loss": 0.6617,
      "step": 177
    },
    {
      "epoch": 0.047701996516146324,
      "grad_norm": 0.7094078777776966,
      "learning_rate": 4.999848877080186e-06,
      "loss": 0.6781,
      "step": 178
    },
    {
      "epoch": 0.047969985260619054,
      "grad_norm": 0.6895194655375466,
      "learning_rate": 4.999844977334427e-06,
      "loss": 0.6905,
      "step": 179
    },
    {
      "epoch": 0.048237974005091784,
      "grad_norm": 0.6968825176478901,
      "learning_rate": 4.999841027912444e-06,
      "loss": 0.6969,
      "step": 180
    },
    {
      "epoch": 0.04850596274956452,
      "grad_norm": 0.7324587611776642,
      "learning_rate": 4.999837028814317e-06,
      "loss": 0.6925,
      "step": 181
    },
    {
      "epoch": 0.04877395149403725,
      "grad_norm": 0.6848383357030712,
      "learning_rate": 4.9998329800401245e-06,
      "loss": 0.6789,
      "step": 182
    },
    {
      "epoch": 0.04904194023850998,
      "grad_norm": 0.6904606748463983,
      "learning_rate": 4.999828881589947e-06,
      "loss": 0.6884,
      "step": 183
    },
    {
      "epoch": 0.04930992898298271,
      "grad_norm": 0.7706376692591392,
      "learning_rate": 4.999824733463866e-06,
      "loss": 0.6869,
      "step": 184
    },
    {
      "epoch": 0.04957791772745545,
      "grad_norm": 0.7578093298529758,
      "learning_rate": 4.999820535661964e-06,
      "loss": 0.7,
      "step": 185
    },
    {
      "epoch": 0.04984590647192818,
      "grad_norm": 0.6964923511143115,
      "learning_rate": 4.999816288184325e-06,
      "loss": 0.6773,
      "step": 186
    },
    {
      "epoch": 0.05011389521640091,
      "grad_norm": 0.6887532773339898,
      "learning_rate": 4.999811991031033e-06,
      "loss": 0.6839,
      "step": 187
    },
    {
      "epoch": 0.05038188396087364,
      "grad_norm": 0.703340258625519,
      "learning_rate": 4.9998076442021725e-06,
      "loss": 0.6479,
      "step": 188
    },
    {
      "epoch": 0.05064987270534638,
      "grad_norm": 0.6473239124242879,
      "learning_rate": 4.999803247697831e-06,
      "loss": 0.6727,
      "step": 189
    },
    {
      "epoch": 0.05091786144981911,
      "grad_norm": 0.7034417595814051,
      "learning_rate": 4.9997988015180955e-06,
      "loss": 0.7006,
      "step": 190
    },
    {
      "epoch": 0.05118585019429184,
      "grad_norm": 0.6905146470773477,
      "learning_rate": 4.999794305663054e-06,
      "loss": 0.7182,
      "step": 191
    },
    {
      "epoch": 0.051453838938764575,
      "grad_norm": 0.6938613902529127,
      "learning_rate": 4.999789760132797e-06,
      "loss": 0.7151,
      "step": 192
    },
    {
      "epoch": 0.051721827683237305,
      "grad_norm": 0.6569148020846134,
      "learning_rate": 4.999785164927413e-06,
      "loss": 0.6596,
      "step": 193
    },
    {
      "epoch": 0.051989816427710035,
      "grad_norm": 0.7050030859145198,
      "learning_rate": 4.999780520046995e-06,
      "loss": 0.6868,
      "step": 194
    },
    {
      "epoch": 0.052257805172182765,
      "grad_norm": 0.6855157319841305,
      "learning_rate": 4.999775825491635e-06,
      "loss": 0.6628,
      "step": 195
    },
    {
      "epoch": 0.0525257939166555,
      "grad_norm": 0.7457258527382111,
      "learning_rate": 4.999771081261425e-06,
      "loss": 0.6333,
      "step": 196
    },
    {
      "epoch": 0.05279378266112823,
      "grad_norm": 0.7251845984036154,
      "learning_rate": 4.999766287356461e-06,
      "loss": 0.6936,
      "step": 197
    },
    {
      "epoch": 0.05306177140560096,
      "grad_norm": 0.6780231563054212,
      "learning_rate": 4.999761443776838e-06,
      "loss": 0.6732,
      "step": 198
    },
    {
      "epoch": 0.0533297601500737,
      "grad_norm": 0.7147963256564284,
      "learning_rate": 4.999756550522651e-06,
      "loss": 0.7032,
      "step": 199
    },
    {
      "epoch": 0.05359774889454643,
      "grad_norm": 0.7180598792183331,
      "learning_rate": 4.999751607593997e-06,
      "loss": 0.6864,
      "step": 200
    },
    {
      "epoch": 0.05386573763901916,
      "grad_norm": 0.7443496428074027,
      "learning_rate": 4.9997466149909765e-06,
      "loss": 0.6971,
      "step": 201
    },
    {
      "epoch": 0.05413372638349189,
      "grad_norm": 0.7115227907078061,
      "learning_rate": 4.999741572713688e-06,
      "loss": 0.7075,
      "step": 202
    },
    {
      "epoch": 0.05440171512796463,
      "grad_norm": 0.6999404078887073,
      "learning_rate": 4.99973648076223e-06,
      "loss": 0.686,
      "step": 203
    },
    {
      "epoch": 0.05466970387243736,
      "grad_norm": 0.6830371546301146,
      "learning_rate": 4.999731339136704e-06,
      "loss": 0.6747,
      "step": 204
    },
    {
      "epoch": 0.05493769261691009,
      "grad_norm": 0.7118027344085432,
      "learning_rate": 4.999726147837215e-06,
      "loss": 0.6802,
      "step": 205
    },
    {
      "epoch": 0.05520568136138282,
      "grad_norm": 0.6878189227896037,
      "learning_rate": 4.999720906863863e-06,
      "loss": 0.6931,
      "step": 206
    },
    {
      "epoch": 0.055473670105855556,
      "grad_norm": 0.7129424657765884,
      "learning_rate": 4.999715616216753e-06,
      "loss": 0.6914,
      "step": 207
    },
    {
      "epoch": 0.055741658850328286,
      "grad_norm": 0.6781144971005258,
      "learning_rate": 4.999710275895991e-06,
      "loss": 0.6795,
      "step": 208
    },
    {
      "epoch": 0.05600964759480102,
      "grad_norm": 0.6853067640471792,
      "learning_rate": 4.999704885901681e-06,
      "loss": 0.671,
      "step": 209
    },
    {
      "epoch": 0.056277636339273754,
      "grad_norm": 0.6857189475339563,
      "learning_rate": 4.999699446233934e-06,
      "loss": 0.6814,
      "step": 210
    },
    {
      "epoch": 0.056545625083746484,
      "grad_norm": 0.6801797785260875,
      "learning_rate": 4.999693956892855e-06,
      "loss": 0.6711,
      "step": 211
    },
    {
      "epoch": 0.056813613828219214,
      "grad_norm": 0.6946512106147108,
      "learning_rate": 4.999688417878553e-06,
      "loss": 0.67,
      "step": 212
    },
    {
      "epoch": 0.057081602572691945,
      "grad_norm": 0.7057396837690668,
      "learning_rate": 4.999682829191139e-06,
      "loss": 0.7007,
      "step": 213
    },
    {
      "epoch": 0.05734959131716468,
      "grad_norm": 0.6754847436531932,
      "learning_rate": 4.999677190830724e-06,
      "loss": 0.6665,
      "step": 214
    },
    {
      "epoch": 0.05761758006163741,
      "grad_norm": 0.7823126116721931,
      "learning_rate": 4.9996715027974205e-06,
      "loss": 0.6942,
      "step": 215
    },
    {
      "epoch": 0.05788556880611014,
      "grad_norm": 0.7390447606138217,
      "learning_rate": 4.9996657650913415e-06,
      "loss": 0.6877,
      "step": 216
    },
    {
      "epoch": 0.05815355755058287,
      "grad_norm": 0.6791298589813068,
      "learning_rate": 4.9996599777126e-06,
      "loss": 0.7054,
      "step": 217
    },
    {
      "epoch": 0.05842154629505561,
      "grad_norm": 0.6782880826405979,
      "learning_rate": 4.999654140661311e-06,
      "loss": 0.6719,
      "step": 218
    },
    {
      "epoch": 0.05868953503952834,
      "grad_norm": 0.6822112418270222,
      "learning_rate": 4.999648253937591e-06,
      "loss": 0.6618,
      "step": 219
    },
    {
      "epoch": 0.05895752378400107,
      "grad_norm": 0.7058146884589738,
      "learning_rate": 4.999642317541558e-06,
      "loss": 0.6873,
      "step": 220
    },
    {
      "epoch": 0.05922551252847381,
      "grad_norm": 0.7093728833671098,
      "learning_rate": 4.999636331473327e-06,
      "loss": 0.6799,
      "step": 221
    },
    {
      "epoch": 0.05949350127294654,
      "grad_norm": 0.6763776032340204,
      "learning_rate": 4.999630295733021e-06,
      "loss": 0.6599,
      "step": 222
    },
    {
      "epoch": 0.05976149001741927,
      "grad_norm": 0.7261238696489111,
      "learning_rate": 4.999624210320757e-06,
      "loss": 0.6835,
      "step": 223
    },
    {
      "epoch": 0.060029478761892,
      "grad_norm": 0.7015968622568214,
      "learning_rate": 4.999618075236658e-06,
      "loss": 0.68,
      "step": 224
    },
    {
      "epoch": 0.060297467506364735,
      "grad_norm": 0.6939853281993197,
      "learning_rate": 4.999611890480843e-06,
      "loss": 0.6998,
      "step": 225
    },
    {
      "epoch": 0.060565456250837466,
      "grad_norm": 0.7025008802198147,
      "learning_rate": 4.999605656053437e-06,
      "loss": 0.6912,
      "step": 226
    },
    {
      "epoch": 0.060833444995310196,
      "grad_norm": 0.6952140463197894,
      "learning_rate": 4.9995993719545645e-06,
      "loss": 0.691,
      "step": 227
    },
    {
      "epoch": 0.061101433739782926,
      "grad_norm": 0.6792197196687223,
      "learning_rate": 4.999593038184348e-06,
      "loss": 0.6788,
      "step": 228
    },
    {
      "epoch": 0.06136942248425566,
      "grad_norm": 0.6694144910151926,
      "learning_rate": 4.999586654742915e-06,
      "loss": 0.6593,
      "step": 229
    },
    {
      "epoch": 0.06163741122872839,
      "grad_norm": 0.6990847283794798,
      "learning_rate": 4.999580221630393e-06,
      "loss": 0.6922,
      "step": 230
    },
    {
      "epoch": 0.061905399973201124,
      "grad_norm": 0.6676651652991394,
      "learning_rate": 4.999573738846908e-06,
      "loss": 0.6562,
      "step": 231
    },
    {
      "epoch": 0.06217338871767386,
      "grad_norm": 0.6991132094177743,
      "learning_rate": 4.999567206392591e-06,
      "loss": 0.652,
      "step": 232
    },
    {
      "epoch": 0.06244137746214659,
      "grad_norm": 0.8367297619180661,
      "learning_rate": 4.999560624267569e-06,
      "loss": 0.6685,
      "step": 233
    },
    {
      "epoch": 0.06270936620661932,
      "grad_norm": 0.6883435942953368,
      "learning_rate": 4.999553992471976e-06,
      "loss": 0.7212,
      "step": 234
    },
    {
      "epoch": 0.06297735495109205,
      "grad_norm": 0.6854633061657015,
      "learning_rate": 4.999547311005941e-06,
      "loss": 0.6598,
      "step": 235
    },
    {
      "epoch": 0.06324534369556478,
      "grad_norm": 0.6870608953443497,
      "learning_rate": 4.999540579869598e-06,
      "loss": 0.6955,
      "step": 236
    },
    {
      "epoch": 0.06351333244003751,
      "grad_norm": 0.6862565467012413,
      "learning_rate": 4.999533799063081e-06,
      "loss": 0.7012,
      "step": 237
    },
    {
      "epoch": 0.06378132118451026,
      "grad_norm": 0.6925314563993943,
      "learning_rate": 4.9995269685865235e-06,
      "loss": 0.6842,
      "step": 238
    },
    {
      "epoch": 0.06404930992898299,
      "grad_norm": 0.6851164515667639,
      "learning_rate": 4.999520088440063e-06,
      "loss": 0.6779,
      "step": 239
    },
    {
      "epoch": 0.06431729867345572,
      "grad_norm": 0.6880391239669099,
      "learning_rate": 4.999513158623835e-06,
      "loss": 0.6832,
      "step": 240
    },
    {
      "epoch": 0.06458528741792845,
      "grad_norm": 0.7239687723923278,
      "learning_rate": 4.999506179137978e-06,
      "loss": 0.668,
      "step": 241
    },
    {
      "epoch": 0.06485327616240118,
      "grad_norm": 0.7153296475293269,
      "learning_rate": 4.9994991499826306e-06,
      "loss": 0.6818,
      "step": 242
    },
    {
      "epoch": 0.06512126490687391,
      "grad_norm": 0.6884011082788689,
      "learning_rate": 4.999492071157932e-06,
      "loss": 0.6473,
      "step": 243
    },
    {
      "epoch": 0.06538925365134664,
      "grad_norm": 0.6894486001897383,
      "learning_rate": 4.999484942664022e-06,
      "loss": 0.685,
      "step": 244
    },
    {
      "epoch": 0.06565724239581938,
      "grad_norm": 0.6774921171476828,
      "learning_rate": 4.999477764501045e-06,
      "loss": 0.6884,
      "step": 245
    },
    {
      "epoch": 0.06592523114029211,
      "grad_norm": 0.6936772347699397,
      "learning_rate": 4.999470536669141e-06,
      "loss": 0.7155,
      "step": 246
    },
    {
      "epoch": 0.06619321988476484,
      "grad_norm": 0.704257811799868,
      "learning_rate": 4.9994632591684545e-06,
      "loss": 0.671,
      "step": 247
    },
    {
      "epoch": 0.06646120862923757,
      "grad_norm": 0.6909532784856128,
      "learning_rate": 4.99945593199913e-06,
      "loss": 0.6511,
      "step": 248
    },
    {
      "epoch": 0.0667291973737103,
      "grad_norm": 0.7290996553505035,
      "learning_rate": 4.999448555161314e-06,
      "loss": 0.6755,
      "step": 249
    },
    {
      "epoch": 0.06699718611818303,
      "grad_norm": 0.7214831180491741,
      "learning_rate": 4.999441128655151e-06,
      "loss": 0.6799,
      "step": 250
    },
    {
      "epoch": 0.06726517486265576,
      "grad_norm": 0.6794384277202605,
      "learning_rate": 4.999433652480791e-06,
      "loss": 0.6807,
      "step": 251
    },
    {
      "epoch": 0.06753316360712851,
      "grad_norm": 0.7233220950404244,
      "learning_rate": 4.999426126638382e-06,
      "loss": 0.6621,
      "step": 252
    },
    {
      "epoch": 0.06780115235160124,
      "grad_norm": 0.6934700135323967,
      "learning_rate": 4.999418551128071e-06,
      "loss": 0.6929,
      "step": 253
    },
    {
      "epoch": 0.06806914109607397,
      "grad_norm": 0.6886244626170983,
      "learning_rate": 4.999410925950012e-06,
      "loss": 0.6822,
      "step": 254
    },
    {
      "epoch": 0.0683371298405467,
      "grad_norm": 0.6838638316502956,
      "learning_rate": 4.999403251104355e-06,
      "loss": 0.6561,
      "step": 255
    },
    {
      "epoch": 0.06860511858501943,
      "grad_norm": 0.7151422941797886,
      "learning_rate": 4.999395526591253e-06,
      "loss": 0.679,
      "step": 256
    },
    {
      "epoch": 0.06887310732949216,
      "grad_norm": 0.6980838843720663,
      "learning_rate": 4.999387752410859e-06,
      "loss": 0.6483,
      "step": 257
    },
    {
      "epoch": 0.06914109607396489,
      "grad_norm": 0.6771135907636605,
      "learning_rate": 4.999379928563328e-06,
      "loss": 0.6878,
      "step": 258
    },
    {
      "epoch": 0.06940908481843762,
      "grad_norm": 0.69104923852252,
      "learning_rate": 4.999372055048814e-06,
      "loss": 0.6691,
      "step": 259
    },
    {
      "epoch": 0.06967707356291036,
      "grad_norm": 0.693817691156683,
      "learning_rate": 4.999364131867476e-06,
      "loss": 0.6704,
      "step": 260
    },
    {
      "epoch": 0.0699450623073831,
      "grad_norm": 0.6995655390938745,
      "learning_rate": 4.999356159019468e-06,
      "loss": 0.6741,
      "step": 261
    },
    {
      "epoch": 0.07021305105185582,
      "grad_norm": 0.6958799917290299,
      "learning_rate": 4.999348136504953e-06,
      "loss": 0.6331,
      "step": 262
    },
    {
      "epoch": 0.07048103979632855,
      "grad_norm": 0.702634962691175,
      "learning_rate": 4.999340064324085e-06,
      "loss": 0.6929,
      "step": 263
    },
    {
      "epoch": 0.07074902854080128,
      "grad_norm": 0.698824407966974,
      "learning_rate": 4.999331942477029e-06,
      "loss": 0.6967,
      "step": 264
    },
    {
      "epoch": 0.07101701728527401,
      "grad_norm": 0.7149630888190478,
      "learning_rate": 4.999323770963945e-06,
      "loss": 0.6677,
      "step": 265
    },
    {
      "epoch": 0.07128500602974674,
      "grad_norm": 0.7300946363934405,
      "learning_rate": 4.999315549784995e-06,
      "loss": 0.6852,
      "step": 266
    },
    {
      "epoch": 0.07155299477421949,
      "grad_norm": 0.6799695808996903,
      "learning_rate": 4.9993072789403405e-06,
      "loss": 0.6504,
      "step": 267
    },
    {
      "epoch": 0.07182098351869222,
      "grad_norm": 0.7406398130010771,
      "learning_rate": 4.99929895843015e-06,
      "loss": 0.6801,
      "step": 268
    },
    {
      "epoch": 0.07208897226316495,
      "grad_norm": 0.7323820946195249,
      "learning_rate": 4.999290588254585e-06,
      "loss": 0.6271,
      "step": 269
    },
    {
      "epoch": 0.07235696100763768,
      "grad_norm": 0.7017959274217292,
      "learning_rate": 4.999282168413814e-06,
      "loss": 0.6841,
      "step": 270
    },
    {
      "epoch": 0.07262494975211041,
      "grad_norm": 0.7199990339962526,
      "learning_rate": 4.999273698908004e-06,
      "loss": 0.6898,
      "step": 271
    },
    {
      "epoch": 0.07289293849658314,
      "grad_norm": 0.7199797696570645,
      "learning_rate": 4.999265179737323e-06,
      "loss": 0.6568,
      "step": 272
    },
    {
      "epoch": 0.07316092724105587,
      "grad_norm": 0.702192736863511,
      "learning_rate": 4.999256610901941e-06,
      "loss": 0.6998,
      "step": 273
    },
    {
      "epoch": 0.07342891598552861,
      "grad_norm": 0.7515528607639175,
      "learning_rate": 4.999247992402026e-06,
      "loss": 0.665,
      "step": 274
    },
    {
      "epoch": 0.07369690473000134,
      "grad_norm": 0.6924024316819809,
      "learning_rate": 4.999239324237752e-06,
      "loss": 0.6683,
      "step": 275
    },
    {
      "epoch": 0.07396489347447407,
      "grad_norm": 0.6992431284469874,
      "learning_rate": 4.99923060640929e-06,
      "loss": 0.6441,
      "step": 276
    },
    {
      "epoch": 0.0742328822189468,
      "grad_norm": 0.7049979971268447,
      "learning_rate": 4.999221838916813e-06,
      "loss": 0.692,
      "step": 277
    },
    {
      "epoch": 0.07450087096341954,
      "grad_norm": 0.7239058188288647,
      "learning_rate": 4.999213021760497e-06,
      "loss": 0.6842,
      "step": 278
    },
    {
      "epoch": 0.07476885970789227,
      "grad_norm": 0.6956107852921796,
      "learning_rate": 4.999204154940515e-06,
      "loss": 0.6726,
      "step": 279
    },
    {
      "epoch": 0.075036848452365,
      "grad_norm": 0.7246354268425289,
      "learning_rate": 4.999195238457043e-06,
      "loss": 0.6419,
      "step": 280
    },
    {
      "epoch": 0.07530483719683773,
      "grad_norm": 0.6939746169953684,
      "learning_rate": 4.99918627231026e-06,
      "loss": 0.6592,
      "step": 281
    },
    {
      "epoch": 0.07557282594131047,
      "grad_norm": 0.6945879300390789,
      "learning_rate": 4.999177256500344e-06,
      "loss": 0.6399,
      "step": 282
    },
    {
      "epoch": 0.0758408146857832,
      "grad_norm": 0.6921135528771797,
      "learning_rate": 4.9991681910274725e-06,
      "loss": 0.679,
      "step": 283
    },
    {
      "epoch": 0.07610880343025593,
      "grad_norm": 0.7426544068244216,
      "learning_rate": 4.999159075891828e-06,
      "loss": 0.6558,
      "step": 284
    },
    {
      "epoch": 0.07637679217472866,
      "grad_norm": 0.7015731180954687,
      "learning_rate": 4.999149911093589e-06,
      "loss": 0.7137,
      "step": 285
    },
    {
      "epoch": 0.07664478091920139,
      "grad_norm": 0.6778383584192442,
      "learning_rate": 4.999140696632939e-06,
      "loss": 0.6466,
      "step": 286
    },
    {
      "epoch": 0.07691276966367412,
      "grad_norm": 0.6656088786424161,
      "learning_rate": 4.999131432510062e-06,
      "loss": 0.6816,
      "step": 287
    },
    {
      "epoch": 0.07718075840814685,
      "grad_norm": 0.7143464374108063,
      "learning_rate": 4.999122118725141e-06,
      "loss": 0.6636,
      "step": 288
    },
    {
      "epoch": 0.0774487471526196,
      "grad_norm": 0.7345662370843171,
      "learning_rate": 4.99911275527836e-06,
      "loss": 0.6535,
      "step": 289
    },
    {
      "epoch": 0.07771673589709233,
      "grad_norm": 0.681940317881374,
      "learning_rate": 4.999103342169908e-06,
      "loss": 0.6761,
      "step": 290
    },
    {
      "epoch": 0.07798472464156506,
      "grad_norm": 0.7235227019964181,
      "learning_rate": 4.999093879399969e-06,
      "loss": 0.6729,
      "step": 291
    },
    {
      "epoch": 0.07825271338603779,
      "grad_norm": 0.7509910993129191,
      "learning_rate": 4.9990843669687325e-06,
      "loss": 0.6409,
      "step": 292
    },
    {
      "epoch": 0.07852070213051052,
      "grad_norm": 0.7220628324955591,
      "learning_rate": 4.9990748048763875e-06,
      "loss": 0.5942,
      "step": 293
    },
    {
      "epoch": 0.07878869087498325,
      "grad_norm": 0.709743582061943,
      "learning_rate": 4.999065193123124e-06,
      "loss": 0.6645,
      "step": 294
    },
    {
      "epoch": 0.07905667961945598,
      "grad_norm": 0.7594892875068282,
      "learning_rate": 4.999055531709134e-06,
      "loss": 0.66,
      "step": 295
    },
    {
      "epoch": 0.07932466836392872,
      "grad_norm": 0.7244605722569276,
      "learning_rate": 4.999045820634607e-06,
      "loss": 0.6704,
      "step": 296
    },
    {
      "epoch": 0.07959265710840145,
      "grad_norm": 0.7262325719221914,
      "learning_rate": 4.999036059899738e-06,
      "loss": 0.6845,
      "step": 297
    },
    {
      "epoch": 0.07986064585287418,
      "grad_norm": 0.7373760618629712,
      "learning_rate": 4.99902624950472e-06,
      "loss": 0.6708,
      "step": 298
    },
    {
      "epoch": 0.08012863459734691,
      "grad_norm": 0.804171472256531,
      "learning_rate": 4.999016389449749e-06,
      "loss": 0.6678,
      "step": 299
    },
    {
      "epoch": 0.08039662334181964,
      "grad_norm": 0.7031117555302927,
      "learning_rate": 4.99900647973502e-06,
      "loss": 0.6774,
      "step": 300
    },
    {
      "epoch": 0.08066461208629237,
      "grad_norm": 0.6837209536802216,
      "learning_rate": 4.998996520360729e-06,
      "loss": 0.6707,
      "step": 301
    },
    {
      "epoch": 0.0809326008307651,
      "grad_norm": 0.6862035574523768,
      "learning_rate": 4.9989865113270765e-06,
      "loss": 0.6357,
      "step": 302
    },
    {
      "epoch": 0.08120058957523783,
      "grad_norm": 0.6929128430334767,
      "learning_rate": 4.99897645263426e-06,
      "loss": 0.655,
      "step": 303
    },
    {
      "epoch": 0.08146857831971058,
      "grad_norm": 0.7054738731190586,
      "learning_rate": 4.9989663442824785e-06,
      "loss": 0.6552,
      "step": 304
    },
    {
      "epoch": 0.08173656706418331,
      "grad_norm": 0.7475047794758737,
      "learning_rate": 4.998956186271934e-06,
      "loss": 0.6502,
      "step": 305
    },
    {
      "epoch": 0.08200455580865604,
      "grad_norm": 0.7306708058904263,
      "learning_rate": 4.9989459786028285e-06,
      "loss": 0.6564,
      "step": 306
    },
    {
      "epoch": 0.08227254455312877,
      "grad_norm": 0.7346621570166151,
      "learning_rate": 4.9989357212753645e-06,
      "loss": 0.6909,
      "step": 307
    },
    {
      "epoch": 0.0825405332976015,
      "grad_norm": 0.7123726537011248,
      "learning_rate": 4.998925414289745e-06,
      "loss": 0.6289,
      "step": 308
    },
    {
      "epoch": 0.08280852204207423,
      "grad_norm": 0.6908108016322119,
      "learning_rate": 4.9989150576461764e-06,
      "loss": 0.6552,
      "step": 309
    },
    {
      "epoch": 0.08307651078654696,
      "grad_norm": 0.7012523864179844,
      "learning_rate": 4.998904651344863e-06,
      "loss": 0.6902,
      "step": 310
    },
    {
      "epoch": 0.0833444995310197,
      "grad_norm": 0.6679044633977735,
      "learning_rate": 4.998894195386012e-06,
      "loss": 0.6368,
      "step": 311
    },
    {
      "epoch": 0.08361248827549243,
      "grad_norm": 0.7153584294750434,
      "learning_rate": 4.998883689769832e-06,
      "loss": 0.6627,
      "step": 312
    },
    {
      "epoch": 0.08388047701996516,
      "grad_norm": 0.6962842628649624,
      "learning_rate": 4.998873134496531e-06,
      "loss": 0.6517,
      "step": 313
    },
    {
      "epoch": 0.0841484657644379,
      "grad_norm": 0.7183530390907331,
      "learning_rate": 4.99886252956632e-06,
      "loss": 0.6345,
      "step": 314
    },
    {
      "epoch": 0.08441645450891062,
      "grad_norm": 0.7247216305959578,
      "learning_rate": 4.998851874979408e-06,
      "loss": 0.6608,
      "step": 315
    },
    {
      "epoch": 0.08468444325338335,
      "grad_norm": 0.7024537118696281,
      "learning_rate": 4.998841170736008e-06,
      "loss": 0.6417,
      "step": 316
    },
    {
      "epoch": 0.08495243199785608,
      "grad_norm": 0.7690200917294031,
      "learning_rate": 4.99883041683633e-06,
      "loss": 0.6348,
      "step": 317
    },
    {
      "epoch": 0.08522042074232883,
      "grad_norm": 0.80047858012253,
      "learning_rate": 4.998819613280593e-06,
      "loss": 0.6715,
      "step": 318
    },
    {
      "epoch": 0.08548840948680156,
      "grad_norm": 0.695003631437708,
      "learning_rate": 4.998808760069006e-06,
      "loss": 0.6748,
      "step": 319
    },
    {
      "epoch": 0.08575639823127429,
      "grad_norm": 0.7198448682908326,
      "learning_rate": 4.9987978572017875e-06,
      "loss": 0.6686,
      "step": 320
    },
    {
      "epoch": 0.08602438697574702,
      "grad_norm": 0.7274830529715616,
      "learning_rate": 4.998786904679155e-06,
      "loss": 0.6969,
      "step": 321
    },
    {
      "epoch": 0.08629237572021975,
      "grad_norm": 0.7319761918136307,
      "learning_rate": 4.998775902501324e-06,
      "loss": 0.6883,
      "step": 322
    },
    {
      "epoch": 0.08656036446469248,
      "grad_norm": 0.7428590143701496,
      "learning_rate": 4.9987648506685145e-06,
      "loss": 0.6548,
      "step": 323
    },
    {
      "epoch": 0.08682835320916521,
      "grad_norm": 0.7419225989264708,
      "learning_rate": 4.998753749180946e-06,
      "loss": 0.6757,
      "step": 324
    },
    {
      "epoch": 0.08709634195363795,
      "grad_norm": 0.7056779544608539,
      "learning_rate": 4.998742598038838e-06,
      "loss": 0.7033,
      "step": 325
    },
    {
      "epoch": 0.08736433069811068,
      "grad_norm": 0.7160289654940805,
      "learning_rate": 4.998731397242414e-06,
      "loss": 0.6828,
      "step": 326
    },
    {
      "epoch": 0.08763231944258341,
      "grad_norm": 0.7273323627719993,
      "learning_rate": 4.998720146791895e-06,
      "loss": 0.6837,
      "step": 327
    },
    {
      "epoch": 0.08790030818705614,
      "grad_norm": 0.6752210778602105,
      "learning_rate": 4.9987088466875055e-06,
      "loss": 0.6728,
      "step": 328
    },
    {
      "epoch": 0.08816829693152888,
      "grad_norm": 0.7341838333984912,
      "learning_rate": 4.998697496929469e-06,
      "loss": 0.6655,
      "step": 329
    },
    {
      "epoch": 0.0884362856760016,
      "grad_norm": 0.7026325797951407,
      "learning_rate": 4.9986860975180114e-06,
      "loss": 0.6484,
      "step": 330
    },
    {
      "epoch": 0.08870427442047434,
      "grad_norm": 0.7032249976043573,
      "learning_rate": 4.998674648453361e-06,
      "loss": 0.6462,
      "step": 331
    },
    {
      "epoch": 0.08897226316494707,
      "grad_norm": 0.6900795502372272,
      "learning_rate": 4.998663149735743e-06,
      "loss": 0.6335,
      "step": 332
    },
    {
      "epoch": 0.08924025190941981,
      "grad_norm": 0.7248561260867296,
      "learning_rate": 4.9986516013653865e-06,
      "loss": 0.6496,
      "step": 333
    },
    {
      "epoch": 0.08950824065389254,
      "grad_norm": 0.7344797117717442,
      "learning_rate": 4.998640003342522e-06,
      "loss": 0.6174,
      "step": 334
    },
    {
      "epoch": 0.08977622939836527,
      "grad_norm": 1.0845724744950842,
      "learning_rate": 4.998628355667378e-06,
      "loss": 0.6852,
      "step": 335
    },
    {
      "epoch": 0.090044218142838,
      "grad_norm": 0.8986483937542907,
      "learning_rate": 4.998616658340187e-06,
      "loss": 0.659,
      "step": 336
    },
    {
      "epoch": 0.09031220688731073,
      "grad_norm": 0.7440485771036833,
      "learning_rate": 4.998604911361184e-06,
      "loss": 0.6633,
      "step": 337
    },
    {
      "epoch": 0.09058019563178346,
      "grad_norm": 0.6926515276625024,
      "learning_rate": 4.998593114730597e-06,
      "loss": 0.6375,
      "step": 338
    },
    {
      "epoch": 0.09084818437625619,
      "grad_norm": 0.6961353004226783,
      "learning_rate": 4.998581268448665e-06,
      "loss": 0.673,
      "step": 339
    },
    {
      "epoch": 0.09111617312072894,
      "grad_norm": 0.7331783412915588,
      "learning_rate": 4.998569372515621e-06,
      "loss": 0.6401,
      "step": 340
    },
    {
      "epoch": 0.09138416186520167,
      "grad_norm": 0.689619820762504,
      "learning_rate": 4.998557426931704e-06,
      "loss": 0.6517,
      "step": 341
    },
    {
      "epoch": 0.0916521506096744,
      "grad_norm": 0.7046729599943601,
      "learning_rate": 4.998545431697149e-06,
      "loss": 0.6359,
      "step": 342
    },
    {
      "epoch": 0.09192013935414713,
      "grad_norm": 0.7343243946321495,
      "learning_rate": 4.998533386812194e-06,
      "loss": 0.6368,
      "step": 343
    },
    {
      "epoch": 0.09218812809861986,
      "grad_norm": 0.7122206678821871,
      "learning_rate": 4.99852129227708e-06,
      "loss": 0.6587,
      "step": 344
    },
    {
      "epoch": 0.09245611684309259,
      "grad_norm": 0.7133760056565577,
      "learning_rate": 4.9985091480920465e-06,
      "loss": 0.6823,
      "step": 345
    },
    {
      "epoch": 0.09272410558756532,
      "grad_norm": 0.7005693353918453,
      "learning_rate": 4.998496954257335e-06,
      "loss": 0.6559,
      "step": 346
    },
    {
      "epoch": 0.09299209433203806,
      "grad_norm": 0.7081715057478484,
      "learning_rate": 4.998484710773187e-06,
      "loss": 0.624,
      "step": 347
    },
    {
      "epoch": 0.09326008307651079,
      "grad_norm": 0.7438900302061195,
      "learning_rate": 4.998472417639848e-06,
      "loss": 0.6694,
      "step": 348
    },
    {
      "epoch": 0.09352807182098352,
      "grad_norm": 0.7139423238861503,
      "learning_rate": 4.99846007485756e-06,
      "loss": 0.6614,
      "step": 349
    },
    {
      "epoch": 0.09379606056545625,
      "grad_norm": 0.7040104198822995,
      "learning_rate": 4.9984476824265694e-06,
      "loss": 0.6283,
      "step": 350
    },
    {
      "epoch": 0.09406404930992898,
      "grad_norm": 0.704752523359492,
      "learning_rate": 4.998435240347123e-06,
      "loss": 0.6475,
      "step": 351
    },
    {
      "epoch": 0.09433203805440171,
      "grad_norm": 0.7051748369870924,
      "learning_rate": 4.998422748619466e-06,
      "loss": 0.6594,
      "step": 352
    },
    {
      "epoch": 0.09460002679887444,
      "grad_norm": 0.7198215571517756,
      "learning_rate": 4.998410207243849e-06,
      "loss": 0.6553,
      "step": 353
    },
    {
      "epoch": 0.09486801554334717,
      "grad_norm": 0.7428709312166283,
      "learning_rate": 4.998397616220519e-06,
      "loss": 0.6488,
      "step": 354
    },
    {
      "epoch": 0.09513600428781992,
      "grad_norm": 0.7170599861091119,
      "learning_rate": 4.998384975549728e-06,
      "loss": 0.671,
      "step": 355
    },
    {
      "epoch": 0.09540399303229265,
      "grad_norm": 0.759483111278123,
      "learning_rate": 4.998372285231727e-06,
      "loss": 0.6729,
      "step": 356
    },
    {
      "epoch": 0.09567198177676538,
      "grad_norm": 0.7414012069857584,
      "learning_rate": 4.9983595452667675e-06,
      "loss": 0.7094,
      "step": 357
    },
    {
      "epoch": 0.09593997052123811,
      "grad_norm": 0.7455240867129616,
      "learning_rate": 4.9983467556551024e-06,
      "loss": 0.6743,
      "step": 358
    },
    {
      "epoch": 0.09620795926571084,
      "grad_norm": 0.6966309985390555,
      "learning_rate": 4.998333916396986e-06,
      "loss": 0.6625,
      "step": 359
    },
    {
      "epoch": 0.09647594801018357,
      "grad_norm": 0.7897859656580913,
      "learning_rate": 4.998321027492674e-06,
      "loss": 0.6425,
      "step": 360
    },
    {
      "epoch": 0.0967439367546563,
      "grad_norm": 0.7359970834605611,
      "learning_rate": 4.998308088942423e-06,
      "loss": 0.6495,
      "step": 361
    },
    {
      "epoch": 0.09701192549912904,
      "grad_norm": 0.699636867957034,
      "learning_rate": 4.998295100746489e-06,
      "loss": 0.6494,
      "step": 362
    },
    {
      "epoch": 0.09727991424360177,
      "grad_norm": 0.715199657267403,
      "learning_rate": 4.9982820629051305e-06,
      "loss": 0.689,
      "step": 363
    },
    {
      "epoch": 0.0975479029880745,
      "grad_norm": 0.7858832809392697,
      "learning_rate": 4.998268975418606e-06,
      "loss": 0.6677,
      "step": 364
    },
    {
      "epoch": 0.09781589173254723,
      "grad_norm": 0.7411035595380772,
      "learning_rate": 4.998255838287177e-06,
      "loss": 0.6777,
      "step": 365
    },
    {
      "epoch": 0.09808388047701996,
      "grad_norm": 0.6920296545552138,
      "learning_rate": 4.998242651511104e-06,
      "loss": 0.6664,
      "step": 366
    },
    {
      "epoch": 0.0983518692214927,
      "grad_norm": 0.6942418591624254,
      "learning_rate": 4.998229415090647e-06,
      "loss": 0.6983,
      "step": 367
    },
    {
      "epoch": 0.09861985796596542,
      "grad_norm": 0.7406225833547359,
      "learning_rate": 4.998216129026072e-06,
      "loss": 0.6274,
      "step": 368
    },
    {
      "epoch": 0.09888784671043817,
      "grad_norm": 0.7437398148203294,
      "learning_rate": 4.9982027933176426e-06,
      "loss": 0.6773,
      "step": 369
    },
    {
      "epoch": 0.0991558354549109,
      "grad_norm": 0.7158340759588109,
      "learning_rate": 4.998189407965622e-06,
      "loss": 0.6484,
      "step": 370
    },
    {
      "epoch": 0.09942382419938363,
      "grad_norm": 0.7842227973978754,
      "learning_rate": 4.9981759729702774e-06,
      "loss": 0.6612,
      "step": 371
    },
    {
      "epoch": 0.09969181294385636,
      "grad_norm": 0.7025373534031218,
      "learning_rate": 4.998162488331875e-06,
      "loss": 0.6765,
      "step": 372
    },
    {
      "epoch": 0.09995980168832909,
      "grad_norm": 0.7361505479384735,
      "learning_rate": 4.998148954050684e-06,
      "loss": 0.6552,
      "step": 373
    },
    {
      "epoch": 0.10022779043280182,
      "grad_norm": 0.6936711775988299,
      "learning_rate": 4.998135370126973e-06,
      "loss": 0.6691,
      "step": 374
    },
    {
      "epoch": 0.10049577917727455,
      "grad_norm": 0.7033966321377483,
      "learning_rate": 4.99812173656101e-06,
      "loss": 0.6579,
      "step": 375
    },
    {
      "epoch": 0.10076376792174728,
      "grad_norm": 0.7321423673306457,
      "learning_rate": 4.998108053353069e-06,
      "loss": 0.6329,
      "step": 376
    },
    {
      "epoch": 0.10103175666622002,
      "grad_norm": 0.7280978448626989,
      "learning_rate": 4.998094320503421e-06,
      "loss": 0.6844,
      "step": 377
    },
    {
      "epoch": 0.10129974541069275,
      "grad_norm": 0.7005747582972888,
      "learning_rate": 4.998080538012336e-06,
      "loss": 0.6458,
      "step": 378
    },
    {
      "epoch": 0.10156773415516548,
      "grad_norm": 0.695939982586724,
      "learning_rate": 4.998066705880092e-06,
      "loss": 0.6989,
      "step": 379
    },
    {
      "epoch": 0.10183572289963821,
      "grad_norm": 0.7356393832190092,
      "learning_rate": 4.998052824106962e-06,
      "loss": 0.6748,
      "step": 380
    },
    {
      "epoch": 0.10210371164411094,
      "grad_norm": 0.7551097535020896,
      "learning_rate": 4.998038892693221e-06,
      "loss": 0.6785,
      "step": 381
    },
    {
      "epoch": 0.10237170038858368,
      "grad_norm": 0.7113378426873269,
      "learning_rate": 4.998024911639148e-06,
      "loss": 0.6661,
      "step": 382
    },
    {
      "epoch": 0.1026396891330564,
      "grad_norm": 0.7184237725916407,
      "learning_rate": 4.998010880945019e-06,
      "loss": 0.6554,
      "step": 383
    },
    {
      "epoch": 0.10290767787752915,
      "grad_norm": 0.6865190676100582,
      "learning_rate": 4.997996800611114e-06,
      "loss": 0.6677,
      "step": 384
    },
    {
      "epoch": 0.10317566662200188,
      "grad_norm": 0.7278817306199018,
      "learning_rate": 4.997982670637711e-06,
      "loss": 0.6231,
      "step": 385
    },
    {
      "epoch": 0.10344365536647461,
      "grad_norm": 0.7215051442175413,
      "learning_rate": 4.997968491025093e-06,
      "loss": 0.6928,
      "step": 386
    },
    {
      "epoch": 0.10371164411094734,
      "grad_norm": 0.7389021547978143,
      "learning_rate": 4.99795426177354e-06,
      "loss": 0.6525,
      "step": 387
    },
    {
      "epoch": 0.10397963285542007,
      "grad_norm": 0.7002116018362358,
      "learning_rate": 4.997939982883336e-06,
      "loss": 0.6655,
      "step": 388
    },
    {
      "epoch": 0.1042476215998928,
      "grad_norm": 0.6971598415534499,
      "learning_rate": 4.997925654354765e-06,
      "loss": 0.6483,
      "step": 389
    },
    {
      "epoch": 0.10451561034436553,
      "grad_norm": 0.7280530449818602,
      "learning_rate": 4.997911276188111e-06,
      "loss": 0.6722,
      "step": 390
    },
    {
      "epoch": 0.10478359908883828,
      "grad_norm": 0.7252675471852773,
      "learning_rate": 4.997896848383659e-06,
      "loss": 0.6922,
      "step": 391
    },
    {
      "epoch": 0.105051587833311,
      "grad_norm": 0.7850097161566785,
      "learning_rate": 4.997882370941696e-06,
      "loss": 0.6583,
      "step": 392
    },
    {
      "epoch": 0.10531957657778374,
      "grad_norm": 0.7070927737511717,
      "learning_rate": 4.99786784386251e-06,
      "loss": 0.6573,
      "step": 393
    },
    {
      "epoch": 0.10558756532225647,
      "grad_norm": 0.7400798253489763,
      "learning_rate": 4.997853267146391e-06,
      "loss": 0.6276,
      "step": 394
    },
    {
      "epoch": 0.1058555540667292,
      "grad_norm": 0.7155834082970496,
      "learning_rate": 4.997838640793628e-06,
      "loss": 0.6623,
      "step": 395
    },
    {
      "epoch": 0.10612354281120193,
      "grad_norm": 0.70258014635928,
      "learning_rate": 4.99782396480451e-06,
      "loss": 0.6638,
      "step": 396
    },
    {
      "epoch": 0.10639153155567466,
      "grad_norm": 0.7018966567350728,
      "learning_rate": 4.9978092391793296e-06,
      "loss": 0.6437,
      "step": 397
    },
    {
      "epoch": 0.1066595203001474,
      "grad_norm": 0.6936807662558884,
      "learning_rate": 4.99779446391838e-06,
      "loss": 0.6554,
      "step": 398
    },
    {
      "epoch": 0.10692750904462013,
      "grad_norm": 0.7054959089322956,
      "learning_rate": 4.9977796390219544e-06,
      "loss": 0.6608,
      "step": 399
    },
    {
      "epoch": 0.10719549778909286,
      "grad_norm": 0.6936877643245225,
      "learning_rate": 4.997764764490347e-06,
      "loss": 0.6479,
      "step": 400
    },
    {
      "epoch": 0.10746348653356559,
      "grad_norm": 0.7388076749122502,
      "learning_rate": 4.9977498403238546e-06,
      "loss": 0.6615,
      "step": 401
    },
    {
      "epoch": 0.10773147527803832,
      "grad_norm": 0.7066382938749577,
      "learning_rate": 4.997734866522772e-06,
      "loss": 0.6193,
      "step": 402
    },
    {
      "epoch": 0.10799946402251105,
      "grad_norm": 0.7249542297660045,
      "learning_rate": 4.997719843087398e-06,
      "loss": 0.703,
      "step": 403
    },
    {
      "epoch": 0.10826745276698378,
      "grad_norm": 0.7185595033744908,
      "learning_rate": 4.997704770018032e-06,
      "loss": 0.6343,
      "step": 404
    },
    {
      "epoch": 0.10853544151145651,
      "grad_norm": 0.7025537119657524,
      "learning_rate": 4.997689647314972e-06,
      "loss": 0.6822,
      "step": 405
    },
    {
      "epoch": 0.10880343025592926,
      "grad_norm": 0.7524201924675584,
      "learning_rate": 4.997674474978518e-06,
      "loss": 0.6792,
      "step": 406
    },
    {
      "epoch": 0.10907141900040199,
      "grad_norm": 0.7421394787095791,
      "learning_rate": 4.9976592530089725e-06,
      "loss": 0.6685,
      "step": 407
    },
    {
      "epoch": 0.10933940774487472,
      "grad_norm": 0.8378250588315985,
      "learning_rate": 4.997643981406638e-06,
      "loss": 0.7046,
      "step": 408
    },
    {
      "epoch": 0.10960739648934745,
      "grad_norm": 0.7216473901210232,
      "learning_rate": 4.9976286601718185e-06,
      "loss": 0.6467,
      "step": 409
    },
    {
      "epoch": 0.10987538523382018,
      "grad_norm": 0.7128034962750008,
      "learning_rate": 4.997613289304817e-06,
      "loss": 0.6547,
      "step": 410
    },
    {
      "epoch": 0.11014337397829291,
      "grad_norm": 0.7234562352680989,
      "learning_rate": 4.9975978688059405e-06,
      "loss": 0.6496,
      "step": 411
    },
    {
      "epoch": 0.11041136272276564,
      "grad_norm": 0.7323251373693421,
      "learning_rate": 4.997582398675494e-06,
      "loss": 0.6642,
      "step": 412
    },
    {
      "epoch": 0.11067935146723838,
      "grad_norm": 0.8081686572442874,
      "learning_rate": 4.997566878913786e-06,
      "loss": 0.6726,
      "step": 413
    },
    {
      "epoch": 0.11094734021171111,
      "grad_norm": 0.6940468331089432,
      "learning_rate": 4.997551309521124e-06,
      "loss": 0.6531,
      "step": 414
    },
    {
      "epoch": 0.11121532895618384,
      "grad_norm": 0.6979633474904187,
      "learning_rate": 4.997535690497819e-06,
      "loss": 0.6467,
      "step": 415
    },
    {
      "epoch": 0.11148331770065657,
      "grad_norm": 0.768429340851774,
      "learning_rate": 4.997520021844179e-06,
      "loss": 0.6604,
      "step": 416
    },
    {
      "epoch": 0.1117513064451293,
      "grad_norm": 0.680543466490247,
      "learning_rate": 4.997504303560517e-06,
      "loss": 0.6699,
      "step": 417
    },
    {
      "epoch": 0.11201929518960203,
      "grad_norm": 0.7174264392318738,
      "learning_rate": 4.9974885356471455e-06,
      "loss": 0.6474,
      "step": 418
    },
    {
      "epoch": 0.11228728393407476,
      "grad_norm": 0.6984198621517772,
      "learning_rate": 4.997472718104377e-06,
      "loss": 0.6499,
      "step": 419
    },
    {
      "epoch": 0.11255527267854751,
      "grad_norm": 0.7120137364493325,
      "learning_rate": 4.9974568509325265e-06,
      "loss": 0.6376,
      "step": 420
    },
    {
      "epoch": 0.11282326142302024,
      "grad_norm": 0.7445367936262278,
      "learning_rate": 4.997440934131909e-06,
      "loss": 0.6751,
      "step": 421
    },
    {
      "epoch": 0.11309125016749297,
      "grad_norm": 0.6827540859070237,
      "learning_rate": 4.997424967702839e-06,
      "loss": 0.6439,
      "step": 422
    },
    {
      "epoch": 0.1133592389119657,
      "grad_norm": 0.7558330456216659,
      "learning_rate": 4.997408951645638e-06,
      "loss": 0.6634,
      "step": 423
    },
    {
      "epoch": 0.11362722765643843,
      "grad_norm": 0.7257456728824587,
      "learning_rate": 4.99739288596062e-06,
      "loss": 0.62,
      "step": 424
    },
    {
      "epoch": 0.11389521640091116,
      "grad_norm": 0.6710636326529411,
      "learning_rate": 4.997376770648107e-06,
      "loss": 0.6226,
      "step": 425
    },
    {
      "epoch": 0.11416320514538389,
      "grad_norm": 0.6559474555184844,
      "learning_rate": 4.997360605708418e-06,
      "loss": 0.6001,
      "step": 426
    },
    {
      "epoch": 0.11443119388985662,
      "grad_norm": 0.7245415671028487,
      "learning_rate": 4.997344391141875e-06,
      "loss": 0.6272,
      "step": 427
    },
    {
      "epoch": 0.11469918263432936,
      "grad_norm": 0.7087362533586415,
      "learning_rate": 4.997328126948799e-06,
      "loss": 0.6677,
      "step": 428
    },
    {
      "epoch": 0.1149671713788021,
      "grad_norm": 0.7268792353740238,
      "learning_rate": 4.997311813129515e-06,
      "loss": 0.6818,
      "step": 429
    },
    {
      "epoch": 0.11523516012327482,
      "grad_norm": 0.7408692322535385,
      "learning_rate": 4.997295449684345e-06,
      "loss": 0.6225,
      "step": 430
    },
    {
      "epoch": 0.11550314886774755,
      "grad_norm": 0.6933045366608425,
      "learning_rate": 4.997279036613615e-06,
      "loss": 0.6307,
      "step": 431
    },
    {
      "epoch": 0.11577113761222028,
      "grad_norm": 0.7333910427052629,
      "learning_rate": 4.997262573917653e-06,
      "loss": 0.6616,
      "step": 432
    },
    {
      "epoch": 0.11603912635669301,
      "grad_norm": 0.7212855524739092,
      "learning_rate": 4.997246061596783e-06,
      "loss": 0.6674,
      "step": 433
    },
    {
      "epoch": 0.11630711510116575,
      "grad_norm": 0.7494771660073004,
      "learning_rate": 4.997229499651336e-06,
      "loss": 0.6551,
      "step": 434
    },
    {
      "epoch": 0.11657510384563849,
      "grad_norm": 0.7054843688445314,
      "learning_rate": 4.997212888081638e-06,
      "loss": 0.6485,
      "step": 435
    },
    {
      "epoch": 0.11684309259011122,
      "grad_norm": 0.7438381741650485,
      "learning_rate": 4.997196226888022e-06,
      "loss": 0.6651,
      "step": 436
    },
    {
      "epoch": 0.11711108133458395,
      "grad_norm": 0.7156244064820284,
      "learning_rate": 4.99717951607082e-06,
      "loss": 0.6919,
      "step": 437
    },
    {
      "epoch": 0.11737907007905668,
      "grad_norm": 0.7537695671039244,
      "learning_rate": 4.997162755630359e-06,
      "loss": 0.6445,
      "step": 438
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.706534012152982,
      "learning_rate": 4.997145945566977e-06,
      "loss": 0.6488,
      "step": 439
    },
    {
      "epoch": 0.11791504756800214,
      "grad_norm": 0.7055924253471998,
      "learning_rate": 4.997129085881005e-06,
      "loss": 0.6801,
      "step": 440
    },
    {
      "epoch": 0.11818303631247487,
      "grad_norm": 0.7248535255973472,
      "learning_rate": 4.99711217657278e-06,
      "loss": 0.6688,
      "step": 441
    },
    {
      "epoch": 0.11845102505694761,
      "grad_norm": 0.7432535537498254,
      "learning_rate": 4.997095217642637e-06,
      "loss": 0.6629,
      "step": 442
    },
    {
      "epoch": 0.11871901380142034,
      "grad_norm": 0.7067744196434438,
      "learning_rate": 4.997078209090913e-06,
      "loss": 0.6559,
      "step": 443
    },
    {
      "epoch": 0.11898700254589308,
      "grad_norm": 0.6966412076480927,
      "learning_rate": 4.997061150917946e-06,
      "loss": 0.6457,
      "step": 444
    },
    {
      "epoch": 0.1192549912903658,
      "grad_norm": 0.7241151361602904,
      "learning_rate": 4.997044043124076e-06,
      "loss": 0.6361,
      "step": 445
    },
    {
      "epoch": 0.11952298003483854,
      "grad_norm": 0.7047653049354295,
      "learning_rate": 4.9970268857096414e-06,
      "loss": 0.6543,
      "step": 446
    },
    {
      "epoch": 0.11979096877931127,
      "grad_norm": 0.6983496544670155,
      "learning_rate": 4.997009678674985e-06,
      "loss": 0.6831,
      "step": 447
    },
    {
      "epoch": 0.120058957523784,
      "grad_norm": 0.7287835922603818,
      "learning_rate": 4.996992422020447e-06,
      "loss": 0.6534,
      "step": 448
    },
    {
      "epoch": 0.12032694626825673,
      "grad_norm": 0.7409323509115785,
      "learning_rate": 4.996975115746371e-06,
      "loss": 0.6469,
      "step": 449
    },
    {
      "epoch": 0.12059493501272947,
      "grad_norm": 0.822815911904995,
      "learning_rate": 4.996957759853101e-06,
      "loss": 0.6505,
      "step": 450
    },
    {
      "epoch": 0.1208629237572022,
      "grad_norm": 0.7487462876408704,
      "learning_rate": 4.996940354340981e-06,
      "loss": 0.6738,
      "step": 451
    },
    {
      "epoch": 0.12113091250167493,
      "grad_norm": 0.7367892608986822,
      "learning_rate": 4.996922899210358e-06,
      "loss": 0.654,
      "step": 452
    },
    {
      "epoch": 0.12139890124614766,
      "grad_norm": 0.6917904897712016,
      "learning_rate": 4.996905394461579e-06,
      "loss": 0.6341,
      "step": 453
    },
    {
      "epoch": 0.12166688999062039,
      "grad_norm": 0.7165094882961384,
      "learning_rate": 4.996887840094991e-06,
      "loss": 0.6753,
      "step": 454
    },
    {
      "epoch": 0.12193487873509312,
      "grad_norm": 0.7539881678729049,
      "learning_rate": 4.996870236110943e-06,
      "loss": 0.6319,
      "step": 455
    },
    {
      "epoch": 0.12220286747956585,
      "grad_norm": 0.7804163737605706,
      "learning_rate": 4.996852582509786e-06,
      "loss": 0.6067,
      "step": 456
    },
    {
      "epoch": 0.1224708562240386,
      "grad_norm": 0.757515897412964,
      "learning_rate": 4.996834879291868e-06,
      "loss": 0.654,
      "step": 457
    },
    {
      "epoch": 0.12273884496851133,
      "grad_norm": 0.7787121213383815,
      "learning_rate": 4.996817126457544e-06,
      "loss": 0.662,
      "step": 458
    },
    {
      "epoch": 0.12300683371298406,
      "grad_norm": 0.7609998246594886,
      "learning_rate": 4.996799324007165e-06,
      "loss": 0.6797,
      "step": 459
    },
    {
      "epoch": 0.12327482245745679,
      "grad_norm": 0.7277082172228565,
      "learning_rate": 4.996781471941085e-06,
      "loss": 0.6447,
      "step": 460
    },
    {
      "epoch": 0.12354281120192952,
      "grad_norm": 0.7270821194080418,
      "learning_rate": 4.996763570259659e-06,
      "loss": 0.6689,
      "step": 461
    },
    {
      "epoch": 0.12381079994640225,
      "grad_norm": 0.797176179177899,
      "learning_rate": 4.9967456189632425e-06,
      "loss": 0.6897,
      "step": 462
    },
    {
      "epoch": 0.12407878869087498,
      "grad_norm": 0.7431074359553652,
      "learning_rate": 4.9967276180521924e-06,
      "loss": 0.6233,
      "step": 463
    },
    {
      "epoch": 0.12434677743534772,
      "grad_norm": 0.734139820758066,
      "learning_rate": 4.996709567526868e-06,
      "loss": 0.6605,
      "step": 464
    },
    {
      "epoch": 0.12461476617982045,
      "grad_norm": 0.69760175326402,
      "learning_rate": 4.996691467387624e-06,
      "loss": 0.6651,
      "step": 465
    },
    {
      "epoch": 0.12488275492429318,
      "grad_norm": 0.7056827468868845,
      "learning_rate": 4.996673317634824e-06,
      "loss": 0.5968,
      "step": 466
    },
    {
      "epoch": 0.1251507436687659,
      "grad_norm": 0.8806741814739946,
      "learning_rate": 4.996655118268826e-06,
      "loss": 0.6033,
      "step": 467
    },
    {
      "epoch": 0.12541873241323864,
      "grad_norm": 0.7371483994635754,
      "learning_rate": 4.996636869289993e-06,
      "loss": 0.6702,
      "step": 468
    },
    {
      "epoch": 0.12568672115771137,
      "grad_norm": 0.7399200679760553,
      "learning_rate": 4.996618570698688e-06,
      "loss": 0.6472,
      "step": 469
    },
    {
      "epoch": 0.1259547099021841,
      "grad_norm": 0.7346138283980898,
      "learning_rate": 4.996600222495274e-06,
      "loss": 0.6783,
      "step": 470
    },
    {
      "epoch": 0.12622269864665683,
      "grad_norm": 0.6956917881899172,
      "learning_rate": 4.996581824680115e-06,
      "loss": 0.6431,
      "step": 471
    },
    {
      "epoch": 0.12649068739112956,
      "grad_norm": 0.681586765753215,
      "learning_rate": 4.996563377253577e-06,
      "loss": 0.6034,
      "step": 472
    },
    {
      "epoch": 0.1267586761356023,
      "grad_norm": 0.7553028499119494,
      "learning_rate": 4.996544880216028e-06,
      "loss": 0.6713,
      "step": 473
    },
    {
      "epoch": 0.12702666488007502,
      "grad_norm": 0.715071351511907,
      "learning_rate": 4.996526333567833e-06,
      "loss": 0.6699,
      "step": 474
    },
    {
      "epoch": 0.12729465362454778,
      "grad_norm": 0.7288467174339455,
      "learning_rate": 4.996507737309363e-06,
      "loss": 0.64,
      "step": 475
    },
    {
      "epoch": 0.1275626423690205,
      "grad_norm": 0.7433868971087181,
      "learning_rate": 4.9964890914409855e-06,
      "loss": 0.6534,
      "step": 476
    },
    {
      "epoch": 0.12783063111349324,
      "grad_norm": 0.6596506452106721,
      "learning_rate": 4.996470395963074e-06,
      "loss": 0.6185,
      "step": 477
    },
    {
      "epoch": 0.12809861985796597,
      "grad_norm": 0.6883312291986582,
      "learning_rate": 4.9964516508759965e-06,
      "loss": 0.6772,
      "step": 478
    },
    {
      "epoch": 0.1283666086024387,
      "grad_norm": 0.7078873590912905,
      "learning_rate": 4.9964328561801265e-06,
      "loss": 0.6242,
      "step": 479
    },
    {
      "epoch": 0.12863459734691143,
      "grad_norm": 0.6903633453624124,
      "learning_rate": 4.996414011875839e-06,
      "loss": 0.6355,
      "step": 480
    },
    {
      "epoch": 0.12890258609138416,
      "grad_norm": 0.6804969252544444,
      "learning_rate": 4.996395117963508e-06,
      "loss": 0.68,
      "step": 481
    },
    {
      "epoch": 0.1291705748358569,
      "grad_norm": 0.6938488748659082,
      "learning_rate": 4.996376174443507e-06,
      "loss": 0.6496,
      "step": 482
    },
    {
      "epoch": 0.12943856358032962,
      "grad_norm": 0.681568358152509,
      "learning_rate": 4.996357181316214e-06,
      "loss": 0.6466,
      "step": 483
    },
    {
      "epoch": 0.12970655232480235,
      "grad_norm": 0.6870766401550165,
      "learning_rate": 4.996338138582007e-06,
      "loss": 0.6432,
      "step": 484
    },
    {
      "epoch": 0.12997454106927508,
      "grad_norm": 0.7100767934498968,
      "learning_rate": 4.9963190462412635e-06,
      "loss": 0.6569,
      "step": 485
    },
    {
      "epoch": 0.13024252981374781,
      "grad_norm": 0.7175021932803748,
      "learning_rate": 4.996299904294363e-06,
      "loss": 0.6578,
      "step": 486
    },
    {
      "epoch": 0.13051051855822055,
      "grad_norm": 0.7028203332504821,
      "learning_rate": 4.996280712741686e-06,
      "loss": 0.6259,
      "step": 487
    },
    {
      "epoch": 0.13077850730269328,
      "grad_norm": 0.7196331070904693,
      "learning_rate": 4.996261471583615e-06,
      "loss": 0.6678,
      "step": 488
    },
    {
      "epoch": 0.131046496047166,
      "grad_norm": 0.714185667508626,
      "learning_rate": 4.996242180820529e-06,
      "loss": 0.6189,
      "step": 489
    },
    {
      "epoch": 0.13131448479163876,
      "grad_norm": 0.7213603644287747,
      "learning_rate": 4.996222840452815e-06,
      "loss": 0.6778,
      "step": 490
    },
    {
      "epoch": 0.1315824735361115,
      "grad_norm": 0.7400343323591807,
      "learning_rate": 4.996203450480856e-06,
      "loss": 0.6414,
      "step": 491
    },
    {
      "epoch": 0.13185046228058422,
      "grad_norm": 0.8344486388208359,
      "learning_rate": 4.996184010905036e-06,
      "loss": 0.6599,
      "step": 492
    },
    {
      "epoch": 0.13211845102505695,
      "grad_norm": 0.704923084180257,
      "learning_rate": 4.996164521725743e-06,
      "loss": 0.6786,
      "step": 493
    },
    {
      "epoch": 0.13238643976952968,
      "grad_norm": 0.7395365805702716,
      "learning_rate": 4.996144982943364e-06,
      "loss": 0.6606,
      "step": 494
    },
    {
      "epoch": 0.13265442851400241,
      "grad_norm": 0.7167572215278386,
      "learning_rate": 4.996125394558287e-06,
      "loss": 0.6498,
      "step": 495
    },
    {
      "epoch": 0.13292241725847515,
      "grad_norm": 0.7137391691534993,
      "learning_rate": 4.9961057565709015e-06,
      "loss": 0.659,
      "step": 496
    },
    {
      "epoch": 0.13319040600294788,
      "grad_norm": 0.6748619768330496,
      "learning_rate": 4.996086068981597e-06,
      "loss": 0.6401,
      "step": 497
    },
    {
      "epoch": 0.1334583947474206,
      "grad_norm": 0.7241750047231857,
      "learning_rate": 4.996066331790766e-06,
      "loss": 0.6395,
      "step": 498
    },
    {
      "epoch": 0.13372638349189334,
      "grad_norm": 0.7680302835474424,
      "learning_rate": 4.9960465449988e-06,
      "loss": 0.6944,
      "step": 499
    },
    {
      "epoch": 0.13399437223636607,
      "grad_norm": 0.6727476756431184,
      "learning_rate": 4.996026708606091e-06,
      "loss": 0.6411,
      "step": 500
    },
    {
      "epoch": 0.1342623609808388,
      "grad_norm": 0.6663133266191728,
      "learning_rate": 4.9960068226130355e-06,
      "loss": 0.6156,
      "step": 501
    },
    {
      "epoch": 0.13453034972531153,
      "grad_norm": 0.6986750598950645,
      "learning_rate": 4.995986887020027e-06,
      "loss": 0.6724,
      "step": 502
    },
    {
      "epoch": 0.13479833846978426,
      "grad_norm": 0.7140566495366323,
      "learning_rate": 4.995966901827462e-06,
      "loss": 0.6144,
      "step": 503
    },
    {
      "epoch": 0.13506632721425701,
      "grad_norm": 0.6913610430858729,
      "learning_rate": 4.995946867035738e-06,
      "loss": 0.6407,
      "step": 504
    },
    {
      "epoch": 0.13533431595872975,
      "grad_norm": 0.6889673927035918,
      "learning_rate": 4.995926782645253e-06,
      "loss": 0.6549,
      "step": 505
    },
    {
      "epoch": 0.13560230470320248,
      "grad_norm": 0.7327251870063456,
      "learning_rate": 4.995906648656406e-06,
      "loss": 0.6704,
      "step": 506
    },
    {
      "epoch": 0.1358702934476752,
      "grad_norm": 0.7994583475039914,
      "learning_rate": 4.995886465069597e-06,
      "loss": 0.6456,
      "step": 507
    },
    {
      "epoch": 0.13613828219214794,
      "grad_norm": 0.6788596659531054,
      "learning_rate": 4.995866231885227e-06,
      "loss": 0.6398,
      "step": 508
    },
    {
      "epoch": 0.13640627093662067,
      "grad_norm": 0.7011523587787191,
      "learning_rate": 4.995845949103699e-06,
      "loss": 0.6292,
      "step": 509
    },
    {
      "epoch": 0.1366742596810934,
      "grad_norm": 0.7435342268861577,
      "learning_rate": 4.9958256167254145e-06,
      "loss": 0.6416,
      "step": 510
    },
    {
      "epoch": 0.13694224842556613,
      "grad_norm": 0.7079517973548682,
      "learning_rate": 4.995805234750779e-06,
      "loss": 0.6534,
      "step": 511
    },
    {
      "epoch": 0.13721023717003886,
      "grad_norm": 0.730707830966801,
      "learning_rate": 4.9957848031801975e-06,
      "loss": 0.656,
      "step": 512
    },
    {
      "epoch": 0.1374782259145116,
      "grad_norm": 0.6893617310281205,
      "learning_rate": 4.995764322014075e-06,
      "loss": 0.6663,
      "step": 513
    },
    {
      "epoch": 0.13774621465898432,
      "grad_norm": 0.7267933208933559,
      "learning_rate": 4.995743791252819e-06,
      "loss": 0.6725,
      "step": 514
    },
    {
      "epoch": 0.13801420340345705,
      "grad_norm": 0.7546681559999151,
      "learning_rate": 4.995723210896836e-06,
      "loss": 0.651,
      "step": 515
    },
    {
      "epoch": 0.13828219214792978,
      "grad_norm": 0.6876884581422139,
      "learning_rate": 4.995702580946538e-06,
      "loss": 0.6241,
      "step": 516
    },
    {
      "epoch": 0.1385501808924025,
      "grad_norm": 0.6866358004956024,
      "learning_rate": 4.995681901402333e-06,
      "loss": 0.6305,
      "step": 517
    },
    {
      "epoch": 0.13881816963687524,
      "grad_norm": 0.7298402240388256,
      "learning_rate": 4.995661172264632e-06,
      "loss": 0.6394,
      "step": 518
    },
    {
      "epoch": 0.139086158381348,
      "grad_norm": 0.7101035119418951,
      "learning_rate": 4.995640393533847e-06,
      "loss": 0.6786,
      "step": 519
    },
    {
      "epoch": 0.13935414712582073,
      "grad_norm": 0.7056852446744254,
      "learning_rate": 4.995619565210392e-06,
      "loss": 0.6683,
      "step": 520
    },
    {
      "epoch": 0.13962213587029346,
      "grad_norm": 0.6972765862356034,
      "learning_rate": 4.9955986872946795e-06,
      "loss": 0.6524,
      "step": 521
    },
    {
      "epoch": 0.1398901246147662,
      "grad_norm": 0.7165767186130404,
      "learning_rate": 4.995577759787124e-06,
      "loss": 0.6405,
      "step": 522
    },
    {
      "epoch": 0.14015811335923892,
      "grad_norm": 0.7384176365968446,
      "learning_rate": 4.995556782688143e-06,
      "loss": 0.6597,
      "step": 523
    },
    {
      "epoch": 0.14042610210371165,
      "grad_norm": 0.7280755926222787,
      "learning_rate": 4.995535755998153e-06,
      "loss": 0.6601,
      "step": 524
    },
    {
      "epoch": 0.14069409084818438,
      "grad_norm": 0.719202461241684,
      "learning_rate": 4.995514679717571e-06,
      "loss": 0.6681,
      "step": 525
    },
    {
      "epoch": 0.1409620795926571,
      "grad_norm": 0.7054731982009363,
      "learning_rate": 4.995493553846817e-06,
      "loss": 0.636,
      "step": 526
    },
    {
      "epoch": 0.14123006833712984,
      "grad_norm": 0.7655663298426513,
      "learning_rate": 4.995472378386308e-06,
      "loss": 0.6477,
      "step": 527
    },
    {
      "epoch": 0.14149805708160257,
      "grad_norm": 0.7137214881664379,
      "learning_rate": 4.995451153336468e-06,
      "loss": 0.6464,
      "step": 528
    },
    {
      "epoch": 0.1417660458260753,
      "grad_norm": 0.7150767656519134,
      "learning_rate": 4.995429878697719e-06,
      "loss": 0.6346,
      "step": 529
    },
    {
      "epoch": 0.14203403457054803,
      "grad_norm": 0.7270902845856231,
      "learning_rate": 4.99540855447048e-06,
      "loss": 0.634,
      "step": 530
    },
    {
      "epoch": 0.14230202331502076,
      "grad_norm": 0.7246033727617203,
      "learning_rate": 4.9953871806551784e-06,
      "loss": 0.6559,
      "step": 531
    },
    {
      "epoch": 0.1425700120594935,
      "grad_norm": 0.7092396423776591,
      "learning_rate": 4.995365757252238e-06,
      "loss": 0.6636,
      "step": 532
    },
    {
      "epoch": 0.14283800080396622,
      "grad_norm": 0.6907605361176238,
      "learning_rate": 4.995344284262083e-06,
      "loss": 0.634,
      "step": 533
    },
    {
      "epoch": 0.14310598954843898,
      "grad_norm": 0.7111780131866282,
      "learning_rate": 4.995322761685142e-06,
      "loss": 0.6857,
      "step": 534
    },
    {
      "epoch": 0.1433739782929117,
      "grad_norm": 0.8148369343221756,
      "learning_rate": 4.995301189521842e-06,
      "loss": 0.6326,
      "step": 535
    },
    {
      "epoch": 0.14364196703738444,
      "grad_norm": 0.7282365605563551,
      "learning_rate": 4.995279567772612e-06,
      "loss": 0.6463,
      "step": 536
    },
    {
      "epoch": 0.14390995578185717,
      "grad_norm": 0.685155365911219,
      "learning_rate": 4.9952578964378805e-06,
      "loss": 0.6654,
      "step": 537
    },
    {
      "epoch": 0.1441779445263299,
      "grad_norm": 0.6945919288185022,
      "learning_rate": 4.99523617551808e-06,
      "loss": 0.6537,
      "step": 538
    },
    {
      "epoch": 0.14444593327080263,
      "grad_norm": 0.7087315972147312,
      "learning_rate": 4.995214405013641e-06,
      "loss": 0.6486,
      "step": 539
    },
    {
      "epoch": 0.14471392201527536,
      "grad_norm": 0.7206406590675641,
      "learning_rate": 4.995192584924995e-06,
      "loss": 0.6389,
      "step": 540
    },
    {
      "epoch": 0.1449819107597481,
      "grad_norm": 0.7424218948959939,
      "learning_rate": 4.995170715252578e-06,
      "loss": 0.6467,
      "step": 541
    },
    {
      "epoch": 0.14524989950422082,
      "grad_norm": 0.7107065570170484,
      "learning_rate": 4.995148795996822e-06,
      "loss": 0.6603,
      "step": 542
    },
    {
      "epoch": 0.14551788824869355,
      "grad_norm": 0.7466852509845948,
      "learning_rate": 4.995126827158166e-06,
      "loss": 0.678,
      "step": 543
    },
    {
      "epoch": 0.14578587699316628,
      "grad_norm": 0.7320839541481875,
      "learning_rate": 4.9951048087370435e-06,
      "loss": 0.69,
      "step": 544
    },
    {
      "epoch": 0.146053865737639,
      "grad_norm": 0.7097044490405184,
      "learning_rate": 4.995082740733893e-06,
      "loss": 0.652,
      "step": 545
    },
    {
      "epoch": 0.14632185448211174,
      "grad_norm": 0.7283690679622784,
      "learning_rate": 4.995060623149153e-06,
      "loss": 0.6498,
      "step": 546
    },
    {
      "epoch": 0.14658984322658447,
      "grad_norm": 0.7398471622236015,
      "learning_rate": 4.995038455983264e-06,
      "loss": 0.6554,
      "step": 547
    },
    {
      "epoch": 0.14685783197105723,
      "grad_norm": 0.7317896432883286,
      "learning_rate": 4.995016239236665e-06,
      "loss": 0.6712,
      "step": 548
    },
    {
      "epoch": 0.14712582071552996,
      "grad_norm": 0.6873652664231841,
      "learning_rate": 4.994993972909798e-06,
      "loss": 0.634,
      "step": 549
    },
    {
      "epoch": 0.1473938094600027,
      "grad_norm": 0.7181089634159745,
      "learning_rate": 4.994971657003106e-06,
      "loss": 0.66,
      "step": 550
    },
    {
      "epoch": 0.14766179820447542,
      "grad_norm": 0.7522043562394417,
      "learning_rate": 4.994949291517032e-06,
      "loss": 0.6292,
      "step": 551
    },
    {
      "epoch": 0.14792978694894815,
      "grad_norm": 0.7456441900803418,
      "learning_rate": 4.99492687645202e-06,
      "loss": 0.6291,
      "step": 552
    },
    {
      "epoch": 0.14819777569342088,
      "grad_norm": 0.7466315465946131,
      "learning_rate": 4.994904411808515e-06,
      "loss": 0.6394,
      "step": 553
    },
    {
      "epoch": 0.1484657644378936,
      "grad_norm": 0.7824047415022475,
      "learning_rate": 4.994881897586966e-06,
      "loss": 0.6687,
      "step": 554
    },
    {
      "epoch": 0.14873375318236634,
      "grad_norm": 0.7282956207958781,
      "learning_rate": 4.9948593337878186e-06,
      "loss": 0.645,
      "step": 555
    },
    {
      "epoch": 0.14900174192683907,
      "grad_norm": 0.7918570663984468,
      "learning_rate": 4.99483672041152e-06,
      "loss": 0.633,
      "step": 556
    },
    {
      "epoch": 0.1492697306713118,
      "grad_norm": 0.733876071356278,
      "learning_rate": 4.994814057458522e-06,
      "loss": 0.6413,
      "step": 557
    },
    {
      "epoch": 0.14953771941578453,
      "grad_norm": 0.6893865793298389,
      "learning_rate": 4.994791344929274e-06,
      "loss": 0.6708,
      "step": 558
    },
    {
      "epoch": 0.14980570816025726,
      "grad_norm": 0.795918530797209,
      "learning_rate": 4.994768582824226e-06,
      "loss": 0.6467,
      "step": 559
    },
    {
      "epoch": 0.15007369690473,
      "grad_norm": 0.7982240793872389,
      "learning_rate": 4.994745771143833e-06,
      "loss": 0.6254,
      "step": 560
    },
    {
      "epoch": 0.15034168564920272,
      "grad_norm": 0.6631054975648678,
      "learning_rate": 4.994722909888546e-06,
      "loss": 0.6252,
      "step": 561
    },
    {
      "epoch": 0.15060967439367545,
      "grad_norm": 0.7318113848865561,
      "learning_rate": 4.99469999905882e-06,
      "loss": 0.6335,
      "step": 562
    },
    {
      "epoch": 0.1508776631381482,
      "grad_norm": 0.8295292442837676,
      "learning_rate": 4.99467703865511e-06,
      "loss": 0.6023,
      "step": 563
    },
    {
      "epoch": 0.15114565188262094,
      "grad_norm": 0.701291576999972,
      "learning_rate": 4.994654028677873e-06,
      "loss": 0.6217,
      "step": 564
    },
    {
      "epoch": 0.15141364062709367,
      "grad_norm": 0.7618202840990439,
      "learning_rate": 4.994630969127566e-06,
      "loss": 0.6167,
      "step": 565
    },
    {
      "epoch": 0.1516816293715664,
      "grad_norm": 0.7159681265114906,
      "learning_rate": 4.9946078600046465e-06,
      "loss": 0.6476,
      "step": 566
    },
    {
      "epoch": 0.15194961811603913,
      "grad_norm": 0.6971641903877472,
      "learning_rate": 4.994584701309575e-06,
      "loss": 0.6599,
      "step": 567
    },
    {
      "epoch": 0.15221760686051186,
      "grad_norm": 0.7176513171098564,
      "learning_rate": 4.994561493042811e-06,
      "loss": 0.666,
      "step": 568
    },
    {
      "epoch": 0.1524855956049846,
      "grad_norm": 0.7094324880900823,
      "learning_rate": 4.994538235204815e-06,
      "loss": 0.6545,
      "step": 569
    },
    {
      "epoch": 0.15275358434945732,
      "grad_norm": 0.6949835785822807,
      "learning_rate": 4.994514927796051e-06,
      "loss": 0.6265,
      "step": 570
    },
    {
      "epoch": 0.15302157309393005,
      "grad_norm": 0.7245173102160389,
      "learning_rate": 4.99449157081698e-06,
      "loss": 0.6184,
      "step": 571
    },
    {
      "epoch": 0.15328956183840278,
      "grad_norm": 0.7221501174303468,
      "learning_rate": 4.9944681642680685e-06,
      "loss": 0.6393,
      "step": 572
    },
    {
      "epoch": 0.1535575505828755,
      "grad_norm": 0.7510979619001918,
      "learning_rate": 4.994444708149779e-06,
      "loss": 0.6756,
      "step": 573
    },
    {
      "epoch": 0.15382553932734824,
      "grad_norm": 0.7262653305980047,
      "learning_rate": 4.9944212024625795e-06,
      "loss": 0.6305,
      "step": 574
    },
    {
      "epoch": 0.15409352807182097,
      "grad_norm": 0.7773500111179588,
      "learning_rate": 4.994397647206936e-06,
      "loss": 0.5731,
      "step": 575
    },
    {
      "epoch": 0.1543615168162937,
      "grad_norm": 0.7246738032650223,
      "learning_rate": 4.994374042383318e-06,
      "loss": 0.6755,
      "step": 576
    },
    {
      "epoch": 0.15462950556076646,
      "grad_norm": 0.7344173209112376,
      "learning_rate": 4.994350387992193e-06,
      "loss": 0.6479,
      "step": 577
    },
    {
      "epoch": 0.1548974943052392,
      "grad_norm": 0.7255533917130642,
      "learning_rate": 4.994326684034031e-06,
      "loss": 0.6406,
      "step": 578
    },
    {
      "epoch": 0.15516548304971192,
      "grad_norm": 0.8124833516901149,
      "learning_rate": 4.994302930509305e-06,
      "loss": 0.6299,
      "step": 579
    },
    {
      "epoch": 0.15543347179418465,
      "grad_norm": 0.7098625141138125,
      "learning_rate": 4.994279127418485e-06,
      "loss": 0.6788,
      "step": 580
    },
    {
      "epoch": 0.15570146053865738,
      "grad_norm": 0.7047559486577673,
      "learning_rate": 4.994255274762045e-06,
      "loss": 0.6646,
      "step": 581
    },
    {
      "epoch": 0.1559694492831301,
      "grad_norm": 0.7180547644029949,
      "learning_rate": 4.99423137254046e-06,
      "loss": 0.6648,
      "step": 582
    },
    {
      "epoch": 0.15623743802760284,
      "grad_norm": 0.8538478778783714,
      "learning_rate": 4.994207420754202e-06,
      "loss": 0.6577,
      "step": 583
    },
    {
      "epoch": 0.15650542677207557,
      "grad_norm": 0.6991665848177537,
      "learning_rate": 4.99418341940375e-06,
      "loss": 0.6112,
      "step": 584
    },
    {
      "epoch": 0.1567734155165483,
      "grad_norm": 0.707503952247657,
      "learning_rate": 4.994159368489579e-06,
      "loss": 0.6664,
      "step": 585
    },
    {
      "epoch": 0.15704140426102103,
      "grad_norm": 3.640412939071144,
      "learning_rate": 4.994135268012168e-06,
      "loss": 0.6116,
      "step": 586
    },
    {
      "epoch": 0.15730939300549376,
      "grad_norm": 0.6986072288296409,
      "learning_rate": 4.994111117971995e-06,
      "loss": 0.6362,
      "step": 587
    },
    {
      "epoch": 0.1575773817499665,
      "grad_norm": 0.7227044223492499,
      "learning_rate": 4.9940869183695405e-06,
      "loss": 0.6599,
      "step": 588
    },
    {
      "epoch": 0.15784537049443922,
      "grad_norm": 0.683958700809136,
      "learning_rate": 4.994062669205285e-06,
      "loss": 0.6332,
      "step": 589
    },
    {
      "epoch": 0.15811335923891195,
      "grad_norm": 0.6858696933907192,
      "learning_rate": 4.994038370479712e-06,
      "loss": 0.62,
      "step": 590
    },
    {
      "epoch": 0.15838134798338468,
      "grad_norm": 0.6690342044672993,
      "learning_rate": 4.994014022193302e-06,
      "loss": 0.6142,
      "step": 591
    },
    {
      "epoch": 0.15864933672785744,
      "grad_norm": 0.757458210104187,
      "learning_rate": 4.99398962434654e-06,
      "loss": 0.6189,
      "step": 592
    },
    {
      "epoch": 0.15891732547233017,
      "grad_norm": 0.733998474490662,
      "learning_rate": 4.993965176939911e-06,
      "loss": 0.6609,
      "step": 593
    },
    {
      "epoch": 0.1591853142168029,
      "grad_norm": 0.7107624560812696,
      "learning_rate": 4.993940679973901e-06,
      "loss": 0.6394,
      "step": 594
    },
    {
      "epoch": 0.15945330296127563,
      "grad_norm": 0.7144454855936371,
      "learning_rate": 4.993916133448996e-06,
      "loss": 0.6449,
      "step": 595
    },
    {
      "epoch": 0.15972129170574836,
      "grad_norm": 0.7233060666828379,
      "learning_rate": 4.9938915373656835e-06,
      "loss": 0.6317,
      "step": 596
    },
    {
      "epoch": 0.1599892804502211,
      "grad_norm": 0.7172556696670124,
      "learning_rate": 4.993866891724454e-06,
      "loss": 0.6861,
      "step": 597
    },
    {
      "epoch": 0.16025726919469382,
      "grad_norm": 0.690316386521837,
      "learning_rate": 4.9938421965257946e-06,
      "loss": 0.6431,
      "step": 598
    },
    {
      "epoch": 0.16052525793916655,
      "grad_norm": 0.677366725332738,
      "learning_rate": 4.993817451770199e-06,
      "loss": 0.5975,
      "step": 599
    },
    {
      "epoch": 0.16079324668363928,
      "grad_norm": 0.7259259250475251,
      "learning_rate": 4.993792657458156e-06,
      "loss": 0.6737,
      "step": 600
    },
    {
      "epoch": 0.16106123542811202,
      "grad_norm": 0.6387002391522255,
      "learning_rate": 4.993767813590161e-06,
      "loss": 0.5818,
      "step": 601
    },
    {
      "epoch": 0.16132922417258475,
      "grad_norm": 0.7117333619159073,
      "learning_rate": 4.993742920166706e-06,
      "loss": 0.6432,
      "step": 602
    },
    {
      "epoch": 0.16159721291705748,
      "grad_norm": 0.911990785157275,
      "learning_rate": 4.9937179771882866e-06,
      "loss": 0.6463,
      "step": 603
    },
    {
      "epoch": 0.1618652016615302,
      "grad_norm": 0.6895773246215932,
      "learning_rate": 4.993692984655397e-06,
      "loss": 0.6404,
      "step": 604
    },
    {
      "epoch": 0.16213319040600294,
      "grad_norm": 0.7204400470832365,
      "learning_rate": 4.993667942568536e-06,
      "loss": 0.6276,
      "step": 605
    },
    {
      "epoch": 0.16240117915047567,
      "grad_norm": 0.6910242580727207,
      "learning_rate": 4.9936428509282e-06,
      "loss": 0.6428,
      "step": 606
    },
    {
      "epoch": 0.16266916789494842,
      "grad_norm": 0.6720278119571813,
      "learning_rate": 4.993617709734887e-06,
      "loss": 0.606,
      "step": 607
    },
    {
      "epoch": 0.16293715663942115,
      "grad_norm": 0.7057095764752408,
      "learning_rate": 4.993592518989097e-06,
      "loss": 0.6244,
      "step": 608
    },
    {
      "epoch": 0.16320514538389388,
      "grad_norm": 0.709699594781715,
      "learning_rate": 4.993567278691332e-06,
      "loss": 0.6771,
      "step": 609
    },
    {
      "epoch": 0.16347313412836662,
      "grad_norm": 0.6841119479410978,
      "learning_rate": 4.993541988842092e-06,
      "loss": 0.641,
      "step": 610
    },
    {
      "epoch": 0.16374112287283935,
      "grad_norm": 0.7172856710496638,
      "learning_rate": 4.9935166494418804e-06,
      "loss": 0.6512,
      "step": 611
    },
    {
      "epoch": 0.16400911161731208,
      "grad_norm": 0.7185485565059917,
      "learning_rate": 4.9934912604912e-06,
      "loss": 0.6321,
      "step": 612
    },
    {
      "epoch": 0.1642771003617848,
      "grad_norm": 0.668515415419898,
      "learning_rate": 4.993465821990555e-06,
      "loss": 0.5695,
      "step": 613
    },
    {
      "epoch": 0.16454508910625754,
      "grad_norm": 0.6711497386451717,
      "learning_rate": 4.993440333940452e-06,
      "loss": 0.6288,
      "step": 614
    },
    {
      "epoch": 0.16481307785073027,
      "grad_norm": 0.6933039932916508,
      "learning_rate": 4.9934147963413985e-06,
      "loss": 0.6142,
      "step": 615
    },
    {
      "epoch": 0.165081066595203,
      "grad_norm": 0.7168790084354827,
      "learning_rate": 4.993389209193898e-06,
      "loss": 0.633,
      "step": 616
    },
    {
      "epoch": 0.16534905533967573,
      "grad_norm": 0.694248617576841,
      "learning_rate": 4.9933635724984636e-06,
      "loss": 0.6381,
      "step": 617
    },
    {
      "epoch": 0.16561704408414846,
      "grad_norm": 0.6915573667085849,
      "learning_rate": 4.993337886255602e-06,
      "loss": 0.663,
      "step": 618
    },
    {
      "epoch": 0.1658850328286212,
      "grad_norm": 0.7150917222588933,
      "learning_rate": 4.9933121504658244e-06,
      "loss": 0.6668,
      "step": 619
    },
    {
      "epoch": 0.16615302157309392,
      "grad_norm": 0.7658691359596284,
      "learning_rate": 4.9932863651296415e-06,
      "loss": 0.677,
      "step": 620
    },
    {
      "epoch": 0.16642101031756668,
      "grad_norm": 0.7065365046081419,
      "learning_rate": 4.9932605302475664e-06,
      "loss": 0.619,
      "step": 621
    },
    {
      "epoch": 0.1666889990620394,
      "grad_norm": 0.6861017796241022,
      "learning_rate": 4.9932346458201135e-06,
      "loss": 0.6436,
      "step": 622
    },
    {
      "epoch": 0.16695698780651214,
      "grad_norm": 0.7128000495166548,
      "learning_rate": 4.993208711847794e-06,
      "loss": 0.6552,
      "step": 623
    },
    {
      "epoch": 0.16722497655098487,
      "grad_norm": 0.6786164604858327,
      "learning_rate": 4.993182728331127e-06,
      "loss": 0.6442,
      "step": 624
    },
    {
      "epoch": 0.1674929652954576,
      "grad_norm": 0.7029226756674728,
      "learning_rate": 4.993156695270627e-06,
      "loss": 0.6257,
      "step": 625
    },
    {
      "epoch": 0.16776095403993033,
      "grad_norm": 0.6926478318783196,
      "learning_rate": 4.993130612666811e-06,
      "loss": 0.6382,
      "step": 626
    },
    {
      "epoch": 0.16802894278440306,
      "grad_norm": 0.7313722999799178,
      "learning_rate": 4.9931044805201975e-06,
      "loss": 0.6714,
      "step": 627
    },
    {
      "epoch": 0.1682969315288758,
      "grad_norm": 0.7135225294651357,
      "learning_rate": 4.9930782988313065e-06,
      "loss": 0.6384,
      "step": 628
    },
    {
      "epoch": 0.16856492027334852,
      "grad_norm": 0.7335803643297969,
      "learning_rate": 4.993052067600657e-06,
      "loss": 0.6635,
      "step": 629
    },
    {
      "epoch": 0.16883290901782125,
      "grad_norm": 0.7463978940015205,
      "learning_rate": 4.993025786828772e-06,
      "loss": 0.642,
      "step": 630
    },
    {
      "epoch": 0.16910089776229398,
      "grad_norm": 0.706487111580625,
      "learning_rate": 4.992999456516173e-06,
      "loss": 0.6229,
      "step": 631
    },
    {
      "epoch": 0.1693688865067667,
      "grad_norm": 0.7205213779655937,
      "learning_rate": 4.992973076663382e-06,
      "loss": 0.6253,
      "step": 632
    },
    {
      "epoch": 0.16963687525123944,
      "grad_norm": 0.666700155509833,
      "learning_rate": 4.992946647270926e-06,
      "loss": 0.582,
      "step": 633
    },
    {
      "epoch": 0.16990486399571217,
      "grad_norm": 0.693833275568337,
      "learning_rate": 4.992920168339327e-06,
      "loss": 0.6514,
      "step": 634
    },
    {
      "epoch": 0.1701728527401849,
      "grad_norm": 0.7317814124793317,
      "learning_rate": 4.992893639869113e-06,
      "loss": 0.6389,
      "step": 635
    },
    {
      "epoch": 0.17044084148465766,
      "grad_norm": 0.702105846516531,
      "learning_rate": 4.992867061860811e-06,
      "loss": 0.6076,
      "step": 636
    },
    {
      "epoch": 0.1707088302291304,
      "grad_norm": 0.6838944155789562,
      "learning_rate": 4.992840434314949e-06,
      "loss": 0.6554,
      "step": 637
    },
    {
      "epoch": 0.17097681897360312,
      "grad_norm": 0.7200632172540031,
      "learning_rate": 4.992813757232055e-06,
      "loss": 0.6332,
      "step": 638
    },
    {
      "epoch": 0.17124480771807585,
      "grad_norm": 0.7391067365880158,
      "learning_rate": 4.992787030612661e-06,
      "loss": 0.6502,
      "step": 639
    },
    {
      "epoch": 0.17151279646254858,
      "grad_norm": 0.7104933254825165,
      "learning_rate": 4.992760254457298e-06,
      "loss": 0.6143,
      "step": 640
    },
    {
      "epoch": 0.1717807852070213,
      "grad_norm": 0.7143145088730471,
      "learning_rate": 4.992733428766497e-06,
      "loss": 0.611,
      "step": 641
    },
    {
      "epoch": 0.17204877395149404,
      "grad_norm": 0.731216355857471,
      "learning_rate": 4.992706553540792e-06,
      "loss": 0.6715,
      "step": 642
    },
    {
      "epoch": 0.17231676269596677,
      "grad_norm": 0.6999531358693072,
      "learning_rate": 4.992679628780715e-06,
      "loss": 0.6425,
      "step": 643
    },
    {
      "epoch": 0.1725847514404395,
      "grad_norm": 0.7080390992608281,
      "learning_rate": 4.9926526544868034e-06,
      "loss": 0.6208,
      "step": 644
    },
    {
      "epoch": 0.17285274018491223,
      "grad_norm": 0.7286756850006472,
      "learning_rate": 4.992625630659593e-06,
      "loss": 0.6764,
      "step": 645
    },
    {
      "epoch": 0.17312072892938496,
      "grad_norm": 0.7548166946735316,
      "learning_rate": 4.992598557299619e-06,
      "loss": 0.6236,
      "step": 646
    },
    {
      "epoch": 0.1733887176738577,
      "grad_norm": 0.7005195354564705,
      "learning_rate": 4.992571434407422e-06,
      "loss": 0.673,
      "step": 647
    },
    {
      "epoch": 0.17365670641833042,
      "grad_norm": 0.7825474716041445,
      "learning_rate": 4.992544261983539e-06,
      "loss": 0.613,
      "step": 648
    },
    {
      "epoch": 0.17392469516280315,
      "grad_norm": 0.7016187122679672,
      "learning_rate": 4.99251704002851e-06,
      "loss": 0.6253,
      "step": 649
    },
    {
      "epoch": 0.1741926839072759,
      "grad_norm": 0.7340567858857756,
      "learning_rate": 4.992489768542877e-06,
      "loss": 0.6636,
      "step": 650
    },
    {
      "epoch": 0.17446067265174864,
      "grad_norm": 0.6990641149495213,
      "learning_rate": 4.992462447527182e-06,
      "loss": 0.6246,
      "step": 651
    },
    {
      "epoch": 0.17472866139622137,
      "grad_norm": 0.691625270894426,
      "learning_rate": 4.992435076981967e-06,
      "loss": 0.6019,
      "step": 652
    },
    {
      "epoch": 0.1749966501406941,
      "grad_norm": 0.7080284593460456,
      "learning_rate": 4.992407656907776e-06,
      "loss": 0.6442,
      "step": 653
    },
    {
      "epoch": 0.17526463888516683,
      "grad_norm": 0.7002771135849868,
      "learning_rate": 4.992380187305154e-06,
      "loss": 0.6485,
      "step": 654
    },
    {
      "epoch": 0.17553262762963956,
      "grad_norm": 0.7172672902146083,
      "learning_rate": 4.992352668174648e-06,
      "loss": 0.6583,
      "step": 655
    },
    {
      "epoch": 0.1758006163741123,
      "grad_norm": 0.6963103362971386,
      "learning_rate": 4.992325099516803e-06,
      "loss": 0.6361,
      "step": 656
    },
    {
      "epoch": 0.17606860511858502,
      "grad_norm": 0.6965731598730267,
      "learning_rate": 4.9922974813321675e-06,
      "loss": 0.6476,
      "step": 657
    },
    {
      "epoch": 0.17633659386305775,
      "grad_norm": 0.6777970291399462,
      "learning_rate": 4.992269813621291e-06,
      "loss": 0.6085,
      "step": 658
    },
    {
      "epoch": 0.17660458260753048,
      "grad_norm": 0.6836199210791316,
      "learning_rate": 4.992242096384722e-06,
      "loss": 0.6644,
      "step": 659
    },
    {
      "epoch": 0.1768725713520032,
      "grad_norm": 0.7051237628986425,
      "learning_rate": 4.992214329623013e-06,
      "loss": 0.6393,
      "step": 660
    },
    {
      "epoch": 0.17714056009647594,
      "grad_norm": 0.680166684404727,
      "learning_rate": 4.992186513336715e-06,
      "loss": 0.6585,
      "step": 661
    },
    {
      "epoch": 0.17740854884094867,
      "grad_norm": 0.7039422401741099,
      "learning_rate": 4.99215864752638e-06,
      "loss": 0.6317,
      "step": 662
    },
    {
      "epoch": 0.1776765375854214,
      "grad_norm": 0.7082697062088105,
      "learning_rate": 4.992130732192563e-06,
      "loss": 0.6121,
      "step": 663
    },
    {
      "epoch": 0.17794452632989413,
      "grad_norm": 0.6799405067674993,
      "learning_rate": 4.992102767335817e-06,
      "loss": 0.6141,
      "step": 664
    },
    {
      "epoch": 0.1782125150743669,
      "grad_norm": 0.6966682748196276,
      "learning_rate": 4.9920747529567e-06,
      "loss": 0.6162,
      "step": 665
    },
    {
      "epoch": 0.17848050381883962,
      "grad_norm": 0.7201283987912631,
      "learning_rate": 4.992046689055767e-06,
      "loss": 0.6538,
      "step": 666
    },
    {
      "epoch": 0.17874849256331235,
      "grad_norm": 0.6753564400439759,
      "learning_rate": 4.992018575633576e-06,
      "loss": 0.6196,
      "step": 667
    },
    {
      "epoch": 0.17901648130778508,
      "grad_norm": 0.6979213293999127,
      "learning_rate": 4.991990412690686e-06,
      "loss": 0.6453,
      "step": 668
    },
    {
      "epoch": 0.1792844700522578,
      "grad_norm": 0.6823524675388535,
      "learning_rate": 4.991962200227657e-06,
      "loss": 0.6132,
      "step": 669
    },
    {
      "epoch": 0.17955245879673054,
      "grad_norm": 0.7255811353128377,
      "learning_rate": 4.991933938245049e-06,
      "loss": 0.6528,
      "step": 670
    },
    {
      "epoch": 0.17982044754120327,
      "grad_norm": 0.6918396190347444,
      "learning_rate": 4.991905626743423e-06,
      "loss": 0.6575,
      "step": 671
    },
    {
      "epoch": 0.180088436285676,
      "grad_norm": 0.7010489749509459,
      "learning_rate": 4.991877265723343e-06,
      "loss": 0.6299,
      "step": 672
    },
    {
      "epoch": 0.18035642503014873,
      "grad_norm": 0.695652145153345,
      "learning_rate": 4.9918488551853716e-06,
      "loss": 0.6501,
      "step": 673
    },
    {
      "epoch": 0.18062441377462146,
      "grad_norm": 0.6795891412881683,
      "learning_rate": 4.9918203951300735e-06,
      "loss": 0.6192,
      "step": 674
    },
    {
      "epoch": 0.1808924025190942,
      "grad_norm": 0.6908328929751038,
      "learning_rate": 4.991791885558015e-06,
      "loss": 0.6171,
      "step": 675
    },
    {
      "epoch": 0.18116039126356692,
      "grad_norm": 0.6753255321460783,
      "learning_rate": 4.9917633264697615e-06,
      "loss": 0.658,
      "step": 676
    },
    {
      "epoch": 0.18142838000803965,
      "grad_norm": 0.6988646143254105,
      "learning_rate": 4.9917347178658824e-06,
      "loss": 0.6484,
      "step": 677
    },
    {
      "epoch": 0.18169636875251238,
      "grad_norm": 0.698780272145876,
      "learning_rate": 4.991706059746943e-06,
      "loss": 0.6382,
      "step": 678
    },
    {
      "epoch": 0.1819643574969851,
      "grad_norm": 0.6927588809842417,
      "learning_rate": 4.991677352113517e-06,
      "loss": 0.644,
      "step": 679
    },
    {
      "epoch": 0.18223234624145787,
      "grad_norm": 0.7011143700946803,
      "learning_rate": 4.99164859496617e-06,
      "loss": 0.6025,
      "step": 680
    },
    {
      "epoch": 0.1825003349859306,
      "grad_norm": 0.6824474971585086,
      "learning_rate": 4.991619788305477e-06,
      "loss": 0.6767,
      "step": 681
    },
    {
      "epoch": 0.18276832373040333,
      "grad_norm": 0.7953390560789294,
      "learning_rate": 4.99159093213201e-06,
      "loss": 0.6517,
      "step": 682
    },
    {
      "epoch": 0.18303631247487606,
      "grad_norm": 0.7140474059024831,
      "learning_rate": 4.991562026446341e-06,
      "loss": 0.6426,
      "step": 683
    },
    {
      "epoch": 0.1833043012193488,
      "grad_norm": 0.6897207960164068,
      "learning_rate": 4.991533071249046e-06,
      "loss": 0.6532,
      "step": 684
    },
    {
      "epoch": 0.18357228996382152,
      "grad_norm": 0.691911909279393,
      "learning_rate": 4.9915040665407e-06,
      "loss": 0.5913,
      "step": 685
    },
    {
      "epoch": 0.18384027870829425,
      "grad_norm": 0.7131992978363232,
      "learning_rate": 4.991475012321878e-06,
      "loss": 0.6448,
      "step": 686
    },
    {
      "epoch": 0.18410826745276698,
      "grad_norm": 0.6802056260330128,
      "learning_rate": 4.991445908593159e-06,
      "loss": 0.5826,
      "step": 687
    },
    {
      "epoch": 0.1843762561972397,
      "grad_norm": 0.6820165772206007,
      "learning_rate": 4.99141675535512e-06,
      "loss": 0.6437,
      "step": 688
    },
    {
      "epoch": 0.18464424494171244,
      "grad_norm": 0.6868977224107226,
      "learning_rate": 4.991387552608342e-06,
      "loss": 0.6307,
      "step": 689
    },
    {
      "epoch": 0.18491223368618517,
      "grad_norm": 0.7090289903901688,
      "learning_rate": 4.9913583003534034e-06,
      "loss": 0.6608,
      "step": 690
    },
    {
      "epoch": 0.1851802224306579,
      "grad_norm": 0.675059316709845,
      "learning_rate": 4.991328998590888e-06,
      "loss": 0.6398,
      "step": 691
    },
    {
      "epoch": 0.18544821117513063,
      "grad_norm": 0.6916765680368226,
      "learning_rate": 4.991299647321375e-06,
      "loss": 0.6159,
      "step": 692
    },
    {
      "epoch": 0.18571619991960336,
      "grad_norm": 0.6900922984597548,
      "learning_rate": 4.99127024654545e-06,
      "loss": 0.6163,
      "step": 693
    },
    {
      "epoch": 0.18598418866407612,
      "grad_norm": 0.7391481566024299,
      "learning_rate": 4.9912407962636965e-06,
      "loss": 0.6454,
      "step": 694
    },
    {
      "epoch": 0.18625217740854885,
      "grad_norm": 0.6980551491761662,
      "learning_rate": 4.991211296476699e-06,
      "loss": 0.6622,
      "step": 695
    },
    {
      "epoch": 0.18652016615302158,
      "grad_norm": 0.7109695967031855,
      "learning_rate": 4.991181747185045e-06,
      "loss": 0.6471,
      "step": 696
    },
    {
      "epoch": 0.1867881548974943,
      "grad_norm": 0.7464738735648744,
      "learning_rate": 4.991152148389321e-06,
      "loss": 0.5948,
      "step": 697
    },
    {
      "epoch": 0.18705614364196704,
      "grad_norm": 0.6874677003112522,
      "learning_rate": 4.991122500090115e-06,
      "loss": 0.6483,
      "step": 698
    },
    {
      "epoch": 0.18732413238643977,
      "grad_norm": 0.7066503795760468,
      "learning_rate": 4.991092802288017e-06,
      "loss": 0.6242,
      "step": 699
    },
    {
      "epoch": 0.1875921211309125,
      "grad_norm": 0.6865011671340356,
      "learning_rate": 4.991063054983616e-06,
      "loss": 0.6188,
      "step": 700
    },
    {
      "epoch": 0.18786010987538523,
      "grad_norm": 0.7144821793742785,
      "learning_rate": 4.991033258177504e-06,
      "loss": 0.6377,
      "step": 701
    },
    {
      "epoch": 0.18812809861985796,
      "grad_norm": 0.7782087941726045,
      "learning_rate": 4.991003411870272e-06,
      "loss": 0.5997,
      "step": 702
    },
    {
      "epoch": 0.1883960873643307,
      "grad_norm": 0.6985061728150461,
      "learning_rate": 4.990973516062515e-06,
      "loss": 0.6196,
      "step": 703
    },
    {
      "epoch": 0.18866407610880342,
      "grad_norm": 0.7640108127318103,
      "learning_rate": 4.990943570754825e-06,
      "loss": 0.6669,
      "step": 704
    },
    {
      "epoch": 0.18893206485327615,
      "grad_norm": 0.7836305349357308,
      "learning_rate": 4.990913575947798e-06,
      "loss": 0.6401,
      "step": 705
    },
    {
      "epoch": 0.18920005359774889,
      "grad_norm": 0.7306161714064305,
      "learning_rate": 4.990883531642031e-06,
      "loss": 0.6676,
      "step": 706
    },
    {
      "epoch": 0.18946804234222162,
      "grad_norm": 0.678608041710362,
      "learning_rate": 4.99085343783812e-06,
      "loss": 0.6407,
      "step": 707
    },
    {
      "epoch": 0.18973603108669435,
      "grad_norm": 0.7215952213995387,
      "learning_rate": 4.990823294536663e-06,
      "loss": 0.6344,
      "step": 708
    },
    {
      "epoch": 0.1900040198311671,
      "grad_norm": 0.7632375317787337,
      "learning_rate": 4.990793101738259e-06,
      "loss": 0.6572,
      "step": 709
    },
    {
      "epoch": 0.19027200857563983,
      "grad_norm": 0.7571042338306371,
      "learning_rate": 4.990762859443508e-06,
      "loss": 0.6602,
      "step": 710
    },
    {
      "epoch": 0.19053999732011256,
      "grad_norm": 0.6932124062747906,
      "learning_rate": 4.99073256765301e-06,
      "loss": 0.6261,
      "step": 711
    },
    {
      "epoch": 0.1908079860645853,
      "grad_norm": 0.7947020368170427,
      "learning_rate": 4.99070222636737e-06,
      "loss": 0.6243,
      "step": 712
    },
    {
      "epoch": 0.19107597480905802,
      "grad_norm": 0.7564128666117115,
      "learning_rate": 4.990671835587189e-06,
      "loss": 0.646,
      "step": 713
    },
    {
      "epoch": 0.19134396355353075,
      "grad_norm": 0.7226581606485349,
      "learning_rate": 4.99064139531307e-06,
      "loss": 0.6256,
      "step": 714
    },
    {
      "epoch": 0.19161195229800349,
      "grad_norm": 0.7058618414194209,
      "learning_rate": 4.990610905545619e-06,
      "loss": 0.6343,
      "step": 715
    },
    {
      "epoch": 0.19187994104247622,
      "grad_norm": 0.7094656338513196,
      "learning_rate": 4.990580366285441e-06,
      "loss": 0.64,
      "step": 716
    },
    {
      "epoch": 0.19214792978694895,
      "grad_norm": 0.6983230883513736,
      "learning_rate": 4.990549777533145e-06,
      "loss": 0.6058,
      "step": 717
    },
    {
      "epoch": 0.19241591853142168,
      "grad_norm": 0.730787860788838,
      "learning_rate": 4.990519139289336e-06,
      "loss": 0.6605,
      "step": 718
    },
    {
      "epoch": 0.1926839072758944,
      "grad_norm": 0.6850742057045152,
      "learning_rate": 4.990488451554625e-06,
      "loss": 0.615,
      "step": 719
    },
    {
      "epoch": 0.19295189602036714,
      "grad_norm": 0.6822886929692724,
      "learning_rate": 4.990457714329621e-06,
      "loss": 0.6337,
      "step": 720
    },
    {
      "epoch": 0.19321988476483987,
      "grad_norm": 0.7163956214815631,
      "learning_rate": 4.9904269276149345e-06,
      "loss": 0.6175,
      "step": 721
    },
    {
      "epoch": 0.1934878735093126,
      "grad_norm": 0.6725718216141872,
      "learning_rate": 4.990396091411178e-06,
      "loss": 0.6248,
      "step": 722
    },
    {
      "epoch": 0.19375586225378535,
      "grad_norm": 0.7005608832802087,
      "learning_rate": 4.990365205718964e-06,
      "loss": 0.6479,
      "step": 723
    },
    {
      "epoch": 0.19402385099825809,
      "grad_norm": 0.7529441137608281,
      "learning_rate": 4.9903342705389056e-06,
      "loss": 0.5905,
      "step": 724
    },
    {
      "epoch": 0.19429183974273082,
      "grad_norm": 0.6873650044732095,
      "learning_rate": 4.990303285871618e-06,
      "loss": 0.6232,
      "step": 725
    },
    {
      "epoch": 0.19455982848720355,
      "grad_norm": 0.721280157696001,
      "learning_rate": 4.990272251717718e-06,
      "loss": 0.6821,
      "step": 726
    },
    {
      "epoch": 0.19482781723167628,
      "grad_norm": 0.7399537031448984,
      "learning_rate": 4.990241168077821e-06,
      "loss": 0.6459,
      "step": 727
    },
    {
      "epoch": 0.195095805976149,
      "grad_norm": 0.6991397908445716,
      "learning_rate": 4.990210034952544e-06,
      "loss": 0.6474,
      "step": 728
    },
    {
      "epoch": 0.19536379472062174,
      "grad_norm": 0.7240355657231803,
      "learning_rate": 4.990178852342508e-06,
      "loss": 0.6337,
      "step": 729
    },
    {
      "epoch": 0.19563178346509447,
      "grad_norm": 0.7315867593566441,
      "learning_rate": 4.99014762024833e-06,
      "loss": 0.6528,
      "step": 730
    },
    {
      "epoch": 0.1958997722095672,
      "grad_norm": 0.693941889458587,
      "learning_rate": 4.990116338670633e-06,
      "loss": 0.6446,
      "step": 731
    },
    {
      "epoch": 0.19616776095403993,
      "grad_norm": 0.6918487782154062,
      "learning_rate": 4.990085007610037e-06,
      "loss": 0.6361,
      "step": 732
    },
    {
      "epoch": 0.19643574969851266,
      "grad_norm": 0.711009955768985,
      "learning_rate": 4.990053627067166e-06,
      "loss": 0.6423,
      "step": 733
    },
    {
      "epoch": 0.1967037384429854,
      "grad_norm": 0.7225429563749267,
      "learning_rate": 4.990022197042642e-06,
      "loss": 0.6482,
      "step": 734
    },
    {
      "epoch": 0.19697172718745812,
      "grad_norm": 0.6899659460303524,
      "learning_rate": 4.989990717537091e-06,
      "loss": 0.6412,
      "step": 735
    },
    {
      "epoch": 0.19723971593193085,
      "grad_norm": 0.6915056688533192,
      "learning_rate": 4.989959188551137e-06,
      "loss": 0.6424,
      "step": 736
    },
    {
      "epoch": 0.19750770467640358,
      "grad_norm": 0.7548243563970931,
      "learning_rate": 4.9899276100854075e-06,
      "loss": 0.675,
      "step": 737
    },
    {
      "epoch": 0.19777569342087634,
      "grad_norm": 0.6989972780789063,
      "learning_rate": 4.98989598214053e-06,
      "loss": 0.6283,
      "step": 738
    },
    {
      "epoch": 0.19804368216534907,
      "grad_norm": 0.7189701351579195,
      "learning_rate": 4.989864304717133e-06,
      "loss": 0.6695,
      "step": 739
    },
    {
      "epoch": 0.1983116709098218,
      "grad_norm": 0.6828933228579979,
      "learning_rate": 4.989832577815846e-06,
      "loss": 0.6464,
      "step": 740
    },
    {
      "epoch": 0.19857965965429453,
      "grad_norm": 0.7237445715208176,
      "learning_rate": 4.9898008014373e-06,
      "loss": 0.6417,
      "step": 741
    },
    {
      "epoch": 0.19884764839876726,
      "grad_norm": 0.7112004905302083,
      "learning_rate": 4.989768975582126e-06,
      "loss": 0.6442,
      "step": 742
    },
    {
      "epoch": 0.19911563714324,
      "grad_norm": 0.6703913908055704,
      "learning_rate": 4.989737100250956e-06,
      "loss": 0.6064,
      "step": 743
    },
    {
      "epoch": 0.19938362588771272,
      "grad_norm": 0.6995140503655988,
      "learning_rate": 4.989705175444423e-06,
      "loss": 0.6465,
      "step": 744
    },
    {
      "epoch": 0.19965161463218545,
      "grad_norm": 0.7257991857042844,
      "learning_rate": 4.989673201163162e-06,
      "loss": 0.6424,
      "step": 745
    },
    {
      "epoch": 0.19991960337665818,
      "grad_norm": 0.7171269575291686,
      "learning_rate": 4.989641177407809e-06,
      "loss": 0.6325,
      "step": 746
    },
    {
      "epoch": 0.2001875921211309,
      "grad_norm": 0.7446441453668453,
      "learning_rate": 4.989609104179e-06,
      "loss": 0.6217,
      "step": 747
    },
    {
      "epoch": 0.20045558086560364,
      "grad_norm": 0.6933843153224584,
      "learning_rate": 4.9895769814773716e-06,
      "loss": 0.6599,
      "step": 748
    },
    {
      "epoch": 0.20072356961007637,
      "grad_norm": 0.7113395767753312,
      "learning_rate": 4.989544809303563e-06,
      "loss": 0.6705,
      "step": 749
    },
    {
      "epoch": 0.2009915583545491,
      "grad_norm": 0.7262941081048949,
      "learning_rate": 4.9895125876582135e-06,
      "loss": 0.6284,
      "step": 750
    },
    {
      "epoch": 0.20125954709902183,
      "grad_norm": 0.6745935202976254,
      "learning_rate": 4.989480316541963e-06,
      "loss": 0.5828,
      "step": 751
    },
    {
      "epoch": 0.20152753584349456,
      "grad_norm": 0.7047079494987045,
      "learning_rate": 4.989447995955453e-06,
      "loss": 0.6553,
      "step": 752
    },
    {
      "epoch": 0.20179552458796732,
      "grad_norm": 0.6988409325751712,
      "learning_rate": 4.989415625899326e-06,
      "loss": 0.6231,
      "step": 753
    },
    {
      "epoch": 0.20206351333244005,
      "grad_norm": 0.7150929953502426,
      "learning_rate": 4.9893832063742246e-06,
      "loss": 0.6508,
      "step": 754
    },
    {
      "epoch": 0.20233150207691278,
      "grad_norm": 0.6750377090364201,
      "learning_rate": 4.9893507373807925e-06,
      "loss": 0.6252,
      "step": 755
    },
    {
      "epoch": 0.2025994908213855,
      "grad_norm": 0.6951383974183751,
      "learning_rate": 4.989318218919677e-06,
      "loss": 0.6736,
      "step": 756
    },
    {
      "epoch": 0.20286747956585824,
      "grad_norm": 0.691947644060252,
      "learning_rate": 4.9892856509915224e-06,
      "loss": 0.6631,
      "step": 757
    },
    {
      "epoch": 0.20313546831033097,
      "grad_norm": 0.6804944775648856,
      "learning_rate": 4.989253033596977e-06,
      "loss": 0.6152,
      "step": 758
    },
    {
      "epoch": 0.2034034570548037,
      "grad_norm": 0.7527874179261579,
      "learning_rate": 4.989220366736688e-06,
      "loss": 0.612,
      "step": 759
    },
    {
      "epoch": 0.20367144579927643,
      "grad_norm": 0.72532055961271,
      "learning_rate": 4.989187650411306e-06,
      "loss": 0.5829,
      "step": 760
    },
    {
      "epoch": 0.20393943454374916,
      "grad_norm": 0.7066704377028078,
      "learning_rate": 4.98915488462148e-06,
      "loss": 0.6448,
      "step": 761
    },
    {
      "epoch": 0.2042074232882219,
      "grad_norm": 0.7236358252715025,
      "learning_rate": 4.98912206936786e-06,
      "loss": 0.63,
      "step": 762
    },
    {
      "epoch": 0.20447541203269462,
      "grad_norm": 0.7240354233494966,
      "learning_rate": 4.989089204651101e-06,
      "loss": 0.632,
      "step": 763
    },
    {
      "epoch": 0.20474340077716735,
      "grad_norm": 0.7472544745829678,
      "learning_rate": 4.989056290471854e-06,
      "loss": 0.6379,
      "step": 764
    },
    {
      "epoch": 0.20501138952164008,
      "grad_norm": 0.6874867716271831,
      "learning_rate": 4.9890233268307734e-06,
      "loss": 0.6395,
      "step": 765
    },
    {
      "epoch": 0.2052793782661128,
      "grad_norm": 0.7475560796551097,
      "learning_rate": 4.988990313728514e-06,
      "loss": 0.661,
      "step": 766
    },
    {
      "epoch": 0.20554736701058557,
      "grad_norm": 0.6964814092923287,
      "learning_rate": 4.988957251165734e-06,
      "loss": 0.6463,
      "step": 767
    },
    {
      "epoch": 0.2058153557550583,
      "grad_norm": 0.7222684232595513,
      "learning_rate": 4.988924139143086e-06,
      "loss": 0.6388,
      "step": 768
    },
    {
      "epoch": 0.20608334449953103,
      "grad_norm": 0.7052393255915468,
      "learning_rate": 4.9888909776612324e-06,
      "loss": 0.6654,
      "step": 769
    },
    {
      "epoch": 0.20635133324400376,
      "grad_norm": 0.7234062728526522,
      "learning_rate": 4.98885776672083e-06,
      "loss": 0.5761,
      "step": 770
    },
    {
      "epoch": 0.2066193219884765,
      "grad_norm": 0.6913124605150095,
      "learning_rate": 4.988824506322539e-06,
      "loss": 0.6302,
      "step": 771
    },
    {
      "epoch": 0.20688731073294922,
      "grad_norm": 0.6982476920853802,
      "learning_rate": 4.988791196467021e-06,
      "loss": 0.6273,
      "step": 772
    },
    {
      "epoch": 0.20715529947742195,
      "grad_norm": 0.7200038443148039,
      "learning_rate": 4.988757837154937e-06,
      "loss": 0.5923,
      "step": 773
    },
    {
      "epoch": 0.20742328822189468,
      "grad_norm": 0.6819695262226855,
      "learning_rate": 4.98872442838695e-06,
      "loss": 0.6371,
      "step": 774
    },
    {
      "epoch": 0.2076912769663674,
      "grad_norm": 0.7225129494636436,
      "learning_rate": 4.988690970163726e-06,
      "loss": 0.6373,
      "step": 775
    },
    {
      "epoch": 0.20795926571084014,
      "grad_norm": 0.6921092747689903,
      "learning_rate": 4.988657462485926e-06,
      "loss": 0.6175,
      "step": 776
    },
    {
      "epoch": 0.20822725445531287,
      "grad_norm": 0.7548687040365631,
      "learning_rate": 4.988623905354219e-06,
      "loss": 0.6419,
      "step": 777
    },
    {
      "epoch": 0.2084952431997856,
      "grad_norm": 0.7106763451590254,
      "learning_rate": 4.98859029876927e-06,
      "loss": 0.6028,
      "step": 778
    },
    {
      "epoch": 0.20876323194425833,
      "grad_norm": 0.7100792098706543,
      "learning_rate": 4.988556642731748e-06,
      "loss": 0.6543,
      "step": 779
    },
    {
      "epoch": 0.20903122068873106,
      "grad_norm": 0.6989615284563013,
      "learning_rate": 4.988522937242322e-06,
      "loss": 0.6617,
      "step": 780
    },
    {
      "epoch": 0.2092992094332038,
      "grad_norm": 0.7068068770155578,
      "learning_rate": 4.98848918230166e-06,
      "loss": 0.6385,
      "step": 781
    },
    {
      "epoch": 0.20956719817767655,
      "grad_norm": 0.6746414065927594,
      "learning_rate": 4.988455377910436e-06,
      "loss": 0.644,
      "step": 782
    },
    {
      "epoch": 0.20983518692214928,
      "grad_norm": 0.7271175067170045,
      "learning_rate": 4.988421524069318e-06,
      "loss": 0.6409,
      "step": 783
    },
    {
      "epoch": 0.210103175666622,
      "grad_norm": 0.7565530760329418,
      "learning_rate": 4.9883876207789806e-06,
      "loss": 0.6141,
      "step": 784
    },
    {
      "epoch": 0.21037116441109474,
      "grad_norm": 0.6644497370151369,
      "learning_rate": 4.988353668040098e-06,
      "loss": 0.6259,
      "step": 785
    },
    {
      "epoch": 0.21063915315556747,
      "grad_norm": 0.7265543625614574,
      "learning_rate": 4.988319665853344e-06,
      "loss": 0.618,
      "step": 786
    },
    {
      "epoch": 0.2109071419000402,
      "grad_norm": 0.7255970490742818,
      "learning_rate": 4.988285614219395e-06,
      "loss": 0.6543,
      "step": 787
    },
    {
      "epoch": 0.21117513064451293,
      "grad_norm": 0.7157902682456567,
      "learning_rate": 4.988251513138926e-06,
      "loss": 0.6275,
      "step": 788
    },
    {
      "epoch": 0.21144311938898566,
      "grad_norm": 0.7253644631721875,
      "learning_rate": 4.988217362612617e-06,
      "loss": 0.6437,
      "step": 789
    },
    {
      "epoch": 0.2117111081334584,
      "grad_norm": 0.7090860186953679,
      "learning_rate": 4.988183162641145e-06,
      "loss": 0.6618,
      "step": 790
    },
    {
      "epoch": 0.21197909687793112,
      "grad_norm": 0.7127589683179473,
      "learning_rate": 4.98814891322519e-06,
      "loss": 0.6206,
      "step": 791
    },
    {
      "epoch": 0.21224708562240385,
      "grad_norm": 0.754987201432984,
      "learning_rate": 4.988114614365433e-06,
      "loss": 0.658,
      "step": 792
    },
    {
      "epoch": 0.21251507436687658,
      "grad_norm": 0.6928715752881008,
      "learning_rate": 4.988080266062555e-06,
      "loss": 0.59,
      "step": 793
    },
    {
      "epoch": 0.2127830631113493,
      "grad_norm": 0.7326801987166989,
      "learning_rate": 4.9880458683172395e-06,
      "loss": 0.6522,
      "step": 794
    },
    {
      "epoch": 0.21305105185582204,
      "grad_norm": 0.679778805469598,
      "learning_rate": 4.988011421130169e-06,
      "loss": 0.6179,
      "step": 795
    },
    {
      "epoch": 0.2133190406002948,
      "grad_norm": 0.7707127018675661,
      "learning_rate": 4.987976924502028e-06,
      "loss": 0.627,
      "step": 796
    },
    {
      "epoch": 0.21358702934476753,
      "grad_norm": 0.6682966564866805,
      "learning_rate": 4.987942378433503e-06,
      "loss": 0.6158,
      "step": 797
    },
    {
      "epoch": 0.21385501808924026,
      "grad_norm": 0.738181085481505,
      "learning_rate": 4.98790778292528e-06,
      "loss": 0.6185,
      "step": 798
    },
    {
      "epoch": 0.214123006833713,
      "grad_norm": 0.7470230994241025,
      "learning_rate": 4.987873137978046e-06,
      "loss": 0.6311,
      "step": 799
    },
    {
      "epoch": 0.21439099557818572,
      "grad_norm": 0.7232082588442644,
      "learning_rate": 4.98783844359249e-06,
      "loss": 0.6027,
      "step": 800
    },
    {
      "epoch": 0.21465898432265845,
      "grad_norm": 0.7108269413719349,
      "learning_rate": 4.987803699769301e-06,
      "loss": 0.6596,
      "step": 801
    },
    {
      "epoch": 0.21492697306713118,
      "grad_norm": 0.7082659941603564,
      "learning_rate": 4.98776890650917e-06,
      "loss": 0.5828,
      "step": 802
    },
    {
      "epoch": 0.2151949618116039,
      "grad_norm": 0.7273459388085469,
      "learning_rate": 4.9877340638127875e-06,
      "loss": 0.6177,
      "step": 803
    },
    {
      "epoch": 0.21546295055607664,
      "grad_norm": 0.7309018153777394,
      "learning_rate": 4.987699171680846e-06,
      "loss": 0.6059,
      "step": 804
    },
    {
      "epoch": 0.21573093930054937,
      "grad_norm": 0.7042042022668088,
      "learning_rate": 4.987664230114041e-06,
      "loss": 0.6364,
      "step": 805
    },
    {
      "epoch": 0.2159989280450221,
      "grad_norm": 0.7798474032128689,
      "learning_rate": 4.987629239113065e-06,
      "loss": 0.6197,
      "step": 806
    },
    {
      "epoch": 0.21626691678949483,
      "grad_norm": 0.7614932479062492,
      "learning_rate": 4.987594198678612e-06,
      "loss": 0.6119,
      "step": 807
    },
    {
      "epoch": 0.21653490553396756,
      "grad_norm": 0.7227523840732901,
      "learning_rate": 4.987559108811382e-06,
      "loss": 0.6112,
      "step": 808
    },
    {
      "epoch": 0.2168028942784403,
      "grad_norm": 0.7039657064625956,
      "learning_rate": 4.987523969512068e-06,
      "loss": 0.6009,
      "step": 809
    },
    {
      "epoch": 0.21707088302291302,
      "grad_norm": 0.7683368446260517,
      "learning_rate": 4.9874887807813714e-06,
      "loss": 0.6494,
      "step": 810
    },
    {
      "epoch": 0.21733887176738578,
      "grad_norm": 0.7428098181594326,
      "learning_rate": 4.987453542619991e-06,
      "loss": 0.6245,
      "step": 811
    },
    {
      "epoch": 0.2176068605118585,
      "grad_norm": 0.7240292552301414,
      "learning_rate": 4.987418255028625e-06,
      "loss": 0.642,
      "step": 812
    },
    {
      "epoch": 0.21787484925633124,
      "grad_norm": 0.7441454717184018,
      "learning_rate": 4.987382918007978e-06,
      "loss": 0.5908,
      "step": 813
    },
    {
      "epoch": 0.21814283800080397,
      "grad_norm": 0.7146459601466089,
      "learning_rate": 4.98734753155875e-06,
      "loss": 0.6139,
      "step": 814
    },
    {
      "epoch": 0.2184108267452767,
      "grad_norm": 0.6812184516906802,
      "learning_rate": 4.987312095681643e-06,
      "loss": 0.6367,
      "step": 815
    },
    {
      "epoch": 0.21867881548974943,
      "grad_norm": 0.7636070736367675,
      "learning_rate": 4.987276610377364e-06,
      "loss": 0.6151,
      "step": 816
    },
    {
      "epoch": 0.21894680423422216,
      "grad_norm": 0.6985235366229124,
      "learning_rate": 4.987241075646616e-06,
      "loss": 0.6367,
      "step": 817
    },
    {
      "epoch": 0.2192147929786949,
      "grad_norm": 0.7171389085439135,
      "learning_rate": 4.987205491490107e-06,
      "loss": 0.6644,
      "step": 818
    },
    {
      "epoch": 0.21948278172316762,
      "grad_norm": 0.7370460326463332,
      "learning_rate": 4.9871698579085425e-06,
      "loss": 0.6156,
      "step": 819
    },
    {
      "epoch": 0.21975077046764036,
      "grad_norm": 0.7026065019102886,
      "learning_rate": 4.987134174902631e-06,
      "loss": 0.6254,
      "step": 820
    },
    {
      "epoch": 0.22001875921211309,
      "grad_norm": 0.7356645747516583,
      "learning_rate": 4.9870984424730824e-06,
      "loss": 0.6144,
      "step": 821
    },
    {
      "epoch": 0.22028674795658582,
      "grad_norm": 0.6952240104953946,
      "learning_rate": 4.987062660620606e-06,
      "loss": 0.6264,
      "step": 822
    },
    {
      "epoch": 0.22055473670105855,
      "grad_norm": 0.6874851880532399,
      "learning_rate": 4.9870268293459135e-06,
      "loss": 0.6378,
      "step": 823
    },
    {
      "epoch": 0.22082272544553128,
      "grad_norm": 0.6902196579958221,
      "learning_rate": 4.986990948649716e-06,
      "loss": 0.6214,
      "step": 824
    },
    {
      "epoch": 0.221090714190004,
      "grad_norm": 0.7034456354458589,
      "learning_rate": 4.986955018532726e-06,
      "loss": 0.6181,
      "step": 825
    },
    {
      "epoch": 0.22135870293447676,
      "grad_norm": 0.6924588973261173,
      "learning_rate": 4.98691903899566e-06,
      "loss": 0.6318,
      "step": 826
    },
    {
      "epoch": 0.2216266916789495,
      "grad_norm": 0.7002626426146673,
      "learning_rate": 4.986883010039231e-06,
      "loss": 0.6773,
      "step": 827
    },
    {
      "epoch": 0.22189468042342222,
      "grad_norm": 0.6971261076926438,
      "learning_rate": 4.986846931664155e-06,
      "loss": 0.6549,
      "step": 828
    },
    {
      "epoch": 0.22216266916789496,
      "grad_norm": 0.6930987652815073,
      "learning_rate": 4.986810803871149e-06,
      "loss": 0.627,
      "step": 829
    },
    {
      "epoch": 0.22243065791236769,
      "grad_norm": 0.7001982922186889,
      "learning_rate": 4.986774626660932e-06,
      "loss": 0.6217,
      "step": 830
    },
    {
      "epoch": 0.22269864665684042,
      "grad_norm": 0.7344754612184736,
      "learning_rate": 4.986738400034222e-06,
      "loss": 0.6662,
      "step": 831
    },
    {
      "epoch": 0.22296663540131315,
      "grad_norm": 0.7334390109536708,
      "learning_rate": 4.9867021239917394e-06,
      "loss": 0.6442,
      "step": 832
    },
    {
      "epoch": 0.22323462414578588,
      "grad_norm": 0.6667259962106943,
      "learning_rate": 4.986665798534204e-06,
      "loss": 0.6047,
      "step": 833
    },
    {
      "epoch": 0.2235026128902586,
      "grad_norm": 0.7232081123382714,
      "learning_rate": 4.986629423662338e-06,
      "loss": 0.5982,
      "step": 834
    },
    {
      "epoch": 0.22377060163473134,
      "grad_norm": 0.7258186829625879,
      "learning_rate": 4.986592999376866e-06,
      "loss": 0.5913,
      "step": 835
    },
    {
      "epoch": 0.22403859037920407,
      "grad_norm": 0.7221244872651278,
      "learning_rate": 4.9865565256785085e-06,
      "loss": 0.6242,
      "step": 836
    },
    {
      "epoch": 0.2243065791236768,
      "grad_norm": 0.7345924275782376,
      "learning_rate": 4.986520002567994e-06,
      "loss": 0.6403,
      "step": 837
    },
    {
      "epoch": 0.22457456786814953,
      "grad_norm": 0.7214596576775447,
      "learning_rate": 4.986483430046045e-06,
      "loss": 0.611,
      "step": 838
    },
    {
      "epoch": 0.22484255661262226,
      "grad_norm": 0.7075221994180405,
      "learning_rate": 4.98644680811339e-06,
      "loss": 0.633,
      "step": 839
    },
    {
      "epoch": 0.22511054535709502,
      "grad_norm": 0.6914195314458719,
      "learning_rate": 4.986410136770757e-06,
      "loss": 0.6441,
      "step": 840
    },
    {
      "epoch": 0.22537853410156775,
      "grad_norm": 0.7418667963251849,
      "learning_rate": 4.986373416018874e-06,
      "loss": 0.5956,
      "step": 841
    },
    {
      "epoch": 0.22564652284604048,
      "grad_norm": 0.7005799414510129,
      "learning_rate": 4.986336645858472e-06,
      "loss": 0.6753,
      "step": 842
    },
    {
      "epoch": 0.2259145115905132,
      "grad_norm": 0.7077710476810983,
      "learning_rate": 4.986299826290278e-06,
      "loss": 0.6126,
      "step": 843
    },
    {
      "epoch": 0.22618250033498594,
      "grad_norm": 0.710936116015065,
      "learning_rate": 4.986262957315029e-06,
      "loss": 0.6527,
      "step": 844
    },
    {
      "epoch": 0.22645048907945867,
      "grad_norm": 0.7176640280423536,
      "learning_rate": 4.986226038933454e-06,
      "loss": 0.6607,
      "step": 845
    },
    {
      "epoch": 0.2267184778239314,
      "grad_norm": 0.7453885521533217,
      "learning_rate": 4.9861890711462865e-06,
      "loss": 0.6884,
      "step": 846
    },
    {
      "epoch": 0.22698646656840413,
      "grad_norm": 0.705045775204173,
      "learning_rate": 4.986152053954263e-06,
      "loss": 0.6246,
      "step": 847
    },
    {
      "epoch": 0.22725445531287686,
      "grad_norm": 0.6927490035074463,
      "learning_rate": 4.986114987358118e-06,
      "loss": 0.6168,
      "step": 848
    },
    {
      "epoch": 0.2275224440573496,
      "grad_norm": 0.7233550897962958,
      "learning_rate": 4.9860778713585875e-06,
      "loss": 0.6404,
      "step": 849
    },
    {
      "epoch": 0.22779043280182232,
      "grad_norm": 0.7167963169812048,
      "learning_rate": 4.986040705956411e-06,
      "loss": 0.6075,
      "step": 850
    },
    {
      "epoch": 0.22805842154629505,
      "grad_norm": 0.7072807393188401,
      "learning_rate": 4.986003491152326e-06,
      "loss": 0.6258,
      "step": 851
    },
    {
      "epoch": 0.22832641029076778,
      "grad_norm": 0.7304487364155695,
      "learning_rate": 4.985966226947071e-06,
      "loss": 0.6428,
      "step": 852
    },
    {
      "epoch": 0.2285943990352405,
      "grad_norm": 0.7280862345016951,
      "learning_rate": 4.985928913341388e-06,
      "loss": 0.6603,
      "step": 853
    },
    {
      "epoch": 0.22886238777971324,
      "grad_norm": 0.7102810525108741,
      "learning_rate": 4.985891550336018e-06,
      "loss": 0.6372,
      "step": 854
    },
    {
      "epoch": 0.229130376524186,
      "grad_norm": 0.6697289309400183,
      "learning_rate": 4.985854137931704e-06,
      "loss": 0.6292,
      "step": 855
    },
    {
      "epoch": 0.22939836526865873,
      "grad_norm": 0.7212407232563104,
      "learning_rate": 4.985816676129188e-06,
      "loss": 0.6449,
      "step": 856
    },
    {
      "epoch": 0.22966635401313146,
      "grad_norm": 0.6903255874239748,
      "learning_rate": 4.9857791649292155e-06,
      "loss": 0.6297,
      "step": 857
    },
    {
      "epoch": 0.2299343427576042,
      "grad_norm": 0.7203544789388783,
      "learning_rate": 4.985741604332532e-06,
      "loss": 0.6055,
      "step": 858
    },
    {
      "epoch": 0.23020233150207692,
      "grad_norm": 0.6847585753228989,
      "learning_rate": 4.985703994339882e-06,
      "loss": 0.635,
      "step": 859
    },
    {
      "epoch": 0.23047032024654965,
      "grad_norm": 0.7120889933606431,
      "learning_rate": 4.985666334952016e-06,
      "loss": 0.6403,
      "step": 860
    },
    {
      "epoch": 0.23073830899102238,
      "grad_norm": 0.6882328290145344,
      "learning_rate": 4.985628626169681e-06,
      "loss": 0.624,
      "step": 861
    },
    {
      "epoch": 0.2310062977354951,
      "grad_norm": 0.7121875645035057,
      "learning_rate": 4.985590867993626e-06,
      "loss": 0.6615,
      "step": 862
    },
    {
      "epoch": 0.23127428647996784,
      "grad_norm": 0.7205946092648623,
      "learning_rate": 4.9855530604246e-06,
      "loss": 0.6861,
      "step": 863
    },
    {
      "epoch": 0.23154227522444057,
      "grad_norm": 0.734656923695149,
      "learning_rate": 4.985515203463358e-06,
      "loss": 0.5958,
      "step": 864
    },
    {
      "epoch": 0.2318102639689133,
      "grad_norm": 0.6927046890087187,
      "learning_rate": 4.985477297110648e-06,
      "loss": 0.6264,
      "step": 865
    },
    {
      "epoch": 0.23207825271338603,
      "grad_norm": 0.6868060137069425,
      "learning_rate": 4.985439341367226e-06,
      "loss": 0.6137,
      "step": 866
    },
    {
      "epoch": 0.23234624145785876,
      "grad_norm": 0.7055958907981201,
      "learning_rate": 4.985401336233846e-06,
      "loss": 0.6556,
      "step": 867
    },
    {
      "epoch": 0.2326142302023315,
      "grad_norm": 0.7104641090424276,
      "learning_rate": 4.985363281711263e-06,
      "loss": 0.6289,
      "step": 868
    },
    {
      "epoch": 0.23288221894680425,
      "grad_norm": 0.7105572466989368,
      "learning_rate": 4.985325177800232e-06,
      "loss": 0.6501,
      "step": 869
    },
    {
      "epoch": 0.23315020769127698,
      "grad_norm": 0.7169393030422722,
      "learning_rate": 4.985287024501512e-06,
      "loss": 0.6322,
      "step": 870
    },
    {
      "epoch": 0.2334181964357497,
      "grad_norm": 0.6954415600537528,
      "learning_rate": 4.98524882181586e-06,
      "loss": 0.642,
      "step": 871
    },
    {
      "epoch": 0.23368618518022244,
      "grad_norm": 0.7287446283867897,
      "learning_rate": 4.985210569744036e-06,
      "loss": 0.6085,
      "step": 872
    },
    {
      "epoch": 0.23395417392469517,
      "grad_norm": 0.7078806916298274,
      "learning_rate": 4.985172268286799e-06,
      "loss": 0.6092,
      "step": 873
    },
    {
      "epoch": 0.2342221626691679,
      "grad_norm": 0.6999058460940814,
      "learning_rate": 4.985133917444912e-06,
      "loss": 0.6028,
      "step": 874
    },
    {
      "epoch": 0.23449015141364063,
      "grad_norm": 0.7229615874884397,
      "learning_rate": 4.985095517219134e-06,
      "loss": 0.6403,
      "step": 875
    },
    {
      "epoch": 0.23475814015811336,
      "grad_norm": 0.7109360035963607,
      "learning_rate": 4.9850570676102315e-06,
      "loss": 0.6203,
      "step": 876
    },
    {
      "epoch": 0.2350261289025861,
      "grad_norm": 0.735428681893653,
      "learning_rate": 4.985018568618966e-06,
      "loss": 0.6326,
      "step": 877
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.690401026466452,
      "learning_rate": 4.984980020246104e-06,
      "loss": 0.6464,
      "step": 878
    },
    {
      "epoch": 0.23556210639153155,
      "grad_norm": 0.6713955303468878,
      "learning_rate": 4.984941422492411e-06,
      "loss": 0.5901,
      "step": 879
    },
    {
      "epoch": 0.23583009513600428,
      "grad_norm": 0.7083563279072498,
      "learning_rate": 4.984902775358654e-06,
      "loss": 0.6365,
      "step": 880
    },
    {
      "epoch": 0.236098083880477,
      "grad_norm": 0.6902860506830981,
      "learning_rate": 4.984864078845601e-06,
      "loss": 0.6357,
      "step": 881
    },
    {
      "epoch": 0.23636607262494974,
      "grad_norm": 0.7052734899061491,
      "learning_rate": 4.984825332954021e-06,
      "loss": 0.6321,
      "step": 882
    },
    {
      "epoch": 0.23663406136942247,
      "grad_norm": 0.7074497247061396,
      "learning_rate": 4.984786537684684e-06,
      "loss": 0.6694,
      "step": 883
    },
    {
      "epoch": 0.23690205011389523,
      "grad_norm": 0.6933055148108048,
      "learning_rate": 4.984747693038361e-06,
      "loss": 0.5977,
      "step": 884
    },
    {
      "epoch": 0.23717003885836796,
      "grad_norm": 0.6952192995250391,
      "learning_rate": 4.984708799015824e-06,
      "loss": 0.614,
      "step": 885
    },
    {
      "epoch": 0.2374380276028407,
      "grad_norm": 0.713274332910376,
      "learning_rate": 4.984669855617845e-06,
      "loss": 0.6309,
      "step": 886
    },
    {
      "epoch": 0.23770601634731342,
      "grad_norm": 0.6767818653069274,
      "learning_rate": 4.984630862845199e-06,
      "loss": 0.6273,
      "step": 887
    },
    {
      "epoch": 0.23797400509178615,
      "grad_norm": 0.7049581818665084,
      "learning_rate": 4.984591820698661e-06,
      "loss": 0.6414,
      "step": 888
    },
    {
      "epoch": 0.23824199383625888,
      "grad_norm": 0.666465681305506,
      "learning_rate": 4.984552729179006e-06,
      "loss": 0.6141,
      "step": 889
    },
    {
      "epoch": 0.2385099825807316,
      "grad_norm": 0.691415580520586,
      "learning_rate": 4.98451358828701e-06,
      "loss": 0.6237,
      "step": 890
    },
    {
      "epoch": 0.23877797132520434,
      "grad_norm": 0.6680103936691001,
      "learning_rate": 4.9844743980234535e-06,
      "loss": 0.6158,
      "step": 891
    },
    {
      "epoch": 0.23904596006967707,
      "grad_norm": 0.6830009549281377,
      "learning_rate": 4.9844351583891125e-06,
      "loss": 0.6433,
      "step": 892
    },
    {
      "epoch": 0.2393139488141498,
      "grad_norm": 0.6929816303207358,
      "learning_rate": 4.984395869384768e-06,
      "loss": 0.6134,
      "step": 893
    },
    {
      "epoch": 0.23958193755862253,
      "grad_norm": 0.7072701261773989,
      "learning_rate": 4.9843565310112e-06,
      "loss": 0.6243,
      "step": 894
    },
    {
      "epoch": 0.23984992630309526,
      "grad_norm": 0.6909602740038907,
      "learning_rate": 4.984317143269193e-06,
      "loss": 0.6256,
      "step": 895
    },
    {
      "epoch": 0.240117915047568,
      "grad_norm": 0.6936807994919197,
      "learning_rate": 4.984277706159526e-06,
      "loss": 0.6293,
      "step": 896
    },
    {
      "epoch": 0.24038590379204072,
      "grad_norm": 0.6719880820313626,
      "learning_rate": 4.984238219682985e-06,
      "loss": 0.5957,
      "step": 897
    },
    {
      "epoch": 0.24065389253651345,
      "grad_norm": 0.6981386696371421,
      "learning_rate": 4.984198683840353e-06,
      "loss": 0.6447,
      "step": 898
    },
    {
      "epoch": 0.2409218812809862,
      "grad_norm": 0.6881767024015881,
      "learning_rate": 4.984159098632417e-06,
      "loss": 0.6555,
      "step": 899
    },
    {
      "epoch": 0.24118987002545894,
      "grad_norm": 0.7184679135740648,
      "learning_rate": 4.984119464059963e-06,
      "loss": 0.6551,
      "step": 900
    },
    {
      "epoch": 0.24145785876993167,
      "grad_norm": 0.726142122771097,
      "learning_rate": 4.98407978012378e-06,
      "loss": 0.6267,
      "step": 901
    },
    {
      "epoch": 0.2417258475144044,
      "grad_norm": 0.7114488620654842,
      "learning_rate": 4.984040046824654e-06,
      "loss": 0.6415,
      "step": 902
    },
    {
      "epoch": 0.24199383625887713,
      "grad_norm": 0.708793461848355,
      "learning_rate": 4.984000264163376e-06,
      "loss": 0.5828,
      "step": 903
    },
    {
      "epoch": 0.24226182500334986,
      "grad_norm": 0.7337425579499977,
      "learning_rate": 4.983960432140737e-06,
      "loss": 0.5992,
      "step": 904
    },
    {
      "epoch": 0.2425298137478226,
      "grad_norm": 0.7153910480764532,
      "learning_rate": 4.983920550757529e-06,
      "loss": 0.6299,
      "step": 905
    },
    {
      "epoch": 0.24279780249229532,
      "grad_norm": 0.6995831064245144,
      "learning_rate": 4.983880620014542e-06,
      "loss": 0.6669,
      "step": 906
    },
    {
      "epoch": 0.24306579123676805,
      "grad_norm": 0.688232721418614,
      "learning_rate": 4.9838406399125715e-06,
      "loss": 0.6575,
      "step": 907
    },
    {
      "epoch": 0.24333377998124078,
      "grad_norm": 0.6766932759060557,
      "learning_rate": 4.98380061045241e-06,
      "loss": 0.6265,
      "step": 908
    },
    {
      "epoch": 0.2436017687257135,
      "grad_norm": 0.7044743488033326,
      "learning_rate": 4.983760531634857e-06,
      "loss": 0.6246,
      "step": 909
    },
    {
      "epoch": 0.24386975747018624,
      "grad_norm": 0.710207454990041,
      "learning_rate": 4.983720403460705e-06,
      "loss": 0.6059,
      "step": 910
    },
    {
      "epoch": 0.24413774621465897,
      "grad_norm": 0.6985045192319745,
      "learning_rate": 4.983680225930754e-06,
      "loss": 0.6258,
      "step": 911
    },
    {
      "epoch": 0.2444057349591317,
      "grad_norm": 0.677364601886501,
      "learning_rate": 4.9836399990458e-06,
      "loss": 0.6308,
      "step": 912
    },
    {
      "epoch": 0.24467372370360446,
      "grad_norm": 0.6941478247239629,
      "learning_rate": 4.983599722806644e-06,
      "loss": 0.6336,
      "step": 913
    },
    {
      "epoch": 0.2449417124480772,
      "grad_norm": 0.701577115889416,
      "learning_rate": 4.983559397214086e-06,
      "loss": 0.6661,
      "step": 914
    },
    {
      "epoch": 0.24520970119254992,
      "grad_norm": 0.6992133759601831,
      "learning_rate": 4.9835190222689274e-06,
      "loss": 0.6556,
      "step": 915
    },
    {
      "epoch": 0.24547768993702265,
      "grad_norm": 0.6789133088742016,
      "learning_rate": 4.98347859797197e-06,
      "loss": 0.6331,
      "step": 916
    },
    {
      "epoch": 0.24574567868149538,
      "grad_norm": 0.7189487228188031,
      "learning_rate": 4.9834381243240174e-06,
      "loss": 0.6119,
      "step": 917
    },
    {
      "epoch": 0.2460136674259681,
      "grad_norm": 0.698851556699704,
      "learning_rate": 4.983397601325874e-06,
      "loss": 0.6531,
      "step": 918
    },
    {
      "epoch": 0.24628165617044084,
      "grad_norm": 0.6772683267469517,
      "learning_rate": 4.983357028978346e-06,
      "loss": 0.645,
      "step": 919
    },
    {
      "epoch": 0.24654964491491357,
      "grad_norm": 0.6887595762406844,
      "learning_rate": 4.983316407282238e-06,
      "loss": 0.6306,
      "step": 920
    },
    {
      "epoch": 0.2468176336593863,
      "grad_norm": 0.656894771867854,
      "learning_rate": 4.983275736238358e-06,
      "loss": 0.6207,
      "step": 921
    },
    {
      "epoch": 0.24708562240385903,
      "grad_norm": 0.6749418050510967,
      "learning_rate": 4.983235015847514e-06,
      "loss": 0.5888,
      "step": 922
    },
    {
      "epoch": 0.24735361114833176,
      "grad_norm": 0.6802951828163503,
      "learning_rate": 4.983194246110516e-06,
      "loss": 0.6106,
      "step": 923
    },
    {
      "epoch": 0.2476215998928045,
      "grad_norm": 0.7192690688354414,
      "learning_rate": 4.983153427028173e-06,
      "loss": 0.6556,
      "step": 924
    },
    {
      "epoch": 0.24788958863727723,
      "grad_norm": 0.688071191598469,
      "learning_rate": 4.983112558601297e-06,
      "loss": 0.6432,
      "step": 925
    },
    {
      "epoch": 0.24815757738174996,
      "grad_norm": 0.6859227425811241,
      "learning_rate": 4.983071640830699e-06,
      "loss": 0.6042,
      "step": 926
    },
    {
      "epoch": 0.24842556612622269,
      "grad_norm": 0.7100678234656254,
      "learning_rate": 4.983030673717193e-06,
      "loss": 0.6288,
      "step": 927
    },
    {
      "epoch": 0.24869355487069544,
      "grad_norm": 0.7082615715158066,
      "learning_rate": 4.982989657261593e-06,
      "loss": 0.5999,
      "step": 928
    },
    {
      "epoch": 0.24896154361516817,
      "grad_norm": 0.705221778940248,
      "learning_rate": 4.9829485914647136e-06,
      "loss": 0.6463,
      "step": 929
    },
    {
      "epoch": 0.2492295323596409,
      "grad_norm": 0.7011576828118872,
      "learning_rate": 4.982907476327372e-06,
      "loss": 0.6391,
      "step": 930
    },
    {
      "epoch": 0.24949752110411363,
      "grad_norm": 0.6939622482684725,
      "learning_rate": 4.982866311850384e-06,
      "loss": 0.6091,
      "step": 931
    },
    {
      "epoch": 0.24976550984858636,
      "grad_norm": 0.7447879693216204,
      "learning_rate": 4.9828250980345684e-06,
      "loss": 0.5874,
      "step": 932
    },
    {
      "epoch": 0.25003349859305907,
      "grad_norm": 0.744409176284046,
      "learning_rate": 4.9827838348807436e-06,
      "loss": 0.6352,
      "step": 933
    },
    {
      "epoch": 0.2503014873375318,
      "grad_norm": 0.688787968534109,
      "learning_rate": 4.9827425223897305e-06,
      "loss": 0.6363,
      "step": 934
    },
    {
      "epoch": 0.2505694760820046,
      "grad_norm": 0.7762519612486921,
      "learning_rate": 4.982701160562349e-06,
      "loss": 0.6168,
      "step": 935
    },
    {
      "epoch": 0.2508374648264773,
      "grad_norm": 0.6861966674808816,
      "learning_rate": 4.982659749399421e-06,
      "loss": 0.6528,
      "step": 936
    },
    {
      "epoch": 0.25110545357095004,
      "grad_norm": 0.6779574610056516,
      "learning_rate": 4.9826182889017706e-06,
      "loss": 0.6321,
      "step": 937
    },
    {
      "epoch": 0.25137344231542275,
      "grad_norm": 0.7009343638753798,
      "learning_rate": 4.98257677907022e-06,
      "loss": 0.6487,
      "step": 938
    },
    {
      "epoch": 0.2516414310598955,
      "grad_norm": 0.6941102602342629,
      "learning_rate": 4.982535219905595e-06,
      "loss": 0.6793,
      "step": 939
    },
    {
      "epoch": 0.2519094198043682,
      "grad_norm": 0.6854670535872863,
      "learning_rate": 4.982493611408722e-06,
      "loss": 0.6057,
      "step": 940
    },
    {
      "epoch": 0.25217740854884096,
      "grad_norm": 0.7121552710621692,
      "learning_rate": 4.9824519535804275e-06,
      "loss": 0.6487,
      "step": 941
    },
    {
      "epoch": 0.25244539729331367,
      "grad_norm": 0.6793283318246764,
      "learning_rate": 4.982410246421539e-06,
      "loss": 0.6,
      "step": 942
    },
    {
      "epoch": 0.2527133860377864,
      "grad_norm": 0.733648950152187,
      "learning_rate": 4.982368489932885e-06,
      "loss": 0.5914,
      "step": 943
    },
    {
      "epoch": 0.2529813747822591,
      "grad_norm": 0.7123450211700185,
      "learning_rate": 4.982326684115296e-06,
      "loss": 0.6352,
      "step": 944
    },
    {
      "epoch": 0.2532493635267319,
      "grad_norm": 0.789848889870531,
      "learning_rate": 4.982284828969602e-06,
      "loss": 0.6539,
      "step": 945
    },
    {
      "epoch": 0.2535173522712046,
      "grad_norm": 0.6930952777197598,
      "learning_rate": 4.982242924496636e-06,
      "loss": 0.5846,
      "step": 946
    },
    {
      "epoch": 0.25378534101567735,
      "grad_norm": 0.7246643738187113,
      "learning_rate": 4.9822009706972295e-06,
      "loss": 0.6479,
      "step": 947
    },
    {
      "epoch": 0.25405332976015005,
      "grad_norm": 0.7762646995750418,
      "learning_rate": 4.982158967572217e-06,
      "loss": 0.6299,
      "step": 948
    },
    {
      "epoch": 0.2543213185046228,
      "grad_norm": 0.7394458433305346,
      "learning_rate": 4.982116915122432e-06,
      "loss": 0.6207,
      "step": 949
    },
    {
      "epoch": 0.25458930724909556,
      "grad_norm": 0.6688358547687572,
      "learning_rate": 4.982074813348711e-06,
      "loss": 0.5879,
      "step": 950
    },
    {
      "epoch": 0.25485729599356827,
      "grad_norm": 0.7078262892139043,
      "learning_rate": 4.98203266225189e-06,
      "loss": 0.609,
      "step": 951
    },
    {
      "epoch": 0.255125284738041,
      "grad_norm": 0.7188833913797128,
      "learning_rate": 4.981990461832808e-06,
      "loss": 0.6228,
      "step": 952
    },
    {
      "epoch": 0.2553932734825137,
      "grad_norm": 0.7493110224846776,
      "learning_rate": 4.9819482120923025e-06,
      "loss": 0.6025,
      "step": 953
    },
    {
      "epoch": 0.2556612622269865,
      "grad_norm": 0.7241802869097672,
      "learning_rate": 4.9819059130312134e-06,
      "loss": 0.6004,
      "step": 954
    },
    {
      "epoch": 0.2559292509714592,
      "grad_norm": 0.7009100543758139,
      "learning_rate": 4.981863564650381e-06,
      "loss": 0.6016,
      "step": 955
    },
    {
      "epoch": 0.25619723971593195,
      "grad_norm": 0.688522069592028,
      "learning_rate": 4.9818211669506465e-06,
      "loss": 0.6277,
      "step": 956
    },
    {
      "epoch": 0.25646522846040465,
      "grad_norm": 0.7032380692696123,
      "learning_rate": 4.981778719932853e-06,
      "loss": 0.6499,
      "step": 957
    },
    {
      "epoch": 0.2567332172048774,
      "grad_norm": 0.7548159084845224,
      "learning_rate": 4.981736223597845e-06,
      "loss": 0.6307,
      "step": 958
    },
    {
      "epoch": 0.2570012059493501,
      "grad_norm": 0.7057825901507616,
      "learning_rate": 4.981693677946465e-06,
      "loss": 0.6211,
      "step": 959
    },
    {
      "epoch": 0.25726919469382287,
      "grad_norm": 0.6858155496072331,
      "learning_rate": 4.98165108297956e-06,
      "loss": 0.615,
      "step": 960
    },
    {
      "epoch": 0.25753718343829557,
      "grad_norm": 0.8134329859346021,
      "learning_rate": 4.981608438697976e-06,
      "loss": 0.5913,
      "step": 961
    },
    {
      "epoch": 0.2578051721827683,
      "grad_norm": 0.7405377869313259,
      "learning_rate": 4.981565745102559e-06,
      "loss": 0.6352,
      "step": 962
    },
    {
      "epoch": 0.25807316092724103,
      "grad_norm": 0.7166230685069928,
      "learning_rate": 4.981523002194159e-06,
      "loss": 0.6134,
      "step": 963
    },
    {
      "epoch": 0.2583411496717138,
      "grad_norm": 0.7495500810267214,
      "learning_rate": 4.981480209973626e-06,
      "loss": 0.5862,
      "step": 964
    },
    {
      "epoch": 0.25860913841618655,
      "grad_norm": 0.7031842198663605,
      "learning_rate": 4.981437368441808e-06,
      "loss": 0.6213,
      "step": 965
    },
    {
      "epoch": 0.25887712716065925,
      "grad_norm": 0.6880066806082806,
      "learning_rate": 4.981394477599558e-06,
      "loss": 0.628,
      "step": 966
    },
    {
      "epoch": 0.259145115905132,
      "grad_norm": 0.7101170054164099,
      "learning_rate": 4.981351537447728e-06,
      "loss": 0.5996,
      "step": 967
    },
    {
      "epoch": 0.2594131046496047,
      "grad_norm": 0.7258816181866747,
      "learning_rate": 4.981308547987172e-06,
      "loss": 0.6271,
      "step": 968
    },
    {
      "epoch": 0.25968109339407747,
      "grad_norm": 0.6848752999316978,
      "learning_rate": 4.981265509218742e-06,
      "loss": 0.6367,
      "step": 969
    },
    {
      "epoch": 0.25994908213855017,
      "grad_norm": 0.7319862946536351,
      "learning_rate": 4.981222421143296e-06,
      "loss": 0.6311,
      "step": 970
    },
    {
      "epoch": 0.2602170708830229,
      "grad_norm": 0.7351018744059107,
      "learning_rate": 4.981179283761689e-06,
      "loss": 0.6049,
      "step": 971
    },
    {
      "epoch": 0.26048505962749563,
      "grad_norm": 0.7014606209605837,
      "learning_rate": 4.981136097074777e-06,
      "loss": 0.62,
      "step": 972
    },
    {
      "epoch": 0.2607530483719684,
      "grad_norm": 0.7140916145923494,
      "learning_rate": 4.981092861083421e-06,
      "loss": 0.617,
      "step": 973
    },
    {
      "epoch": 0.2610210371164411,
      "grad_norm": 0.7464248543377602,
      "learning_rate": 4.981049575788478e-06,
      "loss": 0.6299,
      "step": 974
    },
    {
      "epoch": 0.26128902586091385,
      "grad_norm": 0.6937463492779748,
      "learning_rate": 4.981006241190809e-06,
      "loss": 0.6075,
      "step": 975
    },
    {
      "epoch": 0.26155701460538655,
      "grad_norm": 0.6860235550651806,
      "learning_rate": 4.980962857291274e-06,
      "loss": 0.6495,
      "step": 976
    },
    {
      "epoch": 0.2618250033498593,
      "grad_norm": 0.6720862370766615,
      "learning_rate": 4.980919424090737e-06,
      "loss": 0.6293,
      "step": 977
    },
    {
      "epoch": 0.262092992094332,
      "grad_norm": 0.6711622370534756,
      "learning_rate": 4.98087594159006e-06,
      "loss": 0.6133,
      "step": 978
    },
    {
      "epoch": 0.26236098083880477,
      "grad_norm": 0.6836776183104105,
      "learning_rate": 4.980832409790106e-06,
      "loss": 0.6206,
      "step": 979
    },
    {
      "epoch": 0.2626289695832775,
      "grad_norm": 0.6819286172499913,
      "learning_rate": 4.9807888286917425e-06,
      "loss": 0.6356,
      "step": 980
    },
    {
      "epoch": 0.26289695832775023,
      "grad_norm": 0.7754186612425207,
      "learning_rate": 4.980745198295834e-06,
      "loss": 0.6465,
      "step": 981
    },
    {
      "epoch": 0.263164947072223,
      "grad_norm": 0.6932470593226673,
      "learning_rate": 4.980701518603248e-06,
      "loss": 0.6106,
      "step": 982
    },
    {
      "epoch": 0.2634329358166957,
      "grad_norm": 0.6741892875510037,
      "learning_rate": 4.980657789614852e-06,
      "loss": 0.6234,
      "step": 983
    },
    {
      "epoch": 0.26370092456116845,
      "grad_norm": 0.6980599007196342,
      "learning_rate": 4.980614011331515e-06,
      "loss": 0.6564,
      "step": 984
    },
    {
      "epoch": 0.26396891330564115,
      "grad_norm": 0.7021388977307166,
      "learning_rate": 4.980570183754107e-06,
      "loss": 0.6258,
      "step": 985
    },
    {
      "epoch": 0.2642369020501139,
      "grad_norm": 0.7133578325197556,
      "learning_rate": 4.9805263068835e-06,
      "loss": 0.6278,
      "step": 986
    },
    {
      "epoch": 0.2645048907945866,
      "grad_norm": 0.6961146775694008,
      "learning_rate": 4.980482380720564e-06,
      "loss": 0.6123,
      "step": 987
    },
    {
      "epoch": 0.26477287953905937,
      "grad_norm": 0.690596613353715,
      "learning_rate": 4.980438405266173e-06,
      "loss": 0.6129,
      "step": 988
    },
    {
      "epoch": 0.26504086828353207,
      "grad_norm": 0.685541494885559,
      "learning_rate": 4.9803943805212015e-06,
      "loss": 0.6289,
      "step": 989
    },
    {
      "epoch": 0.26530885702800483,
      "grad_norm": 0.6983793526622475,
      "learning_rate": 4.980350306486524e-06,
      "loss": 0.6502,
      "step": 990
    },
    {
      "epoch": 0.26557684577247753,
      "grad_norm": 0.9408432945326535,
      "learning_rate": 4.9803061831630145e-06,
      "loss": 0.6516,
      "step": 991
    },
    {
      "epoch": 0.2658448345169503,
      "grad_norm": 0.6942077594612021,
      "learning_rate": 4.980262010551552e-06,
      "loss": 0.6435,
      "step": 992
    },
    {
      "epoch": 0.266112823261423,
      "grad_norm": 0.6809032213651688,
      "learning_rate": 4.980217788653013e-06,
      "loss": 0.6034,
      "step": 993
    },
    {
      "epoch": 0.26638081200589575,
      "grad_norm": 0.7093608681494402,
      "learning_rate": 4.980173517468278e-06,
      "loss": 0.6845,
      "step": 994
    },
    {
      "epoch": 0.2666488007503685,
      "grad_norm": 0.6971074506660051,
      "learning_rate": 4.9801291969982245e-06,
      "loss": 0.6418,
      "step": 995
    },
    {
      "epoch": 0.2669167894948412,
      "grad_norm": 0.6686393618662919,
      "learning_rate": 4.980084827243735e-06,
      "loss": 0.5934,
      "step": 996
    },
    {
      "epoch": 0.26718477823931397,
      "grad_norm": 0.6924916549196342,
      "learning_rate": 4.98004040820569e-06,
      "loss": 0.6345,
      "step": 997
    },
    {
      "epoch": 0.26745276698378667,
      "grad_norm": 0.700092407165652,
      "learning_rate": 4.979995939884973e-06,
      "loss": 0.5916,
      "step": 998
    },
    {
      "epoch": 0.26772075572825943,
      "grad_norm": 0.6767353508658231,
      "learning_rate": 4.979951422282466e-06,
      "loss": 0.588,
      "step": 999
    },
    {
      "epoch": 0.26798874447273213,
      "grad_norm": 0.6692210908458796,
      "learning_rate": 4.979906855399056e-06,
      "loss": 0.632,
      "step": 1000
    },
    {
      "epoch": 0.2682567332172049,
      "grad_norm": 0.7199932357304842,
      "learning_rate": 4.979862239235628e-06,
      "loss": 0.6525,
      "step": 1001
    },
    {
      "epoch": 0.2685247219616776,
      "grad_norm": 0.7206370282036285,
      "learning_rate": 4.979817573793068e-06,
      "loss": 0.6213,
      "step": 1002
    },
    {
      "epoch": 0.26879271070615035,
      "grad_norm": 0.7497070125402553,
      "learning_rate": 4.979772859072264e-06,
      "loss": 0.6244,
      "step": 1003
    },
    {
      "epoch": 0.26906069945062305,
      "grad_norm": 0.6987971267488747,
      "learning_rate": 4.979728095074103e-06,
      "loss": 0.6349,
      "step": 1004
    },
    {
      "epoch": 0.2693286881950958,
      "grad_norm": 0.6948871990952534,
      "learning_rate": 4.979683281799477e-06,
      "loss": 0.6365,
      "step": 1005
    },
    {
      "epoch": 0.2695966769395685,
      "grad_norm": 0.6988565733116721,
      "learning_rate": 4.979638419249275e-06,
      "loss": 0.6125,
      "step": 1006
    },
    {
      "epoch": 0.26986466568404127,
      "grad_norm": 0.7339022126764986,
      "learning_rate": 4.97959350742439e-06,
      "loss": 0.6166,
      "step": 1007
    },
    {
      "epoch": 0.27013265442851403,
      "grad_norm": 0.6553376079581585,
      "learning_rate": 4.979548546325712e-06,
      "loss": 0.6426,
      "step": 1008
    },
    {
      "epoch": 0.27040064317298673,
      "grad_norm": 0.744498721618915,
      "learning_rate": 4.979503535954135e-06,
      "loss": 0.5906,
      "step": 1009
    },
    {
      "epoch": 0.2706686319174595,
      "grad_norm": 0.7146153182554145,
      "learning_rate": 4.979458476310555e-06,
      "loss": 0.5989,
      "step": 1010
    },
    {
      "epoch": 0.2709366206619322,
      "grad_norm": 0.7066259907467538,
      "learning_rate": 4.9794133673958685e-06,
      "loss": 0.6358,
      "step": 1011
    },
    {
      "epoch": 0.27120460940640495,
      "grad_norm": 0.7156723259421509,
      "learning_rate": 4.979368209210968e-06,
      "loss": 0.6351,
      "step": 1012
    },
    {
      "epoch": 0.27147259815087765,
      "grad_norm": 0.6988999050926511,
      "learning_rate": 4.979323001756754e-06,
      "loss": 0.6202,
      "step": 1013
    },
    {
      "epoch": 0.2717405868953504,
      "grad_norm": 0.719443491114625,
      "learning_rate": 4.979277745034123e-06,
      "loss": 0.6596,
      "step": 1014
    },
    {
      "epoch": 0.2720085756398231,
      "grad_norm": 0.6843375775808553,
      "learning_rate": 4.979232439043975e-06,
      "loss": 0.6014,
      "step": 1015
    },
    {
      "epoch": 0.27227656438429587,
      "grad_norm": 0.8308807580587616,
      "learning_rate": 4.979187083787212e-06,
      "loss": 0.646,
      "step": 1016
    },
    {
      "epoch": 0.2725445531287686,
      "grad_norm": 0.6815835038179785,
      "learning_rate": 4.979141679264732e-06,
      "loss": 0.6041,
      "step": 1017
    },
    {
      "epoch": 0.27281254187324133,
      "grad_norm": 0.6914943701233427,
      "learning_rate": 4.979096225477439e-06,
      "loss": 0.6613,
      "step": 1018
    },
    {
      "epoch": 0.27308053061771403,
      "grad_norm": 0.6820270558589554,
      "learning_rate": 4.979050722426236e-06,
      "loss": 0.6446,
      "step": 1019
    },
    {
      "epoch": 0.2733485193621868,
      "grad_norm": 0.6993425062476275,
      "learning_rate": 4.979005170112028e-06,
      "loss": 0.6329,
      "step": 1020
    },
    {
      "epoch": 0.2736165081066595,
      "grad_norm": 0.717373604162941,
      "learning_rate": 4.9789595685357195e-06,
      "loss": 0.611,
      "step": 1021
    },
    {
      "epoch": 0.27388449685113225,
      "grad_norm": 0.6958279400344208,
      "learning_rate": 4.978913917698217e-06,
      "loss": 0.6311,
      "step": 1022
    },
    {
      "epoch": 0.274152485595605,
      "grad_norm": 0.6732284999750405,
      "learning_rate": 4.978868217600428e-06,
      "loss": 0.6287,
      "step": 1023
    },
    {
      "epoch": 0.2744204743400777,
      "grad_norm": 0.6970175230141675,
      "learning_rate": 4.978822468243259e-06,
      "loss": 0.6095,
      "step": 1024
    },
    {
      "epoch": 0.27468846308455047,
      "grad_norm": 0.6919397317253391,
      "learning_rate": 4.9787766696276205e-06,
      "loss": 0.615,
      "step": 1025
    },
    {
      "epoch": 0.2749564518290232,
      "grad_norm": 0.6851223596951393,
      "learning_rate": 4.978730821754423e-06,
      "loss": 0.6382,
      "step": 1026
    },
    {
      "epoch": 0.27522444057349593,
      "grad_norm": 0.6780786479302338,
      "learning_rate": 4.978684924624576e-06,
      "loss": 0.6388,
      "step": 1027
    },
    {
      "epoch": 0.27549242931796863,
      "grad_norm": 0.6911325802900693,
      "learning_rate": 4.9786389782389935e-06,
      "loss": 0.5973,
      "step": 1028
    },
    {
      "epoch": 0.2757604180624414,
      "grad_norm": 0.7231173531994355,
      "learning_rate": 4.9785929825985866e-06,
      "loss": 0.6293,
      "step": 1029
    },
    {
      "epoch": 0.2760284068069141,
      "grad_norm": 0.7420104190226369,
      "learning_rate": 4.978546937704271e-06,
      "loss": 0.5756,
      "step": 1030
    },
    {
      "epoch": 0.27629639555138685,
      "grad_norm": 0.7177491354377054,
      "learning_rate": 4.978500843556959e-06,
      "loss": 0.6048,
      "step": 1031
    },
    {
      "epoch": 0.27656438429585956,
      "grad_norm": 0.7117360599073039,
      "learning_rate": 4.9784547001575704e-06,
      "loss": 0.6531,
      "step": 1032
    },
    {
      "epoch": 0.2768323730403323,
      "grad_norm": 0.7591083707851661,
      "learning_rate": 4.97840850750702e-06,
      "loss": 0.6296,
      "step": 1033
    },
    {
      "epoch": 0.277100361784805,
      "grad_norm": 0.6921975410006309,
      "learning_rate": 4.978362265606224e-06,
      "loss": 0.6249,
      "step": 1034
    },
    {
      "epoch": 0.2773683505292778,
      "grad_norm": 0.7228465993275782,
      "learning_rate": 4.978315974456106e-06,
      "loss": 0.5853,
      "step": 1035
    },
    {
      "epoch": 0.2776363392737505,
      "grad_norm": 0.7049288043995434,
      "learning_rate": 4.978269634057582e-06,
      "loss": 0.6148,
      "step": 1036
    },
    {
      "epoch": 0.27790432801822323,
      "grad_norm": 0.6949544902117049,
      "learning_rate": 4.978223244411573e-06,
      "loss": 0.6155,
      "step": 1037
    },
    {
      "epoch": 0.278172316762696,
      "grad_norm": 0.7212880055453204,
      "learning_rate": 4.978176805519003e-06,
      "loss": 0.6411,
      "step": 1038
    },
    {
      "epoch": 0.2784403055071687,
      "grad_norm": 0.760893704537892,
      "learning_rate": 4.978130317380793e-06,
      "loss": 0.5749,
      "step": 1039
    },
    {
      "epoch": 0.27870829425164145,
      "grad_norm": 0.7172470074266377,
      "learning_rate": 4.978083779997869e-06,
      "loss": 0.5941,
      "step": 1040
    },
    {
      "epoch": 0.27897628299611416,
      "grad_norm": 0.7127241812183884,
      "learning_rate": 4.978037193371152e-06,
      "loss": 0.5871,
      "step": 1041
    },
    {
      "epoch": 0.2792442717405869,
      "grad_norm": 0.7221522173710732,
      "learning_rate": 4.977990557501572e-06,
      "loss": 0.6172,
      "step": 1042
    },
    {
      "epoch": 0.2795122604850596,
      "grad_norm": 0.7214874357398022,
      "learning_rate": 4.977943872390052e-06,
      "loss": 0.6277,
      "step": 1043
    },
    {
      "epoch": 0.2797802492295324,
      "grad_norm": 0.730000004434436,
      "learning_rate": 4.977897138037523e-06,
      "loss": 0.6342,
      "step": 1044
    },
    {
      "epoch": 0.2800482379740051,
      "grad_norm": 0.7483414444140956,
      "learning_rate": 4.977850354444911e-06,
      "loss": 0.6361,
      "step": 1045
    },
    {
      "epoch": 0.28031622671847783,
      "grad_norm": 0.7106984580607522,
      "learning_rate": 4.977803521613147e-06,
      "loss": 0.6263,
      "step": 1046
    },
    {
      "epoch": 0.28058421546295054,
      "grad_norm": 0.7014557552427714,
      "learning_rate": 4.9777566395431615e-06,
      "loss": 0.6273,
      "step": 1047
    },
    {
      "epoch": 0.2808522042074233,
      "grad_norm": 0.7890897332823577,
      "learning_rate": 4.977709708235886e-06,
      "loss": 0.6259,
      "step": 1048
    },
    {
      "epoch": 0.281120192951896,
      "grad_norm": 0.6952895149794047,
      "learning_rate": 4.9776627276922536e-06,
      "loss": 0.6335,
      "step": 1049
    },
    {
      "epoch": 0.28138818169636876,
      "grad_norm": 0.7337207676674375,
      "learning_rate": 4.977615697913197e-06,
      "loss": 0.6598,
      "step": 1050
    },
    {
      "epoch": 0.28165617044084146,
      "grad_norm": 0.7485048852867977,
      "learning_rate": 4.977568618899651e-06,
      "loss": 0.6425,
      "step": 1051
    },
    {
      "epoch": 0.2819241591853142,
      "grad_norm": 0.690328637623898,
      "learning_rate": 4.9775214906525525e-06,
      "loss": 0.596,
      "step": 1052
    },
    {
      "epoch": 0.282192147929787,
      "grad_norm": 0.7112119859311132,
      "learning_rate": 4.977474313172836e-06,
      "loss": 0.6294,
      "step": 1053
    },
    {
      "epoch": 0.2824601366742597,
      "grad_norm": 0.7157524022986882,
      "learning_rate": 4.97742708646144e-06,
      "loss": 0.5738,
      "step": 1054
    },
    {
      "epoch": 0.28272812541873243,
      "grad_norm": 0.726330151014833,
      "learning_rate": 4.9773798105193026e-06,
      "loss": 0.614,
      "step": 1055
    },
    {
      "epoch": 0.28299611416320514,
      "grad_norm": 0.7311807203586334,
      "learning_rate": 4.977332485347365e-06,
      "loss": 0.6395,
      "step": 1056
    },
    {
      "epoch": 0.2832641029076779,
      "grad_norm": 0.7225575223427521,
      "learning_rate": 4.9772851109465646e-06,
      "loss": 0.5885,
      "step": 1057
    },
    {
      "epoch": 0.2835320916521506,
      "grad_norm": 0.7226622844108204,
      "learning_rate": 4.9772376873178444e-06,
      "loss": 0.6124,
      "step": 1058
    },
    {
      "epoch": 0.28380008039662336,
      "grad_norm": 0.6910373020539856,
      "learning_rate": 4.977190214462147e-06,
      "loss": 0.6334,
      "step": 1059
    },
    {
      "epoch": 0.28406806914109606,
      "grad_norm": 0.7001444523810225,
      "learning_rate": 4.977142692380416e-06,
      "loss": 0.629,
      "step": 1060
    },
    {
      "epoch": 0.2843360578855688,
      "grad_norm": 0.7310092756437413,
      "learning_rate": 4.977095121073594e-06,
      "loss": 0.6212,
      "step": 1061
    },
    {
      "epoch": 0.2846040466300415,
      "grad_norm": 0.6883139081279732,
      "learning_rate": 4.977047500542629e-06,
      "loss": 0.5955,
      "step": 1062
    },
    {
      "epoch": 0.2848720353745143,
      "grad_norm": 0.6861671642532922,
      "learning_rate": 4.976999830788465e-06,
      "loss": 0.6033,
      "step": 1063
    },
    {
      "epoch": 0.285140024118987,
      "grad_norm": 0.6997271766951108,
      "learning_rate": 4.976952111812051e-06,
      "loss": 0.621,
      "step": 1064
    },
    {
      "epoch": 0.28540801286345974,
      "grad_norm": 0.7179689607306963,
      "learning_rate": 4.9769043436143334e-06,
      "loss": 0.6259,
      "step": 1065
    },
    {
      "epoch": 0.28567600160793244,
      "grad_norm": 0.6851983879140289,
      "learning_rate": 4.976856526196263e-06,
      "loss": 0.6168,
      "step": 1066
    },
    {
      "epoch": 0.2859439903524052,
      "grad_norm": 0.7044836039877804,
      "learning_rate": 4.97680865955879e-06,
      "loss": 0.5914,
      "step": 1067
    },
    {
      "epoch": 0.28621197909687796,
      "grad_norm": 0.7003098097754848,
      "learning_rate": 4.9767607437028645e-06,
      "loss": 0.5954,
      "step": 1068
    },
    {
      "epoch": 0.28647996784135066,
      "grad_norm": 0.659068386011301,
      "learning_rate": 4.976712778629439e-06,
      "loss": 0.5903,
      "step": 1069
    },
    {
      "epoch": 0.2867479565858234,
      "grad_norm": 0.6856099914175252,
      "learning_rate": 4.976664764339467e-06,
      "loss": 0.6638,
      "step": 1070
    },
    {
      "epoch": 0.2870159453302961,
      "grad_norm": 0.7063766157945366,
      "learning_rate": 4.976616700833903e-06,
      "loss": 0.6287,
      "step": 1071
    },
    {
      "epoch": 0.2872839340747689,
      "grad_norm": 0.6919770478154911,
      "learning_rate": 4.976568588113701e-06,
      "loss": 0.6337,
      "step": 1072
    },
    {
      "epoch": 0.2875519228192416,
      "grad_norm": 0.6760694168665161,
      "learning_rate": 4.976520426179818e-06,
      "loss": 0.6117,
      "step": 1073
    },
    {
      "epoch": 0.28781991156371434,
      "grad_norm": 0.680297729269882,
      "learning_rate": 4.976472215033211e-06,
      "loss": 0.6116,
      "step": 1074
    },
    {
      "epoch": 0.28808790030818704,
      "grad_norm": 0.7342206088529074,
      "learning_rate": 4.976423954674837e-06,
      "loss": 0.6196,
      "step": 1075
    },
    {
      "epoch": 0.2883558890526598,
      "grad_norm": 0.7019218516409087,
      "learning_rate": 4.976375645105656e-06,
      "loss": 0.608,
      "step": 1076
    },
    {
      "epoch": 0.2886238777971325,
      "grad_norm": 0.6792518779488169,
      "learning_rate": 4.976327286326629e-06,
      "loss": 0.6269,
      "step": 1077
    },
    {
      "epoch": 0.28889186654160526,
      "grad_norm": 0.7023266691108202,
      "learning_rate": 4.976278878338715e-06,
      "loss": 0.6088,
      "step": 1078
    },
    {
      "epoch": 0.28915985528607796,
      "grad_norm": 0.7291703900734275,
      "learning_rate": 4.976230421142877e-06,
      "loss": 0.6164,
      "step": 1079
    },
    {
      "epoch": 0.2894278440305507,
      "grad_norm": 0.6726272254839488,
      "learning_rate": 4.9761819147400765e-06,
      "loss": 0.627,
      "step": 1080
    },
    {
      "epoch": 0.2896958327750235,
      "grad_norm": 0.6839118945870573,
      "learning_rate": 4.97613335913128e-06,
      "loss": 0.5887,
      "step": 1081
    },
    {
      "epoch": 0.2899638215194962,
      "grad_norm": 1.3339792049486439,
      "learning_rate": 4.97608475431745e-06,
      "loss": 0.6129,
      "step": 1082
    },
    {
      "epoch": 0.29023181026396894,
      "grad_norm": 0.6654674113300207,
      "learning_rate": 4.976036100299554e-06,
      "loss": 0.5943,
      "step": 1083
    },
    {
      "epoch": 0.29049979900844164,
      "grad_norm": 0.6804399295402203,
      "learning_rate": 4.9759873970785574e-06,
      "loss": 0.6407,
      "step": 1084
    },
    {
      "epoch": 0.2907677877529144,
      "grad_norm": 0.707494669062003,
      "learning_rate": 4.975938644655429e-06,
      "loss": 0.6642,
      "step": 1085
    },
    {
      "epoch": 0.2910357764973871,
      "grad_norm": 0.691459542564111,
      "learning_rate": 4.975889843031138e-06,
      "loss": 0.5838,
      "step": 1086
    },
    {
      "epoch": 0.29130376524185986,
      "grad_norm": 0.7078916493881727,
      "learning_rate": 4.975840992206653e-06,
      "loss": 0.5943,
      "step": 1087
    },
    {
      "epoch": 0.29157175398633256,
      "grad_norm": 0.6700037845590132,
      "learning_rate": 4.975792092182946e-06,
      "loss": 0.6173,
      "step": 1088
    },
    {
      "epoch": 0.2918397427308053,
      "grad_norm": 0.6803252233623542,
      "learning_rate": 4.975743142960987e-06,
      "loss": 0.6507,
      "step": 1089
    },
    {
      "epoch": 0.292107731475278,
      "grad_norm": 0.7196666788170226,
      "learning_rate": 4.97569414454175e-06,
      "loss": 0.6448,
      "step": 1090
    },
    {
      "epoch": 0.2923757202197508,
      "grad_norm": 0.7043165579877908,
      "learning_rate": 4.975645096926208e-06,
      "loss": 0.6112,
      "step": 1091
    },
    {
      "epoch": 0.2926437089642235,
      "grad_norm": 0.7060170414463588,
      "learning_rate": 4.975596000115337e-06,
      "loss": 0.6331,
      "step": 1092
    },
    {
      "epoch": 0.29291169770869624,
      "grad_norm": 0.6705002995101021,
      "learning_rate": 4.975546854110112e-06,
      "loss": 0.6076,
      "step": 1093
    },
    {
      "epoch": 0.29317968645316894,
      "grad_norm": 0.69154357828935,
      "learning_rate": 4.975497658911508e-06,
      "loss": 0.5917,
      "step": 1094
    },
    {
      "epoch": 0.2934476751976417,
      "grad_norm": 0.7031433540516612,
      "learning_rate": 4.975448414520505e-06,
      "loss": 0.672,
      "step": 1095
    },
    {
      "epoch": 0.29371566394211446,
      "grad_norm": 0.6948891946029786,
      "learning_rate": 4.975399120938079e-06,
      "loss": 0.6328,
      "step": 1096
    },
    {
      "epoch": 0.29398365268658716,
      "grad_norm": 0.6889155517052818,
      "learning_rate": 4.9753497781652125e-06,
      "loss": 0.6442,
      "step": 1097
    },
    {
      "epoch": 0.2942516414310599,
      "grad_norm": 0.6821925755359569,
      "learning_rate": 4.975300386202884e-06,
      "loss": 0.5628,
      "step": 1098
    },
    {
      "epoch": 0.2945196301755326,
      "grad_norm": 0.6644042544774185,
      "learning_rate": 4.9752509450520754e-06,
      "loss": 0.6226,
      "step": 1099
    },
    {
      "epoch": 0.2947876189200054,
      "grad_norm": 0.6769787460583688,
      "learning_rate": 4.97520145471377e-06,
      "loss": 0.6297,
      "step": 1100
    },
    {
      "epoch": 0.2950556076644781,
      "grad_norm": 0.7307075272228877,
      "learning_rate": 4.97515191518895e-06,
      "loss": 0.6279,
      "step": 1101
    },
    {
      "epoch": 0.29532359640895084,
      "grad_norm": 0.6939340479616387,
      "learning_rate": 4.9751023264786e-06,
      "loss": 0.6347,
      "step": 1102
    },
    {
      "epoch": 0.29559158515342354,
      "grad_norm": 0.7195981594093589,
      "learning_rate": 4.975052688583707e-06,
      "loss": 0.6396,
      "step": 1103
    },
    {
      "epoch": 0.2958595738978963,
      "grad_norm": 0.7099684772974684,
      "learning_rate": 4.975003001505255e-06,
      "loss": 0.6065,
      "step": 1104
    },
    {
      "epoch": 0.296127562642369,
      "grad_norm": 0.669884644159448,
      "learning_rate": 4.974953265244234e-06,
      "loss": 0.6258,
      "step": 1105
    },
    {
      "epoch": 0.29639555138684176,
      "grad_norm": 0.7294828741045941,
      "learning_rate": 4.97490347980163e-06,
      "loss": 0.6174,
      "step": 1106
    },
    {
      "epoch": 0.29666354013131446,
      "grad_norm": 0.6733761568397347,
      "learning_rate": 4.974853645178433e-06,
      "loss": 0.5816,
      "step": 1107
    },
    {
      "epoch": 0.2969315288757872,
      "grad_norm": 0.6829508576742724,
      "learning_rate": 4.974803761375634e-06,
      "loss": 0.613,
      "step": 1108
    },
    {
      "epoch": 0.2971995176202599,
      "grad_norm": 0.683445863436502,
      "learning_rate": 4.974753828394224e-06,
      "loss": 0.6328,
      "step": 1109
    },
    {
      "epoch": 0.2974675063647327,
      "grad_norm": 0.6803934503170418,
      "learning_rate": 4.974703846235195e-06,
      "loss": 0.6037,
      "step": 1110
    },
    {
      "epoch": 0.29773549510920544,
      "grad_norm": 0.6914957133473794,
      "learning_rate": 4.97465381489954e-06,
      "loss": 0.6312,
      "step": 1111
    },
    {
      "epoch": 0.29800348385367814,
      "grad_norm": 0.687078439467001,
      "learning_rate": 4.9746037343882545e-06,
      "loss": 0.6081,
      "step": 1112
    },
    {
      "epoch": 0.2982714725981509,
      "grad_norm": 0.7012521643068056,
      "learning_rate": 4.974553604702332e-06,
      "loss": 0.6384,
      "step": 1113
    },
    {
      "epoch": 0.2985394613426236,
      "grad_norm": 0.6826302305653821,
      "learning_rate": 4.974503425842769e-06,
      "loss": 0.6084,
      "step": 1114
    },
    {
      "epoch": 0.29880745008709636,
      "grad_norm": 0.6689388383141466,
      "learning_rate": 4.974453197810564e-06,
      "loss": 0.6231,
      "step": 1115
    },
    {
      "epoch": 0.29907543883156906,
      "grad_norm": 0.6892007064956545,
      "learning_rate": 4.974402920606713e-06,
      "loss": 0.6125,
      "step": 1116
    },
    {
      "epoch": 0.2993434275760418,
      "grad_norm": 0.6620917635210568,
      "learning_rate": 4.974352594232218e-06,
      "loss": 0.6259,
      "step": 1117
    },
    {
      "epoch": 0.2996114163205145,
      "grad_norm": 0.6795577226143916,
      "learning_rate": 4.974302218688076e-06,
      "loss": 0.6218,
      "step": 1118
    },
    {
      "epoch": 0.2998794050649873,
      "grad_norm": 0.6621755754384973,
      "learning_rate": 4.97425179397529e-06,
      "loss": 0.5929,
      "step": 1119
    },
    {
      "epoch": 0.30014739380946,
      "grad_norm": 0.696773153733667,
      "learning_rate": 4.97420132009486e-06,
      "loss": 0.5799,
      "step": 1120
    },
    {
      "epoch": 0.30041538255393274,
      "grad_norm": 0.7159861820806841,
      "learning_rate": 4.974150797047792e-06,
      "loss": 0.6512,
      "step": 1121
    },
    {
      "epoch": 0.30068337129840544,
      "grad_norm": 0.7176980362482157,
      "learning_rate": 4.9741002248350876e-06,
      "loss": 0.5925,
      "step": 1122
    },
    {
      "epoch": 0.3009513600428782,
      "grad_norm": 0.69066674230552,
      "learning_rate": 4.9740496034577525e-06,
      "loss": 0.6326,
      "step": 1123
    },
    {
      "epoch": 0.3012193487873509,
      "grad_norm": 0.7035429570172205,
      "learning_rate": 4.973998932916793e-06,
      "loss": 0.6828,
      "step": 1124
    },
    {
      "epoch": 0.30148733753182366,
      "grad_norm": 0.6914146214917417,
      "learning_rate": 4.973948213213215e-06,
      "loss": 0.6251,
      "step": 1125
    },
    {
      "epoch": 0.3017553262762964,
      "grad_norm": 0.6853243684334176,
      "learning_rate": 4.9738974443480284e-06,
      "loss": 0.6004,
      "step": 1126
    },
    {
      "epoch": 0.3020233150207691,
      "grad_norm": 0.7148170010500183,
      "learning_rate": 4.973846626322239e-06,
      "loss": 0.6052,
      "step": 1127
    },
    {
      "epoch": 0.3022913037652419,
      "grad_norm": 0.7274159666824814,
      "learning_rate": 4.97379575913686e-06,
      "loss": 0.6205,
      "step": 1128
    },
    {
      "epoch": 0.3025592925097146,
      "grad_norm": 0.6946492347247563,
      "learning_rate": 4.9737448427929e-06,
      "loss": 0.6095,
      "step": 1129
    },
    {
      "epoch": 0.30282728125418734,
      "grad_norm": 0.6827877167301385,
      "learning_rate": 4.973693877291371e-06,
      "loss": 0.6163,
      "step": 1130
    },
    {
      "epoch": 0.30309526999866004,
      "grad_norm": 0.7166677444882789,
      "learning_rate": 4.973642862633287e-06,
      "loss": 0.6084,
      "step": 1131
    },
    {
      "epoch": 0.3033632587431328,
      "grad_norm": 0.7181696077707992,
      "learning_rate": 4.97359179881966e-06,
      "loss": 0.623,
      "step": 1132
    },
    {
      "epoch": 0.3036312474876055,
      "grad_norm": 0.7373209035709622,
      "learning_rate": 4.973540685851505e-06,
      "loss": 0.6073,
      "step": 1133
    },
    {
      "epoch": 0.30389923623207826,
      "grad_norm": 0.732422098673592,
      "learning_rate": 4.97348952372984e-06,
      "loss": 0.6159,
      "step": 1134
    },
    {
      "epoch": 0.30416722497655097,
      "grad_norm": 0.6965675718091127,
      "learning_rate": 4.973438312455679e-06,
      "loss": 0.6478,
      "step": 1135
    },
    {
      "epoch": 0.3044352137210237,
      "grad_norm": 0.723787661046137,
      "learning_rate": 4.973387052030041e-06,
      "loss": 0.6214,
      "step": 1136
    },
    {
      "epoch": 0.3047032024654964,
      "grad_norm": 0.6944410009702949,
      "learning_rate": 4.973335742453944e-06,
      "loss": 0.6222,
      "step": 1137
    },
    {
      "epoch": 0.3049711912099692,
      "grad_norm": 0.6837322054146955,
      "learning_rate": 4.973284383728409e-06,
      "loss": 0.6338,
      "step": 1138
    },
    {
      "epoch": 0.3052391799544419,
      "grad_norm": 0.7093266737890469,
      "learning_rate": 4.973232975854454e-06,
      "loss": 0.5995,
      "step": 1139
    },
    {
      "epoch": 0.30550716869891464,
      "grad_norm": 0.7180327177557932,
      "learning_rate": 4.973181518833102e-06,
      "loss": 0.6115,
      "step": 1140
    },
    {
      "epoch": 0.3057751574433874,
      "grad_norm": 0.6550879376879303,
      "learning_rate": 4.973130012665377e-06,
      "loss": 0.6016,
      "step": 1141
    },
    {
      "epoch": 0.3060431461878601,
      "grad_norm": 0.710643508290811,
      "learning_rate": 4.9730784573523005e-06,
      "loss": 0.6175,
      "step": 1142
    },
    {
      "epoch": 0.30631113493233286,
      "grad_norm": 0.6920131165681692,
      "learning_rate": 4.973026852894897e-06,
      "loss": 0.6327,
      "step": 1143
    },
    {
      "epoch": 0.30657912367680556,
      "grad_norm": 0.6813466629231403,
      "learning_rate": 4.972975199294193e-06,
      "loss": 0.6139,
      "step": 1144
    },
    {
      "epoch": 0.3068471124212783,
      "grad_norm": 0.6925807995966857,
      "learning_rate": 4.972923496551215e-06,
      "loss": 0.6226,
      "step": 1145
    },
    {
      "epoch": 0.307115101165751,
      "grad_norm": 0.7419235153269528,
      "learning_rate": 4.97287174466699e-06,
      "loss": 0.6362,
      "step": 1146
    },
    {
      "epoch": 0.3073830899102238,
      "grad_norm": 0.7159204209576077,
      "learning_rate": 4.972819943642545e-06,
      "loss": 0.6265,
      "step": 1147
    },
    {
      "epoch": 0.3076510786546965,
      "grad_norm": 0.6833648820192563,
      "learning_rate": 4.972768093478913e-06,
      "loss": 0.6407,
      "step": 1148
    },
    {
      "epoch": 0.30791906739916924,
      "grad_norm": 0.7347946159264774,
      "learning_rate": 4.97271619417712e-06,
      "loss": 0.6126,
      "step": 1149
    },
    {
      "epoch": 0.30818705614364195,
      "grad_norm": 0.7082402279116404,
      "learning_rate": 4.9726642457382e-06,
      "loss": 0.6643,
      "step": 1150
    },
    {
      "epoch": 0.3084550448881147,
      "grad_norm": 0.6895620108301537,
      "learning_rate": 4.972612248163184e-06,
      "loss": 0.586,
      "step": 1151
    },
    {
      "epoch": 0.3087230336325874,
      "grad_norm": 0.6892881300775912,
      "learning_rate": 4.972560201453107e-06,
      "loss": 0.6171,
      "step": 1152
    },
    {
      "epoch": 0.30899102237706016,
      "grad_norm": 0.7531463604803571,
      "learning_rate": 4.972508105609003e-06,
      "loss": 0.5915,
      "step": 1153
    },
    {
      "epoch": 0.3092590111215329,
      "grad_norm": 0.6747363052159415,
      "learning_rate": 4.972455960631904e-06,
      "loss": 0.5904,
      "step": 1154
    },
    {
      "epoch": 0.3095269998660056,
      "grad_norm": 0.6831530821544196,
      "learning_rate": 4.972403766522851e-06,
      "loss": 0.5968,
      "step": 1155
    },
    {
      "epoch": 0.3097949886104784,
      "grad_norm": 0.6712555604708271,
      "learning_rate": 4.972351523282878e-06,
      "loss": 0.6286,
      "step": 1156
    },
    {
      "epoch": 0.3100629773549511,
      "grad_norm": 0.777118116400906,
      "learning_rate": 4.972299230913025e-06,
      "loss": 0.5851,
      "step": 1157
    },
    {
      "epoch": 0.31033096609942384,
      "grad_norm": 0.7109527642403315,
      "learning_rate": 4.9722468894143285e-06,
      "loss": 0.6188,
      "step": 1158
    },
    {
      "epoch": 0.31059895484389655,
      "grad_norm": 0.695001806435032,
      "learning_rate": 4.972194498787831e-06,
      "loss": 0.6551,
      "step": 1159
    },
    {
      "epoch": 0.3108669435883693,
      "grad_norm": 0.7132675003556475,
      "learning_rate": 4.972142059034573e-06,
      "loss": 0.6536,
      "step": 1160
    },
    {
      "epoch": 0.311134932332842,
      "grad_norm": 0.6813391011053583,
      "learning_rate": 4.972089570155597e-06,
      "loss": 0.6041,
      "step": 1161
    },
    {
      "epoch": 0.31140292107731476,
      "grad_norm": 0.6982871843020496,
      "learning_rate": 4.972037032151945e-06,
      "loss": 0.6358,
      "step": 1162
    },
    {
      "epoch": 0.31167090982178747,
      "grad_norm": 0.6666143736925849,
      "learning_rate": 4.971984445024662e-06,
      "loss": 0.5936,
      "step": 1163
    },
    {
      "epoch": 0.3119388985662602,
      "grad_norm": 0.6965008119951531,
      "learning_rate": 4.971931808774791e-06,
      "loss": 0.6371,
      "step": 1164
    },
    {
      "epoch": 0.31220688731073293,
      "grad_norm": 0.6976720786678536,
      "learning_rate": 4.97187912340338e-06,
      "loss": 0.6136,
      "step": 1165
    },
    {
      "epoch": 0.3124748760552057,
      "grad_norm": 0.703304711775172,
      "learning_rate": 4.971826388911476e-06,
      "loss": 0.6061,
      "step": 1166
    },
    {
      "epoch": 0.3127428647996784,
      "grad_norm": 0.6840893061117895,
      "learning_rate": 4.971773605300126e-06,
      "loss": 0.6511,
      "step": 1167
    },
    {
      "epoch": 0.31301085354415115,
      "grad_norm": 0.6725394165522178,
      "learning_rate": 4.97172077257038e-06,
      "loss": 0.6074,
      "step": 1168
    },
    {
      "epoch": 0.3132788422886239,
      "grad_norm": 0.6972153472968612,
      "learning_rate": 4.971667890723286e-06,
      "loss": 0.6317,
      "step": 1169
    },
    {
      "epoch": 0.3135468310330966,
      "grad_norm": 0.7456648071917962,
      "learning_rate": 4.9716149597598975e-06,
      "loss": 0.6462,
      "step": 1170
    },
    {
      "epoch": 0.31381481977756936,
      "grad_norm": 0.6875430183655197,
      "learning_rate": 4.971561979681264e-06,
      "loss": 0.6138,
      "step": 1171
    },
    {
      "epoch": 0.31408280852204207,
      "grad_norm": 0.6865031960008801,
      "learning_rate": 4.971508950488439e-06,
      "loss": 0.6048,
      "step": 1172
    },
    {
      "epoch": 0.3143507972665148,
      "grad_norm": 0.7154092472786386,
      "learning_rate": 4.971455872182476e-06,
      "loss": 0.6093,
      "step": 1173
    },
    {
      "epoch": 0.31461878601098753,
      "grad_norm": 0.6753745792734976,
      "learning_rate": 4.9714027447644305e-06,
      "loss": 0.6182,
      "step": 1174
    },
    {
      "epoch": 0.3148867747554603,
      "grad_norm": 0.6845990927523734,
      "learning_rate": 4.971349568235359e-06,
      "loss": 0.61,
      "step": 1175
    },
    {
      "epoch": 0.315154763499933,
      "grad_norm": 0.6748051780127633,
      "learning_rate": 4.971296342596316e-06,
      "loss": 0.6045,
      "step": 1176
    },
    {
      "epoch": 0.31542275224440575,
      "grad_norm": 0.7357574462956152,
      "learning_rate": 4.971243067848361e-06,
      "loss": 0.6458,
      "step": 1177
    },
    {
      "epoch": 0.31569074098887845,
      "grad_norm": 0.6851305007545211,
      "learning_rate": 4.97118974399255e-06,
      "loss": 0.6011,
      "step": 1178
    },
    {
      "epoch": 0.3159587297333512,
      "grad_norm": 0.7174490350271417,
      "learning_rate": 4.9711363710299465e-06,
      "loss": 0.6327,
      "step": 1179
    },
    {
      "epoch": 0.3162267184778239,
      "grad_norm": 0.707465038648979,
      "learning_rate": 4.9710829489616095e-06,
      "loss": 0.6293,
      "step": 1180
    },
    {
      "epoch": 0.31649470722229667,
      "grad_norm": 0.7078520138508283,
      "learning_rate": 4.971029477788599e-06,
      "loss": 0.5884,
      "step": 1181
    },
    {
      "epoch": 0.31676269596676937,
      "grad_norm": 0.6861079977659145,
      "learning_rate": 4.97097595751198e-06,
      "loss": 0.5715,
      "step": 1182
    },
    {
      "epoch": 0.31703068471124213,
      "grad_norm": 0.7574257373133557,
      "learning_rate": 4.9709223881328135e-06,
      "loss": 0.6105,
      "step": 1183
    },
    {
      "epoch": 0.3172986734557149,
      "grad_norm": 0.6695719686930394,
      "learning_rate": 4.970868769652166e-06,
      "loss": 0.628,
      "step": 1184
    },
    {
      "epoch": 0.3175666622001876,
      "grad_norm": 0.736033726512218,
      "learning_rate": 4.970815102071103e-06,
      "loss": 0.6108,
      "step": 1185
    },
    {
      "epoch": 0.31783465094466035,
      "grad_norm": 0.6892561975321416,
      "learning_rate": 4.97076138539069e-06,
      "loss": 0.6559,
      "step": 1186
    },
    {
      "epoch": 0.31810263968913305,
      "grad_norm": 0.7137091331671485,
      "learning_rate": 4.970707619611995e-06,
      "loss": 0.6249,
      "step": 1187
    },
    {
      "epoch": 0.3183706284336058,
      "grad_norm": 0.6931476176478949,
      "learning_rate": 4.9706538047360854e-06,
      "loss": 0.6318,
      "step": 1188
    },
    {
      "epoch": 0.3186386171780785,
      "grad_norm": 0.6747391468864797,
      "learning_rate": 4.970599940764033e-06,
      "loss": 0.5969,
      "step": 1189
    },
    {
      "epoch": 0.31890660592255127,
      "grad_norm": 0.6870573636106231,
      "learning_rate": 4.970546027696905e-06,
      "loss": 0.6066,
      "step": 1190
    },
    {
      "epoch": 0.31917459466702397,
      "grad_norm": 0.6995732500785365,
      "learning_rate": 4.970492065535775e-06,
      "loss": 0.6437,
      "step": 1191
    },
    {
      "epoch": 0.3194425834114967,
      "grad_norm": 0.6975112762470117,
      "learning_rate": 4.970438054281715e-06,
      "loss": 0.6626,
      "step": 1192
    },
    {
      "epoch": 0.31971057215596943,
      "grad_norm": 0.7154527308922997,
      "learning_rate": 4.9703839939357975e-06,
      "loss": 0.6696,
      "step": 1193
    },
    {
      "epoch": 0.3199785609004422,
      "grad_norm": 0.7024282659876223,
      "learning_rate": 4.970329884499097e-06,
      "loss": 0.6277,
      "step": 1194
    },
    {
      "epoch": 0.3202465496449149,
      "grad_norm": 0.6845193159839265,
      "learning_rate": 4.97027572597269e-06,
      "loss": 0.6304,
      "step": 1195
    },
    {
      "epoch": 0.32051453838938765,
      "grad_norm": 0.7085687732337658,
      "learning_rate": 4.970221518357651e-06,
      "loss": 0.6224,
      "step": 1196
    },
    {
      "epoch": 0.32078252713386035,
      "grad_norm": 0.7381650343435271,
      "learning_rate": 4.970167261655058e-06,
      "loss": 0.6003,
      "step": 1197
    },
    {
      "epoch": 0.3210505158783331,
      "grad_norm": 0.7037930261806032,
      "learning_rate": 4.9701129558659884e-06,
      "loss": 0.6349,
      "step": 1198
    },
    {
      "epoch": 0.32131850462280587,
      "grad_norm": 0.7452332243694539,
      "learning_rate": 4.970058600991523e-06,
      "loss": 0.6131,
      "step": 1199
    },
    {
      "epoch": 0.32158649336727857,
      "grad_norm": 0.7220573640580507,
      "learning_rate": 4.970004197032741e-06,
      "loss": 0.6439,
      "step": 1200
    },
    {
      "epoch": 0.3218544821117513,
      "grad_norm": 0.7061763867053845,
      "learning_rate": 4.969949743990723e-06,
      "loss": 0.6381,
      "step": 1201
    },
    {
      "epoch": 0.32212247085622403,
      "grad_norm": 0.6956092822288465,
      "learning_rate": 4.969895241866552e-06,
      "loss": 0.653,
      "step": 1202
    },
    {
      "epoch": 0.3223904596006968,
      "grad_norm": 0.6746221435072538,
      "learning_rate": 4.969840690661309e-06,
      "loss": 0.6114,
      "step": 1203
    },
    {
      "epoch": 0.3226584483451695,
      "grad_norm": 0.6881433391587184,
      "learning_rate": 4.969786090376082e-06,
      "loss": 0.6283,
      "step": 1204
    },
    {
      "epoch": 0.32292643708964225,
      "grad_norm": 0.6958014805272574,
      "learning_rate": 4.9697314410119525e-06,
      "loss": 0.6369,
      "step": 1205
    },
    {
      "epoch": 0.32319442583411495,
      "grad_norm": 0.7282054465308112,
      "learning_rate": 4.969676742570007e-06,
      "loss": 0.6047,
      "step": 1206
    },
    {
      "epoch": 0.3234624145785877,
      "grad_norm": 0.6959750838614075,
      "learning_rate": 4.969621995051334e-06,
      "loss": 0.613,
      "step": 1207
    },
    {
      "epoch": 0.3237304033230604,
      "grad_norm": 0.7216540422481289,
      "learning_rate": 4.96956719845702e-06,
      "loss": 0.624,
      "step": 1208
    },
    {
      "epoch": 0.32399839206753317,
      "grad_norm": 0.6679127786446496,
      "learning_rate": 4.9695123527881535e-06,
      "loss": 0.6413,
      "step": 1209
    },
    {
      "epoch": 0.32426638081200587,
      "grad_norm": 0.7022408454043808,
      "learning_rate": 4.969457458045826e-06,
      "loss": 0.6558,
      "step": 1210
    },
    {
      "epoch": 0.32453436955647863,
      "grad_norm": 0.6847912899848301,
      "learning_rate": 4.969402514231128e-06,
      "loss": 0.6578,
      "step": 1211
    },
    {
      "epoch": 0.32480235830095133,
      "grad_norm": 0.6844076316554042,
      "learning_rate": 4.969347521345151e-06,
      "loss": 0.6201,
      "step": 1212
    },
    {
      "epoch": 0.3250703470454241,
      "grad_norm": 0.7162259393079304,
      "learning_rate": 4.9692924793889875e-06,
      "loss": 0.6321,
      "step": 1213
    },
    {
      "epoch": 0.32533833578989685,
      "grad_norm": 0.6855846001299186,
      "learning_rate": 4.9692373883637315e-06,
      "loss": 0.6163,
      "step": 1214
    },
    {
      "epoch": 0.32560632453436955,
      "grad_norm": 0.6829031105225685,
      "learning_rate": 4.969182248270477e-06,
      "loss": 0.6235,
      "step": 1215
    },
    {
      "epoch": 0.3258743132788423,
      "grad_norm": 0.6887712847715389,
      "learning_rate": 4.969127059110321e-06,
      "loss": 0.6228,
      "step": 1216
    },
    {
      "epoch": 0.326142302023315,
      "grad_norm": 0.7140709418940653,
      "learning_rate": 4.96907182088436e-06,
      "loss": 0.6091,
      "step": 1217
    },
    {
      "epoch": 0.32641029076778777,
      "grad_norm": 0.7068971739020731,
      "learning_rate": 4.969016533593691e-06,
      "loss": 0.6324,
      "step": 1218
    },
    {
      "epoch": 0.32667827951226047,
      "grad_norm": 0.710850470726184,
      "learning_rate": 4.968961197239414e-06,
      "loss": 0.6524,
      "step": 1219
    },
    {
      "epoch": 0.32694626825673323,
      "grad_norm": 0.6970573650436894,
      "learning_rate": 4.9689058118226265e-06,
      "loss": 0.6109,
      "step": 1220
    },
    {
      "epoch": 0.32721425700120593,
      "grad_norm": 0.6651795424029135,
      "learning_rate": 4.9688503773444306e-06,
      "loss": 0.6107,
      "step": 1221
    },
    {
      "epoch": 0.3274822457456787,
      "grad_norm": 0.6798292698189714,
      "learning_rate": 4.968794893805927e-06,
      "loss": 0.6398,
      "step": 1222
    },
    {
      "epoch": 0.3277502344901514,
      "grad_norm": 0.6982371283417861,
      "learning_rate": 4.96873936120822e-06,
      "loss": 0.6239,
      "step": 1223
    },
    {
      "epoch": 0.32801822323462415,
      "grad_norm": 0.714518091041523,
      "learning_rate": 4.968683779552411e-06,
      "loss": 0.6524,
      "step": 1224
    },
    {
      "epoch": 0.32828621197909685,
      "grad_norm": 0.6898007630372274,
      "learning_rate": 4.9686281488396055e-06,
      "loss": 0.6318,
      "step": 1225
    },
    {
      "epoch": 0.3285542007235696,
      "grad_norm": 0.6843774484675859,
      "learning_rate": 4.9685724690709095e-06,
      "loss": 0.565,
      "step": 1226
    },
    {
      "epoch": 0.32882218946804237,
      "grad_norm": 0.705704185179741,
      "learning_rate": 4.968516740247428e-06,
      "loss": 0.6204,
      "step": 1227
    },
    {
      "epoch": 0.32909017821251507,
      "grad_norm": 0.7156548770395114,
      "learning_rate": 4.96846096237027e-06,
      "loss": 0.6081,
      "step": 1228
    },
    {
      "epoch": 0.32935816695698783,
      "grad_norm": 0.7086087421682999,
      "learning_rate": 4.968405135440543e-06,
      "loss": 0.6118,
      "step": 1229
    },
    {
      "epoch": 0.32962615570146053,
      "grad_norm": 0.6900596396242317,
      "learning_rate": 4.9683492594593566e-06,
      "loss": 0.6233,
      "step": 1230
    },
    {
      "epoch": 0.3298941444459333,
      "grad_norm": 0.6854634680894541,
      "learning_rate": 4.96829333442782e-06,
      "loss": 0.6676,
      "step": 1231
    },
    {
      "epoch": 0.330162133190406,
      "grad_norm": 0.6894698566438066,
      "learning_rate": 4.9682373603470465e-06,
      "loss": 0.6133,
      "step": 1232
    },
    {
      "epoch": 0.33043012193487875,
      "grad_norm": 0.6996042021515676,
      "learning_rate": 4.968181337218148e-06,
      "loss": 0.6249,
      "step": 1233
    },
    {
      "epoch": 0.33069811067935145,
      "grad_norm": 0.6704786932551247,
      "learning_rate": 4.968125265042238e-06,
      "loss": 0.6142,
      "step": 1234
    },
    {
      "epoch": 0.3309660994238242,
      "grad_norm": 0.6833767895599795,
      "learning_rate": 4.968069143820428e-06,
      "loss": 0.6209,
      "step": 1235
    },
    {
      "epoch": 0.3312340881682969,
      "grad_norm": 0.6920798824565456,
      "learning_rate": 4.968012973553836e-06,
      "loss": 0.6796,
      "step": 1236
    },
    {
      "epoch": 0.33150207691276967,
      "grad_norm": 0.6791724491495776,
      "learning_rate": 4.967956754243578e-06,
      "loss": 0.6106,
      "step": 1237
    },
    {
      "epoch": 0.3317700656572424,
      "grad_norm": 0.7163595735289434,
      "learning_rate": 4.967900485890771e-06,
      "loss": 0.6239,
      "step": 1238
    },
    {
      "epoch": 0.33203805440171513,
      "grad_norm": 0.69447718575849,
      "learning_rate": 4.967844168496532e-06,
      "loss": 0.6211,
      "step": 1239
    },
    {
      "epoch": 0.33230604314618784,
      "grad_norm": 0.6832055972344524,
      "learning_rate": 4.967787802061981e-06,
      "loss": 0.6041,
      "step": 1240
    },
    {
      "epoch": 0.3325740318906606,
      "grad_norm": 0.6835183772242343,
      "learning_rate": 4.967731386588238e-06,
      "loss": 0.604,
      "step": 1241
    },
    {
      "epoch": 0.33284202063513335,
      "grad_norm": 0.6600870751927108,
      "learning_rate": 4.967674922076425e-06,
      "loss": 0.6016,
      "step": 1242
    },
    {
      "epoch": 0.33311000937960605,
      "grad_norm": 0.6729503605141429,
      "learning_rate": 4.967618408527662e-06,
      "loss": 0.6187,
      "step": 1243
    },
    {
      "epoch": 0.3333779981240788,
      "grad_norm": 0.7570530094113856,
      "learning_rate": 4.967561845943074e-06,
      "loss": 0.6306,
      "step": 1244
    },
    {
      "epoch": 0.3336459868685515,
      "grad_norm": 0.7224773995903083,
      "learning_rate": 4.967505234323784e-06,
      "loss": 0.6334,
      "step": 1245
    },
    {
      "epoch": 0.33391397561302427,
      "grad_norm": 0.6706430121545742,
      "learning_rate": 4.967448573670917e-06,
      "loss": 0.5887,
      "step": 1246
    },
    {
      "epoch": 0.334181964357497,
      "grad_norm": 0.7228814891036968,
      "learning_rate": 4.967391863985599e-06,
      "loss": 0.5927,
      "step": 1247
    },
    {
      "epoch": 0.33444995310196973,
      "grad_norm": 0.6844609067111719,
      "learning_rate": 4.967335105268957e-06,
      "loss": 0.6102,
      "step": 1248
    },
    {
      "epoch": 0.33471794184644243,
      "grad_norm": 0.7240591684235861,
      "learning_rate": 4.967278297522119e-06,
      "loss": 0.6358,
      "step": 1249
    },
    {
      "epoch": 0.3349859305909152,
      "grad_norm": 0.7008121188331397,
      "learning_rate": 4.9672214407462135e-06,
      "loss": 0.6113,
      "step": 1250
    },
    {
      "epoch": 0.3352539193353879,
      "grad_norm": 0.6902061294396283,
      "learning_rate": 4.9671645349423715e-06,
      "loss": 0.6253,
      "step": 1251
    },
    {
      "epoch": 0.33552190807986065,
      "grad_norm": 0.6963125458479713,
      "learning_rate": 4.967107580111722e-06,
      "loss": 0.6755,
      "step": 1252
    },
    {
      "epoch": 0.33578989682433336,
      "grad_norm": 0.7126899164210673,
      "learning_rate": 4.967050576255398e-06,
      "loss": 0.601,
      "step": 1253
    },
    {
      "epoch": 0.3360578855688061,
      "grad_norm": 0.6949321598211755,
      "learning_rate": 4.966993523374533e-06,
      "loss": 0.5783,
      "step": 1254
    },
    {
      "epoch": 0.3363258743132788,
      "grad_norm": 0.6872734367385417,
      "learning_rate": 4.966936421470258e-06,
      "loss": 0.6145,
      "step": 1255
    },
    {
      "epoch": 0.3365938630577516,
      "grad_norm": 0.752484009129018,
      "learning_rate": 4.96687927054371e-06,
      "loss": 0.5579,
      "step": 1256
    },
    {
      "epoch": 0.33686185180222433,
      "grad_norm": 0.7312552483783287,
      "learning_rate": 4.966822070596024e-06,
      "loss": 0.6249,
      "step": 1257
    },
    {
      "epoch": 0.33712984054669703,
      "grad_norm": 0.7132342311212718,
      "learning_rate": 4.966764821628337e-06,
      "loss": 0.5905,
      "step": 1258
    },
    {
      "epoch": 0.3373978292911698,
      "grad_norm": 0.7202767420943886,
      "learning_rate": 4.966707523641787e-06,
      "loss": 0.6413,
      "step": 1259
    },
    {
      "epoch": 0.3376658180356425,
      "grad_norm": 0.7332147154522397,
      "learning_rate": 4.966650176637512e-06,
      "loss": 0.637,
      "step": 1260
    },
    {
      "epoch": 0.33793380678011525,
      "grad_norm": 0.6762721255746075,
      "learning_rate": 4.966592780616651e-06,
      "loss": 0.571,
      "step": 1261
    },
    {
      "epoch": 0.33820179552458796,
      "grad_norm": 0.7130619138399956,
      "learning_rate": 4.966535335580344e-06,
      "loss": 0.5927,
      "step": 1262
    },
    {
      "epoch": 0.3384697842690607,
      "grad_norm": 0.7170648292134464,
      "learning_rate": 4.966477841529736e-06,
      "loss": 0.6584,
      "step": 1263
    },
    {
      "epoch": 0.3387377730135334,
      "grad_norm": 0.6974847367666314,
      "learning_rate": 4.966420298465966e-06,
      "loss": 0.6302,
      "step": 1264
    },
    {
      "epoch": 0.3390057617580062,
      "grad_norm": 0.7429356114421414,
      "learning_rate": 4.9663627063901786e-06,
      "loss": 0.6012,
      "step": 1265
    },
    {
      "epoch": 0.3392737505024789,
      "grad_norm": 0.6996460256956285,
      "learning_rate": 4.966305065303519e-06,
      "loss": 0.6462,
      "step": 1266
    },
    {
      "epoch": 0.33954173924695163,
      "grad_norm": 0.706179261237641,
      "learning_rate": 4.966247375207131e-06,
      "loss": 0.6021,
      "step": 1267
    },
    {
      "epoch": 0.33980972799142434,
      "grad_norm": 0.7166054901907831,
      "learning_rate": 4.966189636102163e-06,
      "loss": 0.6153,
      "step": 1268
    },
    {
      "epoch": 0.3400777167358971,
      "grad_norm": 0.7005742499497354,
      "learning_rate": 4.966131847989761e-06,
      "loss": 0.6403,
      "step": 1269
    },
    {
      "epoch": 0.3403457054803698,
      "grad_norm": 0.6843031469103986,
      "learning_rate": 4.966074010871074e-06,
      "loss": 0.5819,
      "step": 1270
    },
    {
      "epoch": 0.34061369422484256,
      "grad_norm": 0.688997572527427,
      "learning_rate": 4.966016124747251e-06,
      "loss": 0.6356,
      "step": 1271
    },
    {
      "epoch": 0.3408816829693153,
      "grad_norm": 0.7175033131034919,
      "learning_rate": 4.965958189619441e-06,
      "loss": 0.6025,
      "step": 1272
    },
    {
      "epoch": 0.341149671713788,
      "grad_norm": 0.7086563719729244,
      "learning_rate": 4.965900205488798e-06,
      "loss": 0.6171,
      "step": 1273
    },
    {
      "epoch": 0.3414176604582608,
      "grad_norm": 0.7102221408281508,
      "learning_rate": 4.965842172356472e-06,
      "loss": 0.5854,
      "step": 1274
    },
    {
      "epoch": 0.3416856492027335,
      "grad_norm": 0.729357984712417,
      "learning_rate": 4.965784090223618e-06,
      "loss": 0.6298,
      "step": 1275
    },
    {
      "epoch": 0.34195363794720623,
      "grad_norm": 0.6808313092380552,
      "learning_rate": 4.9657259590913885e-06,
      "loss": 0.6276,
      "step": 1276
    },
    {
      "epoch": 0.34222162669167894,
      "grad_norm": 0.7053232824153761,
      "learning_rate": 4.96566777896094e-06,
      "loss": 0.6195,
      "step": 1277
    },
    {
      "epoch": 0.3424896154361517,
      "grad_norm": 0.7270831056831968,
      "learning_rate": 4.965609549833427e-06,
      "loss": 0.6152,
      "step": 1278
    },
    {
      "epoch": 0.3427576041806244,
      "grad_norm": 0.7266124619716366,
      "learning_rate": 4.965551271710009e-06,
      "loss": 0.5923,
      "step": 1279
    },
    {
      "epoch": 0.34302559292509716,
      "grad_norm": 0.6996272770958746,
      "learning_rate": 4.965492944591841e-06,
      "loss": 0.6382,
      "step": 1280
    },
    {
      "epoch": 0.34329358166956986,
      "grad_norm": 0.7074625563302107,
      "learning_rate": 4.9654345684800855e-06,
      "loss": 0.5648,
      "step": 1281
    },
    {
      "epoch": 0.3435615704140426,
      "grad_norm": 0.7174333232298162,
      "learning_rate": 4.9653761433759e-06,
      "loss": 0.6019,
      "step": 1282
    },
    {
      "epoch": 0.3438295591585153,
      "grad_norm": 0.7204362586912983,
      "learning_rate": 4.965317669280446e-06,
      "loss": 0.6128,
      "step": 1283
    },
    {
      "epoch": 0.3440975479029881,
      "grad_norm": 0.7045545629238085,
      "learning_rate": 4.9652591461948866e-06,
      "loss": 0.6599,
      "step": 1284
    },
    {
      "epoch": 0.3443655366474608,
      "grad_norm": 0.695812691901682,
      "learning_rate": 4.965200574120384e-06,
      "loss": 0.6474,
      "step": 1285
    },
    {
      "epoch": 0.34463352539193354,
      "grad_norm": 0.7031943845124915,
      "learning_rate": 4.9651419530581015e-06,
      "loss": 0.6303,
      "step": 1286
    },
    {
      "epoch": 0.3449015141364063,
      "grad_norm": 0.6875037942420337,
      "learning_rate": 4.965083283009205e-06,
      "loss": 0.6026,
      "step": 1287
    },
    {
      "epoch": 0.345169502880879,
      "grad_norm": 0.6938212201040538,
      "learning_rate": 4.96502456397486e-06,
      "loss": 0.6183,
      "step": 1288
    },
    {
      "epoch": 0.34543749162535176,
      "grad_norm": 0.7177937388625115,
      "learning_rate": 4.964965795956233e-06,
      "loss": 0.6099,
      "step": 1289
    },
    {
      "epoch": 0.34570548036982446,
      "grad_norm": 0.7180308921541279,
      "learning_rate": 4.964906978954492e-06,
      "loss": 0.6336,
      "step": 1290
    },
    {
      "epoch": 0.3459734691142972,
      "grad_norm": 0.6768528737471579,
      "learning_rate": 4.964848112970807e-06,
      "loss": 0.6214,
      "step": 1291
    },
    {
      "epoch": 0.3462414578587699,
      "grad_norm": 0.7017281884913774,
      "learning_rate": 4.964789198006345e-06,
      "loss": 0.628,
      "step": 1292
    },
    {
      "epoch": 0.3465094466032427,
      "grad_norm": 0.7242815601757743,
      "learning_rate": 4.96473023406228e-06,
      "loss": 0.6173,
      "step": 1293
    },
    {
      "epoch": 0.3467774353477154,
      "grad_norm": 0.7148410367549236,
      "learning_rate": 4.964671221139782e-06,
      "loss": 0.6431,
      "step": 1294
    },
    {
      "epoch": 0.34704542409218814,
      "grad_norm": 0.7382180365071849,
      "learning_rate": 4.964612159240023e-06,
      "loss": 0.6263,
      "step": 1295
    },
    {
      "epoch": 0.34731341283666084,
      "grad_norm": 0.7097479211584314,
      "learning_rate": 4.964553048364179e-06,
      "loss": 0.5789,
      "step": 1296
    },
    {
      "epoch": 0.3475814015811336,
      "grad_norm": 0.677223718147267,
      "learning_rate": 4.964493888513422e-06,
      "loss": 0.6298,
      "step": 1297
    },
    {
      "epoch": 0.3478493903256063,
      "grad_norm": 0.7095264680241599,
      "learning_rate": 4.9644346796889296e-06,
      "loss": 0.637,
      "step": 1298
    },
    {
      "epoch": 0.34811737907007906,
      "grad_norm": 0.6716476134861098,
      "learning_rate": 4.9643754218918765e-06,
      "loss": 0.6166,
      "step": 1299
    },
    {
      "epoch": 0.3483853678145518,
      "grad_norm": 0.684419440379383,
      "learning_rate": 4.964316115123442e-06,
      "loss": 0.5986,
      "step": 1300
    },
    {
      "epoch": 0.3486533565590245,
      "grad_norm": 0.6942408758635816,
      "learning_rate": 4.964256759384803e-06,
      "loss": 0.6014,
      "step": 1301
    },
    {
      "epoch": 0.3489213453034973,
      "grad_norm": 0.6811886541656065,
      "learning_rate": 4.964197354677141e-06,
      "loss": 0.6301,
      "step": 1302
    },
    {
      "epoch": 0.34918933404797,
      "grad_norm": 0.7145966397147238,
      "learning_rate": 4.964137901001635e-06,
      "loss": 0.6375,
      "step": 1303
    },
    {
      "epoch": 0.34945732279244274,
      "grad_norm": 0.6806064003494597,
      "learning_rate": 4.9640783983594675e-06,
      "loss": 0.6262,
      "step": 1304
    },
    {
      "epoch": 0.34972531153691544,
      "grad_norm": 0.6992709708273775,
      "learning_rate": 4.964018846751819e-06,
      "loss": 0.6234,
      "step": 1305
    },
    {
      "epoch": 0.3499933002813882,
      "grad_norm": 0.6944387939912353,
      "learning_rate": 4.963959246179875e-06,
      "loss": 0.6024,
      "step": 1306
    },
    {
      "epoch": 0.3502612890258609,
      "grad_norm": 0.7272373460071421,
      "learning_rate": 4.963899596644819e-06,
      "loss": 0.6239,
      "step": 1307
    },
    {
      "epoch": 0.35052927777033366,
      "grad_norm": 0.6607306544134169,
      "learning_rate": 4.963839898147836e-06,
      "loss": 0.6059,
      "step": 1308
    },
    {
      "epoch": 0.35079726651480636,
      "grad_norm": 0.704124437176827,
      "learning_rate": 4.963780150690113e-06,
      "loss": 0.6204,
      "step": 1309
    },
    {
      "epoch": 0.3510652552592791,
      "grad_norm": 0.6752363769292302,
      "learning_rate": 4.963720354272837e-06,
      "loss": 0.6045,
      "step": 1310
    },
    {
      "epoch": 0.3513332440037518,
      "grad_norm": 0.706280493986095,
      "learning_rate": 4.963660508897197e-06,
      "loss": 0.6096,
      "step": 1311
    },
    {
      "epoch": 0.3516012327482246,
      "grad_norm": 0.6920132790581562,
      "learning_rate": 4.96360061456438e-06,
      "loss": 0.5919,
      "step": 1312
    },
    {
      "epoch": 0.3518692214926973,
      "grad_norm": 0.696974673237392,
      "learning_rate": 4.963540671275579e-06,
      "loss": 0.638,
      "step": 1313
    },
    {
      "epoch": 0.35213721023717004,
      "grad_norm": 0.6928514997014386,
      "learning_rate": 4.963480679031982e-06,
      "loss": 0.6143,
      "step": 1314
    },
    {
      "epoch": 0.3524051989816428,
      "grad_norm": 0.6891518135981981,
      "learning_rate": 4.963420637834784e-06,
      "loss": 0.6032,
      "step": 1315
    },
    {
      "epoch": 0.3526731877261155,
      "grad_norm": 0.7223161309213283,
      "learning_rate": 4.9633605476851776e-06,
      "loss": 0.6371,
      "step": 1316
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.6919820397998722,
      "learning_rate": 4.963300408584356e-06,
      "loss": 0.6276,
      "step": 1317
    },
    {
      "epoch": 0.35320916521506096,
      "grad_norm": 0.7206197311673072,
      "learning_rate": 4.963240220533515e-06,
      "loss": 0.6107,
      "step": 1318
    },
    {
      "epoch": 0.3534771539595337,
      "grad_norm": 0.6702859776487919,
      "learning_rate": 4.963179983533849e-06,
      "loss": 0.6146,
      "step": 1319
    },
    {
      "epoch": 0.3537451427040064,
      "grad_norm": 0.7161891311984299,
      "learning_rate": 4.963119697586558e-06,
      "loss": 0.6509,
      "step": 1320
    },
    {
      "epoch": 0.3540131314484792,
      "grad_norm": 0.6848375079016027,
      "learning_rate": 4.963059362692837e-06,
      "loss": 0.6002,
      "step": 1321
    },
    {
      "epoch": 0.3542811201929519,
      "grad_norm": 0.686004262378946,
      "learning_rate": 4.962998978853887e-06,
      "loss": 0.6026,
      "step": 1322
    },
    {
      "epoch": 0.35454910893742464,
      "grad_norm": 0.7351307675304116,
      "learning_rate": 4.962938546070907e-06,
      "loss": 0.6043,
      "step": 1323
    },
    {
      "epoch": 0.35481709768189734,
      "grad_norm": 0.7397103681813406,
      "learning_rate": 4.962878064345099e-06,
      "loss": 0.6281,
      "step": 1324
    },
    {
      "epoch": 0.3550850864263701,
      "grad_norm": 0.6734361207632488,
      "learning_rate": 4.962817533677663e-06,
      "loss": 0.6193,
      "step": 1325
    },
    {
      "epoch": 0.3553530751708428,
      "grad_norm": 0.7062948925460725,
      "learning_rate": 4.962756954069803e-06,
      "loss": 0.6042,
      "step": 1326
    },
    {
      "epoch": 0.35562106391531556,
      "grad_norm": 0.7136163759385732,
      "learning_rate": 4.962696325522722e-06,
      "loss": 0.6393,
      "step": 1327
    },
    {
      "epoch": 0.35588905265978826,
      "grad_norm": 0.7220182461255163,
      "learning_rate": 4.9626356480376255e-06,
      "loss": 0.5994,
      "step": 1328
    },
    {
      "epoch": 0.356157041404261,
      "grad_norm": 0.6721558511819188,
      "learning_rate": 4.962574921615721e-06,
      "loss": 0.6031,
      "step": 1329
    },
    {
      "epoch": 0.3564250301487338,
      "grad_norm": 0.680518369344038,
      "learning_rate": 4.962514146258212e-06,
      "loss": 0.6318,
      "step": 1330
    },
    {
      "epoch": 0.3566930188932065,
      "grad_norm": 0.7208154192649147,
      "learning_rate": 4.962453321966307e-06,
      "loss": 0.6215,
      "step": 1331
    },
    {
      "epoch": 0.35696100763767924,
      "grad_norm": 0.693849005147156,
      "learning_rate": 4.962392448741216e-06,
      "loss": 0.6284,
      "step": 1332
    },
    {
      "epoch": 0.35722899638215194,
      "grad_norm": 0.7178118861358344,
      "learning_rate": 4.962331526584149e-06,
      "loss": 0.5848,
      "step": 1333
    },
    {
      "epoch": 0.3574969851266247,
      "grad_norm": 0.6955080939212255,
      "learning_rate": 4.962270555496315e-06,
      "loss": 0.6113,
      "step": 1334
    },
    {
      "epoch": 0.3577649738710974,
      "grad_norm": 0.6962830969819904,
      "learning_rate": 4.962209535478926e-06,
      "loss": 0.6232,
      "step": 1335
    },
    {
      "epoch": 0.35803296261557016,
      "grad_norm": 0.6750093197442688,
      "learning_rate": 4.962148466533194e-06,
      "loss": 0.5896,
      "step": 1336
    },
    {
      "epoch": 0.35830095136004286,
      "grad_norm": 0.6977517043537093,
      "learning_rate": 4.962087348660335e-06,
      "loss": 0.6478,
      "step": 1337
    },
    {
      "epoch": 0.3585689401045156,
      "grad_norm": 0.6970493769765286,
      "learning_rate": 4.962026181861561e-06,
      "loss": 0.6234,
      "step": 1338
    },
    {
      "epoch": 0.3588369288489883,
      "grad_norm": 0.6594362473859643,
      "learning_rate": 4.961964966138088e-06,
      "loss": 0.5643,
      "step": 1339
    },
    {
      "epoch": 0.3591049175934611,
      "grad_norm": 0.7008978929799149,
      "learning_rate": 4.961903701491133e-06,
      "loss": 0.6221,
      "step": 1340
    },
    {
      "epoch": 0.3593729063379338,
      "grad_norm": 0.7062386421195627,
      "learning_rate": 4.961842387921913e-06,
      "loss": 0.5637,
      "step": 1341
    },
    {
      "epoch": 0.35964089508240654,
      "grad_norm": 0.6747968384590821,
      "learning_rate": 4.961781025431648e-06,
      "loss": 0.6128,
      "step": 1342
    },
    {
      "epoch": 0.35990888382687924,
      "grad_norm": 0.6713933726375125,
      "learning_rate": 4.961719614021555e-06,
      "loss": 0.6103,
      "step": 1343
    },
    {
      "epoch": 0.360176872571352,
      "grad_norm": 0.7040979283981067,
      "learning_rate": 4.961658153692856e-06,
      "loss": 0.6372,
      "step": 1344
    },
    {
      "epoch": 0.36044486131582476,
      "grad_norm": 0.6840752477733159,
      "learning_rate": 4.96159664444677e-06,
      "loss": 0.6206,
      "step": 1345
    },
    {
      "epoch": 0.36071285006029746,
      "grad_norm": 0.6996891755553003,
      "learning_rate": 4.961535086284523e-06,
      "loss": 0.5932,
      "step": 1346
    },
    {
      "epoch": 0.3609808388047702,
      "grad_norm": 0.7130564587706029,
      "learning_rate": 4.961473479207336e-06,
      "loss": 0.5975,
      "step": 1347
    },
    {
      "epoch": 0.3612488275492429,
      "grad_norm": 0.6781651018135981,
      "learning_rate": 4.961411823216432e-06,
      "loss": 0.6047,
      "step": 1348
    },
    {
      "epoch": 0.3615168162937157,
      "grad_norm": 0.6466228079450203,
      "learning_rate": 4.961350118313039e-06,
      "loss": 0.5589,
      "step": 1349
    },
    {
      "epoch": 0.3617848050381884,
      "grad_norm": 0.6806597418986668,
      "learning_rate": 4.9612883644983815e-06,
      "loss": 0.615,
      "step": 1350
    },
    {
      "epoch": 0.36205279378266114,
      "grad_norm": 0.7040955780416035,
      "learning_rate": 4.961226561773687e-06,
      "loss": 0.6336,
      "step": 1351
    },
    {
      "epoch": 0.36232078252713384,
      "grad_norm": 0.6654961360228876,
      "learning_rate": 4.961164710140184e-06,
      "loss": 0.6327,
      "step": 1352
    },
    {
      "epoch": 0.3625887712716066,
      "grad_norm": 0.6762144861104048,
      "learning_rate": 4.961102809599101e-06,
      "loss": 0.6259,
      "step": 1353
    },
    {
      "epoch": 0.3628567600160793,
      "grad_norm": 0.7110565720162128,
      "learning_rate": 4.961040860151669e-06,
      "loss": 0.607,
      "step": 1354
    },
    {
      "epoch": 0.36312474876055206,
      "grad_norm": 0.6880597361257116,
      "learning_rate": 4.960978861799117e-06,
      "loss": 0.6477,
      "step": 1355
    },
    {
      "epoch": 0.36339273750502477,
      "grad_norm": 0.7067938416002578,
      "learning_rate": 4.960916814542679e-06,
      "loss": 0.6198,
      "step": 1356
    },
    {
      "epoch": 0.3636607262494975,
      "grad_norm": 0.7102026160279442,
      "learning_rate": 4.960854718383588e-06,
      "loss": 0.6171,
      "step": 1357
    },
    {
      "epoch": 0.3639287149939702,
      "grad_norm": 0.7114943774753165,
      "learning_rate": 4.960792573323076e-06,
      "loss": 0.6313,
      "step": 1358
    },
    {
      "epoch": 0.364196703738443,
      "grad_norm": 0.6662909139252616,
      "learning_rate": 4.960730379362381e-06,
      "loss": 0.6048,
      "step": 1359
    },
    {
      "epoch": 0.36446469248291574,
      "grad_norm": 0.6463860509931972,
      "learning_rate": 4.960668136502736e-06,
      "loss": 0.5943,
      "step": 1360
    },
    {
      "epoch": 0.36473268122738844,
      "grad_norm": 0.6757609452722334,
      "learning_rate": 4.96060584474538e-06,
      "loss": 0.6214,
      "step": 1361
    },
    {
      "epoch": 0.3650006699718612,
      "grad_norm": 0.7030041233677778,
      "learning_rate": 4.9605435040915495e-06,
      "loss": 0.6462,
      "step": 1362
    },
    {
      "epoch": 0.3652686587163339,
      "grad_norm": 0.6485241969298353,
      "learning_rate": 4.960481114542483e-06,
      "loss": 0.6258,
      "step": 1363
    },
    {
      "epoch": 0.36553664746080666,
      "grad_norm": 0.6577020365026685,
      "learning_rate": 4.960418676099422e-06,
      "loss": 0.6116,
      "step": 1364
    },
    {
      "epoch": 0.36580463620527937,
      "grad_norm": 0.6723399837329785,
      "learning_rate": 4.960356188763605e-06,
      "loss": 0.5824,
      "step": 1365
    },
    {
      "epoch": 0.3660726249497521,
      "grad_norm": 0.6900263472751934,
      "learning_rate": 4.960293652536277e-06,
      "loss": 0.6296,
      "step": 1366
    },
    {
      "epoch": 0.3663406136942248,
      "grad_norm": 0.6694115177446796,
      "learning_rate": 4.960231067418677e-06,
      "loss": 0.6153,
      "step": 1367
    },
    {
      "epoch": 0.3666086024386976,
      "grad_norm": 0.6936728031906443,
      "learning_rate": 4.960168433412051e-06,
      "loss": 0.577,
      "step": 1368
    },
    {
      "epoch": 0.3668765911831703,
      "grad_norm": 0.7212049234866901,
      "learning_rate": 4.960105750517642e-06,
      "loss": 0.6016,
      "step": 1369
    },
    {
      "epoch": 0.36714457992764304,
      "grad_norm": 0.7229487962199872,
      "learning_rate": 4.9600430187366975e-06,
      "loss": 0.6131,
      "step": 1370
    },
    {
      "epoch": 0.36741256867211575,
      "grad_norm": 0.6844370686232436,
      "learning_rate": 4.9599802380704634e-06,
      "loss": 0.601,
      "step": 1371
    },
    {
      "epoch": 0.3676805574165885,
      "grad_norm": 0.6830393784785683,
      "learning_rate": 4.959917408520187e-06,
      "loss": 0.5991,
      "step": 1372
    },
    {
      "epoch": 0.36794854616106126,
      "grad_norm": 0.6957086336807969,
      "learning_rate": 4.959854530087117e-06,
      "loss": 0.6026,
      "step": 1373
    },
    {
      "epoch": 0.36821653490553397,
      "grad_norm": 0.6782282386649492,
      "learning_rate": 4.959791602772503e-06,
      "loss": 0.5948,
      "step": 1374
    },
    {
      "epoch": 0.3684845236500067,
      "grad_norm": 0.7131322110346205,
      "learning_rate": 4.959728626577594e-06,
      "loss": 0.5904,
      "step": 1375
    },
    {
      "epoch": 0.3687525123944794,
      "grad_norm": 0.7259444559226251,
      "learning_rate": 4.9596656015036434e-06,
      "loss": 0.6069,
      "step": 1376
    },
    {
      "epoch": 0.3690205011389522,
      "grad_norm": 0.688424428980462,
      "learning_rate": 4.959602527551903e-06,
      "loss": 0.6316,
      "step": 1377
    },
    {
      "epoch": 0.3692884898834249,
      "grad_norm": 0.6650103164907073,
      "learning_rate": 4.959539404723627e-06,
      "loss": 0.5893,
      "step": 1378
    },
    {
      "epoch": 0.36955647862789764,
      "grad_norm": 0.7279417193751908,
      "learning_rate": 4.959476233020067e-06,
      "loss": 0.5809,
      "step": 1379
    },
    {
      "epoch": 0.36982446737237035,
      "grad_norm": 0.6886662781925645,
      "learning_rate": 4.959413012442482e-06,
      "loss": 0.6339,
      "step": 1380
    },
    {
      "epoch": 0.3700924561168431,
      "grad_norm": 0.720400692014373,
      "learning_rate": 4.959349742992126e-06,
      "loss": 0.6191,
      "step": 1381
    },
    {
      "epoch": 0.3703604448613158,
      "grad_norm": 0.7172260319107632,
      "learning_rate": 4.959286424670256e-06,
      "loss": 0.63,
      "step": 1382
    },
    {
      "epoch": 0.37062843360578857,
      "grad_norm": 0.7343621674937547,
      "learning_rate": 4.959223057478132e-06,
      "loss": 0.6344,
      "step": 1383
    },
    {
      "epoch": 0.37089642235026127,
      "grad_norm": 0.6997625639034476,
      "learning_rate": 4.959159641417012e-06,
      "loss": 0.6078,
      "step": 1384
    },
    {
      "epoch": 0.371164411094734,
      "grad_norm": 0.7117742952637222,
      "learning_rate": 4.959096176488156e-06,
      "loss": 0.6121,
      "step": 1385
    },
    {
      "epoch": 0.37143239983920673,
      "grad_norm": 0.6878996756898462,
      "learning_rate": 4.959032662692827e-06,
      "loss": 0.6473,
      "step": 1386
    },
    {
      "epoch": 0.3717003885836795,
      "grad_norm": 0.7035724601940686,
      "learning_rate": 4.958969100032284e-06,
      "loss": 0.5784,
      "step": 1387
    },
    {
      "epoch": 0.37196837732815224,
      "grad_norm": 0.6699646682917652,
      "learning_rate": 4.9589054885077925e-06,
      "loss": 0.5987,
      "step": 1388
    },
    {
      "epoch": 0.37223636607262495,
      "grad_norm": 0.7182209389601302,
      "learning_rate": 4.958841828120616e-06,
      "loss": 0.6448,
      "step": 1389
    },
    {
      "epoch": 0.3725043548170977,
      "grad_norm": 0.6974074783564128,
      "learning_rate": 4.95877811887202e-06,
      "loss": 0.6073,
      "step": 1390
    },
    {
      "epoch": 0.3727723435615704,
      "grad_norm": 0.719966242179184,
      "learning_rate": 4.958714360763269e-06,
      "loss": 0.6039,
      "step": 1391
    },
    {
      "epoch": 0.37304033230604317,
      "grad_norm": 0.7324323058927924,
      "learning_rate": 4.958650553795631e-06,
      "loss": 0.6333,
      "step": 1392
    },
    {
      "epoch": 0.37330832105051587,
      "grad_norm": 0.7191978268961303,
      "learning_rate": 4.958586697970374e-06,
      "loss": 0.6173,
      "step": 1393
    },
    {
      "epoch": 0.3735763097949886,
      "grad_norm": 0.6979904984427665,
      "learning_rate": 4.958522793288767e-06,
      "loss": 0.5926,
      "step": 1394
    },
    {
      "epoch": 0.37384429853946133,
      "grad_norm": 0.6713923251717562,
      "learning_rate": 4.958458839752081e-06,
      "loss": 0.5722,
      "step": 1395
    },
    {
      "epoch": 0.3741122872839341,
      "grad_norm": 0.6961219152054867,
      "learning_rate": 4.958394837361583e-06,
      "loss": 0.5914,
      "step": 1396
    },
    {
      "epoch": 0.3743802760284068,
      "grad_norm": 0.6897360639461934,
      "learning_rate": 4.958330786118549e-06,
      "loss": 0.6283,
      "step": 1397
    },
    {
      "epoch": 0.37464826477287955,
      "grad_norm": 0.6897304632904852,
      "learning_rate": 4.95826668602425e-06,
      "loss": 0.623,
      "step": 1398
    },
    {
      "epoch": 0.37491625351735225,
      "grad_norm": 0.7318359723716161,
      "learning_rate": 4.95820253707996e-06,
      "loss": 0.6362,
      "step": 1399
    },
    {
      "epoch": 0.375184242261825,
      "grad_norm": 0.7297254435534105,
      "learning_rate": 4.958138339286954e-06,
      "loss": 0.5961,
      "step": 1400
    },
    {
      "epoch": 0.3754522310062977,
      "grad_norm": 0.7026193941744211,
      "learning_rate": 4.958074092646507e-06,
      "loss": 0.6428,
      "step": 1401
    },
    {
      "epoch": 0.37572021975077047,
      "grad_norm": 0.7134036610551938,
      "learning_rate": 4.958009797159896e-06,
      "loss": 0.6184,
      "step": 1402
    },
    {
      "epoch": 0.3759882084952432,
      "grad_norm": 0.6766798255142781,
      "learning_rate": 4.957945452828399e-06,
      "loss": 0.5709,
      "step": 1403
    },
    {
      "epoch": 0.37625619723971593,
      "grad_norm": 0.6689354837821605,
      "learning_rate": 4.957881059653294e-06,
      "loss": 0.6005,
      "step": 1404
    },
    {
      "epoch": 0.3765241859841887,
      "grad_norm": 0.683407312247729,
      "learning_rate": 4.957816617635862e-06,
      "loss": 0.6174,
      "step": 1405
    },
    {
      "epoch": 0.3767921747286614,
      "grad_norm": 0.706886271373494,
      "learning_rate": 4.957752126777383e-06,
      "loss": 0.6441,
      "step": 1406
    },
    {
      "epoch": 0.37706016347313415,
      "grad_norm": 0.7311609028589116,
      "learning_rate": 4.957687587079137e-06,
      "loss": 0.5818,
      "step": 1407
    },
    {
      "epoch": 0.37732815221760685,
      "grad_norm": 0.6953134625848754,
      "learning_rate": 4.957622998542408e-06,
      "loss": 0.6332,
      "step": 1408
    },
    {
      "epoch": 0.3775961409620796,
      "grad_norm": 0.6846077252614112,
      "learning_rate": 4.95755836116848e-06,
      "loss": 0.6418,
      "step": 1409
    },
    {
      "epoch": 0.3778641297065523,
      "grad_norm": 0.6935417845453726,
      "learning_rate": 4.9574936749586356e-06,
      "loss": 0.6283,
      "step": 1410
    },
    {
      "epoch": 0.37813211845102507,
      "grad_norm": 0.6867216793867612,
      "learning_rate": 4.957428939914161e-06,
      "loss": 0.655,
      "step": 1411
    },
    {
      "epoch": 0.37840010719549777,
      "grad_norm": 0.720613428990832,
      "learning_rate": 4.957364156036344e-06,
      "loss": 0.6216,
      "step": 1412
    },
    {
      "epoch": 0.37866809593997053,
      "grad_norm": 0.692806371000983,
      "learning_rate": 4.95729932332647e-06,
      "loss": 0.63,
      "step": 1413
    },
    {
      "epoch": 0.37893608468444323,
      "grad_norm": 0.6880866654587533,
      "learning_rate": 4.9572344417858285e-06,
      "loss": 0.5979,
      "step": 1414
    },
    {
      "epoch": 0.379204073428916,
      "grad_norm": 0.7171884240401792,
      "learning_rate": 4.957169511415709e-06,
      "loss": 0.62,
      "step": 1415
    },
    {
      "epoch": 0.3794720621733887,
      "grad_norm": 0.7334883124442227,
      "learning_rate": 4.957104532217401e-06,
      "loss": 0.6203,
      "step": 1416
    },
    {
      "epoch": 0.37974005091786145,
      "grad_norm": 0.6968941502968364,
      "learning_rate": 4.9570395041921955e-06,
      "loss": 0.621,
      "step": 1417
    },
    {
      "epoch": 0.3800080396623342,
      "grad_norm": 0.6933241994252465,
      "learning_rate": 4.9569744273413854e-06,
      "loss": 0.6278,
      "step": 1418
    },
    {
      "epoch": 0.3802760284068069,
      "grad_norm": 0.7272470182584302,
      "learning_rate": 4.956909301666264e-06,
      "loss": 0.5891,
      "step": 1419
    },
    {
      "epoch": 0.38054401715127967,
      "grad_norm": 0.698704593540677,
      "learning_rate": 4.956844127168124e-06,
      "loss": 0.5919,
      "step": 1420
    },
    {
      "epoch": 0.38081200589575237,
      "grad_norm": 0.7516975123567289,
      "learning_rate": 4.956778903848263e-06,
      "loss": 0.6015,
      "step": 1421
    },
    {
      "epoch": 0.38107999464022513,
      "grad_norm": 0.6860613102264022,
      "learning_rate": 4.956713631707976e-06,
      "loss": 0.6397,
      "step": 1422
    },
    {
      "epoch": 0.38134798338469783,
      "grad_norm": 0.6987361401045132,
      "learning_rate": 4.956648310748559e-06,
      "loss": 0.5882,
      "step": 1423
    },
    {
      "epoch": 0.3816159721291706,
      "grad_norm": 0.6960626172544754,
      "learning_rate": 4.956582940971311e-06,
      "loss": 0.6045,
      "step": 1424
    },
    {
      "epoch": 0.3818839608736433,
      "grad_norm": 0.6811655137885066,
      "learning_rate": 4.956517522377531e-06,
      "loss": 0.6008,
      "step": 1425
    },
    {
      "epoch": 0.38215194961811605,
      "grad_norm": 0.6940473804282496,
      "learning_rate": 4.956452054968519e-06,
      "loss": 0.6292,
      "step": 1426
    },
    {
      "epoch": 0.38241993836258875,
      "grad_norm": 0.67464033689844,
      "learning_rate": 4.956386538745576e-06,
      "loss": 0.6117,
      "step": 1427
    },
    {
      "epoch": 0.3826879271070615,
      "grad_norm": 0.7040622843609118,
      "learning_rate": 4.956320973710003e-06,
      "loss": 0.6305,
      "step": 1428
    },
    {
      "epoch": 0.3829559158515342,
      "grad_norm": 0.6953956913733946,
      "learning_rate": 4.956255359863104e-06,
      "loss": 0.644,
      "step": 1429
    },
    {
      "epoch": 0.38322390459600697,
      "grad_norm": 0.6813512699165638,
      "learning_rate": 4.956189697206183e-06,
      "loss": 0.5998,
      "step": 1430
    },
    {
      "epoch": 0.3834918933404797,
      "grad_norm": 0.6967409910470795,
      "learning_rate": 4.956123985740545e-06,
      "loss": 0.649,
      "step": 1431
    },
    {
      "epoch": 0.38375988208495243,
      "grad_norm": 0.7252363076733074,
      "learning_rate": 4.956058225467494e-06,
      "loss": 0.6039,
      "step": 1432
    },
    {
      "epoch": 0.3840278708294252,
      "grad_norm": 0.6959495058586446,
      "learning_rate": 4.955992416388338e-06,
      "loss": 0.6271,
      "step": 1433
    },
    {
      "epoch": 0.3842958595738979,
      "grad_norm": 0.7024553423623674,
      "learning_rate": 4.955926558504385e-06,
      "loss": 0.6411,
      "step": 1434
    },
    {
      "epoch": 0.38456384831837065,
      "grad_norm": 0.6811631885892865,
      "learning_rate": 4.955860651816943e-06,
      "loss": 0.6178,
      "step": 1435
    },
    {
      "epoch": 0.38483183706284335,
      "grad_norm": 0.6969957809757816,
      "learning_rate": 4.955794696327323e-06,
      "loss": 0.6242,
      "step": 1436
    },
    {
      "epoch": 0.3850998258073161,
      "grad_norm": 0.683636408105784,
      "learning_rate": 4.955728692036833e-06,
      "loss": 0.6188,
      "step": 1437
    },
    {
      "epoch": 0.3853678145517888,
      "grad_norm": 0.6716949008636788,
      "learning_rate": 4.955662638946788e-06,
      "loss": 0.5997,
      "step": 1438
    },
    {
      "epoch": 0.38563580329626157,
      "grad_norm": 0.6967588283730739,
      "learning_rate": 4.955596537058498e-06,
      "loss": 0.6103,
      "step": 1439
    },
    {
      "epoch": 0.3859037920407343,
      "grad_norm": 0.6943801081094103,
      "learning_rate": 4.955530386373277e-06,
      "loss": 0.635,
      "step": 1440
    },
    {
      "epoch": 0.38617178078520703,
      "grad_norm": 0.6677781993500754,
      "learning_rate": 4.95546418689244e-06,
      "loss": 0.624,
      "step": 1441
    },
    {
      "epoch": 0.38643976952967973,
      "grad_norm": 0.6779802953156685,
      "learning_rate": 4.955397938617304e-06,
      "loss": 0.6287,
      "step": 1442
    },
    {
      "epoch": 0.3867077582741525,
      "grad_norm": 0.6973490489400892,
      "learning_rate": 4.955331641549183e-06,
      "loss": 0.5767,
      "step": 1443
    },
    {
      "epoch": 0.3869757470186252,
      "grad_norm": 0.701374181962689,
      "learning_rate": 4.955265295689396e-06,
      "loss": 0.5945,
      "step": 1444
    },
    {
      "epoch": 0.38724373576309795,
      "grad_norm": 0.6640235057229564,
      "learning_rate": 4.95519890103926e-06,
      "loss": 0.5917,
      "step": 1445
    },
    {
      "epoch": 0.3875117245075707,
      "grad_norm": 0.7097680566535087,
      "learning_rate": 4.9551324576000946e-06,
      "loss": 0.6245,
      "step": 1446
    },
    {
      "epoch": 0.3877797132520434,
      "grad_norm": 0.6997294646750278,
      "learning_rate": 4.955065965373222e-06,
      "loss": 0.6245,
      "step": 1447
    },
    {
      "epoch": 0.38804770199651617,
      "grad_norm": 0.6876067447833227,
      "learning_rate": 4.954999424359962e-06,
      "loss": 0.6016,
      "step": 1448
    },
    {
      "epoch": 0.3883156907409889,
      "grad_norm": 0.692738380727932,
      "learning_rate": 4.954932834561636e-06,
      "loss": 0.5727,
      "step": 1449
    },
    {
      "epoch": 0.38858367948546163,
      "grad_norm": 0.6963023881429231,
      "learning_rate": 4.95486619597957e-06,
      "loss": 0.6204,
      "step": 1450
    },
    {
      "epoch": 0.38885166822993433,
      "grad_norm": 0.6907965078143631,
      "learning_rate": 4.9547995086150856e-06,
      "loss": 0.6332,
      "step": 1451
    },
    {
      "epoch": 0.3891196569744071,
      "grad_norm": 0.7172668238760591,
      "learning_rate": 4.9547327724695096e-06,
      "loss": 0.6013,
      "step": 1452
    },
    {
      "epoch": 0.3893876457188798,
      "grad_norm": 0.7083714736430115,
      "learning_rate": 4.9546659875441676e-06,
      "loss": 0.5765,
      "step": 1453
    },
    {
      "epoch": 0.38965563446335255,
      "grad_norm": 0.709560405584642,
      "learning_rate": 4.954599153840385e-06,
      "loss": 0.5854,
      "step": 1454
    },
    {
      "epoch": 0.38992362320782525,
      "grad_norm": 0.6620764994194915,
      "learning_rate": 4.954532271359493e-06,
      "loss": 0.5711,
      "step": 1455
    },
    {
      "epoch": 0.390191611952298,
      "grad_norm": 0.6927211074133934,
      "learning_rate": 4.9544653401028194e-06,
      "loss": 0.6239,
      "step": 1456
    },
    {
      "epoch": 0.3904596006967707,
      "grad_norm": 0.6995532075303369,
      "learning_rate": 4.954398360071694e-06,
      "loss": 0.6213,
      "step": 1457
    },
    {
      "epoch": 0.3907275894412435,
      "grad_norm": 0.706503839757894,
      "learning_rate": 4.954331331267448e-06,
      "loss": 0.5852,
      "step": 1458
    },
    {
      "epoch": 0.3909955781857162,
      "grad_norm": 0.7096492769567355,
      "learning_rate": 4.954264253691412e-06,
      "loss": 0.6107,
      "step": 1459
    },
    {
      "epoch": 0.39126356693018893,
      "grad_norm": 0.7258244311915563,
      "learning_rate": 4.954197127344921e-06,
      "loss": 0.6202,
      "step": 1460
    },
    {
      "epoch": 0.3915315556746617,
      "grad_norm": 0.688602986199224,
      "learning_rate": 4.954129952229309e-06,
      "loss": 0.6274,
      "step": 1461
    },
    {
      "epoch": 0.3917995444191344,
      "grad_norm": 0.6769647737654536,
      "learning_rate": 4.954062728345909e-06,
      "loss": 0.6377,
      "step": 1462
    },
    {
      "epoch": 0.39206753316360715,
      "grad_norm": 0.6670932765829616,
      "learning_rate": 4.953995455696058e-06,
      "loss": 0.6126,
      "step": 1463
    },
    {
      "epoch": 0.39233552190807985,
      "grad_norm": 0.6809723141876084,
      "learning_rate": 4.953928134281093e-06,
      "loss": 0.6168,
      "step": 1464
    },
    {
      "epoch": 0.3926035106525526,
      "grad_norm": 0.7335661029430606,
      "learning_rate": 4.953860764102352e-06,
      "loss": 0.6221,
      "step": 1465
    },
    {
      "epoch": 0.3928714993970253,
      "grad_norm": 0.7337751800819784,
      "learning_rate": 4.9537933451611715e-06,
      "loss": 0.6396,
      "step": 1466
    },
    {
      "epoch": 0.3931394881414981,
      "grad_norm": 0.6904584843304169,
      "learning_rate": 4.953725877458895e-06,
      "loss": 0.6269,
      "step": 1467
    },
    {
      "epoch": 0.3934074768859708,
      "grad_norm": 0.6859077352093764,
      "learning_rate": 4.953658360996861e-06,
      "loss": 0.602,
      "step": 1468
    },
    {
      "epoch": 0.39367546563044353,
      "grad_norm": 0.688342693421456,
      "learning_rate": 4.95359079577641e-06,
      "loss": 0.6049,
      "step": 1469
    },
    {
      "epoch": 0.39394345437491624,
      "grad_norm": 0.6840844599301672,
      "learning_rate": 4.953523181798887e-06,
      "loss": 0.6306,
      "step": 1470
    },
    {
      "epoch": 0.394211443119389,
      "grad_norm": 0.7058572194811944,
      "learning_rate": 4.953455519065634e-06,
      "loss": 0.6164,
      "step": 1471
    },
    {
      "epoch": 0.3944794318638617,
      "grad_norm": 0.7130189619046954,
      "learning_rate": 4.953387807577996e-06,
      "loss": 0.6078,
      "step": 1472
    },
    {
      "epoch": 0.39474742060833445,
      "grad_norm": 0.693677706150599,
      "learning_rate": 4.953320047337319e-06,
      "loss": 0.6605,
      "step": 1473
    },
    {
      "epoch": 0.39501540935280716,
      "grad_norm": 0.6744417627349731,
      "learning_rate": 4.95325223834495e-06,
      "loss": 0.5766,
      "step": 1474
    },
    {
      "epoch": 0.3952833980972799,
      "grad_norm": 0.6651786420204094,
      "learning_rate": 4.953184380602235e-06,
      "loss": 0.6005,
      "step": 1475
    },
    {
      "epoch": 0.3955513868417527,
      "grad_norm": 0.6620467050023647,
      "learning_rate": 4.953116474110523e-06,
      "loss": 0.6093,
      "step": 1476
    },
    {
      "epoch": 0.3958193755862254,
      "grad_norm": 0.709227406999839,
      "learning_rate": 4.953048518871164e-06,
      "loss": 0.6315,
      "step": 1477
    },
    {
      "epoch": 0.39608736433069813,
      "grad_norm": 0.6769430973271583,
      "learning_rate": 4.952980514885508e-06,
      "loss": 0.6026,
      "step": 1478
    },
    {
      "epoch": 0.39635535307517084,
      "grad_norm": 0.6779766962451527,
      "learning_rate": 4.9529124621549055e-06,
      "loss": 0.6122,
      "step": 1479
    },
    {
      "epoch": 0.3966233418196436,
      "grad_norm": 0.7023448179049463,
      "learning_rate": 4.952844360680711e-06,
      "loss": 0.6275,
      "step": 1480
    },
    {
      "epoch": 0.3968913305641163,
      "grad_norm": 0.6480018534726579,
      "learning_rate": 4.952776210464275e-06,
      "loss": 0.588,
      "step": 1481
    },
    {
      "epoch": 0.39715931930858905,
      "grad_norm": 0.6899478689836689,
      "learning_rate": 4.952708011506954e-06,
      "loss": 0.6045,
      "step": 1482
    },
    {
      "epoch": 0.39742730805306176,
      "grad_norm": 0.7068741814463114,
      "learning_rate": 4.9526397638101025e-06,
      "loss": 0.5825,
      "step": 1483
    },
    {
      "epoch": 0.3976952967975345,
      "grad_norm": 0.6675873180001877,
      "learning_rate": 4.952571467375077e-06,
      "loss": 0.6008,
      "step": 1484
    },
    {
      "epoch": 0.3979632855420072,
      "grad_norm": 0.6705744101785718,
      "learning_rate": 4.952503122203233e-06,
      "loss": 0.598,
      "step": 1485
    },
    {
      "epoch": 0.39823127428648,
      "grad_norm": 0.684552017186037,
      "learning_rate": 4.952434728295931e-06,
      "loss": 0.6083,
      "step": 1486
    },
    {
      "epoch": 0.3984992630309527,
      "grad_norm": 0.6774125913605262,
      "learning_rate": 4.952366285654529e-06,
      "loss": 0.594,
      "step": 1487
    },
    {
      "epoch": 0.39876725177542544,
      "grad_norm": 0.6864744840586473,
      "learning_rate": 4.952297794280387e-06,
      "loss": 0.642,
      "step": 1488
    },
    {
      "epoch": 0.39903524051989814,
      "grad_norm": 0.6800740896061218,
      "learning_rate": 4.952229254174866e-06,
      "loss": 0.6079,
      "step": 1489
    },
    {
      "epoch": 0.3993032292643709,
      "grad_norm": 0.7467087386760328,
      "learning_rate": 4.952160665339329e-06,
      "loss": 0.629,
      "step": 1490
    },
    {
      "epoch": 0.39957121800884365,
      "grad_norm": 0.6824470693994009,
      "learning_rate": 4.952092027775136e-06,
      "loss": 0.5428,
      "step": 1491
    },
    {
      "epoch": 0.39983920675331636,
      "grad_norm": 0.6822969120053387,
      "learning_rate": 4.952023341483655e-06,
      "loss": 0.5845,
      "step": 1492
    },
    {
      "epoch": 0.4001071954977891,
      "grad_norm": 0.6846922521112133,
      "learning_rate": 4.9519546064662485e-06,
      "loss": 0.5933,
      "step": 1493
    },
    {
      "epoch": 0.4003751842422618,
      "grad_norm": 0.7044751790329054,
      "learning_rate": 4.951885822724283e-06,
      "loss": 0.6021,
      "step": 1494
    },
    {
      "epoch": 0.4006431729867346,
      "grad_norm": 0.6654687304140734,
      "learning_rate": 4.951816990259125e-06,
      "loss": 0.5947,
      "step": 1495
    },
    {
      "epoch": 0.4009111617312073,
      "grad_norm": 0.656385687419314,
      "learning_rate": 4.951748109072143e-06,
      "loss": 0.6143,
      "step": 1496
    },
    {
      "epoch": 0.40117915047568004,
      "grad_norm": 0.6812589766639223,
      "learning_rate": 4.951679179164704e-06,
      "loss": 0.6211,
      "step": 1497
    },
    {
      "epoch": 0.40144713922015274,
      "grad_norm": 0.7019074374666294,
      "learning_rate": 4.951610200538181e-06,
      "loss": 0.6375,
      "step": 1498
    },
    {
      "epoch": 0.4017151279646255,
      "grad_norm": 0.6962410505572806,
      "learning_rate": 4.951541173193941e-06,
      "loss": 0.6155,
      "step": 1499
    },
    {
      "epoch": 0.4019831167090982,
      "grad_norm": 0.6912581938387405,
      "learning_rate": 4.951472097133359e-06,
      "loss": 0.6603,
      "step": 1500
    },
    {
      "epoch": 0.40225110545357096,
      "grad_norm": 0.6617737759508041,
      "learning_rate": 4.951402972357805e-06,
      "loss": 0.5811,
      "step": 1501
    },
    {
      "epoch": 0.40251909419804366,
      "grad_norm": 0.679194687420335,
      "learning_rate": 4.951333798868655e-06,
      "loss": 0.6149,
      "step": 1502
    },
    {
      "epoch": 0.4027870829425164,
      "grad_norm": 0.6635729150590235,
      "learning_rate": 4.951264576667282e-06,
      "loss": 0.5839,
      "step": 1503
    },
    {
      "epoch": 0.4030550716869891,
      "grad_norm": 0.6915261814571173,
      "learning_rate": 4.951195305755062e-06,
      "loss": 0.6131,
      "step": 1504
    },
    {
      "epoch": 0.4033230604314619,
      "grad_norm": 0.7048168730241648,
      "learning_rate": 4.951125986133371e-06,
      "loss": 0.6444,
      "step": 1505
    },
    {
      "epoch": 0.40359104917593464,
      "grad_norm": 0.6924569585183653,
      "learning_rate": 4.951056617803587e-06,
      "loss": 0.5822,
      "step": 1506
    },
    {
      "epoch": 0.40385903792040734,
      "grad_norm": 0.7232591782818794,
      "learning_rate": 4.950987200767089e-06,
      "loss": 0.6007,
      "step": 1507
    },
    {
      "epoch": 0.4041270266648801,
      "grad_norm": 0.6599433406365008,
      "learning_rate": 4.950917735025256e-06,
      "loss": 0.6007,
      "step": 1508
    },
    {
      "epoch": 0.4043950154093528,
      "grad_norm": 0.6975193519054874,
      "learning_rate": 4.950848220579468e-06,
      "loss": 0.6363,
      "step": 1509
    },
    {
      "epoch": 0.40466300415382556,
      "grad_norm": 0.7178415845408564,
      "learning_rate": 4.950778657431107e-06,
      "loss": 0.6308,
      "step": 1510
    },
    {
      "epoch": 0.40493099289829826,
      "grad_norm": 0.6919157309866469,
      "learning_rate": 4.950709045581555e-06,
      "loss": 0.5524,
      "step": 1511
    },
    {
      "epoch": 0.405198981642771,
      "grad_norm": 0.6720307558283162,
      "learning_rate": 4.9506393850321946e-06,
      "loss": 0.5916,
      "step": 1512
    },
    {
      "epoch": 0.4054669703872437,
      "grad_norm": 0.6656155747645174,
      "learning_rate": 4.950569675784412e-06,
      "loss": 0.5703,
      "step": 1513
    },
    {
      "epoch": 0.4057349591317165,
      "grad_norm": 0.6832776860218134,
      "learning_rate": 4.95049991783959e-06,
      "loss": 0.6052,
      "step": 1514
    },
    {
      "epoch": 0.4060029478761892,
      "grad_norm": 0.6765720753173261,
      "learning_rate": 4.950430111199116e-06,
      "loss": 0.6345,
      "step": 1515
    },
    {
      "epoch": 0.40627093662066194,
      "grad_norm": 0.6961568982568855,
      "learning_rate": 4.950360255864377e-06,
      "loss": 0.6275,
      "step": 1516
    },
    {
      "epoch": 0.40653892536513464,
      "grad_norm": 0.6815901760953732,
      "learning_rate": 4.950290351836763e-06,
      "loss": 0.6237,
      "step": 1517
    },
    {
      "epoch": 0.4068069141096074,
      "grad_norm": 0.6568582019603858,
      "learning_rate": 4.950220399117661e-06,
      "loss": 0.6166,
      "step": 1518
    },
    {
      "epoch": 0.40707490285408016,
      "grad_norm": 0.6847154052468997,
      "learning_rate": 4.950150397708461e-06,
      "loss": 0.6449,
      "step": 1519
    },
    {
      "epoch": 0.40734289159855286,
      "grad_norm": 0.6810732748579547,
      "learning_rate": 4.950080347610554e-06,
      "loss": 0.6463,
      "step": 1520
    },
    {
      "epoch": 0.4076108803430256,
      "grad_norm": 0.6635108744137284,
      "learning_rate": 4.9500102488253345e-06,
      "loss": 0.6092,
      "step": 1521
    },
    {
      "epoch": 0.4078788690874983,
      "grad_norm": 0.6713108878468075,
      "learning_rate": 4.949940101354192e-06,
      "loss": 0.5993,
      "step": 1522
    },
    {
      "epoch": 0.4081468578319711,
      "grad_norm": 0.684387134711913,
      "learning_rate": 4.949869905198524e-06,
      "loss": 0.6183,
      "step": 1523
    },
    {
      "epoch": 0.4084148465764438,
      "grad_norm": 0.675959816463184,
      "learning_rate": 4.949799660359721e-06,
      "loss": 0.5836,
      "step": 1524
    },
    {
      "epoch": 0.40868283532091654,
      "grad_norm": 0.6879387458193112,
      "learning_rate": 4.949729366839183e-06,
      "loss": 0.6207,
      "step": 1525
    },
    {
      "epoch": 0.40895082406538924,
      "grad_norm": 0.6917647846006948,
      "learning_rate": 4.949659024638304e-06,
      "loss": 0.638,
      "step": 1526
    },
    {
      "epoch": 0.409218812809862,
      "grad_norm": 0.6942605318704193,
      "learning_rate": 4.949588633758484e-06,
      "loss": 0.6308,
      "step": 1527
    },
    {
      "epoch": 0.4094868015543347,
      "grad_norm": 0.6994935169066968,
      "learning_rate": 4.949518194201121e-06,
      "loss": 0.6139,
      "step": 1528
    },
    {
      "epoch": 0.40975479029880746,
      "grad_norm": 0.6716251994763517,
      "learning_rate": 4.949447705967614e-06,
      "loss": 0.603,
      "step": 1529
    },
    {
      "epoch": 0.41002277904328016,
      "grad_norm": 0.6563011542477527,
      "learning_rate": 4.949377169059365e-06,
      "loss": 0.5874,
      "step": 1530
    },
    {
      "epoch": 0.4102907677877529,
      "grad_norm": 0.6844618326184729,
      "learning_rate": 4.949306583477774e-06,
      "loss": 0.6134,
      "step": 1531
    },
    {
      "epoch": 0.4105587565322256,
      "grad_norm": 0.6732994479753114,
      "learning_rate": 4.949235949224246e-06,
      "loss": 0.6146,
      "step": 1532
    },
    {
      "epoch": 0.4108267452766984,
      "grad_norm": 0.6798963881029182,
      "learning_rate": 4.9491652663001825e-06,
      "loss": 0.5916,
      "step": 1533
    },
    {
      "epoch": 0.41109473402117114,
      "grad_norm": 0.6859467468216063,
      "learning_rate": 4.949094534706989e-06,
      "loss": 0.6197,
      "step": 1534
    },
    {
      "epoch": 0.41136272276564384,
      "grad_norm": 0.6623763697152061,
      "learning_rate": 4.94902375444607e-06,
      "loss": 0.6021,
      "step": 1535
    },
    {
      "epoch": 0.4116307115101166,
      "grad_norm": 0.6815268080045677,
      "learning_rate": 4.948952925518835e-06,
      "loss": 0.5912,
      "step": 1536
    },
    {
      "epoch": 0.4118987002545893,
      "grad_norm": 0.6698329827152624,
      "learning_rate": 4.948882047926688e-06,
      "loss": 0.6023,
      "step": 1537
    },
    {
      "epoch": 0.41216668899906206,
      "grad_norm": 0.6801246013075071,
      "learning_rate": 4.948811121671039e-06,
      "loss": 0.5805,
      "step": 1538
    },
    {
      "epoch": 0.41243467774353476,
      "grad_norm": 0.6733813296128763,
      "learning_rate": 4.948740146753298e-06,
      "loss": 0.5943,
      "step": 1539
    },
    {
      "epoch": 0.4127026664880075,
      "grad_norm": 0.670587844881258,
      "learning_rate": 4.948669123174874e-06,
      "loss": 0.6145,
      "step": 1540
    },
    {
      "epoch": 0.4129706552324802,
      "grad_norm": 0.6767025034222721,
      "learning_rate": 4.948598050937179e-06,
      "loss": 0.6414,
      "step": 1541
    },
    {
      "epoch": 0.413238643976953,
      "grad_norm": 0.6556196140904056,
      "learning_rate": 4.948526930041626e-06,
      "loss": 0.5834,
      "step": 1542
    },
    {
      "epoch": 0.4135066327214257,
      "grad_norm": 0.7151489561032732,
      "learning_rate": 4.948455760489628e-06,
      "loss": 0.594,
      "step": 1543
    },
    {
      "epoch": 0.41377462146589844,
      "grad_norm": 0.6782635291745304,
      "learning_rate": 4.948384542282598e-06,
      "loss": 0.6117,
      "step": 1544
    },
    {
      "epoch": 0.41404261021037114,
      "grad_norm": 0.6594032673621646,
      "learning_rate": 4.948313275421953e-06,
      "loss": 0.575,
      "step": 1545
    },
    {
      "epoch": 0.4143105989548439,
      "grad_norm": 0.7146574470663588,
      "learning_rate": 4.948241959909108e-06,
      "loss": 0.576,
      "step": 1546
    },
    {
      "epoch": 0.4145785876993166,
      "grad_norm": 0.6863762063589541,
      "learning_rate": 4.948170595745481e-06,
      "loss": 0.6383,
      "step": 1547
    },
    {
      "epoch": 0.41484657644378936,
      "grad_norm": 0.6801628271918734,
      "learning_rate": 4.94809918293249e-06,
      "loss": 0.6142,
      "step": 1548
    },
    {
      "epoch": 0.4151145651882621,
      "grad_norm": 0.7010877469670314,
      "learning_rate": 4.9480277214715525e-06,
      "loss": 0.6221,
      "step": 1549
    },
    {
      "epoch": 0.4153825539327348,
      "grad_norm": 0.6997904074619549,
      "learning_rate": 4.947956211364091e-06,
      "loss": 0.5831,
      "step": 1550
    },
    {
      "epoch": 0.4156505426772076,
      "grad_norm": 0.7069091426603658,
      "learning_rate": 4.947884652611525e-06,
      "loss": 0.6532,
      "step": 1551
    },
    {
      "epoch": 0.4159185314216803,
      "grad_norm": 0.6988126280245128,
      "learning_rate": 4.947813045215277e-06,
      "loss": 0.6252,
      "step": 1552
    },
    {
      "epoch": 0.41618652016615304,
      "grad_norm": 0.7017397598477766,
      "learning_rate": 4.947741389176769e-06,
      "loss": 0.5734,
      "step": 1553
    },
    {
      "epoch": 0.41645450891062574,
      "grad_norm": 0.7377902908135004,
      "learning_rate": 4.947669684497427e-06,
      "loss": 0.6133,
      "step": 1554
    },
    {
      "epoch": 0.4167224976550985,
      "grad_norm": 0.687419816752458,
      "learning_rate": 4.947597931178674e-06,
      "loss": 0.606,
      "step": 1555
    },
    {
      "epoch": 0.4169904863995712,
      "grad_norm": 0.6893525611849577,
      "learning_rate": 4.947526129221937e-06,
      "loss": 0.6254,
      "step": 1556
    },
    {
      "epoch": 0.41725847514404396,
      "grad_norm": 0.7025148131040782,
      "learning_rate": 4.947454278628642e-06,
      "loss": 0.6025,
      "step": 1557
    },
    {
      "epoch": 0.41752646388851666,
      "grad_norm": 0.6987837880752006,
      "learning_rate": 4.947382379400217e-06,
      "loss": 0.6022,
      "step": 1558
    },
    {
      "epoch": 0.4177944526329894,
      "grad_norm": 0.760084437310104,
      "learning_rate": 4.947310431538091e-06,
      "loss": 0.6456,
      "step": 1559
    },
    {
      "epoch": 0.4180624413774621,
      "grad_norm": 0.8733667254451583,
      "learning_rate": 4.947238435043694e-06,
      "loss": 0.6182,
      "step": 1560
    },
    {
      "epoch": 0.4183304301219349,
      "grad_norm": 0.7295080027651979,
      "learning_rate": 4.947166389918455e-06,
      "loss": 0.5657,
      "step": 1561
    },
    {
      "epoch": 0.4185984188664076,
      "grad_norm": 0.7485985092715746,
      "learning_rate": 4.947094296163809e-06,
      "loss": 0.6185,
      "step": 1562
    },
    {
      "epoch": 0.41886640761088034,
      "grad_norm": 0.6764616598253556,
      "learning_rate": 4.947022153781185e-06,
      "loss": 0.627,
      "step": 1563
    },
    {
      "epoch": 0.4191343963553531,
      "grad_norm": 0.6707761294066056,
      "learning_rate": 4.9469499627720195e-06,
      "loss": 0.5964,
      "step": 1564
    },
    {
      "epoch": 0.4194023850998258,
      "grad_norm": 0.7183500992667013,
      "learning_rate": 4.946877723137745e-06,
      "loss": 0.5952,
      "step": 1565
    },
    {
      "epoch": 0.41967037384429856,
      "grad_norm": 0.6976061662113799,
      "learning_rate": 4.946805434879798e-06,
      "loss": 0.5798,
      "step": 1566
    },
    {
      "epoch": 0.41993836258877126,
      "grad_norm": 0.6653282022591505,
      "learning_rate": 4.946733097999615e-06,
      "loss": 0.6211,
      "step": 1567
    },
    {
      "epoch": 0.420206351333244,
      "grad_norm": 0.7044437747908975,
      "learning_rate": 4.946660712498632e-06,
      "loss": 0.6218,
      "step": 1568
    },
    {
      "epoch": 0.4204743400777167,
      "grad_norm": 0.7016076533040834,
      "learning_rate": 4.94658827837829e-06,
      "loss": 0.636,
      "step": 1569
    },
    {
      "epoch": 0.4207423288221895,
      "grad_norm": 0.6678505827525385,
      "learning_rate": 4.946515795640026e-06,
      "loss": 0.6277,
      "step": 1570
    },
    {
      "epoch": 0.4210103175666622,
      "grad_norm": 0.6795247888711811,
      "learning_rate": 4.946443264285283e-06,
      "loss": 0.6214,
      "step": 1571
    },
    {
      "epoch": 0.42127830631113494,
      "grad_norm": 0.6609719852761226,
      "learning_rate": 4.9463706843155e-06,
      "loss": 0.6441,
      "step": 1572
    },
    {
      "epoch": 0.42154629505560764,
      "grad_norm": 0.6823451032907328,
      "learning_rate": 4.94629805573212e-06,
      "loss": 0.6143,
      "step": 1573
    },
    {
      "epoch": 0.4218142838000804,
      "grad_norm": 0.6759517367461354,
      "learning_rate": 4.946225378536587e-06,
      "loss": 0.5489,
      "step": 1574
    },
    {
      "epoch": 0.4220822725445531,
      "grad_norm": 0.6372232594481358,
      "learning_rate": 4.946152652730343e-06,
      "loss": 0.5714,
      "step": 1575
    },
    {
      "epoch": 0.42235026128902586,
      "grad_norm": 0.7601369668101134,
      "learning_rate": 4.946079878314835e-06,
      "loss": 0.6247,
      "step": 1576
    },
    {
      "epoch": 0.42261825003349857,
      "grad_norm": 0.6829293116814509,
      "learning_rate": 4.946007055291509e-06,
      "loss": 0.5978,
      "step": 1577
    },
    {
      "epoch": 0.4228862387779713,
      "grad_norm": 0.6816985846541008,
      "learning_rate": 4.945934183661813e-06,
      "loss": 0.5818,
      "step": 1578
    },
    {
      "epoch": 0.4231542275224441,
      "grad_norm": 0.6682121389698455,
      "learning_rate": 4.945861263427193e-06,
      "loss": 0.6062,
      "step": 1579
    },
    {
      "epoch": 0.4234222162669168,
      "grad_norm": 0.6946471277857768,
      "learning_rate": 4.9457882945891e-06,
      "loss": 0.613,
      "step": 1580
    },
    {
      "epoch": 0.42369020501138954,
      "grad_norm": 0.7230621447152797,
      "learning_rate": 4.945715277148982e-06,
      "loss": 0.6696,
      "step": 1581
    },
    {
      "epoch": 0.42395819375586224,
      "grad_norm": 0.6853805711874417,
      "learning_rate": 4.945642211108292e-06,
      "loss": 0.6179,
      "step": 1582
    },
    {
      "epoch": 0.424226182500335,
      "grad_norm": 0.6635486312037006,
      "learning_rate": 4.945569096468481e-06,
      "loss": 0.5785,
      "step": 1583
    },
    {
      "epoch": 0.4244941712448077,
      "grad_norm": 0.6898391411598364,
      "learning_rate": 4.945495933231002e-06,
      "loss": 0.6044,
      "step": 1584
    },
    {
      "epoch": 0.42476215998928046,
      "grad_norm": 0.6881602800477862,
      "learning_rate": 4.9454227213973074e-06,
      "loss": 0.6137,
      "step": 1585
    },
    {
      "epoch": 0.42503014873375317,
      "grad_norm": 0.6892092866273943,
      "learning_rate": 4.945349460968855e-06,
      "loss": 0.6195,
      "step": 1586
    },
    {
      "epoch": 0.4252981374782259,
      "grad_norm": 0.6921061280726335,
      "learning_rate": 4.945276151947098e-06,
      "loss": 0.6566,
      "step": 1587
    },
    {
      "epoch": 0.4255661262226986,
      "grad_norm": 0.6599957153601214,
      "learning_rate": 4.945202794333496e-06,
      "loss": 0.6053,
      "step": 1588
    },
    {
      "epoch": 0.4258341149671714,
      "grad_norm": 0.6998022635298694,
      "learning_rate": 4.9451293881295035e-06,
      "loss": 0.6148,
      "step": 1589
    },
    {
      "epoch": 0.4261021037116441,
      "grad_norm": 0.7226866783509778,
      "learning_rate": 4.945055933336582e-06,
      "loss": 0.6654,
      "step": 1590
    },
    {
      "epoch": 0.42637009245611684,
      "grad_norm": 0.6922215204800024,
      "learning_rate": 4.944982429956189e-06,
      "loss": 0.604,
      "step": 1591
    },
    {
      "epoch": 0.4266380812005896,
      "grad_norm": 0.6816765084963986,
      "learning_rate": 4.944908877989786e-06,
      "loss": 0.609,
      "step": 1592
    },
    {
      "epoch": 0.4269060699450623,
      "grad_norm": 0.6959925445570135,
      "learning_rate": 4.944835277438835e-06,
      "loss": 0.6162,
      "step": 1593
    },
    {
      "epoch": 0.42717405868953506,
      "grad_norm": 0.7028188176571819,
      "learning_rate": 4.944761628304798e-06,
      "loss": 0.5977,
      "step": 1594
    },
    {
      "epoch": 0.42744204743400777,
      "grad_norm": 0.652197003143818,
      "learning_rate": 4.944687930589138e-06,
      "loss": 0.5853,
      "step": 1595
    },
    {
      "epoch": 0.4277100361784805,
      "grad_norm": 0.700676326861139,
      "learning_rate": 4.944614184293321e-06,
      "loss": 0.5999,
      "step": 1596
    },
    {
      "epoch": 0.4279780249229532,
      "grad_norm": 0.7149971233131409,
      "learning_rate": 4.944540389418813e-06,
      "loss": 0.6141,
      "step": 1597
    },
    {
      "epoch": 0.428246013667426,
      "grad_norm": 0.6572527937973329,
      "learning_rate": 4.944466545967077e-06,
      "loss": 0.5931,
      "step": 1598
    },
    {
      "epoch": 0.4285140024118987,
      "grad_norm": 0.6768537869577023,
      "learning_rate": 4.944392653939584e-06,
      "loss": 0.6449,
      "step": 1599
    },
    {
      "epoch": 0.42878199115637144,
      "grad_norm": 0.7009929316639637,
      "learning_rate": 4.9443187133378e-06,
      "loss": 0.5897,
      "step": 1600
    },
    {
      "epoch": 0.42904997990084415,
      "grad_norm": 0.6724918793341044,
      "learning_rate": 4.944244724163195e-06,
      "loss": 0.5937,
      "step": 1601
    },
    {
      "epoch": 0.4293179686453169,
      "grad_norm": 0.6854046011412739,
      "learning_rate": 4.944170686417241e-06,
      "loss": 0.5836,
      "step": 1602
    },
    {
      "epoch": 0.4295859573897896,
      "grad_norm": 0.680802963886409,
      "learning_rate": 4.944096600101407e-06,
      "loss": 0.6181,
      "step": 1603
    },
    {
      "epoch": 0.42985394613426237,
      "grad_norm": 0.7103796380369579,
      "learning_rate": 4.944022465217165e-06,
      "loss": 0.6076,
      "step": 1604
    },
    {
      "epoch": 0.43012193487873507,
      "grad_norm": 0.6773002717558532,
      "learning_rate": 4.943948281765991e-06,
      "loss": 0.6164,
      "step": 1605
    },
    {
      "epoch": 0.4303899236232078,
      "grad_norm": 0.7011355778552792,
      "learning_rate": 4.943874049749356e-06,
      "loss": 0.6263,
      "step": 1606
    },
    {
      "epoch": 0.4306579123676806,
      "grad_norm": 0.6744577665862905,
      "learning_rate": 4.943799769168737e-06,
      "loss": 0.6253,
      "step": 1607
    },
    {
      "epoch": 0.4309259011121533,
      "grad_norm": 0.715827224186663,
      "learning_rate": 4.94372544002561e-06,
      "loss": 0.5967,
      "step": 1608
    },
    {
      "epoch": 0.43119388985662604,
      "grad_norm": 0.6852330893085719,
      "learning_rate": 4.943651062321451e-06,
      "loss": 0.6165,
      "step": 1609
    },
    {
      "epoch": 0.43146187860109875,
      "grad_norm": 0.6845849884852723,
      "learning_rate": 4.943576636057738e-06,
      "loss": 0.6051,
      "step": 1610
    },
    {
      "epoch": 0.4317298673455715,
      "grad_norm": 0.6684481152650782,
      "learning_rate": 4.943502161235951e-06,
      "loss": 0.5576,
      "step": 1611
    },
    {
      "epoch": 0.4319978560900442,
      "grad_norm": 0.6622310838944198,
      "learning_rate": 4.943427637857569e-06,
      "loss": 0.5826,
      "step": 1612
    },
    {
      "epoch": 0.43226584483451697,
      "grad_norm": 0.6834601926251982,
      "learning_rate": 4.943353065924075e-06,
      "loss": 0.6337,
      "step": 1613
    },
    {
      "epoch": 0.43253383357898967,
      "grad_norm": 0.6991861795044021,
      "learning_rate": 4.943278445436949e-06,
      "loss": 0.585,
      "step": 1614
    },
    {
      "epoch": 0.4328018223234624,
      "grad_norm": 0.6595780507335989,
      "learning_rate": 4.943203776397674e-06,
      "loss": 0.6356,
      "step": 1615
    },
    {
      "epoch": 0.43306981106793513,
      "grad_norm": 0.6818083076945957,
      "learning_rate": 4.943129058807733e-06,
      "loss": 0.621,
      "step": 1616
    },
    {
      "epoch": 0.4333377998124079,
      "grad_norm": 0.7594540960284507,
      "learning_rate": 4.943054292668612e-06,
      "loss": 0.5991,
      "step": 1617
    },
    {
      "epoch": 0.4336057885568806,
      "grad_norm": 0.6645589605958434,
      "learning_rate": 4.942979477981797e-06,
      "loss": 0.6313,
      "step": 1618
    },
    {
      "epoch": 0.43387377730135335,
      "grad_norm": 0.6936103375724134,
      "learning_rate": 4.9429046147487735e-06,
      "loss": 0.5889,
      "step": 1619
    },
    {
      "epoch": 0.43414176604582605,
      "grad_norm": 0.7644737831548164,
      "learning_rate": 4.94282970297103e-06,
      "loss": 0.5633,
      "step": 1620
    },
    {
      "epoch": 0.4344097547902988,
      "grad_norm": 0.688293224598141,
      "learning_rate": 4.942754742650055e-06,
      "loss": 0.6039,
      "step": 1621
    },
    {
      "epoch": 0.43467774353477157,
      "grad_norm": 0.7494002139123886,
      "learning_rate": 4.9426797337873385e-06,
      "loss": 0.6071,
      "step": 1622
    },
    {
      "epoch": 0.43494573227924427,
      "grad_norm": 0.7013636292178671,
      "learning_rate": 4.94260467638437e-06,
      "loss": 0.6485,
      "step": 1623
    },
    {
      "epoch": 0.435213721023717,
      "grad_norm": 0.7086352325716078,
      "learning_rate": 4.9425295704426425e-06,
      "loss": 0.6414,
      "step": 1624
    },
    {
      "epoch": 0.43548170976818973,
      "grad_norm": 0.6876082959478133,
      "learning_rate": 4.9424544159636465e-06,
      "loss": 0.6182,
      "step": 1625
    },
    {
      "epoch": 0.4357496985126625,
      "grad_norm": 0.7348884756988033,
      "learning_rate": 4.942379212948877e-06,
      "loss": 0.6098,
      "step": 1626
    },
    {
      "epoch": 0.4360176872571352,
      "grad_norm": 0.7022367062434572,
      "learning_rate": 4.942303961399828e-06,
      "loss": 0.6125,
      "step": 1627
    },
    {
      "epoch": 0.43628567600160795,
      "grad_norm": 0.6959696199816366,
      "learning_rate": 4.942228661317995e-06,
      "loss": 0.6005,
      "step": 1628
    },
    {
      "epoch": 0.43655366474608065,
      "grad_norm": 0.6667034951833071,
      "learning_rate": 4.942153312704874e-06,
      "loss": 0.5924,
      "step": 1629
    },
    {
      "epoch": 0.4368216534905534,
      "grad_norm": 0.676654058061013,
      "learning_rate": 4.942077915561962e-06,
      "loss": 0.6471,
      "step": 1630
    },
    {
      "epoch": 0.4370896422350261,
      "grad_norm": 0.7305793564006674,
      "learning_rate": 4.942002469890758e-06,
      "loss": 0.6457,
      "step": 1631
    },
    {
      "epoch": 0.43735763097949887,
      "grad_norm": 0.6871096134944399,
      "learning_rate": 4.941926975692761e-06,
      "loss": 0.6174,
      "step": 1632
    },
    {
      "epoch": 0.43762561972397157,
      "grad_norm": 0.6593989983647107,
      "learning_rate": 4.941851432969471e-06,
      "loss": 0.5976,
      "step": 1633
    },
    {
      "epoch": 0.43789360846844433,
      "grad_norm": 0.6774611039446992,
      "learning_rate": 4.94177584172239e-06,
      "loss": 0.6298,
      "step": 1634
    },
    {
      "epoch": 0.43816159721291703,
      "grad_norm": 0.812534117844718,
      "learning_rate": 4.9417002019530195e-06,
      "loss": 0.5644,
      "step": 1635
    },
    {
      "epoch": 0.4384295859573898,
      "grad_norm": 0.6764373352375778,
      "learning_rate": 4.9416245136628614e-06,
      "loss": 0.6411,
      "step": 1636
    },
    {
      "epoch": 0.43869757470186255,
      "grad_norm": 0.6780764636247945,
      "learning_rate": 4.941548776853422e-06,
      "loss": 0.6173,
      "step": 1637
    },
    {
      "epoch": 0.43896556344633525,
      "grad_norm": 0.7206284992060128,
      "learning_rate": 4.941472991526205e-06,
      "loss": 0.6287,
      "step": 1638
    },
    {
      "epoch": 0.439233552190808,
      "grad_norm": 0.7261640121191082,
      "learning_rate": 4.941397157682716e-06,
      "loss": 0.5958,
      "step": 1639
    },
    {
      "epoch": 0.4395015409352807,
      "grad_norm": 0.6801257195310181,
      "learning_rate": 4.941321275324463e-06,
      "loss": 0.6461,
      "step": 1640
    },
    {
      "epoch": 0.43976952967975347,
      "grad_norm": 0.6790947086850402,
      "learning_rate": 4.941245344452954e-06,
      "loss": 0.6041,
      "step": 1641
    },
    {
      "epoch": 0.44003751842422617,
      "grad_norm": 0.7625942360725276,
      "learning_rate": 4.941169365069697e-06,
      "loss": 0.5908,
      "step": 1642
    },
    {
      "epoch": 0.44030550716869893,
      "grad_norm": 0.6737309712479358,
      "learning_rate": 4.941093337176201e-06,
      "loss": 0.6187,
      "step": 1643
    },
    {
      "epoch": 0.44057349591317163,
      "grad_norm": 0.6771993004473351,
      "learning_rate": 4.94101726077398e-06,
      "loss": 0.6001,
      "step": 1644
    },
    {
      "epoch": 0.4408414846576444,
      "grad_norm": 0.7900599912134855,
      "learning_rate": 4.940941135864542e-06,
      "loss": 0.5771,
      "step": 1645
    },
    {
      "epoch": 0.4411094734021171,
      "grad_norm": 0.6796077275661464,
      "learning_rate": 4.940864962449403e-06,
      "loss": 0.5907,
      "step": 1646
    },
    {
      "epoch": 0.44137746214658985,
      "grad_norm": 0.6792763338377636,
      "learning_rate": 4.9407887405300745e-06,
      "loss": 0.6172,
      "step": 1647
    },
    {
      "epoch": 0.44164545089106255,
      "grad_norm": 0.6704506483208029,
      "learning_rate": 4.940712470108073e-06,
      "loss": 0.604,
      "step": 1648
    },
    {
      "epoch": 0.4419134396355353,
      "grad_norm": 0.6847603883984216,
      "learning_rate": 4.940636151184911e-06,
      "loss": 0.5885,
      "step": 1649
    },
    {
      "epoch": 0.442181428380008,
      "grad_norm": 0.6708649408166751,
      "learning_rate": 4.940559783762108e-06,
      "loss": 0.5833,
      "step": 1650
    },
    {
      "epoch": 0.44244941712448077,
      "grad_norm": 0.6629715918403254,
      "learning_rate": 4.940483367841181e-06,
      "loss": 0.6371,
      "step": 1651
    },
    {
      "epoch": 0.44271740586895353,
      "grad_norm": 0.6655994652877665,
      "learning_rate": 4.940406903423648e-06,
      "loss": 0.5898,
      "step": 1652
    },
    {
      "epoch": 0.44298539461342623,
      "grad_norm": 0.6730799849742499,
      "learning_rate": 4.940330390511028e-06,
      "loss": 0.6173,
      "step": 1653
    },
    {
      "epoch": 0.443253383357899,
      "grad_norm": 0.664265384324033,
      "learning_rate": 4.940253829104843e-06,
      "loss": 0.6297,
      "step": 1654
    },
    {
      "epoch": 0.4435213721023717,
      "grad_norm": 0.6757407125507914,
      "learning_rate": 4.940177219206612e-06,
      "loss": 0.6113,
      "step": 1655
    },
    {
      "epoch": 0.44378936084684445,
      "grad_norm": 0.6718647362645105,
      "learning_rate": 4.94010056081786e-06,
      "loss": 0.5982,
      "step": 1656
    },
    {
      "epoch": 0.44405734959131715,
      "grad_norm": 0.6905728459813067,
      "learning_rate": 4.940023853940109e-06,
      "loss": 0.6039,
      "step": 1657
    },
    {
      "epoch": 0.4443253383357899,
      "grad_norm": 0.6942191228125795,
      "learning_rate": 4.939947098574884e-06,
      "loss": 0.6238,
      "step": 1658
    },
    {
      "epoch": 0.4445933270802626,
      "grad_norm": 0.664151855467319,
      "learning_rate": 4.939870294723709e-06,
      "loss": 0.6101,
      "step": 1659
    },
    {
      "epoch": 0.44486131582473537,
      "grad_norm": 0.6707131206820287,
      "learning_rate": 4.939793442388111e-06,
      "loss": 0.5857,
      "step": 1660
    },
    {
      "epoch": 0.4451293045692081,
      "grad_norm": 0.6757587151270269,
      "learning_rate": 4.939716541569618e-06,
      "loss": 0.6248,
      "step": 1661
    },
    {
      "epoch": 0.44539729331368083,
      "grad_norm": 0.6899295314469285,
      "learning_rate": 4.939639592269757e-06,
      "loss": 0.6305,
      "step": 1662
    },
    {
      "epoch": 0.44566528205815353,
      "grad_norm": 0.6556892886335683,
      "learning_rate": 4.9395625944900566e-06,
      "loss": 0.5895,
      "step": 1663
    },
    {
      "epoch": 0.4459332708026263,
      "grad_norm": 0.6693576589111068,
      "learning_rate": 4.9394855482320485e-06,
      "loss": 0.5585,
      "step": 1664
    },
    {
      "epoch": 0.44620125954709905,
      "grad_norm": 0.6751218149448487,
      "learning_rate": 4.939408453497262e-06,
      "loss": 0.6181,
      "step": 1665
    },
    {
      "epoch": 0.44646924829157175,
      "grad_norm": 0.662547899534758,
      "learning_rate": 4.939331310287231e-06,
      "loss": 0.6174,
      "step": 1666
    },
    {
      "epoch": 0.4467372370360445,
      "grad_norm": 0.6827862149104941,
      "learning_rate": 4.939254118603487e-06,
      "loss": 0.5952,
      "step": 1667
    },
    {
      "epoch": 0.4470052257805172,
      "grad_norm": 0.6655778776543869,
      "learning_rate": 4.939176878447565e-06,
      "loss": 0.61,
      "step": 1668
    },
    {
      "epoch": 0.44727321452498997,
      "grad_norm": 0.6700983360631835,
      "learning_rate": 4.939099589820999e-06,
      "loss": 0.5942,
      "step": 1669
    },
    {
      "epoch": 0.4475412032694627,
      "grad_norm": 0.7012230337907904,
      "learning_rate": 4.939022252725325e-06,
      "loss": 0.5931,
      "step": 1670
    },
    {
      "epoch": 0.44780919201393543,
      "grad_norm": 0.6497580522286626,
      "learning_rate": 4.93894486716208e-06,
      "loss": 0.6141,
      "step": 1671
    },
    {
      "epoch": 0.44807718075840813,
      "grad_norm": 0.6979522044680219,
      "learning_rate": 4.938867433132802e-06,
      "loss": 0.6126,
      "step": 1672
    },
    {
      "epoch": 0.4483451695028809,
      "grad_norm": 0.6695639263371472,
      "learning_rate": 4.93878995063903e-06,
      "loss": 0.5926,
      "step": 1673
    },
    {
      "epoch": 0.4486131582473536,
      "grad_norm": 0.6830076544710246,
      "learning_rate": 4.938712419682303e-06,
      "loss": 0.6218,
      "step": 1674
    },
    {
      "epoch": 0.44888114699182635,
      "grad_norm": 0.6712952044048092,
      "learning_rate": 4.938634840264161e-06,
      "loss": 0.6463,
      "step": 1675
    },
    {
      "epoch": 0.44914913573629905,
      "grad_norm": 0.6789795578196055,
      "learning_rate": 4.938557212386148e-06,
      "loss": 0.6223,
      "step": 1676
    },
    {
      "epoch": 0.4494171244807718,
      "grad_norm": 0.685021136918899,
      "learning_rate": 4.938479536049805e-06,
      "loss": 0.6082,
      "step": 1677
    },
    {
      "epoch": 0.4496851132252445,
      "grad_norm": 0.6736717598250317,
      "learning_rate": 4.9384018112566755e-06,
      "loss": 0.6068,
      "step": 1678
    },
    {
      "epoch": 0.4499531019697173,
      "grad_norm": 0.6766913750170841,
      "learning_rate": 4.9383240380083034e-06,
      "loss": 0.5874,
      "step": 1679
    },
    {
      "epoch": 0.45022109071419003,
      "grad_norm": 0.6943741826175166,
      "learning_rate": 4.938246216306236e-06,
      "loss": 0.5956,
      "step": 1680
    },
    {
      "epoch": 0.45048907945866273,
      "grad_norm": 0.6878318710964046,
      "learning_rate": 4.938168346152018e-06,
      "loss": 0.5735,
      "step": 1681
    },
    {
      "epoch": 0.4507570682031355,
      "grad_norm": 0.680814172008617,
      "learning_rate": 4.938090427547199e-06,
      "loss": 0.6163,
      "step": 1682
    },
    {
      "epoch": 0.4510250569476082,
      "grad_norm": 0.6619677159882119,
      "learning_rate": 4.9380124604933246e-06,
      "loss": 0.5878,
      "step": 1683
    },
    {
      "epoch": 0.45129304569208095,
      "grad_norm": 0.6609742024576787,
      "learning_rate": 4.9379344449919465e-06,
      "loss": 0.5966,
      "step": 1684
    },
    {
      "epoch": 0.45156103443655365,
      "grad_norm": 0.6804870279922313,
      "learning_rate": 4.937856381044613e-06,
      "loss": 0.604,
      "step": 1685
    },
    {
      "epoch": 0.4518290231810264,
      "grad_norm": 0.6609380426828414,
      "learning_rate": 4.937778268652878e-06,
      "loss": 0.6049,
      "step": 1686
    },
    {
      "epoch": 0.4520970119254991,
      "grad_norm": 0.6727736323776541,
      "learning_rate": 4.93770010781829e-06,
      "loss": 0.5732,
      "step": 1687
    },
    {
      "epoch": 0.4523650006699719,
      "grad_norm": 0.6581988098613242,
      "learning_rate": 4.9376218985424056e-06,
      "loss": 0.5874,
      "step": 1688
    },
    {
      "epoch": 0.4526329894144446,
      "grad_norm": 0.6531675325693753,
      "learning_rate": 4.9375436408267775e-06,
      "loss": 0.597,
      "step": 1689
    },
    {
      "epoch": 0.45290097815891733,
      "grad_norm": 0.7484536128278311,
      "learning_rate": 4.9374653346729605e-06,
      "loss": 0.6195,
      "step": 1690
    },
    {
      "epoch": 0.45316896690339004,
      "grad_norm": 0.697717567759986,
      "learning_rate": 4.937386980082512e-06,
      "loss": 0.5864,
      "step": 1691
    },
    {
      "epoch": 0.4534369556478628,
      "grad_norm": 0.6717398118864883,
      "learning_rate": 4.937308577056988e-06,
      "loss": 0.6269,
      "step": 1692
    },
    {
      "epoch": 0.4537049443923355,
      "grad_norm": 0.6835680426545228,
      "learning_rate": 4.937230125597947e-06,
      "loss": 0.5901,
      "step": 1693
    },
    {
      "epoch": 0.45397293313680825,
      "grad_norm": 0.6788532667249177,
      "learning_rate": 4.9371516257069465e-06,
      "loss": 0.5891,
      "step": 1694
    },
    {
      "epoch": 0.454240921881281,
      "grad_norm": 0.6750640554242073,
      "learning_rate": 4.937073077385548e-06,
      "loss": 0.6054,
      "step": 1695
    },
    {
      "epoch": 0.4545089106257537,
      "grad_norm": 0.6695472130679707,
      "learning_rate": 4.9369944806353135e-06,
      "loss": 0.5719,
      "step": 1696
    },
    {
      "epoch": 0.4547768993702265,
      "grad_norm": 0.6903464958412189,
      "learning_rate": 4.936915835457802e-06,
      "loss": 0.6241,
      "step": 1697
    },
    {
      "epoch": 0.4550448881146992,
      "grad_norm": 0.6813824967121885,
      "learning_rate": 4.936837141854577e-06,
      "loss": 0.643,
      "step": 1698
    },
    {
      "epoch": 0.45531287685917193,
      "grad_norm": 0.6786398828746963,
      "learning_rate": 4.936758399827204e-06,
      "loss": 0.6337,
      "step": 1699
    },
    {
      "epoch": 0.45558086560364464,
      "grad_norm": 0.6577439146415763,
      "learning_rate": 4.936679609377246e-06,
      "loss": 0.5876,
      "step": 1700
    },
    {
      "epoch": 0.4558488543481174,
      "grad_norm": 0.6768953689024712,
      "learning_rate": 4.93660077050627e-06,
      "loss": 0.5989,
      "step": 1701
    },
    {
      "epoch": 0.4561168430925901,
      "grad_norm": 0.7294923288716217,
      "learning_rate": 4.936521883215843e-06,
      "loss": 0.6209,
      "step": 1702
    },
    {
      "epoch": 0.45638483183706285,
      "grad_norm": 0.6669545125369524,
      "learning_rate": 4.93644294750753e-06,
      "loss": 0.59,
      "step": 1703
    },
    {
      "epoch": 0.45665282058153556,
      "grad_norm": 0.7254915547066578,
      "learning_rate": 4.936363963382902e-06,
      "loss": 0.6145,
      "step": 1704
    },
    {
      "epoch": 0.4569208093260083,
      "grad_norm": 0.6799824624145457,
      "learning_rate": 4.936284930843527e-06,
      "loss": 0.6149,
      "step": 1705
    },
    {
      "epoch": 0.457188798070481,
      "grad_norm": 0.6814980850527398,
      "learning_rate": 4.936205849890977e-06,
      "loss": 0.598,
      "step": 1706
    },
    {
      "epoch": 0.4574567868149538,
      "grad_norm": 0.6587986398485778,
      "learning_rate": 4.936126720526823e-06,
      "loss": 0.5752,
      "step": 1707
    },
    {
      "epoch": 0.4577247755594265,
      "grad_norm": 0.6954932811090386,
      "learning_rate": 4.936047542752637e-06,
      "loss": 0.6155,
      "step": 1708
    },
    {
      "epoch": 0.45799276430389924,
      "grad_norm": 0.6982802442091444,
      "learning_rate": 4.935968316569993e-06,
      "loss": 0.5918,
      "step": 1709
    },
    {
      "epoch": 0.458260753048372,
      "grad_norm": 0.7004056252217641,
      "learning_rate": 4.935889041980464e-06,
      "loss": 0.5924,
      "step": 1710
    },
    {
      "epoch": 0.4585287417928447,
      "grad_norm": 0.7012585329581128,
      "learning_rate": 4.935809718985627e-06,
      "loss": 0.5965,
      "step": 1711
    },
    {
      "epoch": 0.45879673053731745,
      "grad_norm": 0.6808208415938942,
      "learning_rate": 4.9357303475870575e-06,
      "loss": 0.6218,
      "step": 1712
    },
    {
      "epoch": 0.45906471928179016,
      "grad_norm": 0.6763890350727461,
      "learning_rate": 4.9356509277863324e-06,
      "loss": 0.5916,
      "step": 1713
    },
    {
      "epoch": 0.4593327080262629,
      "grad_norm": 0.6716759331422166,
      "learning_rate": 4.935571459585031e-06,
      "loss": 0.6116,
      "step": 1714
    },
    {
      "epoch": 0.4596006967707356,
      "grad_norm": 0.6930003821257882,
      "learning_rate": 4.935491942984731e-06,
      "loss": 0.6334,
      "step": 1715
    },
    {
      "epoch": 0.4598686855152084,
      "grad_norm": 0.6445525802066998,
      "learning_rate": 4.935412377987014e-06,
      "loss": 0.6031,
      "step": 1716
    },
    {
      "epoch": 0.4601366742596811,
      "grad_norm": 0.6931757929648853,
      "learning_rate": 4.935332764593461e-06,
      "loss": 0.6146,
      "step": 1717
    },
    {
      "epoch": 0.46040466300415384,
      "grad_norm": 0.6594697414685718,
      "learning_rate": 4.935253102805652e-06,
      "loss": 0.585,
      "step": 1718
    },
    {
      "epoch": 0.46067265174862654,
      "grad_norm": 0.6839851448588998,
      "learning_rate": 4.935173392625174e-06,
      "loss": 0.6283,
      "step": 1719
    },
    {
      "epoch": 0.4609406404930993,
      "grad_norm": 0.6680033852483375,
      "learning_rate": 4.935093634053606e-06,
      "loss": 0.5692,
      "step": 1720
    },
    {
      "epoch": 0.461208629237572,
      "grad_norm": 0.6711733957253422,
      "learning_rate": 4.935013827092536e-06,
      "loss": 0.5483,
      "step": 1721
    },
    {
      "epoch": 0.46147661798204476,
      "grad_norm": 0.6853875875200814,
      "learning_rate": 4.93493397174355e-06,
      "loss": 0.6066,
      "step": 1722
    },
    {
      "epoch": 0.46174460672651746,
      "grad_norm": 0.7270774657306919,
      "learning_rate": 4.934854068008234e-06,
      "loss": 0.6325,
      "step": 1723
    },
    {
      "epoch": 0.4620125954709902,
      "grad_norm": 0.6618309484142373,
      "learning_rate": 4.934774115888176e-06,
      "loss": 0.604,
      "step": 1724
    },
    {
      "epoch": 0.462280584215463,
      "grad_norm": 0.6766148411057036,
      "learning_rate": 4.934694115384964e-06,
      "loss": 0.579,
      "step": 1725
    },
    {
      "epoch": 0.4625485729599357,
      "grad_norm": 0.6570171266207241,
      "learning_rate": 4.934614066500189e-06,
      "loss": 0.5957,
      "step": 1726
    },
    {
      "epoch": 0.46281656170440844,
      "grad_norm": 0.6883987864040384,
      "learning_rate": 4.934533969235442e-06,
      "loss": 0.6111,
      "step": 1727
    },
    {
      "epoch": 0.46308455044888114,
      "grad_norm": 0.6761748624358689,
      "learning_rate": 4.934453823592313e-06,
      "loss": 0.6077,
      "step": 1728
    },
    {
      "epoch": 0.4633525391933539,
      "grad_norm": 0.7121591993922748,
      "learning_rate": 4.934373629572397e-06,
      "loss": 0.6124,
      "step": 1729
    },
    {
      "epoch": 0.4636205279378266,
      "grad_norm": 0.6801953745056609,
      "learning_rate": 4.934293387177285e-06,
      "loss": 0.5779,
      "step": 1730
    },
    {
      "epoch": 0.46388851668229936,
      "grad_norm": 0.7316805484544365,
      "learning_rate": 4.934213096408573e-06,
      "loss": 0.6308,
      "step": 1731
    },
    {
      "epoch": 0.46415650542677206,
      "grad_norm": 0.7147457920573597,
      "learning_rate": 4.934132757267856e-06,
      "loss": 0.5733,
      "step": 1732
    },
    {
      "epoch": 0.4644244941712448,
      "grad_norm": 0.6689458787537145,
      "learning_rate": 4.934052369756731e-06,
      "loss": 0.6029,
      "step": 1733
    },
    {
      "epoch": 0.4646924829157175,
      "grad_norm": 0.6688171893315649,
      "learning_rate": 4.933971933876795e-06,
      "loss": 0.6165,
      "step": 1734
    },
    {
      "epoch": 0.4649604716601903,
      "grad_norm": 0.7010909095077938,
      "learning_rate": 4.933891449629646e-06,
      "loss": 0.5987,
      "step": 1735
    },
    {
      "epoch": 0.465228460404663,
      "grad_norm": 0.67030721366231,
      "learning_rate": 4.933810917016885e-06,
      "loss": 0.6007,
      "step": 1736
    },
    {
      "epoch": 0.46549644914913574,
      "grad_norm": 0.661302299781105,
      "learning_rate": 4.933730336040111e-06,
      "loss": 0.5849,
      "step": 1737
    },
    {
      "epoch": 0.4657644378936085,
      "grad_norm": 0.6478994276290064,
      "learning_rate": 4.933649706700925e-06,
      "loss": 0.6216,
      "step": 1738
    },
    {
      "epoch": 0.4660324266380812,
      "grad_norm": 0.6680749922929449,
      "learning_rate": 4.933569029000931e-06,
      "loss": 0.6095,
      "step": 1739
    },
    {
      "epoch": 0.46630041538255396,
      "grad_norm": 0.6901370221673068,
      "learning_rate": 4.933488302941729e-06,
      "loss": 0.6043,
      "step": 1740
    },
    {
      "epoch": 0.46656840412702666,
      "grad_norm": 0.6910776041851143,
      "learning_rate": 4.933407528524927e-06,
      "loss": 0.6144,
      "step": 1741
    },
    {
      "epoch": 0.4668363928714994,
      "grad_norm": 0.6808602869581035,
      "learning_rate": 4.933326705752128e-06,
      "loss": 0.6203,
      "step": 1742
    },
    {
      "epoch": 0.4671043816159721,
      "grad_norm": 0.683372852191312,
      "learning_rate": 4.933245834624937e-06,
      "loss": 0.5937,
      "step": 1743
    },
    {
      "epoch": 0.4673723703604449,
      "grad_norm": 0.6657921182617819,
      "learning_rate": 4.933164915144963e-06,
      "loss": 0.5705,
      "step": 1744
    },
    {
      "epoch": 0.4676403591049176,
      "grad_norm": 0.6926448431031337,
      "learning_rate": 4.9330839473138136e-06,
      "loss": 0.6567,
      "step": 1745
    },
    {
      "epoch": 0.46790834784939034,
      "grad_norm": 0.7199929992905001,
      "learning_rate": 4.9330029311330974e-06,
      "loss": 0.612,
      "step": 1746
    },
    {
      "epoch": 0.46817633659386304,
      "grad_norm": 0.7508479919382314,
      "learning_rate": 4.932921866604424e-06,
      "loss": 0.6023,
      "step": 1747
    },
    {
      "epoch": 0.4684443253383358,
      "grad_norm": 0.7553343952182853,
      "learning_rate": 4.932840753729406e-06,
      "loss": 0.5652,
      "step": 1748
    },
    {
      "epoch": 0.4687123140828085,
      "grad_norm": 0.6813554581441624,
      "learning_rate": 4.932759592509653e-06,
      "loss": 0.6305,
      "step": 1749
    },
    {
      "epoch": 0.46898030282728126,
      "grad_norm": 0.6633864288708611,
      "learning_rate": 4.9326783829467795e-06,
      "loss": 0.5978,
      "step": 1750
    },
    {
      "epoch": 0.46924829157175396,
      "grad_norm": 0.6687600523035152,
      "learning_rate": 4.932597125042399e-06,
      "loss": 0.5907,
      "step": 1751
    },
    {
      "epoch": 0.4695162803162267,
      "grad_norm": 0.7565133757025216,
      "learning_rate": 4.932515818798125e-06,
      "loss": 0.5773,
      "step": 1752
    },
    {
      "epoch": 0.4697842690606995,
      "grad_norm": 0.6429354962696685,
      "learning_rate": 4.932434464215574e-06,
      "loss": 0.5822,
      "step": 1753
    },
    {
      "epoch": 0.4700522578051722,
      "grad_norm": 0.706203592070594,
      "learning_rate": 4.9323530612963636e-06,
      "loss": 0.609,
      "step": 1754
    },
    {
      "epoch": 0.47032024654964494,
      "grad_norm": 0.7234557424327794,
      "learning_rate": 4.93227161004211e-06,
      "loss": 0.5955,
      "step": 1755
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.6985597226726307,
      "learning_rate": 4.932190110454432e-06,
      "loss": 0.6077,
      "step": 1756
    },
    {
      "epoch": 0.4708562240385904,
      "grad_norm": 0.6903276771779718,
      "learning_rate": 4.9321085625349505e-06,
      "loss": 0.5867,
      "step": 1757
    },
    {
      "epoch": 0.4711242127830631,
      "grad_norm": 0.7016191560362115,
      "learning_rate": 4.932026966285283e-06,
      "loss": 0.5859,
      "step": 1758
    },
    {
      "epoch": 0.47139220152753586,
      "grad_norm": 0.6957412039419166,
      "learning_rate": 4.931945321707055e-06,
      "loss": 0.5989,
      "step": 1759
    },
    {
      "epoch": 0.47166019027200856,
      "grad_norm": 0.7021494125758576,
      "learning_rate": 4.931863628801886e-06,
      "loss": 0.6154,
      "step": 1760
    },
    {
      "epoch": 0.4719281790164813,
      "grad_norm": 0.6693260338514824,
      "learning_rate": 4.9317818875714e-06,
      "loss": 0.5733,
      "step": 1761
    },
    {
      "epoch": 0.472196167760954,
      "grad_norm": 0.7048243420530681,
      "learning_rate": 4.931700098017222e-06,
      "loss": 0.6253,
      "step": 1762
    },
    {
      "epoch": 0.4724641565054268,
      "grad_norm": 0.6772030911428129,
      "learning_rate": 4.931618260140977e-06,
      "loss": 0.5898,
      "step": 1763
    },
    {
      "epoch": 0.4727321452498995,
      "grad_norm": 0.6847664080921768,
      "learning_rate": 4.93153637394429e-06,
      "loss": 0.6135,
      "step": 1764
    },
    {
      "epoch": 0.47300013399437224,
      "grad_norm": 0.6813816901076019,
      "learning_rate": 4.93145443942879e-06,
      "loss": 0.6192,
      "step": 1765
    },
    {
      "epoch": 0.47326812273884494,
      "grad_norm": 0.7103466123695682,
      "learning_rate": 4.9313724565961045e-06,
      "loss": 0.6359,
      "step": 1766
    },
    {
      "epoch": 0.4735361114833177,
      "grad_norm": 0.6775408052892581,
      "learning_rate": 4.931290425447863e-06,
      "loss": 0.6072,
      "step": 1767
    },
    {
      "epoch": 0.47380410022779046,
      "grad_norm": 0.6750202866681545,
      "learning_rate": 4.931208345985694e-06,
      "loss": 0.6136,
      "step": 1768
    },
    {
      "epoch": 0.47407208897226316,
      "grad_norm": 0.6580661726218692,
      "learning_rate": 4.9311262182112315e-06,
      "loss": 0.6369,
      "step": 1769
    },
    {
      "epoch": 0.4743400777167359,
      "grad_norm": 0.683679300467441,
      "learning_rate": 4.931044042126106e-06,
      "loss": 0.589,
      "step": 1770
    },
    {
      "epoch": 0.4746080664612086,
      "grad_norm": 0.6884652094465448,
      "learning_rate": 4.930961817731949e-06,
      "loss": 0.6075,
      "step": 1771
    },
    {
      "epoch": 0.4748760552056814,
      "grad_norm": 0.7273155463395196,
      "learning_rate": 4.930879545030395e-06,
      "loss": 0.5822,
      "step": 1772
    },
    {
      "epoch": 0.4751440439501541,
      "grad_norm": 0.6631033982199616,
      "learning_rate": 4.93079722402308e-06,
      "loss": 0.589,
      "step": 1773
    },
    {
      "epoch": 0.47541203269462684,
      "grad_norm": 0.6847771191085934,
      "learning_rate": 4.9307148547116415e-06,
      "loss": 0.6473,
      "step": 1774
    },
    {
      "epoch": 0.47568002143909954,
      "grad_norm": 0.6533396199070762,
      "learning_rate": 4.9306324370977125e-06,
      "loss": 0.6331,
      "step": 1775
    },
    {
      "epoch": 0.4759480101835723,
      "grad_norm": 0.6630716526273638,
      "learning_rate": 4.930549971182933e-06,
      "loss": 0.5949,
      "step": 1776
    },
    {
      "epoch": 0.476215998928045,
      "grad_norm": 0.7967585291057978,
      "learning_rate": 4.9304674569689415e-06,
      "loss": 0.6085,
      "step": 1777
    },
    {
      "epoch": 0.47648398767251776,
      "grad_norm": 0.6701140298298285,
      "learning_rate": 4.930384894457377e-06,
      "loss": 0.62,
      "step": 1778
    },
    {
      "epoch": 0.47675197641699046,
      "grad_norm": 0.6568197763774131,
      "learning_rate": 4.930302283649882e-06,
      "loss": 0.593,
      "step": 1779
    },
    {
      "epoch": 0.4770199651614632,
      "grad_norm": 0.6844588140710407,
      "learning_rate": 4.930219624548096e-06,
      "loss": 0.6139,
      "step": 1780
    },
    {
      "epoch": 0.4772879539059359,
      "grad_norm": 0.6870456228890186,
      "learning_rate": 4.930136917153662e-06,
      "loss": 0.6145,
      "step": 1781
    },
    {
      "epoch": 0.4775559426504087,
      "grad_norm": 0.6712812820980464,
      "learning_rate": 4.930054161468224e-06,
      "loss": 0.615,
      "step": 1782
    },
    {
      "epoch": 0.47782393139488144,
      "grad_norm": 0.6850094184402671,
      "learning_rate": 4.929971357493427e-06,
      "loss": 0.6169,
      "step": 1783
    },
    {
      "epoch": 0.47809192013935414,
      "grad_norm": 0.70111411458747,
      "learning_rate": 4.9298885052309166e-06,
      "loss": 0.5971,
      "step": 1784
    },
    {
      "epoch": 0.4783599088838269,
      "grad_norm": 0.6563994019254609,
      "learning_rate": 4.929805604682337e-06,
      "loss": 0.5912,
      "step": 1785
    },
    {
      "epoch": 0.4786278976282996,
      "grad_norm": 0.6878525989026527,
      "learning_rate": 4.929722655849338e-06,
      "loss": 0.6062,
      "step": 1786
    },
    {
      "epoch": 0.47889588637277236,
      "grad_norm": 0.6721302030268519,
      "learning_rate": 4.929639658733567e-06,
      "loss": 0.6081,
      "step": 1787
    },
    {
      "epoch": 0.47916387511724506,
      "grad_norm": 0.648341338895909,
      "learning_rate": 4.929556613336672e-06,
      "loss": 0.5986,
      "step": 1788
    },
    {
      "epoch": 0.4794318638617178,
      "grad_norm": 0.6798163055033773,
      "learning_rate": 4.929473519660306e-06,
      "loss": 0.5812,
      "step": 1789
    },
    {
      "epoch": 0.4796998526061905,
      "grad_norm": 0.6850296234183723,
      "learning_rate": 4.929390377706118e-06,
      "loss": 0.61,
      "step": 1790
    },
    {
      "epoch": 0.4799678413506633,
      "grad_norm": 0.6495698892771424,
      "learning_rate": 4.929307187475762e-06,
      "loss": 0.5729,
      "step": 1791
    },
    {
      "epoch": 0.480235830095136,
      "grad_norm": 0.6914844937775279,
      "learning_rate": 4.92922394897089e-06,
      "loss": 0.5908,
      "step": 1792
    },
    {
      "epoch": 0.48050381883960874,
      "grad_norm": 0.6927327173338623,
      "learning_rate": 4.929140662193155e-06,
      "loss": 0.6107,
      "step": 1793
    },
    {
      "epoch": 0.48077180758408145,
      "grad_norm": 0.687012482983764,
      "learning_rate": 4.929057327144213e-06,
      "loss": 0.6006,
      "step": 1794
    },
    {
      "epoch": 0.4810397963285542,
      "grad_norm": 0.7446016929124284,
      "learning_rate": 4.928973943825721e-06,
      "loss": 0.5728,
      "step": 1795
    },
    {
      "epoch": 0.4813077850730269,
      "grad_norm": 0.7164101357167812,
      "learning_rate": 4.928890512239336e-06,
      "loss": 0.6304,
      "step": 1796
    },
    {
      "epoch": 0.48157577381749966,
      "grad_norm": 0.7172100822426628,
      "learning_rate": 4.928807032386714e-06,
      "loss": 0.6151,
      "step": 1797
    },
    {
      "epoch": 0.4818437625619724,
      "grad_norm": 0.7145472125378158,
      "learning_rate": 4.928723504269516e-06,
      "loss": 0.6058,
      "step": 1798
    },
    {
      "epoch": 0.4821117513064451,
      "grad_norm": 0.6860140872792941,
      "learning_rate": 4.9286399278894e-06,
      "loss": 0.5874,
      "step": 1799
    },
    {
      "epoch": 0.4823797400509179,
      "grad_norm": 0.6892593112839018,
      "learning_rate": 4.928556303248028e-06,
      "loss": 0.5728,
      "step": 1800
    },
    {
      "epoch": 0.4826477287953906,
      "grad_norm": 0.7295440245860517,
      "learning_rate": 4.928472630347062e-06,
      "loss": 0.6473,
      "step": 1801
    },
    {
      "epoch": 0.48291571753986334,
      "grad_norm": 0.7253540054100537,
      "learning_rate": 4.928388909188164e-06,
      "loss": 0.5982,
      "step": 1802
    },
    {
      "epoch": 0.48318370628433605,
      "grad_norm": 0.67967805200905,
      "learning_rate": 4.928305139772997e-06,
      "loss": 0.6046,
      "step": 1803
    },
    {
      "epoch": 0.4834516950288088,
      "grad_norm": 0.6962635627671402,
      "learning_rate": 4.928221322103227e-06,
      "loss": 0.5633,
      "step": 1804
    },
    {
      "epoch": 0.4837196837732815,
      "grad_norm": 0.6660881730962666,
      "learning_rate": 4.9281374561805195e-06,
      "loss": 0.6096,
      "step": 1805
    },
    {
      "epoch": 0.48398767251775426,
      "grad_norm": 0.6815882970389768,
      "learning_rate": 4.9280535420065405e-06,
      "loss": 0.6368,
      "step": 1806
    },
    {
      "epoch": 0.48425566126222697,
      "grad_norm": 0.6697662137905043,
      "learning_rate": 4.927969579582958e-06,
      "loss": 0.5725,
      "step": 1807
    },
    {
      "epoch": 0.4845236500066997,
      "grad_norm": 0.7388332113710988,
      "learning_rate": 4.927885568911439e-06,
      "loss": 0.5887,
      "step": 1808
    },
    {
      "epoch": 0.4847916387511724,
      "grad_norm": 0.6661920728931112,
      "learning_rate": 4.927801509993655e-06,
      "loss": 0.5943,
      "step": 1809
    },
    {
      "epoch": 0.4850596274956452,
      "grad_norm": 0.6862647074921667,
      "learning_rate": 4.927717402831274e-06,
      "loss": 0.5869,
      "step": 1810
    },
    {
      "epoch": 0.48532761624011794,
      "grad_norm": 0.7639933674359767,
      "learning_rate": 4.927633247425971e-06,
      "loss": 0.6242,
      "step": 1811
    },
    {
      "epoch": 0.48559560498459065,
      "grad_norm": 0.7659426539108167,
      "learning_rate": 4.927549043779415e-06,
      "loss": 0.5798,
      "step": 1812
    },
    {
      "epoch": 0.4858635937290634,
      "grad_norm": 0.6821316911423244,
      "learning_rate": 4.9274647918932815e-06,
      "loss": 0.6022,
      "step": 1813
    },
    {
      "epoch": 0.4861315824735361,
      "grad_norm": 0.6647163296668012,
      "learning_rate": 4.927380491769242e-06,
      "loss": 0.632,
      "step": 1814
    },
    {
      "epoch": 0.48639957121800886,
      "grad_norm": 0.7641316531543394,
      "learning_rate": 4.927296143408975e-06,
      "loss": 0.575,
      "step": 1815
    },
    {
      "epoch": 0.48666755996248157,
      "grad_norm": 0.7115528540810168,
      "learning_rate": 4.927211746814155e-06,
      "loss": 0.5795,
      "step": 1816
    },
    {
      "epoch": 0.4869355487069543,
      "grad_norm": 0.6772707338388667,
      "learning_rate": 4.927127301986458e-06,
      "loss": 0.6065,
      "step": 1817
    },
    {
      "epoch": 0.487203537451427,
      "grad_norm": 0.6923584774141827,
      "learning_rate": 4.927042808927563e-06,
      "loss": 0.6116,
      "step": 1818
    },
    {
      "epoch": 0.4874715261958998,
      "grad_norm": 0.6990220620942276,
      "learning_rate": 4.92695826763915e-06,
      "loss": 0.5853,
      "step": 1819
    },
    {
      "epoch": 0.4877395149403725,
      "grad_norm": 0.7463713687838336,
      "learning_rate": 4.9268736781228975e-06,
      "loss": 0.5878,
      "step": 1820
    },
    {
      "epoch": 0.48800750368484525,
      "grad_norm": 0.6923579452156734,
      "learning_rate": 4.926789040380487e-06,
      "loss": 0.59,
      "step": 1821
    },
    {
      "epoch": 0.48827549242931795,
      "grad_norm": 0.7300047934809715,
      "learning_rate": 4.926704354413602e-06,
      "loss": 0.5728,
      "step": 1822
    },
    {
      "epoch": 0.4885434811737907,
      "grad_norm": 0.7127338619876431,
      "learning_rate": 4.9266196202239225e-06,
      "loss": 0.5771,
      "step": 1823
    },
    {
      "epoch": 0.4888114699182634,
      "grad_norm": 0.7177699105106334,
      "learning_rate": 4.926534837813134e-06,
      "loss": 0.6124,
      "step": 1824
    },
    {
      "epoch": 0.48907945866273617,
      "grad_norm": 0.7094322431429576,
      "learning_rate": 4.926450007182922e-06,
      "loss": 0.6101,
      "step": 1825
    },
    {
      "epoch": 0.4893474474072089,
      "grad_norm": 0.657550154785613,
      "learning_rate": 4.92636512833497e-06,
      "loss": 0.5734,
      "step": 1826
    },
    {
      "epoch": 0.4896154361516816,
      "grad_norm": 0.6831889144554187,
      "learning_rate": 4.926280201270965e-06,
      "loss": 0.6334,
      "step": 1827
    },
    {
      "epoch": 0.4898834248961544,
      "grad_norm": 0.7095127427326774,
      "learning_rate": 4.9261952259925976e-06,
      "loss": 0.6143,
      "step": 1828
    },
    {
      "epoch": 0.4901514136406271,
      "grad_norm": 0.7247883792881268,
      "learning_rate": 4.9261102025015524e-06,
      "loss": 0.6152,
      "step": 1829
    },
    {
      "epoch": 0.49041940238509985,
      "grad_norm": 0.6554227352330229,
      "learning_rate": 4.926025130799522e-06,
      "loss": 0.5886,
      "step": 1830
    },
    {
      "epoch": 0.49068739112957255,
      "grad_norm": 0.6829905230800284,
      "learning_rate": 4.925940010888196e-06,
      "loss": 0.6363,
      "step": 1831
    },
    {
      "epoch": 0.4909553798740453,
      "grad_norm": 0.6937315311273836,
      "learning_rate": 4.925854842769265e-06,
      "loss": 0.5767,
      "step": 1832
    },
    {
      "epoch": 0.491223368618518,
      "grad_norm": 0.7300552415494079,
      "learning_rate": 4.925769626444423e-06,
      "loss": 0.5998,
      "step": 1833
    },
    {
      "epoch": 0.49149135736299077,
      "grad_norm": 0.7065759536778445,
      "learning_rate": 4.925684361915362e-06,
      "loss": 0.5806,
      "step": 1834
    },
    {
      "epoch": 0.49175934610746347,
      "grad_norm": 0.6529770098394563,
      "learning_rate": 4.925599049183777e-06,
      "loss": 0.5761,
      "step": 1835
    },
    {
      "epoch": 0.4920273348519362,
      "grad_norm": 0.6854605479543532,
      "learning_rate": 4.925513688251362e-06,
      "loss": 0.5694,
      "step": 1836
    },
    {
      "epoch": 0.49229532359640893,
      "grad_norm": 0.7098362171238946,
      "learning_rate": 4.925428279119816e-06,
      "loss": 0.5812,
      "step": 1837
    },
    {
      "epoch": 0.4925633123408817,
      "grad_norm": 0.6666969749624096,
      "learning_rate": 4.925342821790834e-06,
      "loss": 0.5472,
      "step": 1838
    },
    {
      "epoch": 0.4928313010853544,
      "grad_norm": 0.6710351087265205,
      "learning_rate": 4.9252573162661156e-06,
      "loss": 0.6271,
      "step": 1839
    },
    {
      "epoch": 0.49309928982982715,
      "grad_norm": 0.6871464963365448,
      "learning_rate": 4.925171762547358e-06,
      "loss": 0.62,
      "step": 1840
    },
    {
      "epoch": 0.4933672785742999,
      "grad_norm": 0.6749567138114979,
      "learning_rate": 4.9250861606362635e-06,
      "loss": 0.5999,
      "step": 1841
    },
    {
      "epoch": 0.4936352673187726,
      "grad_norm": 0.7447576673648232,
      "learning_rate": 4.925000510534532e-06,
      "loss": 0.5756,
      "step": 1842
    },
    {
      "epoch": 0.49390325606324537,
      "grad_norm": 0.6685927435382841,
      "learning_rate": 4.924914812243865e-06,
      "loss": 0.611,
      "step": 1843
    },
    {
      "epoch": 0.49417124480771807,
      "grad_norm": 0.7260572988375826,
      "learning_rate": 4.924829065765967e-06,
      "loss": 0.6153,
      "step": 1844
    },
    {
      "epoch": 0.4944392335521908,
      "grad_norm": 0.6726176218807591,
      "learning_rate": 4.92474327110254e-06,
      "loss": 0.5511,
      "step": 1845
    },
    {
      "epoch": 0.49470722229666353,
      "grad_norm": 0.6805875615265515,
      "learning_rate": 4.924657428255291e-06,
      "loss": 0.5763,
      "step": 1846
    },
    {
      "epoch": 0.4949752110411363,
      "grad_norm": 0.7508431081134296,
      "learning_rate": 4.924571537225925e-06,
      "loss": 0.5948,
      "step": 1847
    },
    {
      "epoch": 0.495243199785609,
      "grad_norm": 0.677709942734862,
      "learning_rate": 4.9244855980161475e-06,
      "loss": 0.5657,
      "step": 1848
    },
    {
      "epoch": 0.49551118853008175,
      "grad_norm": 0.6996995704922868,
      "learning_rate": 4.924399610627668e-06,
      "loss": 0.6352,
      "step": 1849
    },
    {
      "epoch": 0.49577917727455445,
      "grad_norm": 0.6864266480388194,
      "learning_rate": 4.9243135750621945e-06,
      "loss": 0.618,
      "step": 1850
    },
    {
      "epoch": 0.4960471660190272,
      "grad_norm": 0.6843457266186204,
      "learning_rate": 4.924227491321437e-06,
      "loss": 0.632,
      "step": 1851
    },
    {
      "epoch": 0.4963151547634999,
      "grad_norm": 0.6860839972127676,
      "learning_rate": 4.924141359407106e-06,
      "loss": 0.5834,
      "step": 1852
    },
    {
      "epoch": 0.49658314350797267,
      "grad_norm": 0.6825103788609804,
      "learning_rate": 4.924055179320912e-06,
      "loss": 0.5829,
      "step": 1853
    },
    {
      "epoch": 0.49685113225244537,
      "grad_norm": 0.7153850004469181,
      "learning_rate": 4.92396895106457e-06,
      "loss": 0.6118,
      "step": 1854
    },
    {
      "epoch": 0.49711912099691813,
      "grad_norm": 0.6691912747719649,
      "learning_rate": 4.923882674639791e-06,
      "loss": 0.6042,
      "step": 1855
    },
    {
      "epoch": 0.4973871097413909,
      "grad_norm": 0.6641093885741687,
      "learning_rate": 4.923796350048291e-06,
      "loss": 0.612,
      "step": 1856
    },
    {
      "epoch": 0.4976550984858636,
      "grad_norm": 0.7336438187689613,
      "learning_rate": 4.923709977291784e-06,
      "loss": 0.5751,
      "step": 1857
    },
    {
      "epoch": 0.49792308723033635,
      "grad_norm": 0.7141167962299675,
      "learning_rate": 4.923623556371989e-06,
      "loss": 0.6199,
      "step": 1858
    },
    {
      "epoch": 0.49819107597480905,
      "grad_norm": 0.6927804726256696,
      "learning_rate": 4.92353708729062e-06,
      "loss": 0.6178,
      "step": 1859
    },
    {
      "epoch": 0.4984590647192818,
      "grad_norm": 0.6970445513666474,
      "learning_rate": 4.923450570049398e-06,
      "loss": 0.5721,
      "step": 1860
    },
    {
      "epoch": 0.4987270534637545,
      "grad_norm": 0.6744299736102546,
      "learning_rate": 4.923364004650041e-06,
      "loss": 0.5751,
      "step": 1861
    },
    {
      "epoch": 0.49899504220822727,
      "grad_norm": 0.7101313089069801,
      "learning_rate": 4.923277391094269e-06,
      "loss": 0.6135,
      "step": 1862
    },
    {
      "epoch": 0.49926303095269997,
      "grad_norm": 0.6933510386198476,
      "learning_rate": 4.923190729383804e-06,
      "loss": 0.5782,
      "step": 1863
    },
    {
      "epoch": 0.49953101969717273,
      "grad_norm": 0.6538405403065393,
      "learning_rate": 4.923104019520367e-06,
      "loss": 0.5933,
      "step": 1864
    },
    {
      "epoch": 0.49979900844164543,
      "grad_norm": 0.6440490453694707,
      "learning_rate": 4.923017261505681e-06,
      "loss": 0.5681,
      "step": 1865
    },
    {
      "epoch": 0.5000669971861181,
      "grad_norm": 0.6716842986449055,
      "learning_rate": 4.922930455341473e-06,
      "loss": 0.5858,
      "step": 1866
    },
    {
      "epoch": 0.5003349859305909,
      "grad_norm": 0.6954993247203629,
      "learning_rate": 4.922843601029463e-06,
      "loss": 0.5731,
      "step": 1867
    },
    {
      "epoch": 0.5006029746750636,
      "grad_norm": 0.6632780823538208,
      "learning_rate": 4.922756698571381e-06,
      "loss": 0.6181,
      "step": 1868
    },
    {
      "epoch": 0.5008709634195364,
      "grad_norm": 0.6758264044280636,
      "learning_rate": 4.922669747968953e-06,
      "loss": 0.5871,
      "step": 1869
    },
    {
      "epoch": 0.5011389521640092,
      "grad_norm": 0.6733033328350713,
      "learning_rate": 4.9225827492239065e-06,
      "loss": 0.6075,
      "step": 1870
    },
    {
      "epoch": 0.5014069409084818,
      "grad_norm": 0.6837948821419588,
      "learning_rate": 4.92249570233797e-06,
      "loss": 0.5948,
      "step": 1871
    },
    {
      "epoch": 0.5016749296529546,
      "grad_norm": 0.7349506571421083,
      "learning_rate": 4.922408607312874e-06,
      "loss": 0.6191,
      "step": 1872
    },
    {
      "epoch": 0.5019429183974273,
      "grad_norm": 0.6772089257689896,
      "learning_rate": 4.922321464150348e-06,
      "loss": 0.6047,
      "step": 1873
    },
    {
      "epoch": 0.5022109071419001,
      "grad_norm": 0.6636552269641303,
      "learning_rate": 4.922234272852124e-06,
      "loss": 0.583,
      "step": 1874
    },
    {
      "epoch": 0.5024788958863727,
      "grad_norm": 0.6665599071236875,
      "learning_rate": 4.922147033419936e-06,
      "loss": 0.564,
      "step": 1875
    },
    {
      "epoch": 0.5027468846308455,
      "grad_norm": 0.6745289232474666,
      "learning_rate": 4.922059745855517e-06,
      "loss": 0.6086,
      "step": 1876
    },
    {
      "epoch": 0.5030148733753182,
      "grad_norm": 0.6853965606122696,
      "learning_rate": 4.9219724101606e-06,
      "loss": 0.6256,
      "step": 1877
    },
    {
      "epoch": 0.503282862119791,
      "grad_norm": 0.6641179380552773,
      "learning_rate": 4.921885026336923e-06,
      "loss": 0.5924,
      "step": 1878
    },
    {
      "epoch": 0.5035508508642637,
      "grad_norm": 0.6643787310401014,
      "learning_rate": 4.92179759438622e-06,
      "loss": 0.6158,
      "step": 1879
    },
    {
      "epoch": 0.5038188396087364,
      "grad_norm": 0.6852478110438123,
      "learning_rate": 4.921710114310231e-06,
      "loss": 0.6035,
      "step": 1880
    },
    {
      "epoch": 0.5040868283532092,
      "grad_norm": 0.6713337163863268,
      "learning_rate": 4.921622586110692e-06,
      "loss": 0.5759,
      "step": 1881
    },
    {
      "epoch": 0.5043548170976819,
      "grad_norm": 0.6948714420986007,
      "learning_rate": 4.921535009789344e-06,
      "loss": 0.5985,
      "step": 1882
    },
    {
      "epoch": 0.5046228058421546,
      "grad_norm": 0.6757881284045256,
      "learning_rate": 4.921447385347926e-06,
      "loss": 0.6254,
      "step": 1883
    },
    {
      "epoch": 0.5048907945866273,
      "grad_norm": 0.674640942203849,
      "learning_rate": 4.92135971278818e-06,
      "loss": 0.6593,
      "step": 1884
    },
    {
      "epoch": 0.5051587833311001,
      "grad_norm": 0.6774220886709643,
      "learning_rate": 4.921271992111849e-06,
      "loss": 0.6112,
      "step": 1885
    },
    {
      "epoch": 0.5054267720755728,
      "grad_norm": 0.6761793527356578,
      "learning_rate": 4.9211842233206744e-06,
      "loss": 0.6402,
      "step": 1886
    },
    {
      "epoch": 0.5056947608200456,
      "grad_norm": 0.6811225786139145,
      "learning_rate": 4.921096406416401e-06,
      "loss": 0.6023,
      "step": 1887
    },
    {
      "epoch": 0.5059627495645183,
      "grad_norm": 0.6847685290829604,
      "learning_rate": 4.921008541400774e-06,
      "loss": 0.5891,
      "step": 1888
    },
    {
      "epoch": 0.506230738308991,
      "grad_norm": 0.6690372378778577,
      "learning_rate": 4.92092062827554e-06,
      "loss": 0.6115,
      "step": 1889
    },
    {
      "epoch": 0.5064987270534638,
      "grad_norm": 0.6754289623903724,
      "learning_rate": 4.920832667042445e-06,
      "loss": 0.5746,
      "step": 1890
    },
    {
      "epoch": 0.5067667157979365,
      "grad_norm": 0.7028013312740907,
      "learning_rate": 4.920744657703238e-06,
      "loss": 0.5888,
      "step": 1891
    },
    {
      "epoch": 0.5070347045424092,
      "grad_norm": 0.7043086410425156,
      "learning_rate": 4.920656600259667e-06,
      "loss": 0.5858,
      "step": 1892
    },
    {
      "epoch": 0.5073026932868819,
      "grad_norm": 0.6861428896947537,
      "learning_rate": 4.920568494713482e-06,
      "loss": 0.6021,
      "step": 1893
    },
    {
      "epoch": 0.5075706820313547,
      "grad_norm": 0.6653368692261603,
      "learning_rate": 4.920480341066434e-06,
      "loss": 0.5984,
      "step": 1894
    },
    {
      "epoch": 0.5078386707758274,
      "grad_norm": 0.692537445414287,
      "learning_rate": 4.920392139320275e-06,
      "loss": 0.6214,
      "step": 1895
    },
    {
      "epoch": 0.5081066595203001,
      "grad_norm": 0.6904406682775183,
      "learning_rate": 4.920303889476758e-06,
      "loss": 0.6279,
      "step": 1896
    },
    {
      "epoch": 0.5083746482647729,
      "grad_norm": 0.6763469299643168,
      "learning_rate": 4.920215591537635e-06,
      "loss": 0.6111,
      "step": 1897
    },
    {
      "epoch": 0.5086426370092456,
      "grad_norm": 0.6699930089745397,
      "learning_rate": 4.920127245504662e-06,
      "loss": 0.6067,
      "step": 1898
    },
    {
      "epoch": 0.5089106257537184,
      "grad_norm": 0.7062329259243172,
      "learning_rate": 4.920038851379594e-06,
      "loss": 0.5922,
      "step": 1899
    },
    {
      "epoch": 0.5091786144981911,
      "grad_norm": 0.6884680316196266,
      "learning_rate": 4.919950409164189e-06,
      "loss": 0.629,
      "step": 1900
    },
    {
      "epoch": 0.5094466032426638,
      "grad_norm": 0.6931578333109122,
      "learning_rate": 4.919861918860203e-06,
      "loss": 0.5895,
      "step": 1901
    },
    {
      "epoch": 0.5097145919871365,
      "grad_norm": 0.6511789960152771,
      "learning_rate": 4.919773380469395e-06,
      "loss": 0.5812,
      "step": 1902
    },
    {
      "epoch": 0.5099825807316093,
      "grad_norm": 0.6526290574323884,
      "learning_rate": 4.919684793993523e-06,
      "loss": 0.6047,
      "step": 1903
    },
    {
      "epoch": 0.510250569476082,
      "grad_norm": 0.6528226784362516,
      "learning_rate": 4.91959615943435e-06,
      "loss": 0.5999,
      "step": 1904
    },
    {
      "epoch": 0.5105185582205547,
      "grad_norm": 0.6635509606148909,
      "learning_rate": 4.919507476793635e-06,
      "loss": 0.5769,
      "step": 1905
    },
    {
      "epoch": 0.5107865469650275,
      "grad_norm": 0.6787676090412086,
      "learning_rate": 4.919418746073141e-06,
      "loss": 0.5883,
      "step": 1906
    },
    {
      "epoch": 0.5110545357095002,
      "grad_norm": 0.6795717923436618,
      "learning_rate": 4.919329967274633e-06,
      "loss": 0.6113,
      "step": 1907
    },
    {
      "epoch": 0.511322524453973,
      "grad_norm": 0.6844577209921386,
      "learning_rate": 4.919241140399872e-06,
      "loss": 0.6027,
      "step": 1908
    },
    {
      "epoch": 0.5115905131984456,
      "grad_norm": 0.6670619189302658,
      "learning_rate": 4.919152265450626e-06,
      "loss": 0.5988,
      "step": 1909
    },
    {
      "epoch": 0.5118585019429184,
      "grad_norm": 0.6980371929650776,
      "learning_rate": 4.919063342428659e-06,
      "loss": 0.6173,
      "step": 1910
    },
    {
      "epoch": 0.5121264906873911,
      "grad_norm": 0.6820830845305791,
      "learning_rate": 4.9189743713357394e-06,
      "loss": 0.5897,
      "step": 1911
    },
    {
      "epoch": 0.5123944794318639,
      "grad_norm": 0.7268730077410498,
      "learning_rate": 4.918885352173635e-06,
      "loss": 0.5705,
      "step": 1912
    },
    {
      "epoch": 0.5126624681763365,
      "grad_norm": 0.6888184182996131,
      "learning_rate": 4.918796284944114e-06,
      "loss": 0.6177,
      "step": 1913
    },
    {
      "epoch": 0.5129304569208093,
      "grad_norm": 0.6846035988038659,
      "learning_rate": 4.918707169648947e-06,
      "loss": 0.5989,
      "step": 1914
    },
    {
      "epoch": 0.513198445665282,
      "grad_norm": 0.6716738578525039,
      "learning_rate": 4.918618006289905e-06,
      "loss": 0.611,
      "step": 1915
    },
    {
      "epoch": 0.5134664344097548,
      "grad_norm": 0.709793211089424,
      "learning_rate": 4.918528794868759e-06,
      "loss": 0.6071,
      "step": 1916
    },
    {
      "epoch": 0.5137344231542276,
      "grad_norm": 0.678119437471238,
      "learning_rate": 4.918439535387282e-06,
      "loss": 0.6084,
      "step": 1917
    },
    {
      "epoch": 0.5140024118987002,
      "grad_norm": 0.6747649899126351,
      "learning_rate": 4.918350227847248e-06,
      "loss": 0.5848,
      "step": 1918
    },
    {
      "epoch": 0.514270400643173,
      "grad_norm": 0.6641990769584981,
      "learning_rate": 4.918260872250433e-06,
      "loss": 0.5512,
      "step": 1919
    },
    {
      "epoch": 0.5145383893876457,
      "grad_norm": 0.6908615069602553,
      "learning_rate": 4.91817146859861e-06,
      "loss": 0.6032,
      "step": 1920
    },
    {
      "epoch": 0.5148063781321185,
      "grad_norm": 0.7174389604946939,
      "learning_rate": 4.918082016893558e-06,
      "loss": 0.6262,
      "step": 1921
    },
    {
      "epoch": 0.5150743668765911,
      "grad_norm": 0.6826566138757871,
      "learning_rate": 4.917992517137053e-06,
      "loss": 0.6238,
      "step": 1922
    },
    {
      "epoch": 0.5153423556210639,
      "grad_norm": 0.6986926354015428,
      "learning_rate": 4.917902969330875e-06,
      "loss": 0.5926,
      "step": 1923
    },
    {
      "epoch": 0.5156103443655367,
      "grad_norm": 0.6793795340695832,
      "learning_rate": 4.9178133734768015e-06,
      "loss": 0.6226,
      "step": 1924
    },
    {
      "epoch": 0.5158783331100094,
      "grad_norm": 0.7217137437204021,
      "learning_rate": 4.917723729576615e-06,
      "loss": 0.6378,
      "step": 1925
    },
    {
      "epoch": 0.5161463218544821,
      "grad_norm": 0.6874839597170429,
      "learning_rate": 4.917634037632095e-06,
      "loss": 0.5779,
      "step": 1926
    },
    {
      "epoch": 0.5164143105989548,
      "grad_norm": 0.7121225000848759,
      "learning_rate": 4.917544297645024e-06,
      "loss": 0.6295,
      "step": 1927
    },
    {
      "epoch": 0.5166822993434276,
      "grad_norm": 0.7151579438617389,
      "learning_rate": 4.917454509617187e-06,
      "loss": 0.5805,
      "step": 1928
    },
    {
      "epoch": 0.5169502880879003,
      "grad_norm": 0.667724348271962,
      "learning_rate": 4.917364673550367e-06,
      "loss": 0.5768,
      "step": 1929
    },
    {
      "epoch": 0.5172182768323731,
      "grad_norm": 0.6680216285827486,
      "learning_rate": 4.917274789446349e-06,
      "loss": 0.5943,
      "step": 1930
    },
    {
      "epoch": 0.5174862655768457,
      "grad_norm": 0.6869008111384549,
      "learning_rate": 4.91718485730692e-06,
      "loss": 0.5588,
      "step": 1931
    },
    {
      "epoch": 0.5177542543213185,
      "grad_norm": 0.6880918012566767,
      "learning_rate": 4.9170948771338675e-06,
      "loss": 0.6086,
      "step": 1932
    },
    {
      "epoch": 0.5180222430657913,
      "grad_norm": 0.7590773420519474,
      "learning_rate": 4.917004848928978e-06,
      "loss": 0.5946,
      "step": 1933
    },
    {
      "epoch": 0.518290231810264,
      "grad_norm": 0.6836555553721839,
      "learning_rate": 4.91691477269404e-06,
      "loss": 0.5871,
      "step": 1934
    },
    {
      "epoch": 0.5185582205547367,
      "grad_norm": 0.6814185298544256,
      "learning_rate": 4.916824648430846e-06,
      "loss": 0.6044,
      "step": 1935
    },
    {
      "epoch": 0.5188262092992094,
      "grad_norm": 0.6773423287153642,
      "learning_rate": 4.916734476141186e-06,
      "loss": 0.6019,
      "step": 1936
    },
    {
      "epoch": 0.5190941980436822,
      "grad_norm": 0.7194414573401141,
      "learning_rate": 4.9166442558268505e-06,
      "loss": 0.5762,
      "step": 1937
    },
    {
      "epoch": 0.5193621867881549,
      "grad_norm": 0.6972874573880838,
      "learning_rate": 4.916553987489634e-06,
      "loss": 0.6136,
      "step": 1938
    },
    {
      "epoch": 0.5196301755326276,
      "grad_norm": 0.6773522213598032,
      "learning_rate": 4.916463671131329e-06,
      "loss": 0.5984,
      "step": 1939
    },
    {
      "epoch": 0.5198981642771003,
      "grad_norm": 0.7266890150923501,
      "learning_rate": 4.916373306753732e-06,
      "loss": 0.6078,
      "step": 1940
    },
    {
      "epoch": 0.5201661530215731,
      "grad_norm": 0.699399349633916,
      "learning_rate": 4.916282894358637e-06,
      "loss": 0.6088,
      "step": 1941
    },
    {
      "epoch": 0.5204341417660459,
      "grad_norm": 0.7027530346904746,
      "learning_rate": 4.9161924339478405e-06,
      "loss": 0.5928,
      "step": 1942
    },
    {
      "epoch": 0.5207021305105186,
      "grad_norm": 0.6693803024347358,
      "learning_rate": 4.9161019255231414e-06,
      "loss": 0.5723,
      "step": 1943
    },
    {
      "epoch": 0.5209701192549913,
      "grad_norm": 0.6846988974348753,
      "learning_rate": 4.916011369086337e-06,
      "loss": 0.6297,
      "step": 1944
    },
    {
      "epoch": 0.521238107999464,
      "grad_norm": 0.6866009725807293,
      "learning_rate": 4.91592076463923e-06,
      "loss": 0.6266,
      "step": 1945
    },
    {
      "epoch": 0.5215060967439368,
      "grad_norm": 0.6422312014671561,
      "learning_rate": 4.915830112183615e-06,
      "loss": 0.6003,
      "step": 1946
    },
    {
      "epoch": 0.5217740854884095,
      "grad_norm": 0.7518495521790985,
      "learning_rate": 4.915739411721299e-06,
      "loss": 0.5661,
      "step": 1947
    },
    {
      "epoch": 0.5220420742328822,
      "grad_norm": 0.6663590259680006,
      "learning_rate": 4.915648663254081e-06,
      "loss": 0.5536,
      "step": 1948
    },
    {
      "epoch": 0.5223100629773549,
      "grad_norm": 0.667336677563985,
      "learning_rate": 4.915557866783766e-06,
      "loss": 0.6104,
      "step": 1949
    },
    {
      "epoch": 0.5225780517218277,
      "grad_norm": 0.6802720959420188,
      "learning_rate": 4.915467022312158e-06,
      "loss": 0.6041,
      "step": 1950
    },
    {
      "epoch": 0.5228460404663005,
      "grad_norm": 0.7014396095712542,
      "learning_rate": 4.915376129841062e-06,
      "loss": 0.6152,
      "step": 1951
    },
    {
      "epoch": 0.5231140292107731,
      "grad_norm": 0.689425510893113,
      "learning_rate": 4.915285189372283e-06,
      "loss": 0.6308,
      "step": 1952
    },
    {
      "epoch": 0.5233820179552459,
      "grad_norm": 0.6764430106013337,
      "learning_rate": 4.915194200907631e-06,
      "loss": 0.6252,
      "step": 1953
    },
    {
      "epoch": 0.5236500066997186,
      "grad_norm": 0.671855684792156,
      "learning_rate": 4.915103164448911e-06,
      "loss": 0.6042,
      "step": 1954
    },
    {
      "epoch": 0.5239179954441914,
      "grad_norm": 0.6703542029898869,
      "learning_rate": 4.915012079997934e-06,
      "loss": 0.6187,
      "step": 1955
    },
    {
      "epoch": 0.524185984188664,
      "grad_norm": 0.6775010746340953,
      "learning_rate": 4.91492094755651e-06,
      "loss": 0.5923,
      "step": 1956
    },
    {
      "epoch": 0.5244539729331368,
      "grad_norm": 0.6770466359335293,
      "learning_rate": 4.9148297671264485e-06,
      "loss": 0.5897,
      "step": 1957
    },
    {
      "epoch": 0.5247219616776095,
      "grad_norm": 0.6515469840544164,
      "learning_rate": 4.914738538709563e-06,
      "loss": 0.5731,
      "step": 1958
    },
    {
      "epoch": 0.5249899504220823,
      "grad_norm": 0.671556250225073,
      "learning_rate": 4.914647262307666e-06,
      "loss": 0.5918,
      "step": 1959
    },
    {
      "epoch": 0.525257939166555,
      "grad_norm": 0.6567505569600476,
      "learning_rate": 4.914555937922571e-06,
      "loss": 0.5964,
      "step": 1960
    },
    {
      "epoch": 0.5255259279110277,
      "grad_norm": 0.6986417202244178,
      "learning_rate": 4.914464565556093e-06,
      "loss": 0.6216,
      "step": 1961
    },
    {
      "epoch": 0.5257939166555005,
      "grad_norm": 0.7197168898472559,
      "learning_rate": 4.914373145210047e-06,
      "loss": 0.5902,
      "step": 1962
    },
    {
      "epoch": 0.5260619053999732,
      "grad_norm": 0.6938459141055117,
      "learning_rate": 4.91428167688625e-06,
      "loss": 0.5929,
      "step": 1963
    },
    {
      "epoch": 0.526329894144446,
      "grad_norm": 0.6819873825525299,
      "learning_rate": 4.9141901605865205e-06,
      "loss": 0.6234,
      "step": 1964
    },
    {
      "epoch": 0.5265978828889186,
      "grad_norm": 0.6559263194057094,
      "learning_rate": 4.914098596312676e-06,
      "loss": 0.6226,
      "step": 1965
    },
    {
      "epoch": 0.5268658716333914,
      "grad_norm": 0.6760873442821377,
      "learning_rate": 4.914006984066536e-06,
      "loss": 0.6321,
      "step": 1966
    },
    {
      "epoch": 0.5271338603778641,
      "grad_norm": 0.671660869361851,
      "learning_rate": 4.9139153238499225e-06,
      "loss": 0.5995,
      "step": 1967
    },
    {
      "epoch": 0.5274018491223369,
      "grad_norm": 0.6591609408980788,
      "learning_rate": 4.913823615664656e-06,
      "loss": 0.608,
      "step": 1968
    },
    {
      "epoch": 0.5276698378668095,
      "grad_norm": 0.6505304994195519,
      "learning_rate": 4.913731859512558e-06,
      "loss": 0.5908,
      "step": 1969
    },
    {
      "epoch": 0.5279378266112823,
      "grad_norm": 0.6597692401453974,
      "learning_rate": 4.9136400553954526e-06,
      "loss": 0.6103,
      "step": 1970
    },
    {
      "epoch": 0.5282058153557551,
      "grad_norm": 0.6862886310142831,
      "learning_rate": 4.913548203315165e-06,
      "loss": 0.5877,
      "step": 1971
    },
    {
      "epoch": 0.5284738041002278,
      "grad_norm": 0.6973638922078622,
      "learning_rate": 4.913456303273518e-06,
      "loss": 0.6125,
      "step": 1972
    },
    {
      "epoch": 0.5287417928447006,
      "grad_norm": 0.6549506431258372,
      "learning_rate": 4.913364355272341e-06,
      "loss": 0.5847,
      "step": 1973
    },
    {
      "epoch": 0.5290097815891732,
      "grad_norm": 0.6615389921677716,
      "learning_rate": 4.913272359313459e-06,
      "loss": 0.6288,
      "step": 1974
    },
    {
      "epoch": 0.529277770333646,
      "grad_norm": 0.683355522523139,
      "learning_rate": 4.9131803153987015e-06,
      "loss": 0.6046,
      "step": 1975
    },
    {
      "epoch": 0.5295457590781187,
      "grad_norm": 0.6393543579855859,
      "learning_rate": 4.913088223529896e-06,
      "loss": 0.6026,
      "step": 1976
    },
    {
      "epoch": 0.5298137478225915,
      "grad_norm": 0.6694018033296458,
      "learning_rate": 4.912996083708874e-06,
      "loss": 0.6242,
      "step": 1977
    },
    {
      "epoch": 0.5300817365670641,
      "grad_norm": 0.6662733646339505,
      "learning_rate": 4.912903895937465e-06,
      "loss": 0.5775,
      "step": 1978
    },
    {
      "epoch": 0.5303497253115369,
      "grad_norm": 0.6876087566188026,
      "learning_rate": 4.912811660217502e-06,
      "loss": 0.5945,
      "step": 1979
    },
    {
      "epoch": 0.5306177140560097,
      "grad_norm": 0.6647897702141633,
      "learning_rate": 4.912719376550818e-06,
      "loss": 0.5828,
      "step": 1980
    },
    {
      "epoch": 0.5308857028004824,
      "grad_norm": 0.6939441011346367,
      "learning_rate": 4.9126270449392465e-06,
      "loss": 0.5982,
      "step": 1981
    },
    {
      "epoch": 0.5311536915449551,
      "grad_norm": 0.69340470816096,
      "learning_rate": 4.912534665384621e-06,
      "loss": 0.5967,
      "step": 1982
    },
    {
      "epoch": 0.5314216802894278,
      "grad_norm": 0.709151998306196,
      "learning_rate": 4.91244223788878e-06,
      "loss": 0.6157,
      "step": 1983
    },
    {
      "epoch": 0.5316896690339006,
      "grad_norm": 0.6707055196442323,
      "learning_rate": 4.9123497624535585e-06,
      "loss": 0.6001,
      "step": 1984
    },
    {
      "epoch": 0.5319576577783733,
      "grad_norm": 0.6859174419987475,
      "learning_rate": 4.912257239080793e-06,
      "loss": 0.625,
      "step": 1985
    },
    {
      "epoch": 0.532225646522846,
      "grad_norm": 0.6806178282267489,
      "learning_rate": 4.912164667772325e-06,
      "loss": 0.5857,
      "step": 1986
    },
    {
      "epoch": 0.5324936352673187,
      "grad_norm": 0.6829215208636135,
      "learning_rate": 4.912072048529992e-06,
      "loss": 0.5861,
      "step": 1987
    },
    {
      "epoch": 0.5327616240117915,
      "grad_norm": 0.6524370819495954,
      "learning_rate": 4.911979381355635e-06,
      "loss": 0.5746,
      "step": 1988
    },
    {
      "epoch": 0.5330296127562643,
      "grad_norm": 0.6757503179942262,
      "learning_rate": 4.911886666251095e-06,
      "loss": 0.6075,
      "step": 1989
    },
    {
      "epoch": 0.533297601500737,
      "grad_norm": 0.6685608242836899,
      "learning_rate": 4.911793903218215e-06,
      "loss": 0.5978,
      "step": 1990
    },
    {
      "epoch": 0.5335655902452097,
      "grad_norm": 0.6718087125676748,
      "learning_rate": 4.911701092258838e-06,
      "loss": 0.6008,
      "step": 1991
    },
    {
      "epoch": 0.5338335789896824,
      "grad_norm": 0.7102056657510527,
      "learning_rate": 4.91160823337481e-06,
      "loss": 0.615,
      "step": 1992
    },
    {
      "epoch": 0.5341015677341552,
      "grad_norm": 0.6666699710669354,
      "learning_rate": 4.911515326567973e-06,
      "loss": 0.5688,
      "step": 1993
    },
    {
      "epoch": 0.5343695564786279,
      "grad_norm": 0.6663116689687447,
      "learning_rate": 4.911422371840175e-06,
      "loss": 0.5717,
      "step": 1994
    },
    {
      "epoch": 0.5346375452231006,
      "grad_norm": 0.6629966625867514,
      "learning_rate": 4.911329369193264e-06,
      "loss": 0.6036,
      "step": 1995
    },
    {
      "epoch": 0.5349055339675733,
      "grad_norm": 0.7179509595939403,
      "learning_rate": 4.911236318629087e-06,
      "loss": 0.6082,
      "step": 1996
    },
    {
      "epoch": 0.5351735227120461,
      "grad_norm": 0.707852832014802,
      "learning_rate": 4.911143220149494e-06,
      "loss": 0.6168,
      "step": 1997
    },
    {
      "epoch": 0.5354415114565189,
      "grad_norm": 0.6777755811651912,
      "learning_rate": 4.911050073756333e-06,
      "loss": 0.6185,
      "step": 1998
    },
    {
      "epoch": 0.5357095002009915,
      "grad_norm": 0.6682476173931508,
      "learning_rate": 4.910956879451458e-06,
      "loss": 0.5592,
      "step": 1999
    },
    {
      "epoch": 0.5359774889454643,
      "grad_norm": 0.6633784636174194,
      "learning_rate": 4.910863637236718e-06,
      "loss": 0.586,
      "step": 2000
    },
    {
      "epoch": 0.536245477689937,
      "grad_norm": 0.6636363566366982,
      "learning_rate": 4.910770347113967e-06,
      "loss": 0.6245,
      "step": 2001
    },
    {
      "epoch": 0.5365134664344098,
      "grad_norm": 0.6718699534638887,
      "learning_rate": 4.91067700908506e-06,
      "loss": 0.5948,
      "step": 2002
    },
    {
      "epoch": 0.5367814551788825,
      "grad_norm": 0.6638979707379371,
      "learning_rate": 4.910583623151851e-06,
      "loss": 0.5746,
      "step": 2003
    },
    {
      "epoch": 0.5370494439233552,
      "grad_norm": 0.6723416229144853,
      "learning_rate": 4.9104901893161946e-06,
      "loss": 0.586,
      "step": 2004
    },
    {
      "epoch": 0.5373174326678279,
      "grad_norm": 0.6630066037529619,
      "learning_rate": 4.910396707579949e-06,
      "loss": 0.5762,
      "step": 2005
    },
    {
      "epoch": 0.5375854214123007,
      "grad_norm": 0.6793395203155606,
      "learning_rate": 4.9103031779449705e-06,
      "loss": 0.5882,
      "step": 2006
    },
    {
      "epoch": 0.5378534101567735,
      "grad_norm": 0.7033167336006408,
      "learning_rate": 4.910209600413119e-06,
      "loss": 0.5902,
      "step": 2007
    },
    {
      "epoch": 0.5381213989012461,
      "grad_norm": 0.6899351667644836,
      "learning_rate": 4.910115974986254e-06,
      "loss": 0.5918,
      "step": 2008
    },
    {
      "epoch": 0.5383893876457189,
      "grad_norm": 0.6702461561923475,
      "learning_rate": 4.910022301666235e-06,
      "loss": 0.5903,
      "step": 2009
    },
    {
      "epoch": 0.5386573763901916,
      "grad_norm": 0.6794694276911194,
      "learning_rate": 4.9099285804549246e-06,
      "loss": 0.594,
      "step": 2010
    },
    {
      "epoch": 0.5389253651346644,
      "grad_norm": 0.665118603764784,
      "learning_rate": 4.909834811354184e-06,
      "loss": 0.6072,
      "step": 2011
    },
    {
      "epoch": 0.539193353879137,
      "grad_norm": 0.6412286243898685,
      "learning_rate": 4.909740994365877e-06,
      "loss": 0.5686,
      "step": 2012
    },
    {
      "epoch": 0.5394613426236098,
      "grad_norm": 0.6933867012022309,
      "learning_rate": 4.909647129491868e-06,
      "loss": 0.5995,
      "step": 2013
    },
    {
      "epoch": 0.5397293313680825,
      "grad_norm": 0.6559861084090948,
      "learning_rate": 4.909553216734024e-06,
      "loss": 0.6136,
      "step": 2014
    },
    {
      "epoch": 0.5399973201125553,
      "grad_norm": 0.6688057590952757,
      "learning_rate": 4.909459256094208e-06,
      "loss": 0.5782,
      "step": 2015
    },
    {
      "epoch": 0.5402653088570281,
      "grad_norm": 0.6595041889980257,
      "learning_rate": 4.909365247574288e-06,
      "loss": 0.564,
      "step": 2016
    },
    {
      "epoch": 0.5405332976015007,
      "grad_norm": 0.6781731582949737,
      "learning_rate": 4.909271191176134e-06,
      "loss": 0.615,
      "step": 2017
    },
    {
      "epoch": 0.5408012863459735,
      "grad_norm": 0.6777978924027859,
      "learning_rate": 4.909177086901613e-06,
      "loss": 0.5658,
      "step": 2018
    },
    {
      "epoch": 0.5410692750904462,
      "grad_norm": 0.6844617850896818,
      "learning_rate": 4.909082934752596e-06,
      "loss": 0.6165,
      "step": 2019
    },
    {
      "epoch": 0.541337263834919,
      "grad_norm": 0.6832680311745681,
      "learning_rate": 4.908988734730954e-06,
      "loss": 0.5978,
      "step": 2020
    },
    {
      "epoch": 0.5416052525793916,
      "grad_norm": 0.7182663202410363,
      "learning_rate": 4.9088944868385595e-06,
      "loss": 0.6015,
      "step": 2021
    },
    {
      "epoch": 0.5418732413238644,
      "grad_norm": 0.6973314093942917,
      "learning_rate": 4.908800191077283e-06,
      "loss": 0.5706,
      "step": 2022
    },
    {
      "epoch": 0.5421412300683371,
      "grad_norm": 0.6718900175479703,
      "learning_rate": 4.908705847449001e-06,
      "loss": 0.5577,
      "step": 2023
    },
    {
      "epoch": 0.5424092188128099,
      "grad_norm": 0.6760920003668903,
      "learning_rate": 4.908611455955587e-06,
      "loss": 0.6049,
      "step": 2024
    },
    {
      "epoch": 0.5426772075572825,
      "grad_norm": 0.6914775241403285,
      "learning_rate": 4.908517016598916e-06,
      "loss": 0.6052,
      "step": 2025
    },
    {
      "epoch": 0.5429451963017553,
      "grad_norm": 0.6806327201103145,
      "learning_rate": 4.908422529380867e-06,
      "loss": 0.6381,
      "step": 2026
    },
    {
      "epoch": 0.5432131850462281,
      "grad_norm": 0.6701923323298916,
      "learning_rate": 4.9083279943033155e-06,
      "loss": 0.5953,
      "step": 2027
    },
    {
      "epoch": 0.5434811737907008,
      "grad_norm": 0.6967363666358184,
      "learning_rate": 4.908233411368141e-06,
      "loss": 0.5786,
      "step": 2028
    },
    {
      "epoch": 0.5437491625351735,
      "grad_norm": 0.6589579817945047,
      "learning_rate": 4.908138780577223e-06,
      "loss": 0.5886,
      "step": 2029
    },
    {
      "epoch": 0.5440171512796462,
      "grad_norm": 0.6614923530911566,
      "learning_rate": 4.908044101932441e-06,
      "loss": 0.5957,
      "step": 2030
    },
    {
      "epoch": 0.544285140024119,
      "grad_norm": 0.6361627152401645,
      "learning_rate": 4.9079493754356774e-06,
      "loss": 0.5617,
      "step": 2031
    },
    {
      "epoch": 0.5445531287685917,
      "grad_norm": 0.6953556868319949,
      "learning_rate": 4.907854601088815e-06,
      "loss": 0.6116,
      "step": 2032
    },
    {
      "epoch": 0.5448211175130645,
      "grad_norm": 0.6525109706521365,
      "learning_rate": 4.907759778893736e-06,
      "loss": 0.6083,
      "step": 2033
    },
    {
      "epoch": 0.5450891062575371,
      "grad_norm": 0.6928985269954416,
      "learning_rate": 4.907664908852325e-06,
      "loss": 0.6023,
      "step": 2034
    },
    {
      "epoch": 0.5453570950020099,
      "grad_norm": 0.7334534016139918,
      "learning_rate": 4.907569990966468e-06,
      "loss": 0.6009,
      "step": 2035
    },
    {
      "epoch": 0.5456250837464827,
      "grad_norm": 0.6532938719673854,
      "learning_rate": 4.907475025238051e-06,
      "loss": 0.5892,
      "step": 2036
    },
    {
      "epoch": 0.5458930724909554,
      "grad_norm": 0.677640024383669,
      "learning_rate": 4.90738001166896e-06,
      "loss": 0.5796,
      "step": 2037
    },
    {
      "epoch": 0.5461610612354281,
      "grad_norm": 0.6714107855293038,
      "learning_rate": 4.907284950261084e-06,
      "loss": 0.5613,
      "step": 2038
    },
    {
      "epoch": 0.5464290499799008,
      "grad_norm": 0.669964065081849,
      "learning_rate": 4.907189841016311e-06,
      "loss": 0.6134,
      "step": 2039
    },
    {
      "epoch": 0.5466970387243736,
      "grad_norm": 0.658309798355565,
      "learning_rate": 4.907094683936533e-06,
      "loss": 0.5867,
      "step": 2040
    },
    {
      "epoch": 0.5469650274688463,
      "grad_norm": 0.6627900654308646,
      "learning_rate": 4.90699947902364e-06,
      "loss": 0.6109,
      "step": 2041
    },
    {
      "epoch": 0.547233016213319,
      "grad_norm": 0.6676879708132591,
      "learning_rate": 4.906904226279523e-06,
      "loss": 0.5987,
      "step": 2042
    },
    {
      "epoch": 0.5475010049577917,
      "grad_norm": 0.6865944129120005,
      "learning_rate": 4.906808925706075e-06,
      "loss": 0.6159,
      "step": 2043
    },
    {
      "epoch": 0.5477689937022645,
      "grad_norm": 0.6686496189320613,
      "learning_rate": 4.9067135773051915e-06,
      "loss": 0.6156,
      "step": 2044
    },
    {
      "epoch": 0.5480369824467373,
      "grad_norm": 0.6652678320227149,
      "learning_rate": 4.906618181078765e-06,
      "loss": 0.5917,
      "step": 2045
    },
    {
      "epoch": 0.54830497119121,
      "grad_norm": 0.6598853147360066,
      "learning_rate": 4.9065227370286926e-06,
      "loss": 0.5871,
      "step": 2046
    },
    {
      "epoch": 0.5485729599356827,
      "grad_norm": 0.6914443485523798,
      "learning_rate": 4.90642724515687e-06,
      "loss": 0.6245,
      "step": 2047
    },
    {
      "epoch": 0.5488409486801554,
      "grad_norm": 0.6555162702345201,
      "learning_rate": 4.906331705465197e-06,
      "loss": 0.5694,
      "step": 2048
    },
    {
      "epoch": 0.5491089374246282,
      "grad_norm": 0.6609844321515436,
      "learning_rate": 4.906236117955569e-06,
      "loss": 0.5656,
      "step": 2049
    },
    {
      "epoch": 0.5493769261691009,
      "grad_norm": 0.6761198432951011,
      "learning_rate": 4.906140482629887e-06,
      "loss": 0.5851,
      "step": 2050
    },
    {
      "epoch": 0.5496449149135736,
      "grad_norm": 0.6588309317506655,
      "learning_rate": 4.906044799490052e-06,
      "loss": 0.5999,
      "step": 2051
    },
    {
      "epoch": 0.5499129036580463,
      "grad_norm": 0.6538041712169054,
      "learning_rate": 4.905949068537965e-06,
      "loss": 0.583,
      "step": 2052
    },
    {
      "epoch": 0.5501808924025191,
      "grad_norm": 0.6768294943392315,
      "learning_rate": 4.9058532897755275e-06,
      "loss": 0.5938,
      "step": 2053
    },
    {
      "epoch": 0.5504488811469919,
      "grad_norm": 0.6736611201304477,
      "learning_rate": 4.905757463204643e-06,
      "loss": 0.6261,
      "step": 2054
    },
    {
      "epoch": 0.5507168698914645,
      "grad_norm": 0.6699491049477648,
      "learning_rate": 4.905661588827218e-06,
      "loss": 0.6063,
      "step": 2055
    },
    {
      "epoch": 0.5509848586359373,
      "grad_norm": 0.6542814945270636,
      "learning_rate": 4.9055656666451535e-06,
      "loss": 0.5694,
      "step": 2056
    },
    {
      "epoch": 0.55125284738041,
      "grad_norm": 0.6492328318678013,
      "learning_rate": 4.90546969666036e-06,
      "loss": 0.6243,
      "step": 2057
    },
    {
      "epoch": 0.5515208361248828,
      "grad_norm": 0.695514070391006,
      "learning_rate": 4.905373678874741e-06,
      "loss": 0.6014,
      "step": 2058
    },
    {
      "epoch": 0.5517888248693554,
      "grad_norm": 0.6627386527294605,
      "learning_rate": 4.905277613290207e-06,
      "loss": 0.6066,
      "step": 2059
    },
    {
      "epoch": 0.5520568136138282,
      "grad_norm": 0.692466511995778,
      "learning_rate": 4.905181499908666e-06,
      "loss": 0.6109,
      "step": 2060
    },
    {
      "epoch": 0.552324802358301,
      "grad_norm": 0.6540144021060108,
      "learning_rate": 4.905085338732028e-06,
      "loss": 0.5832,
      "step": 2061
    },
    {
      "epoch": 0.5525927911027737,
      "grad_norm": 0.653362347569939,
      "learning_rate": 4.904989129762203e-06,
      "loss": 0.5855,
      "step": 2062
    },
    {
      "epoch": 0.5528607798472465,
      "grad_norm": 0.6436548565203599,
      "learning_rate": 4.904892873001105e-06,
      "loss": 0.5872,
      "step": 2063
    },
    {
      "epoch": 0.5531287685917191,
      "grad_norm": 0.6491028577848617,
      "learning_rate": 4.904796568450645e-06,
      "loss": 0.5995,
      "step": 2064
    },
    {
      "epoch": 0.5533967573361919,
      "grad_norm": 0.6875930975304434,
      "learning_rate": 4.904700216112738e-06,
      "loss": 0.5709,
      "step": 2065
    },
    {
      "epoch": 0.5536647460806646,
      "grad_norm": 0.6717386925470101,
      "learning_rate": 4.904603815989297e-06,
      "loss": 0.6007,
      "step": 2066
    },
    {
      "epoch": 0.5539327348251374,
      "grad_norm": 0.6692985885464802,
      "learning_rate": 4.904507368082238e-06,
      "loss": 0.6068,
      "step": 2067
    },
    {
      "epoch": 0.55420072356961,
      "grad_norm": 0.679087220347972,
      "learning_rate": 4.90441087239348e-06,
      "loss": 0.6173,
      "step": 2068
    },
    {
      "epoch": 0.5544687123140828,
      "grad_norm": 0.6615168596495721,
      "learning_rate": 4.904314328924938e-06,
      "loss": 0.6174,
      "step": 2069
    },
    {
      "epoch": 0.5547367010585555,
      "grad_norm": 0.657541220947205,
      "learning_rate": 4.904217737678531e-06,
      "loss": 0.5659,
      "step": 2070
    },
    {
      "epoch": 0.5550046898030283,
      "grad_norm": 0.648843237672258,
      "learning_rate": 4.904121098656179e-06,
      "loss": 0.5889,
      "step": 2071
    },
    {
      "epoch": 0.555272678547501,
      "grad_norm": 0.6625034201559062,
      "learning_rate": 4.904024411859802e-06,
      "loss": 0.5851,
      "step": 2072
    },
    {
      "epoch": 0.5555406672919737,
      "grad_norm": 0.6609046633590937,
      "learning_rate": 4.903927677291321e-06,
      "loss": 0.594,
      "step": 2073
    },
    {
      "epoch": 0.5558086560364465,
      "grad_norm": 0.6734959337719271,
      "learning_rate": 4.903830894952659e-06,
      "loss": 0.5906,
      "step": 2074
    },
    {
      "epoch": 0.5560766447809192,
      "grad_norm": 0.6970385882308603,
      "learning_rate": 4.903734064845739e-06,
      "loss": 0.5995,
      "step": 2075
    },
    {
      "epoch": 0.556344633525392,
      "grad_norm": 0.673893032641133,
      "learning_rate": 4.903637186972484e-06,
      "loss": 0.618,
      "step": 2076
    },
    {
      "epoch": 0.5566126222698646,
      "grad_norm": 0.6838632892185236,
      "learning_rate": 4.903540261334822e-06,
      "loss": 0.5796,
      "step": 2077
    },
    {
      "epoch": 0.5568806110143374,
      "grad_norm": 0.6725244627499904,
      "learning_rate": 4.903443287934676e-06,
      "loss": 0.6049,
      "step": 2078
    },
    {
      "epoch": 0.5571485997588101,
      "grad_norm": 0.6701263458239054,
      "learning_rate": 4.903346266773974e-06,
      "loss": 0.6148,
      "step": 2079
    },
    {
      "epoch": 0.5574165885032829,
      "grad_norm": 0.6916762224924238,
      "learning_rate": 4.903249197854645e-06,
      "loss": 0.5912,
      "step": 2080
    },
    {
      "epoch": 0.5576845772477556,
      "grad_norm": 0.6875189124362431,
      "learning_rate": 4.903152081178617e-06,
      "loss": 0.6118,
      "step": 2081
    },
    {
      "epoch": 0.5579525659922283,
      "grad_norm": 0.6622944787782153,
      "learning_rate": 4.90305491674782e-06,
      "loss": 0.6123,
      "step": 2082
    },
    {
      "epoch": 0.5582205547367011,
      "grad_norm": 0.6954032994361642,
      "learning_rate": 4.902957704564185e-06,
      "loss": 0.5812,
      "step": 2083
    },
    {
      "epoch": 0.5584885434811738,
      "grad_norm": 0.6608009744430056,
      "learning_rate": 4.902860444629644e-06,
      "loss": 0.6017,
      "step": 2084
    },
    {
      "epoch": 0.5587565322256465,
      "grad_norm": 0.7041197713396986,
      "learning_rate": 4.902763136946129e-06,
      "loss": 0.5908,
      "step": 2085
    },
    {
      "epoch": 0.5590245209701192,
      "grad_norm": 0.6953248714459597,
      "learning_rate": 4.9026657815155745e-06,
      "loss": 0.5757,
      "step": 2086
    },
    {
      "epoch": 0.559292509714592,
      "grad_norm": 0.6777135218605428,
      "learning_rate": 4.902568378339914e-06,
      "loss": 0.5812,
      "step": 2087
    },
    {
      "epoch": 0.5595604984590647,
      "grad_norm": 0.6748489190146864,
      "learning_rate": 4.902470927421085e-06,
      "loss": 0.6183,
      "step": 2088
    },
    {
      "epoch": 0.5598284872035375,
      "grad_norm": 0.6570058972192931,
      "learning_rate": 4.9023734287610214e-06,
      "loss": 0.6124,
      "step": 2089
    },
    {
      "epoch": 0.5600964759480102,
      "grad_norm": 0.68977994644619,
      "learning_rate": 4.902275882361662e-06,
      "loss": 0.6075,
      "step": 2090
    },
    {
      "epoch": 0.5603644646924829,
      "grad_norm": 0.6872403138910607,
      "learning_rate": 4.902178288224946e-06,
      "loss": 0.613,
      "step": 2091
    },
    {
      "epoch": 0.5606324534369557,
      "grad_norm": 0.689665402338895,
      "learning_rate": 4.902080646352812e-06,
      "loss": 0.6198,
      "step": 2092
    },
    {
      "epoch": 0.5609004421814284,
      "grad_norm": 0.6709807623587006,
      "learning_rate": 4.9019829567472e-06,
      "loss": 0.6228,
      "step": 2093
    },
    {
      "epoch": 0.5611684309259011,
      "grad_norm": 0.6871062454410857,
      "learning_rate": 4.9018852194100515e-06,
      "loss": 0.6163,
      "step": 2094
    },
    {
      "epoch": 0.5614364196703738,
      "grad_norm": 0.6621609356368225,
      "learning_rate": 4.901787434343309e-06,
      "loss": 0.6164,
      "step": 2095
    },
    {
      "epoch": 0.5617044084148466,
      "grad_norm": 0.7248157014916586,
      "learning_rate": 4.901689601548916e-06,
      "loss": 0.5743,
      "step": 2096
    },
    {
      "epoch": 0.5619723971593193,
      "grad_norm": 0.6568945143186629,
      "learning_rate": 4.901591721028816e-06,
      "loss": 0.5869,
      "step": 2097
    },
    {
      "epoch": 0.562240385903792,
      "grad_norm": 0.6866854535475333,
      "learning_rate": 4.901493792784954e-06,
      "loss": 0.5996,
      "step": 2098
    },
    {
      "epoch": 0.5625083746482648,
      "grad_norm": 0.7256338791394306,
      "learning_rate": 4.901395816819275e-06,
      "loss": 0.5866,
      "step": 2099
    },
    {
      "epoch": 0.5627763633927375,
      "grad_norm": 0.6776574364068231,
      "learning_rate": 4.901297793133729e-06,
      "loss": 0.6119,
      "step": 2100
    },
    {
      "epoch": 0.5630443521372103,
      "grad_norm": 0.7142939482374187,
      "learning_rate": 4.901199721730261e-06,
      "loss": 0.5543,
      "step": 2101
    },
    {
      "epoch": 0.5633123408816829,
      "grad_norm": 0.6614689892553564,
      "learning_rate": 4.90110160261082e-06,
      "loss": 0.5837,
      "step": 2102
    },
    {
      "epoch": 0.5635803296261557,
      "grad_norm": 0.6736946432082692,
      "learning_rate": 4.901003435777358e-06,
      "loss": 0.6198,
      "step": 2103
    },
    {
      "epoch": 0.5638483183706284,
      "grad_norm": 0.7039469341414348,
      "learning_rate": 4.900905221231823e-06,
      "loss": 0.6093,
      "step": 2104
    },
    {
      "epoch": 0.5641163071151012,
      "grad_norm": 0.6952322923227892,
      "learning_rate": 4.900806958976169e-06,
      "loss": 0.5692,
      "step": 2105
    },
    {
      "epoch": 0.564384295859574,
      "grad_norm": 0.6701287336813074,
      "learning_rate": 4.900708649012348e-06,
      "loss": 0.5511,
      "step": 2106
    },
    {
      "epoch": 0.5646522846040466,
      "grad_norm": 0.6447513263842523,
      "learning_rate": 4.900610291342313e-06,
      "loss": 0.5802,
      "step": 2107
    },
    {
      "epoch": 0.5649202733485194,
      "grad_norm": 0.6677215905502977,
      "learning_rate": 4.900511885968018e-06,
      "loss": 0.6075,
      "step": 2108
    },
    {
      "epoch": 0.5651882620929921,
      "grad_norm": 0.6916700911908645,
      "learning_rate": 4.90041343289142e-06,
      "loss": 0.6088,
      "step": 2109
    },
    {
      "epoch": 0.5654562508374649,
      "grad_norm": 0.6590598540411297,
      "learning_rate": 4.900314932114475e-06,
      "loss": 0.6259,
      "step": 2110
    },
    {
      "epoch": 0.5657242395819375,
      "grad_norm": 0.6570224046351665,
      "learning_rate": 4.900216383639139e-06,
      "loss": 0.588,
      "step": 2111
    },
    {
      "epoch": 0.5659922283264103,
      "grad_norm": 0.7067699223336168,
      "learning_rate": 4.900117787467373e-06,
      "loss": 0.6208,
      "step": 2112
    },
    {
      "epoch": 0.566260217070883,
      "grad_norm": 0.6717174558301434,
      "learning_rate": 4.9000191436011335e-06,
      "loss": 0.5951,
      "step": 2113
    },
    {
      "epoch": 0.5665282058153558,
      "grad_norm": 0.6547049507581008,
      "learning_rate": 4.8999204520423825e-06,
      "loss": 0.6199,
      "step": 2114
    },
    {
      "epoch": 0.5667961945598284,
      "grad_norm": 0.7080436959428729,
      "learning_rate": 4.89982171279308e-06,
      "loss": 0.604,
      "step": 2115
    },
    {
      "epoch": 0.5670641833043012,
      "grad_norm": 0.6428278200886632,
      "learning_rate": 4.89972292585519e-06,
      "loss": 0.6032,
      "step": 2116
    },
    {
      "epoch": 0.567332172048774,
      "grad_norm": 0.6597001894182936,
      "learning_rate": 4.899624091230673e-06,
      "loss": 0.6115,
      "step": 2117
    },
    {
      "epoch": 0.5676001607932467,
      "grad_norm": 0.6820137948851258,
      "learning_rate": 4.899525208921495e-06,
      "loss": 0.6054,
      "step": 2118
    },
    {
      "epoch": 0.5678681495377195,
      "grad_norm": 0.657213491769215,
      "learning_rate": 4.89942627892962e-06,
      "loss": 0.5703,
      "step": 2119
    },
    {
      "epoch": 0.5681361382821921,
      "grad_norm": 0.6444085139799238,
      "learning_rate": 4.899327301257015e-06,
      "loss": 0.5951,
      "step": 2120
    },
    {
      "epoch": 0.5684041270266649,
      "grad_norm": 0.6814055585565628,
      "learning_rate": 4.899228275905646e-06,
      "loss": 0.6302,
      "step": 2121
    },
    {
      "epoch": 0.5686721157711376,
      "grad_norm": 0.651284533260244,
      "learning_rate": 4.89912920287748e-06,
      "loss": 0.6161,
      "step": 2122
    },
    {
      "epoch": 0.5689401045156104,
      "grad_norm": 0.6811149191452193,
      "learning_rate": 4.899030082174486e-06,
      "loss": 0.6052,
      "step": 2123
    },
    {
      "epoch": 0.569208093260083,
      "grad_norm": 0.6885602329938352,
      "learning_rate": 4.898930913798635e-06,
      "loss": 0.5738,
      "step": 2124
    },
    {
      "epoch": 0.5694760820045558,
      "grad_norm": 0.6715853033266578,
      "learning_rate": 4.898831697751897e-06,
      "loss": 0.5783,
      "step": 2125
    },
    {
      "epoch": 0.5697440707490286,
      "grad_norm": 0.6648108046504755,
      "learning_rate": 4.8987324340362445e-06,
      "loss": 0.6233,
      "step": 2126
    },
    {
      "epoch": 0.5700120594935013,
      "grad_norm": 0.6557345217851549,
      "learning_rate": 4.898633122653647e-06,
      "loss": 0.5731,
      "step": 2127
    },
    {
      "epoch": 0.570280048237974,
      "grad_norm": 0.6529930233183555,
      "learning_rate": 4.898533763606081e-06,
      "loss": 0.6124,
      "step": 2128
    },
    {
      "epoch": 0.5705480369824467,
      "grad_norm": 0.6940816184613767,
      "learning_rate": 4.8984343568955214e-06,
      "loss": 0.5985,
      "step": 2129
    },
    {
      "epoch": 0.5708160257269195,
      "grad_norm": 0.6648924898114589,
      "learning_rate": 4.898334902523941e-06,
      "loss": 0.5979,
      "step": 2130
    },
    {
      "epoch": 0.5710840144713922,
      "grad_norm": 0.6912070232506113,
      "learning_rate": 4.898235400493317e-06,
      "loss": 0.6276,
      "step": 2131
    },
    {
      "epoch": 0.5713520032158649,
      "grad_norm": 0.6819010408528677,
      "learning_rate": 4.898135850805627e-06,
      "loss": 0.5951,
      "step": 2132
    },
    {
      "epoch": 0.5716199919603376,
      "grad_norm": 0.6568683864966646,
      "learning_rate": 4.89803625346285e-06,
      "loss": 0.5767,
      "step": 2133
    },
    {
      "epoch": 0.5718879807048104,
      "grad_norm": 0.6672291711848143,
      "learning_rate": 4.897936608466964e-06,
      "loss": 0.5777,
      "step": 2134
    },
    {
      "epoch": 0.5721559694492832,
      "grad_norm": 0.6912559138792752,
      "learning_rate": 4.897836915819949e-06,
      "loss": 0.5853,
      "step": 2135
    },
    {
      "epoch": 0.5724239581937559,
      "grad_norm": 0.6603606563330003,
      "learning_rate": 4.897737175523787e-06,
      "loss": 0.5966,
      "step": 2136
    },
    {
      "epoch": 0.5726919469382286,
      "grad_norm": 0.6624251648169562,
      "learning_rate": 4.8976373875804595e-06,
      "loss": 0.5846,
      "step": 2137
    },
    {
      "epoch": 0.5729599356827013,
      "grad_norm": 0.6809194718846988,
      "learning_rate": 4.89753755199195e-06,
      "loss": 0.606,
      "step": 2138
    },
    {
      "epoch": 0.5732279244271741,
      "grad_norm": 0.7333907060157929,
      "learning_rate": 4.8974376687602415e-06,
      "loss": 0.5785,
      "step": 2139
    },
    {
      "epoch": 0.5734959131716468,
      "grad_norm": 0.6714225095307104,
      "learning_rate": 4.897337737887319e-06,
      "loss": 0.5939,
      "step": 2140
    },
    {
      "epoch": 0.5737639019161195,
      "grad_norm": 0.717436916061112,
      "learning_rate": 4.897237759375169e-06,
      "loss": 0.6015,
      "step": 2141
    },
    {
      "epoch": 0.5740318906605922,
      "grad_norm": 0.6898805981294611,
      "learning_rate": 4.897137733225778e-06,
      "loss": 0.6293,
      "step": 2142
    },
    {
      "epoch": 0.574299879405065,
      "grad_norm": 0.7000287343716173,
      "learning_rate": 4.897037659441133e-06,
      "loss": 0.6292,
      "step": 2143
    },
    {
      "epoch": 0.5745678681495378,
      "grad_norm": 0.673654628959138,
      "learning_rate": 4.896937538023224e-06,
      "loss": 0.5765,
      "step": 2144
    },
    {
      "epoch": 0.5748358568940104,
      "grad_norm": 0.6817991147665108,
      "learning_rate": 4.896837368974039e-06,
      "loss": 0.6305,
      "step": 2145
    },
    {
      "epoch": 0.5751038456384832,
      "grad_norm": 0.6706861499979841,
      "learning_rate": 4.89673715229557e-06,
      "loss": 0.622,
      "step": 2146
    },
    {
      "epoch": 0.5753718343829559,
      "grad_norm": 0.6765022469674955,
      "learning_rate": 4.896636887989807e-06,
      "loss": 0.6067,
      "step": 2147
    },
    {
      "epoch": 0.5756398231274287,
      "grad_norm": 0.6641864663126548,
      "learning_rate": 4.896536576058744e-06,
      "loss": 0.5523,
      "step": 2148
    },
    {
      "epoch": 0.5759078118719014,
      "grad_norm": 0.679548158583295,
      "learning_rate": 4.896436216504372e-06,
      "loss": 0.6056,
      "step": 2149
    },
    {
      "epoch": 0.5761758006163741,
      "grad_norm": 0.6554560953154205,
      "learning_rate": 4.896335809328688e-06,
      "loss": 0.5751,
      "step": 2150
    },
    {
      "epoch": 0.5764437893608468,
      "grad_norm": 0.6534947148893575,
      "learning_rate": 4.896235354533686e-06,
      "loss": 0.5637,
      "step": 2151
    },
    {
      "epoch": 0.5767117781053196,
      "grad_norm": 0.7003160111509731,
      "learning_rate": 4.896134852121362e-06,
      "loss": 0.5884,
      "step": 2152
    },
    {
      "epoch": 0.5769797668497924,
      "grad_norm": 0.6607766894395563,
      "learning_rate": 4.896034302093714e-06,
      "loss": 0.5907,
      "step": 2153
    },
    {
      "epoch": 0.577247755594265,
      "grad_norm": 0.6979789178776113,
      "learning_rate": 4.89593370445274e-06,
      "loss": 0.6513,
      "step": 2154
    },
    {
      "epoch": 0.5775157443387378,
      "grad_norm": 0.7038274287097491,
      "learning_rate": 4.895833059200438e-06,
      "loss": 0.5693,
      "step": 2155
    },
    {
      "epoch": 0.5777837330832105,
      "grad_norm": 0.6837174070156566,
      "learning_rate": 4.895732366338809e-06,
      "loss": 0.6113,
      "step": 2156
    },
    {
      "epoch": 0.5780517218276833,
      "grad_norm": 0.7752672167892782,
      "learning_rate": 4.895631625869853e-06,
      "loss": 0.5947,
      "step": 2157
    },
    {
      "epoch": 0.5783197105721559,
      "grad_norm": 0.655248866751445,
      "learning_rate": 4.895530837795573e-06,
      "loss": 0.6031,
      "step": 2158
    },
    {
      "epoch": 0.5785876993166287,
      "grad_norm": 0.6807659057734405,
      "learning_rate": 4.895430002117971e-06,
      "loss": 0.5969,
      "step": 2159
    },
    {
      "epoch": 0.5788556880611014,
      "grad_norm": 0.6716042302263979,
      "learning_rate": 4.895329118839052e-06,
      "loss": 0.5817,
      "step": 2160
    },
    {
      "epoch": 0.5791236768055742,
      "grad_norm": 0.6691409227462164,
      "learning_rate": 4.895228187960819e-06,
      "loss": 0.6051,
      "step": 2161
    },
    {
      "epoch": 0.579391665550047,
      "grad_norm": 0.6680563933163105,
      "learning_rate": 4.895127209485279e-06,
      "loss": 0.5784,
      "step": 2162
    },
    {
      "epoch": 0.5796596542945196,
      "grad_norm": 0.703439514182853,
      "learning_rate": 4.895026183414437e-06,
      "loss": 0.6037,
      "step": 2163
    },
    {
      "epoch": 0.5799276430389924,
      "grad_norm": 0.6692807672297726,
      "learning_rate": 4.8949251097503025e-06,
      "loss": 0.6115,
      "step": 2164
    },
    {
      "epoch": 0.5801956317834651,
      "grad_norm": 0.6742755669042336,
      "learning_rate": 4.894823988494882e-06,
      "loss": 0.6095,
      "step": 2165
    },
    {
      "epoch": 0.5804636205279379,
      "grad_norm": 0.676594005212124,
      "learning_rate": 4.894722819650188e-06,
      "loss": 0.6013,
      "step": 2166
    },
    {
      "epoch": 0.5807316092724105,
      "grad_norm": 0.6824690003982876,
      "learning_rate": 4.894621603218228e-06,
      "loss": 0.615,
      "step": 2167
    },
    {
      "epoch": 0.5809995980168833,
      "grad_norm": 0.670697748024148,
      "learning_rate": 4.894520339201014e-06,
      "loss": 0.603,
      "step": 2168
    },
    {
      "epoch": 0.581267586761356,
      "grad_norm": 0.6958957042177477,
      "learning_rate": 4.89441902760056e-06,
      "loss": 0.5764,
      "step": 2169
    },
    {
      "epoch": 0.5815355755058288,
      "grad_norm": 0.6937703693430919,
      "learning_rate": 4.8943176684188764e-06,
      "loss": 0.6201,
      "step": 2170
    },
    {
      "epoch": 0.5818035642503014,
      "grad_norm": 0.6849018766240899,
      "learning_rate": 4.8942162616579795e-06,
      "loss": 0.5824,
      "step": 2171
    },
    {
      "epoch": 0.5820715529947742,
      "grad_norm": 0.687403561829552,
      "learning_rate": 4.894114807319884e-06,
      "loss": 0.5866,
      "step": 2172
    },
    {
      "epoch": 0.582339541739247,
      "grad_norm": 0.659031335993403,
      "learning_rate": 4.894013305406605e-06,
      "loss": 0.5823,
      "step": 2173
    },
    {
      "epoch": 0.5826075304837197,
      "grad_norm": 0.656383855900534,
      "learning_rate": 4.893911755920161e-06,
      "loss": 0.5951,
      "step": 2174
    },
    {
      "epoch": 0.5828755192281924,
      "grad_norm": 0.6840533198672095,
      "learning_rate": 4.8938101588625695e-06,
      "loss": 0.5934,
      "step": 2175
    },
    {
      "epoch": 0.5831435079726651,
      "grad_norm": 0.6815760049943419,
      "learning_rate": 4.893708514235848e-06,
      "loss": 0.6022,
      "step": 2176
    },
    {
      "epoch": 0.5834114967171379,
      "grad_norm": 0.6631919730191358,
      "learning_rate": 4.893606822042018e-06,
      "loss": 0.6044,
      "step": 2177
    },
    {
      "epoch": 0.5836794854616106,
      "grad_norm": 0.7000447614863983,
      "learning_rate": 4.893505082283101e-06,
      "loss": 0.6535,
      "step": 2178
    },
    {
      "epoch": 0.5839474742060834,
      "grad_norm": 0.647193809036964,
      "learning_rate": 4.893403294961116e-06,
      "loss": 0.6024,
      "step": 2179
    },
    {
      "epoch": 0.584215462950556,
      "grad_norm": 0.6780150354411124,
      "learning_rate": 4.893301460078088e-06,
      "loss": 0.5981,
      "step": 2180
    },
    {
      "epoch": 0.5844834516950288,
      "grad_norm": 0.6724917489266912,
      "learning_rate": 4.89319957763604e-06,
      "loss": 0.581,
      "step": 2181
    },
    {
      "epoch": 0.5847514404395016,
      "grad_norm": 0.6530435993225557,
      "learning_rate": 4.893097647636996e-06,
      "loss": 0.5973,
      "step": 2182
    },
    {
      "epoch": 0.5850194291839743,
      "grad_norm": 0.645559989776435,
      "learning_rate": 4.892995670082983e-06,
      "loss": 0.5945,
      "step": 2183
    },
    {
      "epoch": 0.585287417928447,
      "grad_norm": 0.656048545127488,
      "learning_rate": 4.8928936449760255e-06,
      "loss": 0.5874,
      "step": 2184
    },
    {
      "epoch": 0.5855554066729197,
      "grad_norm": 0.6753793998826668,
      "learning_rate": 4.892791572318152e-06,
      "loss": 0.599,
      "step": 2185
    },
    {
      "epoch": 0.5858233954173925,
      "grad_norm": 0.6467446498290901,
      "learning_rate": 4.892689452111391e-06,
      "loss": 0.5737,
      "step": 2186
    },
    {
      "epoch": 0.5860913841618652,
      "grad_norm": 0.6637548516368396,
      "learning_rate": 4.892587284357771e-06,
      "loss": 0.6171,
      "step": 2187
    },
    {
      "epoch": 0.5863593729063379,
      "grad_norm": 0.6997150022038775,
      "learning_rate": 4.892485069059323e-06,
      "loss": 0.59,
      "step": 2188
    },
    {
      "epoch": 0.5866273616508106,
      "grad_norm": 0.6745531828097318,
      "learning_rate": 4.8923828062180776e-06,
      "loss": 0.6104,
      "step": 2189
    },
    {
      "epoch": 0.5868953503952834,
      "grad_norm": 0.6661868322747297,
      "learning_rate": 4.892280495836068e-06,
      "loss": 0.5686,
      "step": 2190
    },
    {
      "epoch": 0.5871633391397562,
      "grad_norm": 0.6531306702749947,
      "learning_rate": 4.892178137915327e-06,
      "loss": 0.5854,
      "step": 2191
    },
    {
      "epoch": 0.5874313278842289,
      "grad_norm": 0.6703196050862497,
      "learning_rate": 4.892075732457887e-06,
      "loss": 0.6095,
      "step": 2192
    },
    {
      "epoch": 0.5876993166287016,
      "grad_norm": 0.6931415874433409,
      "learning_rate": 4.891973279465785e-06,
      "loss": 0.5752,
      "step": 2193
    },
    {
      "epoch": 0.5879673053731743,
      "grad_norm": 0.6814732044081497,
      "learning_rate": 4.8918707789410555e-06,
      "loss": 0.5963,
      "step": 2194
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.6711411708739151,
      "learning_rate": 4.891768230885737e-06,
      "loss": 0.6051,
      "step": 2195
    },
    {
      "epoch": 0.5885032828621198,
      "grad_norm": 0.6509255045326636,
      "learning_rate": 4.891665635301866e-06,
      "loss": 0.604,
      "step": 2196
    },
    {
      "epoch": 0.5887712716065925,
      "grad_norm": 0.6644904677152259,
      "learning_rate": 4.891562992191481e-06,
      "loss": 0.6144,
      "step": 2197
    },
    {
      "epoch": 0.5890392603510652,
      "grad_norm": 0.6784991279888428,
      "learning_rate": 4.891460301556622e-06,
      "loss": 0.6,
      "step": 2198
    },
    {
      "epoch": 0.589307249095538,
      "grad_norm": 0.6710758082362293,
      "learning_rate": 4.89135756339933e-06,
      "loss": 0.5827,
      "step": 2199
    },
    {
      "epoch": 0.5895752378400108,
      "grad_norm": 0.6734629973388073,
      "learning_rate": 4.8912547777216465e-06,
      "loss": 0.6217,
      "step": 2200
    },
    {
      "epoch": 0.5898432265844834,
      "grad_norm": 0.6478363855107202,
      "learning_rate": 4.891151944525615e-06,
      "loss": 0.6078,
      "step": 2201
    },
    {
      "epoch": 0.5901112153289562,
      "grad_norm": 0.6792544489440322,
      "learning_rate": 4.891049063813277e-06,
      "loss": 0.6199,
      "step": 2202
    },
    {
      "epoch": 0.5903792040734289,
      "grad_norm": 0.706943051895551,
      "learning_rate": 4.890946135586678e-06,
      "loss": 0.5552,
      "step": 2203
    },
    {
      "epoch": 0.5906471928179017,
      "grad_norm": 0.6775391473751594,
      "learning_rate": 4.890843159847863e-06,
      "loss": 0.6276,
      "step": 2204
    },
    {
      "epoch": 0.5909151815623743,
      "grad_norm": 0.7729992133092818,
      "learning_rate": 4.89074013659888e-06,
      "loss": 0.569,
      "step": 2205
    },
    {
      "epoch": 0.5911831703068471,
      "grad_norm": 0.668188837543304,
      "learning_rate": 4.890637065841773e-06,
      "loss": 0.5458,
      "step": 2206
    },
    {
      "epoch": 0.5914511590513198,
      "grad_norm": 0.6926691093972152,
      "learning_rate": 4.8905339475785926e-06,
      "loss": 0.6187,
      "step": 2207
    },
    {
      "epoch": 0.5917191477957926,
      "grad_norm": 0.6585539357156212,
      "learning_rate": 4.890430781811387e-06,
      "loss": 0.6038,
      "step": 2208
    },
    {
      "epoch": 0.5919871365402654,
      "grad_norm": 0.7784604316817246,
      "learning_rate": 4.890327568542207e-06,
      "loss": 0.5952,
      "step": 2209
    },
    {
      "epoch": 0.592255125284738,
      "grad_norm": 0.679740701711694,
      "learning_rate": 4.8902243077731035e-06,
      "loss": 0.6054,
      "step": 2210
    },
    {
      "epoch": 0.5925231140292108,
      "grad_norm": 0.6775800705049149,
      "learning_rate": 4.890120999506128e-06,
      "loss": 0.5686,
      "step": 2211
    },
    {
      "epoch": 0.5927911027736835,
      "grad_norm": 0.7596408562946968,
      "learning_rate": 4.890017643743334e-06,
      "loss": 0.6083,
      "step": 2212
    },
    {
      "epoch": 0.5930590915181563,
      "grad_norm": 0.6919170691714128,
      "learning_rate": 4.889914240486774e-06,
      "loss": 0.6438,
      "step": 2213
    },
    {
      "epoch": 0.5933270802626289,
      "grad_norm": 0.6824697868687069,
      "learning_rate": 4.889810789738504e-06,
      "loss": 0.589,
      "step": 2214
    },
    {
      "epoch": 0.5935950690071017,
      "grad_norm": 0.6675511794499281,
      "learning_rate": 4.88970729150058e-06,
      "loss": 0.617,
      "step": 2215
    },
    {
      "epoch": 0.5938630577515744,
      "grad_norm": 0.6595952250598258,
      "learning_rate": 4.889603745775058e-06,
      "loss": 0.5909,
      "step": 2216
    },
    {
      "epoch": 0.5941310464960472,
      "grad_norm": 0.6588566321009074,
      "learning_rate": 4.889500152563995e-06,
      "loss": 0.5763,
      "step": 2217
    },
    {
      "epoch": 0.5943990352405198,
      "grad_norm": 0.6461532787714592,
      "learning_rate": 4.889396511869451e-06,
      "loss": 0.562,
      "step": 2218
    },
    {
      "epoch": 0.5946670239849926,
      "grad_norm": 0.6735442667584167,
      "learning_rate": 4.889292823693485e-06,
      "loss": 0.6341,
      "step": 2219
    },
    {
      "epoch": 0.5949350127294654,
      "grad_norm": 0.7172978871485828,
      "learning_rate": 4.889189088038158e-06,
      "loss": 0.5712,
      "step": 2220
    },
    {
      "epoch": 0.5952030014739381,
      "grad_norm": 0.6821010445703443,
      "learning_rate": 4.8890853049055294e-06,
      "loss": 0.6137,
      "step": 2221
    },
    {
      "epoch": 0.5954709902184109,
      "grad_norm": 0.6572109581574579,
      "learning_rate": 4.888981474297665e-06,
      "loss": 0.5844,
      "step": 2222
    },
    {
      "epoch": 0.5957389789628835,
      "grad_norm": 0.6840899533645368,
      "learning_rate": 4.888877596216625e-06,
      "loss": 0.5957,
      "step": 2223
    },
    {
      "epoch": 0.5960069677073563,
      "grad_norm": 0.6969513338324049,
      "learning_rate": 4.888773670664474e-06,
      "loss": 0.5968,
      "step": 2224
    },
    {
      "epoch": 0.596274956451829,
      "grad_norm": 0.6931749123961447,
      "learning_rate": 4.888669697643278e-06,
      "loss": 0.5973,
      "step": 2225
    },
    {
      "epoch": 0.5965429451963018,
      "grad_norm": 0.6653795096864439,
      "learning_rate": 4.888565677155104e-06,
      "loss": 0.578,
      "step": 2226
    },
    {
      "epoch": 0.5968109339407744,
      "grad_norm": 0.6678991338410492,
      "learning_rate": 4.888461609202016e-06,
      "loss": 0.6013,
      "step": 2227
    },
    {
      "epoch": 0.5970789226852472,
      "grad_norm": 0.6851293663258797,
      "learning_rate": 4.888357493786086e-06,
      "loss": 0.6036,
      "step": 2228
    },
    {
      "epoch": 0.59734691142972,
      "grad_norm": 0.6717587989766441,
      "learning_rate": 4.888253330909381e-06,
      "loss": 0.6064,
      "step": 2229
    },
    {
      "epoch": 0.5976149001741927,
      "grad_norm": 0.6689182537867212,
      "learning_rate": 4.888149120573969e-06,
      "loss": 0.5869,
      "step": 2230
    },
    {
      "epoch": 0.5978828889186654,
      "grad_norm": 0.6453980744635444,
      "learning_rate": 4.888044862781925e-06,
      "loss": 0.5529,
      "step": 2231
    },
    {
      "epoch": 0.5981508776631381,
      "grad_norm": 0.6560443267214439,
      "learning_rate": 4.887940557535318e-06,
      "loss": 0.5709,
      "step": 2232
    },
    {
      "epoch": 0.5984188664076109,
      "grad_norm": 0.6961481086663459,
      "learning_rate": 4.88783620483622e-06,
      "loss": 0.5531,
      "step": 2233
    },
    {
      "epoch": 0.5986868551520836,
      "grad_norm": 0.6518925885994858,
      "learning_rate": 4.887731804686707e-06,
      "loss": 0.5819,
      "step": 2234
    },
    {
      "epoch": 0.5989548438965564,
      "grad_norm": 0.7052462791299815,
      "learning_rate": 4.8876273570888525e-06,
      "loss": 0.5701,
      "step": 2235
    },
    {
      "epoch": 0.599222832641029,
      "grad_norm": 0.678416984603414,
      "learning_rate": 4.887522862044731e-06,
      "loss": 0.5736,
      "step": 2236
    },
    {
      "epoch": 0.5994908213855018,
      "grad_norm": 0.6690652825772638,
      "learning_rate": 4.887418319556422e-06,
      "loss": 0.6148,
      "step": 2237
    },
    {
      "epoch": 0.5997588101299746,
      "grad_norm": 0.6747212911151714,
      "learning_rate": 4.887313729626e-06,
      "loss": 0.6275,
      "step": 2238
    },
    {
      "epoch": 0.6000267988744473,
      "grad_norm": 0.6966343923960371,
      "learning_rate": 4.887209092255544e-06,
      "loss": 0.5748,
      "step": 2239
    },
    {
      "epoch": 0.60029478761892,
      "grad_norm": 0.6800387293338381,
      "learning_rate": 4.887104407447133e-06,
      "loss": 0.6084,
      "step": 2240
    },
    {
      "epoch": 0.6005627763633927,
      "grad_norm": 0.6716252802842345,
      "learning_rate": 4.886999675202849e-06,
      "loss": 0.5696,
      "step": 2241
    },
    {
      "epoch": 0.6008307651078655,
      "grad_norm": 0.6740606756639105,
      "learning_rate": 4.886894895524773e-06,
      "loss": 0.5835,
      "step": 2242
    },
    {
      "epoch": 0.6010987538523382,
      "grad_norm": 0.649984494873452,
      "learning_rate": 4.886790068414985e-06,
      "loss": 0.5914,
      "step": 2243
    },
    {
      "epoch": 0.6013667425968109,
      "grad_norm": 0.6847890236570661,
      "learning_rate": 4.88668519387557e-06,
      "loss": 0.6071,
      "step": 2244
    },
    {
      "epoch": 0.6016347313412836,
      "grad_norm": 0.6861585586931701,
      "learning_rate": 4.886580271908612e-06,
      "loss": 0.5722,
      "step": 2245
    },
    {
      "epoch": 0.6019027200857564,
      "grad_norm": 0.644255254664118,
      "learning_rate": 4.886475302516195e-06,
      "loss": 0.59,
      "step": 2246
    },
    {
      "epoch": 0.6021707088302292,
      "grad_norm": 0.69734538006788,
      "learning_rate": 4.886370285700405e-06,
      "loss": 0.6135,
      "step": 2247
    },
    {
      "epoch": 0.6024386975747018,
      "grad_norm": 0.6456369912146671,
      "learning_rate": 4.88626522146333e-06,
      "loss": 0.5807,
      "step": 2248
    },
    {
      "epoch": 0.6027066863191746,
      "grad_norm": 0.6770992893602991,
      "learning_rate": 4.886160109807056e-06,
      "loss": 0.6069,
      "step": 2249
    },
    {
      "epoch": 0.6029746750636473,
      "grad_norm": 0.6635535521132341,
      "learning_rate": 4.886054950733674e-06,
      "loss": 0.5716,
      "step": 2250
    },
    {
      "epoch": 0.6032426638081201,
      "grad_norm": 0.6519921898701568,
      "learning_rate": 4.885949744245271e-06,
      "loss": 0.5867,
      "step": 2251
    },
    {
      "epoch": 0.6035106525525928,
      "grad_norm": 0.6689699550754813,
      "learning_rate": 4.88584449034394e-06,
      "loss": 0.5885,
      "step": 2252
    },
    {
      "epoch": 0.6037786412970655,
      "grad_norm": 0.702835527031719,
      "learning_rate": 4.885739189031772e-06,
      "loss": 0.5743,
      "step": 2253
    },
    {
      "epoch": 0.6040466300415382,
      "grad_norm": 0.6862106661411155,
      "learning_rate": 4.88563384031086e-06,
      "loss": 0.5915,
      "step": 2254
    },
    {
      "epoch": 0.604314618786011,
      "grad_norm": 0.6833276343281863,
      "learning_rate": 4.885528444183295e-06,
      "loss": 0.5766,
      "step": 2255
    },
    {
      "epoch": 0.6045826075304838,
      "grad_norm": 0.6605418460436224,
      "learning_rate": 4.885423000651174e-06,
      "loss": 0.5911,
      "step": 2256
    },
    {
      "epoch": 0.6048505962749564,
      "grad_norm": 0.7100890127434504,
      "learning_rate": 4.88531750971659e-06,
      "loss": 0.6353,
      "step": 2257
    },
    {
      "epoch": 0.6051185850194292,
      "grad_norm": 0.6645757205251945,
      "learning_rate": 4.885211971381642e-06,
      "loss": 0.6167,
      "step": 2258
    },
    {
      "epoch": 0.6053865737639019,
      "grad_norm": 0.6518262305081123,
      "learning_rate": 4.885106385648426e-06,
      "loss": 0.6252,
      "step": 2259
    },
    {
      "epoch": 0.6056545625083747,
      "grad_norm": 0.6633803946223704,
      "learning_rate": 4.88500075251904e-06,
      "loss": 0.5752,
      "step": 2260
    },
    {
      "epoch": 0.6059225512528473,
      "grad_norm": 0.6456478045116785,
      "learning_rate": 4.884895071995583e-06,
      "loss": 0.5853,
      "step": 2261
    },
    {
      "epoch": 0.6061905399973201,
      "grad_norm": 0.66492184595212,
      "learning_rate": 4.884789344080156e-06,
      "loss": 0.6083,
      "step": 2262
    },
    {
      "epoch": 0.6064585287417928,
      "grad_norm": 0.6677317611446311,
      "learning_rate": 4.884683568774859e-06,
      "loss": 0.6097,
      "step": 2263
    },
    {
      "epoch": 0.6067265174862656,
      "grad_norm": 0.6803647845820445,
      "learning_rate": 4.884577746081794e-06,
      "loss": 0.5912,
      "step": 2264
    },
    {
      "epoch": 0.6069945062307384,
      "grad_norm": 0.671493386607265,
      "learning_rate": 4.884471876003064e-06,
      "loss": 0.6238,
      "step": 2265
    },
    {
      "epoch": 0.607262494975211,
      "grad_norm": 0.673474197898069,
      "learning_rate": 4.8843659585407735e-06,
      "loss": 0.5871,
      "step": 2266
    },
    {
      "epoch": 0.6075304837196838,
      "grad_norm": 0.6620091207982662,
      "learning_rate": 4.884259993697026e-06,
      "loss": 0.5757,
      "step": 2267
    },
    {
      "epoch": 0.6077984724641565,
      "grad_norm": 0.6742970745070924,
      "learning_rate": 4.884153981473927e-06,
      "loss": 0.5846,
      "step": 2268
    },
    {
      "epoch": 0.6080664612086293,
      "grad_norm": 0.6866886848738287,
      "learning_rate": 4.8840479218735856e-06,
      "loss": 0.588,
      "step": 2269
    },
    {
      "epoch": 0.6083344499531019,
      "grad_norm": 0.6552750052173407,
      "learning_rate": 4.8839418148981076e-06,
      "loss": 0.6059,
      "step": 2270
    },
    {
      "epoch": 0.6086024386975747,
      "grad_norm": 0.7010828536246829,
      "learning_rate": 4.8838356605496015e-06,
      "loss": 0.6096,
      "step": 2271
    },
    {
      "epoch": 0.6088704274420474,
      "grad_norm": 0.6610781987090645,
      "learning_rate": 4.883729458830176e-06,
      "loss": 0.6109,
      "step": 2272
    },
    {
      "epoch": 0.6091384161865202,
      "grad_norm": 0.6809086401678967,
      "learning_rate": 4.883623209741944e-06,
      "loss": 0.6222,
      "step": 2273
    },
    {
      "epoch": 0.6094064049309929,
      "grad_norm": 0.6537522790279061,
      "learning_rate": 4.883516913287015e-06,
      "loss": 0.6155,
      "step": 2274
    },
    {
      "epoch": 0.6096743936754656,
      "grad_norm": 0.6778890394722559,
      "learning_rate": 4.8834105694675014e-06,
      "loss": 0.575,
      "step": 2275
    },
    {
      "epoch": 0.6099423824199384,
      "grad_norm": 0.6674786596087089,
      "learning_rate": 4.883304178285517e-06,
      "loss": 0.5842,
      "step": 2276
    },
    {
      "epoch": 0.6102103711644111,
      "grad_norm": 0.6856475860945523,
      "learning_rate": 4.883197739743175e-06,
      "loss": 0.603,
      "step": 2277
    },
    {
      "epoch": 0.6104783599088838,
      "grad_norm": 0.6711809553754265,
      "learning_rate": 4.883091253842592e-06,
      "loss": 0.5825,
      "step": 2278
    },
    {
      "epoch": 0.6107463486533565,
      "grad_norm": 0.6773765746876741,
      "learning_rate": 4.882984720585883e-06,
      "loss": 0.5951,
      "step": 2279
    },
    {
      "epoch": 0.6110143373978293,
      "grad_norm": 0.6931747676453043,
      "learning_rate": 4.882878139975165e-06,
      "loss": 0.5742,
      "step": 2280
    },
    {
      "epoch": 0.611282326142302,
      "grad_norm": 0.7006806491462001,
      "learning_rate": 4.8827715120125565e-06,
      "loss": 0.6352,
      "step": 2281
    },
    {
      "epoch": 0.6115503148867748,
      "grad_norm": 0.7023820432862601,
      "learning_rate": 4.882664836700176e-06,
      "loss": 0.5623,
      "step": 2282
    },
    {
      "epoch": 0.6118183036312475,
      "grad_norm": 0.6438124124077471,
      "learning_rate": 4.882558114040143e-06,
      "loss": 0.5582,
      "step": 2283
    },
    {
      "epoch": 0.6120862923757202,
      "grad_norm": 0.6491780159753425,
      "learning_rate": 4.882451344034579e-06,
      "loss": 0.5438,
      "step": 2284
    },
    {
      "epoch": 0.612354281120193,
      "grad_norm": 0.6860523719535062,
      "learning_rate": 4.882344526685606e-06,
      "loss": 0.6095,
      "step": 2285
    },
    {
      "epoch": 0.6126222698646657,
      "grad_norm": 0.6959467310163309,
      "learning_rate": 4.882237661995346e-06,
      "loss": 0.6106,
      "step": 2286
    },
    {
      "epoch": 0.6128902586091384,
      "grad_norm": 0.6802939583734594,
      "learning_rate": 4.882130749965921e-06,
      "loss": 0.5921,
      "step": 2287
    },
    {
      "epoch": 0.6131582473536111,
      "grad_norm": 0.6849074427549567,
      "learning_rate": 4.882023790599458e-06,
      "loss": 0.6166,
      "step": 2288
    },
    {
      "epoch": 0.6134262360980839,
      "grad_norm": 0.6617353589140417,
      "learning_rate": 4.881916783898082e-06,
      "loss": 0.6062,
      "step": 2289
    },
    {
      "epoch": 0.6136942248425566,
      "grad_norm": 0.6722515426124415,
      "learning_rate": 4.881809729863919e-06,
      "loss": 0.5719,
      "step": 2290
    },
    {
      "epoch": 0.6139622135870293,
      "grad_norm": 0.6666554725294989,
      "learning_rate": 4.8817026284990955e-06,
      "loss": 0.5959,
      "step": 2291
    },
    {
      "epoch": 0.614230202331502,
      "grad_norm": 0.6806901864287939,
      "learning_rate": 4.881595479805742e-06,
      "loss": 0.6049,
      "step": 2292
    },
    {
      "epoch": 0.6144981910759748,
      "grad_norm": 0.7672165556442216,
      "learning_rate": 4.881488283785985e-06,
      "loss": 0.613,
      "step": 2293
    },
    {
      "epoch": 0.6147661798204476,
      "grad_norm": 0.6793470643275331,
      "learning_rate": 4.8813810404419575e-06,
      "loss": 0.5792,
      "step": 2294
    },
    {
      "epoch": 0.6150341685649203,
      "grad_norm": 0.6642699839734191,
      "learning_rate": 4.881273749775788e-06,
      "loss": 0.573,
      "step": 2295
    },
    {
      "epoch": 0.615302157309393,
      "grad_norm": 0.6827881975043264,
      "learning_rate": 4.881166411789609e-06,
      "loss": 0.5722,
      "step": 2296
    },
    {
      "epoch": 0.6155701460538657,
      "grad_norm": 0.6769902388467274,
      "learning_rate": 4.881059026485556e-06,
      "loss": 0.6152,
      "step": 2297
    },
    {
      "epoch": 0.6158381347983385,
      "grad_norm": 0.6533782522831686,
      "learning_rate": 4.8809515938657596e-06,
      "loss": 0.5883,
      "step": 2298
    },
    {
      "epoch": 0.6161061235428112,
      "grad_norm": 0.6710133811926237,
      "learning_rate": 4.880844113932357e-06,
      "loss": 0.5909,
      "step": 2299
    },
    {
      "epoch": 0.6163741122872839,
      "grad_norm": 0.6864815421806396,
      "learning_rate": 4.8807365866874825e-06,
      "loss": 0.6103,
      "step": 2300
    },
    {
      "epoch": 0.6166421010317567,
      "grad_norm": 0.6733361100892103,
      "learning_rate": 4.880629012133275e-06,
      "loss": 0.5778,
      "step": 2301
    },
    {
      "epoch": 0.6169100897762294,
      "grad_norm": 0.661455604297965,
      "learning_rate": 4.88052139027187e-06,
      "loss": 0.5988,
      "step": 2302
    },
    {
      "epoch": 0.6171780785207022,
      "grad_norm": 0.6731709968538163,
      "learning_rate": 4.880413721105406e-06,
      "loss": 0.6014,
      "step": 2303
    },
    {
      "epoch": 0.6174460672651748,
      "grad_norm": 0.6641508306150136,
      "learning_rate": 4.880306004636024e-06,
      "loss": 0.6161,
      "step": 2304
    },
    {
      "epoch": 0.6177140560096476,
      "grad_norm": 0.6900014400016905,
      "learning_rate": 4.880198240865864e-06,
      "loss": 0.6187,
      "step": 2305
    },
    {
      "epoch": 0.6179820447541203,
      "grad_norm": 0.7061420508560217,
      "learning_rate": 4.880090429797067e-06,
      "loss": 0.6241,
      "step": 2306
    },
    {
      "epoch": 0.6182500334985931,
      "grad_norm": 0.6595603595952824,
      "learning_rate": 4.879982571431777e-06,
      "loss": 0.5984,
      "step": 2307
    },
    {
      "epoch": 0.6185180222430658,
      "grad_norm": 0.6859989009283594,
      "learning_rate": 4.879874665772135e-06,
      "loss": 0.5672,
      "step": 2308
    },
    {
      "epoch": 0.6187860109875385,
      "grad_norm": 0.6532136862569382,
      "learning_rate": 4.879766712820286e-06,
      "loss": 0.5874,
      "step": 2309
    },
    {
      "epoch": 0.6190539997320113,
      "grad_norm": 0.6503681694266883,
      "learning_rate": 4.879658712578377e-06,
      "loss": 0.5944,
      "step": 2310
    },
    {
      "epoch": 0.619321988476484,
      "grad_norm": 0.6879389534221694,
      "learning_rate": 4.879550665048553e-06,
      "loss": 0.6091,
      "step": 2311
    },
    {
      "epoch": 0.6195899772209568,
      "grad_norm": 0.676447309597376,
      "learning_rate": 4.87944257023296e-06,
      "loss": 0.5912,
      "step": 2312
    },
    {
      "epoch": 0.6198579659654294,
      "grad_norm": 0.6682531112687405,
      "learning_rate": 4.8793344281337465e-06,
      "loss": 0.5534,
      "step": 2313
    },
    {
      "epoch": 0.6201259547099022,
      "grad_norm": 0.6742552521938272,
      "learning_rate": 4.879226238753062e-06,
      "loss": 0.5921,
      "step": 2314
    },
    {
      "epoch": 0.6203939434543749,
      "grad_norm": 0.6762292557758686,
      "learning_rate": 4.879118002093056e-06,
      "loss": 0.572,
      "step": 2315
    },
    {
      "epoch": 0.6206619321988477,
      "grad_norm": 0.6791702290054499,
      "learning_rate": 4.87900971815588e-06,
      "loss": 0.6122,
      "step": 2316
    },
    {
      "epoch": 0.6209299209433203,
      "grad_norm": 0.651255578021185,
      "learning_rate": 4.878901386943686e-06,
      "loss": 0.5581,
      "step": 2317
    },
    {
      "epoch": 0.6211979096877931,
      "grad_norm": 0.6876639130932992,
      "learning_rate": 4.878793008458626e-06,
      "loss": 0.5922,
      "step": 2318
    },
    {
      "epoch": 0.6214658984322659,
      "grad_norm": 0.6504519988280814,
      "learning_rate": 4.8786845827028524e-06,
      "loss": 0.5944,
      "step": 2319
    },
    {
      "epoch": 0.6217338871767386,
      "grad_norm": 0.6616276543846621,
      "learning_rate": 4.878576109678522e-06,
      "loss": 0.5652,
      "step": 2320
    },
    {
      "epoch": 0.6220018759212113,
      "grad_norm": 0.6792311356130372,
      "learning_rate": 4.87846758938779e-06,
      "loss": 0.5852,
      "step": 2321
    },
    {
      "epoch": 0.622269864665684,
      "grad_norm": 0.7089875891530175,
      "learning_rate": 4.878359021832812e-06,
      "loss": 0.6164,
      "step": 2322
    },
    {
      "epoch": 0.6225378534101568,
      "grad_norm": 0.6802261239264417,
      "learning_rate": 4.878250407015747e-06,
      "loss": 0.6152,
      "step": 2323
    },
    {
      "epoch": 0.6228058421546295,
      "grad_norm": 0.7159311116702722,
      "learning_rate": 4.87814174493875e-06,
      "loss": 0.5691,
      "step": 2324
    },
    {
      "epoch": 0.6230738308991023,
      "grad_norm": 0.7245045120590773,
      "learning_rate": 4.878033035603984e-06,
      "loss": 0.5928,
      "step": 2325
    },
    {
      "epoch": 0.6233418196435749,
      "grad_norm": 0.7620926818742177,
      "learning_rate": 4.877924279013608e-06,
      "loss": 0.6045,
      "step": 2326
    },
    {
      "epoch": 0.6236098083880477,
      "grad_norm": 0.678487029191511,
      "learning_rate": 4.877815475169782e-06,
      "loss": 0.5854,
      "step": 2327
    },
    {
      "epoch": 0.6238777971325205,
      "grad_norm": 0.6755793119218372,
      "learning_rate": 4.8777066240746696e-06,
      "loss": 0.5706,
      "step": 2328
    },
    {
      "epoch": 0.6241457858769932,
      "grad_norm": 0.7000700035069624,
      "learning_rate": 4.877597725730433e-06,
      "loss": 0.6149,
      "step": 2329
    },
    {
      "epoch": 0.6244137746214659,
      "grad_norm": 0.6867717651491584,
      "learning_rate": 4.877488780139237e-06,
      "loss": 0.5941,
      "step": 2330
    },
    {
      "epoch": 0.6246817633659386,
      "grad_norm": 0.6816734421578333,
      "learning_rate": 4.877379787303246e-06,
      "loss": 0.5916,
      "step": 2331
    },
    {
      "epoch": 0.6249497521104114,
      "grad_norm": 0.6586839251480443,
      "learning_rate": 4.877270747224625e-06,
      "loss": 0.5855,
      "step": 2332
    },
    {
      "epoch": 0.6252177408548841,
      "grad_norm": 0.679125504448386,
      "learning_rate": 4.877161659905543e-06,
      "loss": 0.5712,
      "step": 2333
    },
    {
      "epoch": 0.6254857295993568,
      "grad_norm": 0.6824040725104618,
      "learning_rate": 4.877052525348166e-06,
      "loss": 0.5719,
      "step": 2334
    },
    {
      "epoch": 0.6257537183438295,
      "grad_norm": 0.673775447827636,
      "learning_rate": 4.8769433435546635e-06,
      "loss": 0.5856,
      "step": 2335
    },
    {
      "epoch": 0.6260217070883023,
      "grad_norm": 0.685841534095724,
      "learning_rate": 4.876834114527204e-06,
      "loss": 0.6134,
      "step": 2336
    },
    {
      "epoch": 0.626289695832775,
      "grad_norm": 0.6714025516538628,
      "learning_rate": 4.87672483826796e-06,
      "loss": 0.5907,
      "step": 2337
    },
    {
      "epoch": 0.6265576845772478,
      "grad_norm": 0.6945805541800867,
      "learning_rate": 4.8766155147791e-06,
      "loss": 0.6108,
      "step": 2338
    },
    {
      "epoch": 0.6268256733217205,
      "grad_norm": 0.6833871258593722,
      "learning_rate": 4.8765061440628e-06,
      "loss": 0.6262,
      "step": 2339
    },
    {
      "epoch": 0.6270936620661932,
      "grad_norm": 0.6738782026178215,
      "learning_rate": 4.876396726121232e-06,
      "loss": 0.6183,
      "step": 2340
    },
    {
      "epoch": 0.627361650810666,
      "grad_norm": 0.6596072831715613,
      "learning_rate": 4.876287260956569e-06,
      "loss": 0.5886,
      "step": 2341
    },
    {
      "epoch": 0.6276296395551387,
      "grad_norm": 0.6645967820230655,
      "learning_rate": 4.876177748570988e-06,
      "loss": 0.6205,
      "step": 2342
    },
    {
      "epoch": 0.6278976282996114,
      "grad_norm": 0.6527748372462235,
      "learning_rate": 4.876068188966664e-06,
      "loss": 0.6156,
      "step": 2343
    },
    {
      "epoch": 0.6281656170440841,
      "grad_norm": 0.6767834455098094,
      "learning_rate": 4.875958582145775e-06,
      "loss": 0.5843,
      "step": 2344
    },
    {
      "epoch": 0.6284336057885569,
      "grad_norm": 0.6511697405780744,
      "learning_rate": 4.875848928110498e-06,
      "loss": 0.5955,
      "step": 2345
    },
    {
      "epoch": 0.6287015945330297,
      "grad_norm": 0.7050080313486351,
      "learning_rate": 4.8757392268630134e-06,
      "loss": 0.5962,
      "step": 2346
    },
    {
      "epoch": 0.6289695832775023,
      "grad_norm": 0.669395980120362,
      "learning_rate": 4.8756294784055e-06,
      "loss": 0.5956,
      "step": 2347
    },
    {
      "epoch": 0.6292375720219751,
      "grad_norm": 0.6884824650726011,
      "learning_rate": 4.875519682740139e-06,
      "loss": 0.5941,
      "step": 2348
    },
    {
      "epoch": 0.6295055607664478,
      "grad_norm": 0.6619312363728095,
      "learning_rate": 4.875409839869114e-06,
      "loss": 0.6062,
      "step": 2349
    },
    {
      "epoch": 0.6297735495109206,
      "grad_norm": 0.7225933141693721,
      "learning_rate": 4.875299949794605e-06,
      "loss": 0.5813,
      "step": 2350
    },
    {
      "epoch": 0.6300415382553932,
      "grad_norm": 0.6861132745824762,
      "learning_rate": 4.875190012518797e-06,
      "loss": 0.5906,
      "step": 2351
    },
    {
      "epoch": 0.630309526999866,
      "grad_norm": 0.7109160179751696,
      "learning_rate": 4.875080028043875e-06,
      "loss": 0.6026,
      "step": 2352
    },
    {
      "epoch": 0.6305775157443387,
      "grad_norm": 0.6964083053499286,
      "learning_rate": 4.8749699963720236e-06,
      "loss": 0.5995,
      "step": 2353
    },
    {
      "epoch": 0.6308455044888115,
      "grad_norm": 0.6614747885110188,
      "learning_rate": 4.874859917505429e-06,
      "loss": 0.5795,
      "step": 2354
    },
    {
      "epoch": 0.6311134932332843,
      "grad_norm": 0.6710026859940015,
      "learning_rate": 4.8747497914462805e-06,
      "loss": 0.6064,
      "step": 2355
    },
    {
      "epoch": 0.6313814819777569,
      "grad_norm": 0.6782453023631347,
      "learning_rate": 4.874639618196765e-06,
      "loss": 0.5664,
      "step": 2356
    },
    {
      "epoch": 0.6316494707222297,
      "grad_norm": 0.7225725105342274,
      "learning_rate": 4.874529397759073e-06,
      "loss": 0.6421,
      "step": 2357
    },
    {
      "epoch": 0.6319174594667024,
      "grad_norm": 0.6993306622920641,
      "learning_rate": 4.874419130135394e-06,
      "loss": 0.5761,
      "step": 2358
    },
    {
      "epoch": 0.6321854482111752,
      "grad_norm": 0.6949548043643072,
      "learning_rate": 4.874308815327918e-06,
      "loss": 0.5957,
      "step": 2359
    },
    {
      "epoch": 0.6324534369556478,
      "grad_norm": 0.7573830973643939,
      "learning_rate": 4.8741984533388395e-06,
      "loss": 0.5773,
      "step": 2360
    },
    {
      "epoch": 0.6327214257001206,
      "grad_norm": 0.688415741495808,
      "learning_rate": 4.874088044170351e-06,
      "loss": 0.593,
      "step": 2361
    },
    {
      "epoch": 0.6329894144445933,
      "grad_norm": 0.6702667044202975,
      "learning_rate": 4.873977587824645e-06,
      "loss": 0.626,
      "step": 2362
    },
    {
      "epoch": 0.6332574031890661,
      "grad_norm": 0.6794872909509493,
      "learning_rate": 4.8738670843039185e-06,
      "loss": 0.5894,
      "step": 2363
    },
    {
      "epoch": 0.6335253919335387,
      "grad_norm": 0.6626849562414807,
      "learning_rate": 4.873756533610366e-06,
      "loss": 0.5964,
      "step": 2364
    },
    {
      "epoch": 0.6337933806780115,
      "grad_norm": 0.688218739082941,
      "learning_rate": 4.873645935746184e-06,
      "loss": 0.5956,
      "step": 2365
    },
    {
      "epoch": 0.6340613694224843,
      "grad_norm": 0.6534593671862222,
      "learning_rate": 4.873535290713571e-06,
      "loss": 0.5834,
      "step": 2366
    },
    {
      "epoch": 0.634329358166957,
      "grad_norm": 0.677352306564312,
      "learning_rate": 4.873424598514727e-06,
      "loss": 0.6292,
      "step": 2367
    },
    {
      "epoch": 0.6345973469114298,
      "grad_norm": 0.6688778145903165,
      "learning_rate": 4.873313859151849e-06,
      "loss": 0.5723,
      "step": 2368
    },
    {
      "epoch": 0.6348653356559024,
      "grad_norm": 0.6929996883183087,
      "learning_rate": 4.87320307262714e-06,
      "loss": 0.594,
      "step": 2369
    },
    {
      "epoch": 0.6351333244003752,
      "grad_norm": 0.6704055691226555,
      "learning_rate": 4.8730922389428e-06,
      "loss": 0.6028,
      "step": 2370
    },
    {
      "epoch": 0.6354013131448479,
      "grad_norm": 0.7009672304866306,
      "learning_rate": 4.872981358101032e-06,
      "loss": 0.6212,
      "step": 2371
    },
    {
      "epoch": 0.6356693018893207,
      "grad_norm": 0.6834182908426989,
      "learning_rate": 4.872870430104039e-06,
      "loss": 0.5987,
      "step": 2372
    },
    {
      "epoch": 0.6359372906337933,
      "grad_norm": 0.6932861014822315,
      "learning_rate": 4.872759454954026e-06,
      "loss": 0.606,
      "step": 2373
    },
    {
      "epoch": 0.6362052793782661,
      "grad_norm": 0.6697117952492786,
      "learning_rate": 4.872648432653197e-06,
      "loss": 0.6178,
      "step": 2374
    },
    {
      "epoch": 0.6364732681227389,
      "grad_norm": 0.6791001745014753,
      "learning_rate": 4.87253736320376e-06,
      "loss": 0.6083,
      "step": 2375
    },
    {
      "epoch": 0.6367412568672116,
      "grad_norm": 0.6716927592288594,
      "learning_rate": 4.8724262466079205e-06,
      "loss": 0.5845,
      "step": 2376
    },
    {
      "epoch": 0.6370092456116843,
      "grad_norm": 0.6656114601917962,
      "learning_rate": 4.872315082867888e-06,
      "loss": 0.6363,
      "step": 2377
    },
    {
      "epoch": 0.637277234356157,
      "grad_norm": 0.6561683743545167,
      "learning_rate": 4.872203871985871e-06,
      "loss": 0.5497,
      "step": 2378
    },
    {
      "epoch": 0.6375452231006298,
      "grad_norm": 0.6787568973780367,
      "learning_rate": 4.872092613964078e-06,
      "loss": 0.6153,
      "step": 2379
    },
    {
      "epoch": 0.6378132118451025,
      "grad_norm": 0.6754319339158427,
      "learning_rate": 4.871981308804722e-06,
      "loss": 0.6215,
      "step": 2380
    },
    {
      "epoch": 0.6380812005895753,
      "grad_norm": 0.6558806036958015,
      "learning_rate": 4.871869956510014e-06,
      "loss": 0.6321,
      "step": 2381
    },
    {
      "epoch": 0.6383491893340479,
      "grad_norm": 0.6749754679963227,
      "learning_rate": 4.8717585570821664e-06,
      "loss": 0.5804,
      "step": 2382
    },
    {
      "epoch": 0.6386171780785207,
      "grad_norm": 0.6806535356285817,
      "learning_rate": 4.8716471105233944e-06,
      "loss": 0.5851,
      "step": 2383
    },
    {
      "epoch": 0.6388851668229935,
      "grad_norm": 0.6534414968446095,
      "learning_rate": 4.871535616835911e-06,
      "loss": 0.5967,
      "step": 2384
    },
    {
      "epoch": 0.6391531555674662,
      "grad_norm": 0.6600167417663889,
      "learning_rate": 4.8714240760219315e-06,
      "loss": 0.6138,
      "step": 2385
    },
    {
      "epoch": 0.6394211443119389,
      "grad_norm": 0.6865220953359288,
      "learning_rate": 4.871312488083674e-06,
      "loss": 0.5861,
      "step": 2386
    },
    {
      "epoch": 0.6396891330564116,
      "grad_norm": 0.665007202593661,
      "learning_rate": 4.871200853023355e-06,
      "loss": 0.5969,
      "step": 2387
    },
    {
      "epoch": 0.6399571218008844,
      "grad_norm": 0.6585333018037774,
      "learning_rate": 4.871089170843192e-06,
      "loss": 0.6128,
      "step": 2388
    },
    {
      "epoch": 0.6402251105453571,
      "grad_norm": 0.6719340716757665,
      "learning_rate": 4.870977441545406e-06,
      "loss": 0.5912,
      "step": 2389
    },
    {
      "epoch": 0.6404930992898298,
      "grad_norm": 0.6427389978426518,
      "learning_rate": 4.8708656651322164e-06,
      "loss": 0.5759,
      "step": 2390
    },
    {
      "epoch": 0.6407610880343025,
      "grad_norm": 0.6870322636005295,
      "learning_rate": 4.870753841605845e-06,
      "loss": 0.5823,
      "step": 2391
    },
    {
      "epoch": 0.6410290767787753,
      "grad_norm": 0.7017574294378549,
      "learning_rate": 4.870641970968513e-06,
      "loss": 0.6093,
      "step": 2392
    },
    {
      "epoch": 0.641297065523248,
      "grad_norm": 0.6472165954465623,
      "learning_rate": 4.870530053222444e-06,
      "loss": 0.5891,
      "step": 2393
    },
    {
      "epoch": 0.6415650542677207,
      "grad_norm": 0.6707929153423974,
      "learning_rate": 4.870418088369862e-06,
      "loss": 0.6055,
      "step": 2394
    },
    {
      "epoch": 0.6418330430121935,
      "grad_norm": 0.6570167133554726,
      "learning_rate": 4.8703060764129915e-06,
      "loss": 0.5495,
      "step": 2395
    },
    {
      "epoch": 0.6421010317566662,
      "grad_norm": 0.6712861132578475,
      "learning_rate": 4.87019401735406e-06,
      "loss": 0.6013,
      "step": 2396
    },
    {
      "epoch": 0.642369020501139,
      "grad_norm": 0.7213601907535392,
      "learning_rate": 4.870081911195292e-06,
      "loss": 0.5945,
      "step": 2397
    },
    {
      "epoch": 0.6426370092456117,
      "grad_norm": 0.648257126828715,
      "learning_rate": 4.869969757938916e-06,
      "loss": 0.5877,
      "step": 2398
    },
    {
      "epoch": 0.6429049979900844,
      "grad_norm": 0.6698503813738509,
      "learning_rate": 4.869857557587162e-06,
      "loss": 0.6071,
      "step": 2399
    },
    {
      "epoch": 0.6431729867345571,
      "grad_norm": 0.6753114436687865,
      "learning_rate": 4.869745310142258e-06,
      "loss": 0.6034,
      "step": 2400
    },
    {
      "epoch": 0.6434409754790299,
      "grad_norm": 0.6626819795536614,
      "learning_rate": 4.869633015606436e-06,
      "loss": 0.5823,
      "step": 2401
    },
    {
      "epoch": 0.6437089642235027,
      "grad_norm": 0.6795055693098965,
      "learning_rate": 4.869520673981924e-06,
      "loss": 0.5848,
      "step": 2402
    },
    {
      "epoch": 0.6439769529679753,
      "grad_norm": 0.6861545040544184,
      "learning_rate": 4.86940828527096e-06,
      "loss": 0.6333,
      "step": 2403
    },
    {
      "epoch": 0.6442449417124481,
      "grad_norm": 0.6688246795962695,
      "learning_rate": 4.869295849475773e-06,
      "loss": 0.5809,
      "step": 2404
    },
    {
      "epoch": 0.6445129304569208,
      "grad_norm": 0.6802829058657853,
      "learning_rate": 4.869183366598599e-06,
      "loss": 0.5567,
      "step": 2405
    },
    {
      "epoch": 0.6447809192013936,
      "grad_norm": 0.6901291369991357,
      "learning_rate": 4.8690708366416725e-06,
      "loss": 0.5914,
      "step": 2406
    },
    {
      "epoch": 0.6450489079458662,
      "grad_norm": 0.6814820476727791,
      "learning_rate": 4.86895825960723e-06,
      "loss": 0.5707,
      "step": 2407
    },
    {
      "epoch": 0.645316896690339,
      "grad_norm": 0.674681119106682,
      "learning_rate": 4.86884563549751e-06,
      "loss": 0.5977,
      "step": 2408
    },
    {
      "epoch": 0.6455848854348117,
      "grad_norm": 0.6678968096149063,
      "learning_rate": 4.868732964314748e-06,
      "loss": 0.5875,
      "step": 2409
    },
    {
      "epoch": 0.6458528741792845,
      "grad_norm": 0.6826589741587842,
      "learning_rate": 4.868620246061185e-06,
      "loss": 0.5886,
      "step": 2410
    },
    {
      "epoch": 0.6461208629237573,
      "grad_norm": 0.6874766204261055,
      "learning_rate": 4.868507480739059e-06,
      "loss": 0.6127,
      "step": 2411
    },
    {
      "epoch": 0.6463888516682299,
      "grad_norm": 0.6633124224621277,
      "learning_rate": 4.868394668350613e-06,
      "loss": 0.6087,
      "step": 2412
    },
    {
      "epoch": 0.6466568404127027,
      "grad_norm": 0.6693733618827283,
      "learning_rate": 4.868281808898087e-06,
      "loss": 0.5928,
      "step": 2413
    },
    {
      "epoch": 0.6469248291571754,
      "grad_norm": 0.6905411647444757,
      "learning_rate": 4.868168902383725e-06,
      "loss": 0.5672,
      "step": 2414
    },
    {
      "epoch": 0.6471928179016482,
      "grad_norm": 0.6767020091822495,
      "learning_rate": 4.868055948809769e-06,
      "loss": 0.5783,
      "step": 2415
    },
    {
      "epoch": 0.6474608066461208,
      "grad_norm": 0.665971606153033,
      "learning_rate": 4.8679429481784656e-06,
      "loss": 0.6014,
      "step": 2416
    },
    {
      "epoch": 0.6477287953905936,
      "grad_norm": 0.6871527761467723,
      "learning_rate": 4.867829900492059e-06,
      "loss": 0.6199,
      "step": 2417
    },
    {
      "epoch": 0.6479967841350663,
      "grad_norm": 0.6874346474148005,
      "learning_rate": 4.867716805752797e-06,
      "loss": 0.598,
      "step": 2418
    },
    {
      "epoch": 0.6482647728795391,
      "grad_norm": 0.6891109825671496,
      "learning_rate": 4.867603663962925e-06,
      "loss": 0.625,
      "step": 2419
    },
    {
      "epoch": 0.6485327616240117,
      "grad_norm": 0.6516294103124263,
      "learning_rate": 4.867490475124692e-06,
      "loss": 0.5774,
      "step": 2420
    },
    {
      "epoch": 0.6488007503684845,
      "grad_norm": 0.6655239455172913,
      "learning_rate": 4.8673772392403485e-06,
      "loss": 0.568,
      "step": 2421
    },
    {
      "epoch": 0.6490687391129573,
      "grad_norm": 0.7669372629660421,
      "learning_rate": 4.867263956312144e-06,
      "loss": 0.5165,
      "step": 2422
    },
    {
      "epoch": 0.64933672785743,
      "grad_norm": 0.6811780862392681,
      "learning_rate": 4.867150626342328e-06,
      "loss": 0.6192,
      "step": 2423
    },
    {
      "epoch": 0.6496047166019027,
      "grad_norm": 0.6717042648279103,
      "learning_rate": 4.867037249333155e-06,
      "loss": 0.6004,
      "step": 2424
    },
    {
      "epoch": 0.6498727053463754,
      "grad_norm": 0.6831660115151046,
      "learning_rate": 4.866923825286877e-06,
      "loss": 0.6392,
      "step": 2425
    },
    {
      "epoch": 0.6501406940908482,
      "grad_norm": 0.6999530966174727,
      "learning_rate": 4.8668103542057475e-06,
      "loss": 0.6002,
      "step": 2426
    },
    {
      "epoch": 0.6504086828353209,
      "grad_norm": 0.6898235047092262,
      "learning_rate": 4.866696836092022e-06,
      "loss": 0.6232,
      "step": 2427
    },
    {
      "epoch": 0.6506766715797937,
      "grad_norm": 0.6736757682649441,
      "learning_rate": 4.866583270947956e-06,
      "loss": 0.6254,
      "step": 2428
    },
    {
      "epoch": 0.6509446603242663,
      "grad_norm": 0.6847849479246999,
      "learning_rate": 4.866469658775806e-06,
      "loss": 0.5943,
      "step": 2429
    },
    {
      "epoch": 0.6512126490687391,
      "grad_norm": 0.6795056935303888,
      "learning_rate": 4.86635599957783e-06,
      "loss": 0.5944,
      "step": 2430
    },
    {
      "epoch": 0.6514806378132119,
      "grad_norm": 0.6776160691801473,
      "learning_rate": 4.866242293356287e-06,
      "loss": 0.5714,
      "step": 2431
    },
    {
      "epoch": 0.6517486265576846,
      "grad_norm": 0.6726886375611963,
      "learning_rate": 4.866128540113436e-06,
      "loss": 0.5905,
      "step": 2432
    },
    {
      "epoch": 0.6520166153021573,
      "grad_norm": 0.6716359623466359,
      "learning_rate": 4.8660147398515365e-06,
      "loss": 0.6244,
      "step": 2433
    },
    {
      "epoch": 0.65228460404663,
      "grad_norm": 0.7006782142807356,
      "learning_rate": 4.865900892572852e-06,
      "loss": 0.5374,
      "step": 2434
    },
    {
      "epoch": 0.6525525927911028,
      "grad_norm": 0.7229415179757707,
      "learning_rate": 4.865786998279644e-06,
      "loss": 0.5831,
      "step": 2435
    },
    {
      "epoch": 0.6528205815355755,
      "grad_norm": 0.6773369197437245,
      "learning_rate": 4.8656730569741755e-06,
      "loss": 0.6091,
      "step": 2436
    },
    {
      "epoch": 0.6530885702800482,
      "grad_norm": 0.6824877567664337,
      "learning_rate": 4.86555906865871e-06,
      "loss": 0.603,
      "step": 2437
    },
    {
      "epoch": 0.6533565590245209,
      "grad_norm": 0.6509429636536106,
      "learning_rate": 4.865445033335515e-06,
      "loss": 0.5536,
      "step": 2438
    },
    {
      "epoch": 0.6536245477689937,
      "grad_norm": 0.6601435074373608,
      "learning_rate": 4.865330951006853e-06,
      "loss": 0.5633,
      "step": 2439
    },
    {
      "epoch": 0.6538925365134665,
      "grad_norm": 0.6808120791469614,
      "learning_rate": 4.865216821674995e-06,
      "loss": 0.5845,
      "step": 2440
    },
    {
      "epoch": 0.6541605252579392,
      "grad_norm": 0.6631587329990942,
      "learning_rate": 4.865102645342207e-06,
      "loss": 0.5989,
      "step": 2441
    },
    {
      "epoch": 0.6544285140024119,
      "grad_norm": 0.6440721412557678,
      "learning_rate": 4.864988422010757e-06,
      "loss": 0.5814,
      "step": 2442
    },
    {
      "epoch": 0.6546965027468846,
      "grad_norm": 0.6804796917208786,
      "learning_rate": 4.864874151682916e-06,
      "loss": 0.5726,
      "step": 2443
    },
    {
      "epoch": 0.6549644914913574,
      "grad_norm": 0.6915706870246106,
      "learning_rate": 4.864759834360954e-06,
      "loss": 0.608,
      "step": 2444
    },
    {
      "epoch": 0.6552324802358301,
      "grad_norm": 0.6818572278910944,
      "learning_rate": 4.8646454700471424e-06,
      "loss": 0.6188,
      "step": 2445
    },
    {
      "epoch": 0.6555004689803028,
      "grad_norm": 0.7479101897517219,
      "learning_rate": 4.8645310587437555e-06,
      "loss": 0.6091,
      "step": 2446
    },
    {
      "epoch": 0.6557684577247755,
      "grad_norm": 0.6618679679988017,
      "learning_rate": 4.8644166004530665e-06,
      "loss": 0.5935,
      "step": 2447
    },
    {
      "epoch": 0.6560364464692483,
      "grad_norm": 0.6642141172041679,
      "learning_rate": 4.864302095177348e-06,
      "loss": 0.5852,
      "step": 2448
    },
    {
      "epoch": 0.6563044352137211,
      "grad_norm": 0.6625935999932244,
      "learning_rate": 4.864187542918878e-06,
      "loss": 0.6033,
      "step": 2449
    },
    {
      "epoch": 0.6565724239581937,
      "grad_norm": 0.6608447306667584,
      "learning_rate": 4.864072943679931e-06,
      "loss": 0.5873,
      "step": 2450
    },
    {
      "epoch": 0.6568404127026665,
      "grad_norm": 0.6697255954309829,
      "learning_rate": 4.863958297462784e-06,
      "loss": 0.6048,
      "step": 2451
    },
    {
      "epoch": 0.6571084014471392,
      "grad_norm": 0.6797477652555215,
      "learning_rate": 4.863843604269717e-06,
      "loss": 0.5543,
      "step": 2452
    },
    {
      "epoch": 0.657376390191612,
      "grad_norm": 0.6807876769619902,
      "learning_rate": 4.863728864103008e-06,
      "loss": 0.6043,
      "step": 2453
    },
    {
      "epoch": 0.6576443789360847,
      "grad_norm": 0.6581827804607454,
      "learning_rate": 4.863614076964937e-06,
      "loss": 0.6113,
      "step": 2454
    },
    {
      "epoch": 0.6579123676805574,
      "grad_norm": 0.6761837271785587,
      "learning_rate": 4.863499242857786e-06,
      "loss": 0.5873,
      "step": 2455
    },
    {
      "epoch": 0.6581803564250301,
      "grad_norm": 0.6767639633965561,
      "learning_rate": 4.863384361783835e-06,
      "loss": 0.6381,
      "step": 2456
    },
    {
      "epoch": 0.6584483451695029,
      "grad_norm": 0.6437459197820887,
      "learning_rate": 4.863269433745369e-06,
      "loss": 0.6138,
      "step": 2457
    },
    {
      "epoch": 0.6587163339139757,
      "grad_norm": 0.6747955210511125,
      "learning_rate": 4.863154458744671e-06,
      "loss": 0.6074,
      "step": 2458
    },
    {
      "epoch": 0.6589843226584483,
      "grad_norm": 0.6476839927222694,
      "learning_rate": 4.863039436784025e-06,
      "loss": 0.5674,
      "step": 2459
    },
    {
      "epoch": 0.6592523114029211,
      "grad_norm": 0.6606517660118391,
      "learning_rate": 4.862924367865718e-06,
      "loss": 0.6248,
      "step": 2460
    },
    {
      "epoch": 0.6595203001473938,
      "grad_norm": 0.6400583259920154,
      "learning_rate": 4.862809251992036e-06,
      "loss": 0.6041,
      "step": 2461
    },
    {
      "epoch": 0.6597882888918666,
      "grad_norm": 0.6722925097954028,
      "learning_rate": 4.862694089165266e-06,
      "loss": 0.6013,
      "step": 2462
    },
    {
      "epoch": 0.6600562776363392,
      "grad_norm": 0.675244494738191,
      "learning_rate": 4.862578879387697e-06,
      "loss": 0.6052,
      "step": 2463
    },
    {
      "epoch": 0.660324266380812,
      "grad_norm": 0.6776124198419725,
      "learning_rate": 4.862463622661619e-06,
      "loss": 0.6164,
      "step": 2464
    },
    {
      "epoch": 0.6605922551252847,
      "grad_norm": 0.6644817174128181,
      "learning_rate": 4.862348318989322e-06,
      "loss": 0.5458,
      "step": 2465
    },
    {
      "epoch": 0.6608602438697575,
      "grad_norm": 0.6793932096735128,
      "learning_rate": 4.862232968373096e-06,
      "loss": 0.6234,
      "step": 2466
    },
    {
      "epoch": 0.6611282326142301,
      "grad_norm": 0.6626120786322985,
      "learning_rate": 4.862117570815236e-06,
      "loss": 0.614,
      "step": 2467
    },
    {
      "epoch": 0.6613962213587029,
      "grad_norm": 0.6863650862777178,
      "learning_rate": 4.862002126318032e-06,
      "loss": 0.6026,
      "step": 2468
    },
    {
      "epoch": 0.6616642101031757,
      "grad_norm": 0.6670022382077287,
      "learning_rate": 4.86188663488378e-06,
      "loss": 0.5996,
      "step": 2469
    },
    {
      "epoch": 0.6619321988476484,
      "grad_norm": 0.6703478293066605,
      "learning_rate": 4.861771096514775e-06,
      "loss": 0.5969,
      "step": 2470
    },
    {
      "epoch": 0.6622001875921212,
      "grad_norm": 0.6607386786540337,
      "learning_rate": 4.861655511213312e-06,
      "loss": 0.5717,
      "step": 2471
    },
    {
      "epoch": 0.6624681763365938,
      "grad_norm": 0.6732755527485391,
      "learning_rate": 4.861539878981688e-06,
      "loss": 0.6364,
      "step": 2472
    },
    {
      "epoch": 0.6627361650810666,
      "grad_norm": 0.6577522914296923,
      "learning_rate": 4.861424199822201e-06,
      "loss": 0.5881,
      "step": 2473
    },
    {
      "epoch": 0.6630041538255393,
      "grad_norm": 0.6624376558066998,
      "learning_rate": 4.861308473737151e-06,
      "loss": 0.5774,
      "step": 2474
    },
    {
      "epoch": 0.6632721425700121,
      "grad_norm": 0.661904005199402,
      "learning_rate": 4.861192700728837e-06,
      "loss": 0.5692,
      "step": 2475
    },
    {
      "epoch": 0.6635401313144847,
      "grad_norm": 0.6403262159654343,
      "learning_rate": 4.8610768807995575e-06,
      "loss": 0.5802,
      "step": 2476
    },
    {
      "epoch": 0.6638081200589575,
      "grad_norm": 0.6585895029934615,
      "learning_rate": 4.860961013951616e-06,
      "loss": 0.6199,
      "step": 2477
    },
    {
      "epoch": 0.6640761088034303,
      "grad_norm": 0.6670527216994843,
      "learning_rate": 4.860845100187316e-06,
      "loss": 0.6033,
      "step": 2478
    },
    {
      "epoch": 0.664344097547903,
      "grad_norm": 0.6976679378195201,
      "learning_rate": 4.860729139508958e-06,
      "loss": 0.5668,
      "step": 2479
    },
    {
      "epoch": 0.6646120862923757,
      "grad_norm": 0.662017899015491,
      "learning_rate": 4.8606131319188495e-06,
      "loss": 0.6049,
      "step": 2480
    },
    {
      "epoch": 0.6648800750368484,
      "grad_norm": 0.662205084126266,
      "learning_rate": 4.860497077419294e-06,
      "loss": 0.5959,
      "step": 2481
    },
    {
      "epoch": 0.6651480637813212,
      "grad_norm": 0.686154806880967,
      "learning_rate": 4.860380976012597e-06,
      "loss": 0.5881,
      "step": 2482
    },
    {
      "epoch": 0.6654160525257939,
      "grad_norm": 0.6699254499849804,
      "learning_rate": 4.8602648277010674e-06,
      "loss": 0.5957,
      "step": 2483
    },
    {
      "epoch": 0.6656840412702667,
      "grad_norm": 0.6579736657774793,
      "learning_rate": 4.860148632487011e-06,
      "loss": 0.5987,
      "step": 2484
    },
    {
      "epoch": 0.6659520300147393,
      "grad_norm": 0.6683023426301878,
      "learning_rate": 4.86003239037274e-06,
      "loss": 0.5678,
      "step": 2485
    },
    {
      "epoch": 0.6662200187592121,
      "grad_norm": 0.6530676215220408,
      "learning_rate": 4.859916101360562e-06,
      "loss": 0.5653,
      "step": 2486
    },
    {
      "epoch": 0.6664880075036849,
      "grad_norm": 0.6825024511922683,
      "learning_rate": 4.859799765452789e-06,
      "loss": 0.5919,
      "step": 2487
    },
    {
      "epoch": 0.6667559962481576,
      "grad_norm": 0.6822428120463035,
      "learning_rate": 4.859683382651731e-06,
      "loss": 0.6083,
      "step": 2488
    },
    {
      "epoch": 0.6670239849926303,
      "grad_norm": 0.6887213826482967,
      "learning_rate": 4.859566952959702e-06,
      "loss": 0.5877,
      "step": 2489
    },
    {
      "epoch": 0.667291973737103,
      "grad_norm": 0.6642011837548922,
      "learning_rate": 4.859450476379017e-06,
      "loss": 0.5734,
      "step": 2490
    },
    {
      "epoch": 0.6675599624815758,
      "grad_norm": 0.6750982331583146,
      "learning_rate": 4.859333952911988e-06,
      "loss": 0.6086,
      "step": 2491
    },
    {
      "epoch": 0.6678279512260485,
      "grad_norm": 0.67251101987444,
      "learning_rate": 4.859217382560932e-06,
      "loss": 0.5884,
      "step": 2492
    },
    {
      "epoch": 0.6680959399705212,
      "grad_norm": 0.6421731502339928,
      "learning_rate": 4.859100765328165e-06,
      "loss": 0.6077,
      "step": 2493
    },
    {
      "epoch": 0.668363928714994,
      "grad_norm": 0.697775031008447,
      "learning_rate": 4.8589841012160055e-06,
      "loss": 0.6321,
      "step": 2494
    },
    {
      "epoch": 0.6686319174594667,
      "grad_norm": 0.6619631095286712,
      "learning_rate": 4.85886739022677e-06,
      "loss": 0.579,
      "step": 2495
    },
    {
      "epoch": 0.6688999062039395,
      "grad_norm": 0.6721423144685701,
      "learning_rate": 4.858750632362779e-06,
      "loss": 0.5973,
      "step": 2496
    },
    {
      "epoch": 0.6691678949484121,
      "grad_norm": 0.6421511822383734,
      "learning_rate": 4.8586338276263524e-06,
      "loss": 0.5777,
      "step": 2497
    },
    {
      "epoch": 0.6694358836928849,
      "grad_norm": 0.6637891437471325,
      "learning_rate": 4.85851697601981e-06,
      "loss": 0.5986,
      "step": 2498
    },
    {
      "epoch": 0.6697038724373576,
      "grad_norm": 0.6737476020261842,
      "learning_rate": 4.858400077545477e-06,
      "loss": 0.6123,
      "step": 2499
    },
    {
      "epoch": 0.6699718611818304,
      "grad_norm": 0.7073131155073951,
      "learning_rate": 4.858283132205673e-06,
      "loss": 0.6105,
      "step": 2500
    },
    {
      "epoch": 0.6702398499263031,
      "grad_norm": 0.6880260880453746,
      "learning_rate": 4.858166140002724e-06,
      "loss": 0.6007,
      "step": 2501
    },
    {
      "epoch": 0.6705078386707758,
      "grad_norm": 0.6784591154290162,
      "learning_rate": 4.858049100938954e-06,
      "loss": 0.5921,
      "step": 2502
    },
    {
      "epoch": 0.6707758274152485,
      "grad_norm": 0.6746766583902135,
      "learning_rate": 4.8579320150166885e-06,
      "loss": 0.553,
      "step": 2503
    },
    {
      "epoch": 0.6710438161597213,
      "grad_norm": 0.6579367482207009,
      "learning_rate": 4.857814882238255e-06,
      "loss": 0.6134,
      "step": 2504
    },
    {
      "epoch": 0.6713118049041941,
      "grad_norm": 0.6487517481800434,
      "learning_rate": 4.857697702605981e-06,
      "loss": 0.5717,
      "step": 2505
    },
    {
      "epoch": 0.6715797936486667,
      "grad_norm": 0.6536157187488336,
      "learning_rate": 4.8575804761221944e-06,
      "loss": 0.5321,
      "step": 2506
    },
    {
      "epoch": 0.6718477823931395,
      "grad_norm": 0.6584091005177616,
      "learning_rate": 4.857463202789225e-06,
      "loss": 0.6053,
      "step": 2507
    },
    {
      "epoch": 0.6721157711376122,
      "grad_norm": 0.6557048645398158,
      "learning_rate": 4.857345882609404e-06,
      "loss": 0.5733,
      "step": 2508
    },
    {
      "epoch": 0.672383759882085,
      "grad_norm": 0.6576372482000459,
      "learning_rate": 4.857228515585063e-06,
      "loss": 0.593,
      "step": 2509
    },
    {
      "epoch": 0.6726517486265576,
      "grad_norm": 0.6779882634663464,
      "learning_rate": 4.857111101718531e-06,
      "loss": 0.5876,
      "step": 2510
    },
    {
      "epoch": 0.6729197373710304,
      "grad_norm": 0.6612261306102958,
      "learning_rate": 4.856993641012145e-06,
      "loss": 0.5882,
      "step": 2511
    },
    {
      "epoch": 0.6731877261155031,
      "grad_norm": 0.6478348338151604,
      "learning_rate": 4.856876133468238e-06,
      "loss": 0.6095,
      "step": 2512
    },
    {
      "epoch": 0.6734557148599759,
      "grad_norm": 0.6481919538844253,
      "learning_rate": 4.856758579089145e-06,
      "loss": 0.5897,
      "step": 2513
    },
    {
      "epoch": 0.6737237036044487,
      "grad_norm": 0.6456860728354603,
      "learning_rate": 4.856640977877201e-06,
      "loss": 0.5657,
      "step": 2514
    },
    {
      "epoch": 0.6739916923489213,
      "grad_norm": 0.6478596388312234,
      "learning_rate": 4.856523329834744e-06,
      "loss": 0.6242,
      "step": 2515
    },
    {
      "epoch": 0.6742596810933941,
      "grad_norm": 0.6738501692494997,
      "learning_rate": 4.856405634964111e-06,
      "loss": 0.606,
      "step": 2516
    },
    {
      "epoch": 0.6745276698378668,
      "grad_norm": 0.6514349256384586,
      "learning_rate": 4.8562878932676425e-06,
      "loss": 0.5708,
      "step": 2517
    },
    {
      "epoch": 0.6747956585823396,
      "grad_norm": 0.6757077104481876,
      "learning_rate": 4.856170104747677e-06,
      "loss": 0.5862,
      "step": 2518
    },
    {
      "epoch": 0.6750636473268122,
      "grad_norm": 0.6570003144693087,
      "learning_rate": 4.856052269406555e-06,
      "loss": 0.5936,
      "step": 2519
    },
    {
      "epoch": 0.675331636071285,
      "grad_norm": 0.643655154189106,
      "learning_rate": 4.855934387246619e-06,
      "loss": 0.5888,
      "step": 2520
    },
    {
      "epoch": 0.6755996248157577,
      "grad_norm": 0.6501799306962293,
      "learning_rate": 4.85581645827021e-06,
      "loss": 0.5785,
      "step": 2521
    },
    {
      "epoch": 0.6758676135602305,
      "grad_norm": 0.6862345387097251,
      "learning_rate": 4.855698482479672e-06,
      "loss": 0.5815,
      "step": 2522
    },
    {
      "epoch": 0.6761356023047032,
      "grad_norm": 0.6725515471732948,
      "learning_rate": 4.855580459877352e-06,
      "loss": 0.5753,
      "step": 2523
    },
    {
      "epoch": 0.6764035910491759,
      "grad_norm": 0.6674373421920102,
      "learning_rate": 4.855462390465591e-06,
      "loss": 0.5981,
      "step": 2524
    },
    {
      "epoch": 0.6766715797936487,
      "grad_norm": 0.680492129580542,
      "learning_rate": 4.855344274246738e-06,
      "loss": 0.6121,
      "step": 2525
    },
    {
      "epoch": 0.6769395685381214,
      "grad_norm": 0.6640316964040206,
      "learning_rate": 4.85522611122314e-06,
      "loss": 0.5901,
      "step": 2526
    },
    {
      "epoch": 0.6772075572825942,
      "grad_norm": 0.6618494251774601,
      "learning_rate": 4.855107901397144e-06,
      "loss": 0.591,
      "step": 2527
    },
    {
      "epoch": 0.6774755460270668,
      "grad_norm": 0.6527887891893167,
      "learning_rate": 4.8549896447711e-06,
      "loss": 0.5891,
      "step": 2528
    },
    {
      "epoch": 0.6777435347715396,
      "grad_norm": 0.6559370170582656,
      "learning_rate": 4.854871341347357e-06,
      "loss": 0.6043,
      "step": 2529
    },
    {
      "epoch": 0.6780115235160123,
      "grad_norm": 0.680335437029257,
      "learning_rate": 4.854752991128268e-06,
      "loss": 0.6125,
      "step": 2530
    },
    {
      "epoch": 0.6782795122604851,
      "grad_norm": 0.6747811305230645,
      "learning_rate": 4.854634594116182e-06,
      "loss": 0.5858,
      "step": 2531
    },
    {
      "epoch": 0.6785475010049578,
      "grad_norm": 0.6796488445257324,
      "learning_rate": 4.854516150313453e-06,
      "loss": 0.6281,
      "step": 2532
    },
    {
      "epoch": 0.6788154897494305,
      "grad_norm": 0.6477274055274949,
      "learning_rate": 4.854397659722435e-06,
      "loss": 0.5952,
      "step": 2533
    },
    {
      "epoch": 0.6790834784939033,
      "grad_norm": 0.6495980731740536,
      "learning_rate": 4.854279122345482e-06,
      "loss": 0.5764,
      "step": 2534
    },
    {
      "epoch": 0.679351467238376,
      "grad_norm": 0.6567480530667632,
      "learning_rate": 4.854160538184951e-06,
      "loss": 0.5688,
      "step": 2535
    },
    {
      "epoch": 0.6796194559828487,
      "grad_norm": 0.6533065729038855,
      "learning_rate": 4.854041907243197e-06,
      "loss": 0.5736,
      "step": 2536
    },
    {
      "epoch": 0.6798874447273214,
      "grad_norm": 0.6868804856368254,
      "learning_rate": 4.8539232295225775e-06,
      "loss": 0.5759,
      "step": 2537
    },
    {
      "epoch": 0.6801554334717942,
      "grad_norm": 0.6621316693291577,
      "learning_rate": 4.853804505025451e-06,
      "loss": 0.6199,
      "step": 2538
    },
    {
      "epoch": 0.680423422216267,
      "grad_norm": 0.6558892269196795,
      "learning_rate": 4.853685733754177e-06,
      "loss": 0.5642,
      "step": 2539
    },
    {
      "epoch": 0.6806914109607396,
      "grad_norm": 0.6522784827600088,
      "learning_rate": 4.8535669157111154e-06,
      "loss": 0.5714,
      "step": 2540
    },
    {
      "epoch": 0.6809593997052124,
      "grad_norm": 0.7468148324395066,
      "learning_rate": 4.853448050898628e-06,
      "loss": 0.6246,
      "step": 2541
    },
    {
      "epoch": 0.6812273884496851,
      "grad_norm": 0.6792262553251798,
      "learning_rate": 4.853329139319076e-06,
      "loss": 0.5867,
      "step": 2542
    },
    {
      "epoch": 0.6814953771941579,
      "grad_norm": 0.6641150682279979,
      "learning_rate": 4.853210180974823e-06,
      "loss": 0.5896,
      "step": 2543
    },
    {
      "epoch": 0.6817633659386306,
      "grad_norm": 0.6557236127341773,
      "learning_rate": 4.853091175868232e-06,
      "loss": 0.5868,
      "step": 2544
    },
    {
      "epoch": 0.6820313546831033,
      "grad_norm": 0.6593360112392035,
      "learning_rate": 4.852972124001669e-06,
      "loss": 0.6165,
      "step": 2545
    },
    {
      "epoch": 0.682299343427576,
      "grad_norm": 0.6616777998590199,
      "learning_rate": 4.852853025377499e-06,
      "loss": 0.6276,
      "step": 2546
    },
    {
      "epoch": 0.6825673321720488,
      "grad_norm": 0.6651008193440886,
      "learning_rate": 4.8527338799980885e-06,
      "loss": 0.5867,
      "step": 2547
    },
    {
      "epoch": 0.6828353209165215,
      "grad_norm": 0.6716912105725182,
      "learning_rate": 4.852614687865806e-06,
      "loss": 0.5722,
      "step": 2548
    },
    {
      "epoch": 0.6831033096609942,
      "grad_norm": 0.6441120052221373,
      "learning_rate": 4.852495448983019e-06,
      "loss": 0.6034,
      "step": 2549
    },
    {
      "epoch": 0.683371298405467,
      "grad_norm": 0.6717748957380288,
      "learning_rate": 4.852376163352098e-06,
      "loss": 0.6121,
      "step": 2550
    },
    {
      "epoch": 0.6836392871499397,
      "grad_norm": 0.676320280766163,
      "learning_rate": 4.852256830975414e-06,
      "loss": 0.6167,
      "step": 2551
    },
    {
      "epoch": 0.6839072758944125,
      "grad_norm": 0.6787922444542865,
      "learning_rate": 4.8521374518553365e-06,
      "loss": 0.6004,
      "step": 2552
    },
    {
      "epoch": 0.6841752646388851,
      "grad_norm": 0.6837083661498069,
      "learning_rate": 4.852018025994239e-06,
      "loss": 0.6147,
      "step": 2553
    },
    {
      "epoch": 0.6844432533833579,
      "grad_norm": 0.66386482807909,
      "learning_rate": 4.851898553394493e-06,
      "loss": 0.5655,
      "step": 2554
    },
    {
      "epoch": 0.6847112421278306,
      "grad_norm": 0.6569182944947033,
      "learning_rate": 4.851779034058475e-06,
      "loss": 0.5758,
      "step": 2555
    },
    {
      "epoch": 0.6849792308723034,
      "grad_norm": 0.6338182165665993,
      "learning_rate": 4.851659467988558e-06,
      "loss": 0.5531,
      "step": 2556
    },
    {
      "epoch": 0.6852472196167761,
      "grad_norm": 0.6476164299362088,
      "learning_rate": 4.851539855187119e-06,
      "loss": 0.5672,
      "step": 2557
    },
    {
      "epoch": 0.6855152083612488,
      "grad_norm": 0.6746304709985154,
      "learning_rate": 4.8514201956565364e-06,
      "loss": 0.5959,
      "step": 2558
    },
    {
      "epoch": 0.6857831971057216,
      "grad_norm": 0.6511760210266698,
      "learning_rate": 4.8513004893991855e-06,
      "loss": 0.5502,
      "step": 2559
    },
    {
      "epoch": 0.6860511858501943,
      "grad_norm": 0.6863755614072066,
      "learning_rate": 4.8511807364174465e-06,
      "loss": 0.5972,
      "step": 2560
    },
    {
      "epoch": 0.6863191745946671,
      "grad_norm": 0.6919239787272385,
      "learning_rate": 4.851060936713698e-06,
      "loss": 0.5659,
      "step": 2561
    },
    {
      "epoch": 0.6865871633391397,
      "grad_norm": 0.6592514021856213,
      "learning_rate": 4.850941090290322e-06,
      "loss": 0.5636,
      "step": 2562
    },
    {
      "epoch": 0.6868551520836125,
      "grad_norm": 0.6717563090625515,
      "learning_rate": 4.850821197149699e-06,
      "loss": 0.6094,
      "step": 2563
    },
    {
      "epoch": 0.6871231408280852,
      "grad_norm": 0.657310958346881,
      "learning_rate": 4.850701257294212e-06,
      "loss": 0.5754,
      "step": 2564
    },
    {
      "epoch": 0.687391129572558,
      "grad_norm": 0.6689186074773306,
      "learning_rate": 4.850581270726243e-06,
      "loss": 0.6083,
      "step": 2565
    },
    {
      "epoch": 0.6876591183170306,
      "grad_norm": 0.6414863705852704,
      "learning_rate": 4.850461237448179e-06,
      "loss": 0.5857,
      "step": 2566
    },
    {
      "epoch": 0.6879271070615034,
      "grad_norm": 0.6752095025893469,
      "learning_rate": 4.850341157462404e-06,
      "loss": 0.5525,
      "step": 2567
    },
    {
      "epoch": 0.6881950958059762,
      "grad_norm": 0.6672801386509692,
      "learning_rate": 4.850221030771303e-06,
      "loss": 0.5767,
      "step": 2568
    },
    {
      "epoch": 0.6884630845504489,
      "grad_norm": 0.6930774914257243,
      "learning_rate": 4.850100857377265e-06,
      "loss": 0.5979,
      "step": 2569
    },
    {
      "epoch": 0.6887310732949216,
      "grad_norm": 0.6594576794069845,
      "learning_rate": 4.849980637282677e-06,
      "loss": 0.591,
      "step": 2570
    },
    {
      "epoch": 0.6889990620393943,
      "grad_norm": 0.6865722908512877,
      "learning_rate": 4.849860370489928e-06,
      "loss": 0.5977,
      "step": 2571
    },
    {
      "epoch": 0.6892670507838671,
      "grad_norm": 0.692509476828724,
      "learning_rate": 4.849740057001408e-06,
      "loss": 0.5819,
      "step": 2572
    },
    {
      "epoch": 0.6895350395283398,
      "grad_norm": 0.6850362174929904,
      "learning_rate": 4.849619696819509e-06,
      "loss": 0.5798,
      "step": 2573
    },
    {
      "epoch": 0.6898030282728126,
      "grad_norm": 0.7075235518443284,
      "learning_rate": 4.8494992899466205e-06,
      "loss": 0.6041,
      "step": 2574
    },
    {
      "epoch": 0.6900710170172852,
      "grad_norm": 0.7054036652745108,
      "learning_rate": 4.849378836385137e-06,
      "loss": 0.6196,
      "step": 2575
    },
    {
      "epoch": 0.690339005761758,
      "grad_norm": 0.7119669605409736,
      "learning_rate": 4.849258336137451e-06,
      "loss": 0.5713,
      "step": 2576
    },
    {
      "epoch": 0.6906069945062308,
      "grad_norm": 0.6533896113599172,
      "learning_rate": 4.849137789205957e-06,
      "loss": 0.5985,
      "step": 2577
    },
    {
      "epoch": 0.6908749832507035,
      "grad_norm": 0.6568521483942803,
      "learning_rate": 4.849017195593053e-06,
      "loss": 0.5853,
      "step": 2578
    },
    {
      "epoch": 0.6911429719951762,
      "grad_norm": 0.6678025844863266,
      "learning_rate": 4.848896555301133e-06,
      "loss": 0.5941,
      "step": 2579
    },
    {
      "epoch": 0.6914109607396489,
      "grad_norm": 0.704295486722959,
      "learning_rate": 4.848775868332594e-06,
      "loss": 0.5778,
      "step": 2580
    },
    {
      "epoch": 0.6916789494841217,
      "grad_norm": 0.6693421556139377,
      "learning_rate": 4.8486551346898346e-06,
      "loss": 0.6384,
      "step": 2581
    },
    {
      "epoch": 0.6919469382285944,
      "grad_norm": 0.6715099715425945,
      "learning_rate": 4.848534354375255e-06,
      "loss": 0.5762,
      "step": 2582
    },
    {
      "epoch": 0.6922149269730671,
      "grad_norm": 0.6765332281391849,
      "learning_rate": 4.848413527391255e-06,
      "loss": 0.5627,
      "step": 2583
    },
    {
      "epoch": 0.6924829157175398,
      "grad_norm": 0.6702812568732157,
      "learning_rate": 4.848292653740235e-06,
      "loss": 0.5734,
      "step": 2584
    },
    {
      "epoch": 0.6927509044620126,
      "grad_norm": 0.6788785024389318,
      "learning_rate": 4.848171733424598e-06,
      "loss": 0.6015,
      "step": 2585
    },
    {
      "epoch": 0.6930188932064854,
      "grad_norm": 0.6718625086010538,
      "learning_rate": 4.848050766446746e-06,
      "loss": 0.5992,
      "step": 2586
    },
    {
      "epoch": 0.6932868819509581,
      "grad_norm": 0.6480684223540609,
      "learning_rate": 4.847929752809083e-06,
      "loss": 0.5978,
      "step": 2587
    },
    {
      "epoch": 0.6935548706954308,
      "grad_norm": 0.6608948107474599,
      "learning_rate": 4.847808692514014e-06,
      "loss": 0.546,
      "step": 2588
    },
    {
      "epoch": 0.6938228594399035,
      "grad_norm": 0.6823876727497372,
      "learning_rate": 4.847687585563944e-06,
      "loss": 0.5996,
      "step": 2589
    },
    {
      "epoch": 0.6940908481843763,
      "grad_norm": 0.6838150107285762,
      "learning_rate": 4.84756643196128e-06,
      "loss": 0.6056,
      "step": 2590
    },
    {
      "epoch": 0.694358836928849,
      "grad_norm": 0.6485436798736592,
      "learning_rate": 4.847445231708429e-06,
      "loss": 0.6038,
      "step": 2591
    },
    {
      "epoch": 0.6946268256733217,
      "grad_norm": 0.6485272820787401,
      "learning_rate": 4.847323984807802e-06,
      "loss": 0.5871,
      "step": 2592
    },
    {
      "epoch": 0.6948948144177944,
      "grad_norm": 0.655457451991121,
      "learning_rate": 4.847202691261806e-06,
      "loss": 0.6252,
      "step": 2593
    },
    {
      "epoch": 0.6951628031622672,
      "grad_norm": 0.6543961087501741,
      "learning_rate": 4.84708135107285e-06,
      "loss": 0.6014,
      "step": 2594
    },
    {
      "epoch": 0.69543079190674,
      "grad_norm": 0.6605574875788125,
      "learning_rate": 4.846959964243349e-06,
      "loss": 0.6228,
      "step": 2595
    },
    {
      "epoch": 0.6956987806512126,
      "grad_norm": 0.6626763716263318,
      "learning_rate": 4.846838530775712e-06,
      "loss": 0.6206,
      "step": 2596
    },
    {
      "epoch": 0.6959667693956854,
      "grad_norm": 0.6443652124211312,
      "learning_rate": 4.846717050672354e-06,
      "loss": 0.5748,
      "step": 2597
    },
    {
      "epoch": 0.6962347581401581,
      "grad_norm": 0.6548636505711205,
      "learning_rate": 4.846595523935688e-06,
      "loss": 0.598,
      "step": 2598
    },
    {
      "epoch": 0.6965027468846309,
      "grad_norm": 0.6623113967319123,
      "learning_rate": 4.846473950568129e-06,
      "loss": 0.6053,
      "step": 2599
    },
    {
      "epoch": 0.6967707356291036,
      "grad_norm": 0.6760455418183109,
      "learning_rate": 4.846352330572092e-06,
      "loss": 0.5954,
      "step": 2600
    },
    {
      "epoch": 0.6970387243735763,
      "grad_norm": 0.666374590142564,
      "learning_rate": 4.846230663949998e-06,
      "loss": 0.606,
      "step": 2601
    },
    {
      "epoch": 0.697306713118049,
      "grad_norm": 0.6790410705701511,
      "learning_rate": 4.846108950704259e-06,
      "loss": 0.6086,
      "step": 2602
    },
    {
      "epoch": 0.6975747018625218,
      "grad_norm": 0.6723255428761172,
      "learning_rate": 4.8459871908372965e-06,
      "loss": 0.5793,
      "step": 2603
    },
    {
      "epoch": 0.6978426906069946,
      "grad_norm": 0.6585892369528579,
      "learning_rate": 4.845865384351532e-06,
      "loss": 0.618,
      "step": 2604
    },
    {
      "epoch": 0.6981106793514672,
      "grad_norm": 0.6566319085828848,
      "learning_rate": 4.845743531249381e-06,
      "loss": 0.5863,
      "step": 2605
    },
    {
      "epoch": 0.69837866809594,
      "grad_norm": 0.6966931288365673,
      "learning_rate": 4.84562163153327e-06,
      "loss": 0.582,
      "step": 2606
    },
    {
      "epoch": 0.6986466568404127,
      "grad_norm": 0.6679821108238697,
      "learning_rate": 4.845499685205618e-06,
      "loss": 0.5648,
      "step": 2607
    },
    {
      "epoch": 0.6989146455848855,
      "grad_norm": 0.6914143537832459,
      "learning_rate": 4.84537769226885e-06,
      "loss": 0.5352,
      "step": 2608
    },
    {
      "epoch": 0.6991826343293581,
      "grad_norm": 0.6777304972724039,
      "learning_rate": 4.845255652725389e-06,
      "loss": 0.5724,
      "step": 2609
    },
    {
      "epoch": 0.6994506230738309,
      "grad_norm": 0.7161353422298719,
      "learning_rate": 4.845133566577661e-06,
      "loss": 0.5825,
      "step": 2610
    },
    {
      "epoch": 0.6997186118183036,
      "grad_norm": 0.6578113221947194,
      "learning_rate": 4.845011433828093e-06,
      "loss": 0.5864,
      "step": 2611
    },
    {
      "epoch": 0.6999866005627764,
      "grad_norm": 0.6526267667459,
      "learning_rate": 4.844889254479109e-06,
      "loss": 0.6042,
      "step": 2612
    },
    {
      "epoch": 0.700254589307249,
      "grad_norm": 0.6593956534409463,
      "learning_rate": 4.84476702853314e-06,
      "loss": 0.5944,
      "step": 2613
    },
    {
      "epoch": 0.7005225780517218,
      "grad_norm": 0.660984743881293,
      "learning_rate": 4.844644755992613e-06,
      "loss": 0.6138,
      "step": 2614
    },
    {
      "epoch": 0.7007905667961946,
      "grad_norm": 0.6950918254458353,
      "learning_rate": 4.844522436859958e-06,
      "loss": 0.5966,
      "step": 2615
    },
    {
      "epoch": 0.7010585555406673,
      "grad_norm": 0.6579121653867644,
      "learning_rate": 4.844400071137607e-06,
      "loss": 0.5945,
      "step": 2616
    },
    {
      "epoch": 0.7013265442851401,
      "grad_norm": 0.65615094455226,
      "learning_rate": 4.84427765882799e-06,
      "loss": 0.581,
      "step": 2617
    },
    {
      "epoch": 0.7015945330296127,
      "grad_norm": 0.6667553350586981,
      "learning_rate": 4.844155199933541e-06,
      "loss": 0.5906,
      "step": 2618
    },
    {
      "epoch": 0.7018625217740855,
      "grad_norm": 0.6791466809272683,
      "learning_rate": 4.844032694456692e-06,
      "loss": 0.5897,
      "step": 2619
    },
    {
      "epoch": 0.7021305105185582,
      "grad_norm": 0.6664766155718668,
      "learning_rate": 4.843910142399878e-06,
      "loss": 0.5827,
      "step": 2620
    },
    {
      "epoch": 0.702398499263031,
      "grad_norm": 0.6533081220924032,
      "learning_rate": 4.843787543765535e-06,
      "loss": 0.5913,
      "step": 2621
    },
    {
      "epoch": 0.7026664880075036,
      "grad_norm": 0.7006971725579735,
      "learning_rate": 4.8436648985560986e-06,
      "loss": 0.6217,
      "step": 2622
    },
    {
      "epoch": 0.7029344767519764,
      "grad_norm": 0.6570347349363096,
      "learning_rate": 4.8435422067740055e-06,
      "loss": 0.6128,
      "step": 2623
    },
    {
      "epoch": 0.7032024654964492,
      "grad_norm": 0.668891668604825,
      "learning_rate": 4.8434194684216945e-06,
      "loss": 0.562,
      "step": 2624
    },
    {
      "epoch": 0.7034704542409219,
      "grad_norm": 0.638823914123726,
      "learning_rate": 4.843296683501604e-06,
      "loss": 0.5573,
      "step": 2625
    },
    {
      "epoch": 0.7037384429853946,
      "grad_norm": 0.6524035157200643,
      "learning_rate": 4.843173852016175e-06,
      "loss": 0.5684,
      "step": 2626
    },
    {
      "epoch": 0.7040064317298673,
      "grad_norm": 0.6527209831560449,
      "learning_rate": 4.843050973967848e-06,
      "loss": 0.6,
      "step": 2627
    },
    {
      "epoch": 0.7042744204743401,
      "grad_norm": 0.654165984454427,
      "learning_rate": 4.8429280493590634e-06,
      "loss": 0.5953,
      "step": 2628
    },
    {
      "epoch": 0.7045424092188128,
      "grad_norm": 0.6363036139409866,
      "learning_rate": 4.842805078192267e-06,
      "loss": 0.5681,
      "step": 2629
    },
    {
      "epoch": 0.7048103979632856,
      "grad_norm": 0.663632434579929,
      "learning_rate": 4.842682060469899e-06,
      "loss": 0.5924,
      "step": 2630
    },
    {
      "epoch": 0.7050783867077582,
      "grad_norm": 0.6650850375149893,
      "learning_rate": 4.842558996194405e-06,
      "loss": 0.5902,
      "step": 2631
    },
    {
      "epoch": 0.705346375452231,
      "grad_norm": 0.6739780850862143,
      "learning_rate": 4.8424358853682315e-06,
      "loss": 0.6077,
      "step": 2632
    },
    {
      "epoch": 0.7056143641967038,
      "grad_norm": 0.6741020129049475,
      "learning_rate": 4.842312727993826e-06,
      "loss": 0.5821,
      "step": 2633
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.6470697385549887,
      "learning_rate": 4.842189524073631e-06,
      "loss": 0.5653,
      "step": 2634
    },
    {
      "epoch": 0.7061503416856492,
      "grad_norm": 0.6753098729753906,
      "learning_rate": 4.842066273610101e-06,
      "loss": 0.5939,
      "step": 2635
    },
    {
      "epoch": 0.7064183304301219,
      "grad_norm": 0.666843781353632,
      "learning_rate": 4.841942976605681e-06,
      "loss": 0.584,
      "step": 2636
    },
    {
      "epoch": 0.7066863191745947,
      "grad_norm": 0.7063099901696865,
      "learning_rate": 4.841819633062823e-06,
      "loss": 0.5823,
      "step": 2637
    },
    {
      "epoch": 0.7069543079190674,
      "grad_norm": 0.6865487214063042,
      "learning_rate": 4.8416962429839764e-06,
      "loss": 0.5749,
      "step": 2638
    },
    {
      "epoch": 0.7072222966635401,
      "grad_norm": 0.6794514917602433,
      "learning_rate": 4.841572806371595e-06,
      "loss": 0.5952,
      "step": 2639
    },
    {
      "epoch": 0.7074902854080128,
      "grad_norm": 0.6774611929552347,
      "learning_rate": 4.841449323228131e-06,
      "loss": 0.5994,
      "step": 2640
    },
    {
      "epoch": 0.7077582741524856,
      "grad_norm": 0.6513754832804897,
      "learning_rate": 4.841325793556038e-06,
      "loss": 0.568,
      "step": 2641
    },
    {
      "epoch": 0.7080262628969584,
      "grad_norm": 0.6400300357204026,
      "learning_rate": 4.84120221735777e-06,
      "loss": 0.6024,
      "step": 2642
    },
    {
      "epoch": 0.708294251641431,
      "grad_norm": 0.6533732297833068,
      "learning_rate": 4.841078594635784e-06,
      "loss": 0.6026,
      "step": 2643
    },
    {
      "epoch": 0.7085622403859038,
      "grad_norm": 0.6990039189525638,
      "learning_rate": 4.840954925392536e-06,
      "loss": 0.6293,
      "step": 2644
    },
    {
      "epoch": 0.7088302291303765,
      "grad_norm": 0.6598468889892585,
      "learning_rate": 4.840831209630484e-06,
      "loss": 0.5712,
      "step": 2645
    },
    {
      "epoch": 0.7090982178748493,
      "grad_norm": 0.6818700133798227,
      "learning_rate": 4.840707447352087e-06,
      "loss": 0.5981,
      "step": 2646
    },
    {
      "epoch": 0.709366206619322,
      "grad_norm": 0.6886688530977887,
      "learning_rate": 4.8405836385598016e-06,
      "loss": 0.5842,
      "step": 2647
    },
    {
      "epoch": 0.7096341953637947,
      "grad_norm": 0.6531289713422886,
      "learning_rate": 4.840459783256091e-06,
      "loss": 0.6024,
      "step": 2648
    },
    {
      "epoch": 0.7099021841082674,
      "grad_norm": 0.6880702556633296,
      "learning_rate": 4.840335881443415e-06,
      "loss": 0.6044,
      "step": 2649
    },
    {
      "epoch": 0.7101701728527402,
      "grad_norm": 0.6978795630805755,
      "learning_rate": 4.840211933124237e-06,
      "loss": 0.5774,
      "step": 2650
    },
    {
      "epoch": 0.710438161597213,
      "grad_norm": 0.6705580387676805,
      "learning_rate": 4.840087938301018e-06,
      "loss": 0.5917,
      "step": 2651
    },
    {
      "epoch": 0.7107061503416856,
      "grad_norm": 0.6464966147449296,
      "learning_rate": 4.839963896976223e-06,
      "loss": 0.5865,
      "step": 2652
    },
    {
      "epoch": 0.7109741390861584,
      "grad_norm": 0.7072774276220654,
      "learning_rate": 4.839839809152318e-06,
      "loss": 0.5529,
      "step": 2653
    },
    {
      "epoch": 0.7112421278306311,
      "grad_norm": 0.6670382462646588,
      "learning_rate": 4.839715674831767e-06,
      "loss": 0.5358,
      "step": 2654
    },
    {
      "epoch": 0.7115101165751039,
      "grad_norm": 0.6430123223781367,
      "learning_rate": 4.839591494017038e-06,
      "loss": 0.573,
      "step": 2655
    },
    {
      "epoch": 0.7117781053195765,
      "grad_norm": 0.6808603591181945,
      "learning_rate": 4.839467266710599e-06,
      "loss": 0.5486,
      "step": 2656
    },
    {
      "epoch": 0.7120460940640493,
      "grad_norm": 0.6555049139999665,
      "learning_rate": 4.839342992914918e-06,
      "loss": 0.5806,
      "step": 2657
    },
    {
      "epoch": 0.712314082808522,
      "grad_norm": 0.6828035654704242,
      "learning_rate": 4.839218672632464e-06,
      "loss": 0.5886,
      "step": 2658
    },
    {
      "epoch": 0.7125820715529948,
      "grad_norm": 0.6798422037734261,
      "learning_rate": 4.839094305865708e-06,
      "loss": 0.5756,
      "step": 2659
    },
    {
      "epoch": 0.7128500602974676,
      "grad_norm": 0.6896323961754109,
      "learning_rate": 4.838969892617122e-06,
      "loss": 0.5795,
      "step": 2660
    },
    {
      "epoch": 0.7131180490419402,
      "grad_norm": 0.6642657896268868,
      "learning_rate": 4.838845432889177e-06,
      "loss": 0.5838,
      "step": 2661
    },
    {
      "epoch": 0.713386037786413,
      "grad_norm": 0.6647546553554954,
      "learning_rate": 4.838720926684347e-06,
      "loss": 0.5789,
      "step": 2662
    },
    {
      "epoch": 0.7136540265308857,
      "grad_norm": 0.7431283136987687,
      "learning_rate": 4.838596374005106e-06,
      "loss": 0.5788,
      "step": 2663
    },
    {
      "epoch": 0.7139220152753585,
      "grad_norm": 0.65688726748899,
      "learning_rate": 4.838471774853929e-06,
      "loss": 0.5819,
      "step": 2664
    },
    {
      "epoch": 0.7141900040198311,
      "grad_norm": 0.6430908343037002,
      "learning_rate": 4.8383471292332924e-06,
      "loss": 0.5857,
      "step": 2665
    },
    {
      "epoch": 0.7144579927643039,
      "grad_norm": 0.6614629787367357,
      "learning_rate": 4.838222437145673e-06,
      "loss": 0.6229,
      "step": 2666
    },
    {
      "epoch": 0.7147259815087766,
      "grad_norm": 0.6493382704984886,
      "learning_rate": 4.8380976985935476e-06,
      "loss": 0.5706,
      "step": 2667
    },
    {
      "epoch": 0.7149939702532494,
      "grad_norm": 0.6539538160694472,
      "learning_rate": 4.837972913579397e-06,
      "loss": 0.5744,
      "step": 2668
    },
    {
      "epoch": 0.715261958997722,
      "grad_norm": 0.6355226129552618,
      "learning_rate": 4.8378480821057e-06,
      "loss": 0.5352,
      "step": 2669
    },
    {
      "epoch": 0.7155299477421948,
      "grad_norm": 0.6655552072966543,
      "learning_rate": 4.837723204174935e-06,
      "loss": 0.5733,
      "step": 2670
    },
    {
      "epoch": 0.7157979364866676,
      "grad_norm": 0.6574365304720134,
      "learning_rate": 4.8375982797895875e-06,
      "loss": 0.5883,
      "step": 2671
    },
    {
      "epoch": 0.7160659252311403,
      "grad_norm": 0.6534237990797156,
      "learning_rate": 4.837473308952136e-06,
      "loss": 0.5824,
      "step": 2672
    },
    {
      "epoch": 0.7163339139756131,
      "grad_norm": 0.661755701083592,
      "learning_rate": 4.837348291665068e-06,
      "loss": 0.5674,
      "step": 2673
    },
    {
      "epoch": 0.7166019027200857,
      "grad_norm": 0.6535142761080797,
      "learning_rate": 4.837223227930864e-06,
      "loss": 0.5675,
      "step": 2674
    },
    {
      "epoch": 0.7168698914645585,
      "grad_norm": 0.6604375487007299,
      "learning_rate": 4.8370981177520124e-06,
      "loss": 0.5725,
      "step": 2675
    },
    {
      "epoch": 0.7171378802090312,
      "grad_norm": 0.6585430546747404,
      "learning_rate": 4.836972961130997e-06,
      "loss": 0.5996,
      "step": 2676
    },
    {
      "epoch": 0.717405868953504,
      "grad_norm": 0.6733462207892329,
      "learning_rate": 4.836847758070305e-06,
      "loss": 0.6349,
      "step": 2677
    },
    {
      "epoch": 0.7176738576979766,
      "grad_norm": 0.6534365784378243,
      "learning_rate": 4.836722508572426e-06,
      "loss": 0.5903,
      "step": 2678
    },
    {
      "epoch": 0.7179418464424494,
      "grad_norm": 0.6627193025879372,
      "learning_rate": 4.836597212639848e-06,
      "loss": 0.5798,
      "step": 2679
    },
    {
      "epoch": 0.7182098351869222,
      "grad_norm": 0.6737486732134451,
      "learning_rate": 4.836471870275061e-06,
      "loss": 0.619,
      "step": 2680
    },
    {
      "epoch": 0.7184778239313949,
      "grad_norm": 0.6701018905269942,
      "learning_rate": 4.836346481480556e-06,
      "loss": 0.5513,
      "step": 2681
    },
    {
      "epoch": 0.7187458126758676,
      "grad_norm": 0.6570688402543312,
      "learning_rate": 4.8362210462588235e-06,
      "loss": 0.6035,
      "step": 2682
    },
    {
      "epoch": 0.7190138014203403,
      "grad_norm": 0.658619087888974,
      "learning_rate": 4.836095564612357e-06,
      "loss": 0.59,
      "step": 2683
    },
    {
      "epoch": 0.7192817901648131,
      "grad_norm": 0.6704135703508517,
      "learning_rate": 4.83597003654365e-06,
      "loss": 0.5838,
      "step": 2684
    },
    {
      "epoch": 0.7195497789092858,
      "grad_norm": 0.6969634560369278,
      "learning_rate": 4.835844462055198e-06,
      "loss": 0.5817,
      "step": 2685
    },
    {
      "epoch": 0.7198177676537585,
      "grad_norm": 0.6640317854923744,
      "learning_rate": 4.835718841149494e-06,
      "loss": 0.5709,
      "step": 2686
    },
    {
      "epoch": 0.7200857563982312,
      "grad_norm": 0.6547190287233914,
      "learning_rate": 4.8355931738290365e-06,
      "loss": 0.6004,
      "step": 2687
    },
    {
      "epoch": 0.720353745142704,
      "grad_norm": 0.6946791548414575,
      "learning_rate": 4.835467460096321e-06,
      "loss": 0.5981,
      "step": 2688
    },
    {
      "epoch": 0.7206217338871768,
      "grad_norm": 0.6421648637101216,
      "learning_rate": 4.835341699953848e-06,
      "loss": 0.5989,
      "step": 2689
    },
    {
      "epoch": 0.7208897226316495,
      "grad_norm": 0.6389219621740597,
      "learning_rate": 4.835215893404114e-06,
      "loss": 0.5551,
      "step": 2690
    },
    {
      "epoch": 0.7211577113761222,
      "grad_norm": 0.6974130807885339,
      "learning_rate": 4.83509004044962e-06,
      "loss": 0.6215,
      "step": 2691
    },
    {
      "epoch": 0.7214257001205949,
      "grad_norm": 0.6563401706951528,
      "learning_rate": 4.8349641410928676e-06,
      "loss": 0.5926,
      "step": 2692
    },
    {
      "epoch": 0.7216936888650677,
      "grad_norm": 0.6531401722226804,
      "learning_rate": 4.834838195336358e-06,
      "loss": 0.555,
      "step": 2693
    },
    {
      "epoch": 0.7219616776095404,
      "grad_norm": 0.6378141446240123,
      "learning_rate": 4.834712203182593e-06,
      "loss": 0.5741,
      "step": 2694
    },
    {
      "epoch": 0.7222296663540131,
      "grad_norm": 0.6673345529565252,
      "learning_rate": 4.834586164634078e-06,
      "loss": 0.5902,
      "step": 2695
    },
    {
      "epoch": 0.7224976550984858,
      "grad_norm": 0.6880586760648543,
      "learning_rate": 4.834460079693317e-06,
      "loss": 0.5878,
      "step": 2696
    },
    {
      "epoch": 0.7227656438429586,
      "grad_norm": 0.6391373619484333,
      "learning_rate": 4.834333948362816e-06,
      "loss": 0.5839,
      "step": 2697
    },
    {
      "epoch": 0.7230336325874314,
      "grad_norm": 0.7176737874288575,
      "learning_rate": 4.83420777064508e-06,
      "loss": 0.6098,
      "step": 2698
    },
    {
      "epoch": 0.723301621331904,
      "grad_norm": 0.6615883637705746,
      "learning_rate": 4.8340815465426176e-06,
      "loss": 0.5773,
      "step": 2699
    },
    {
      "epoch": 0.7235696100763768,
      "grad_norm": 0.6698431625496825,
      "learning_rate": 4.833955276057937e-06,
      "loss": 0.5805,
      "step": 2700
    },
    {
      "epoch": 0.7238375988208495,
      "grad_norm": 0.663602208054803,
      "learning_rate": 4.833828959193546e-06,
      "loss": 0.5949,
      "step": 2701
    },
    {
      "epoch": 0.7241055875653223,
      "grad_norm": 0.6800344635630144,
      "learning_rate": 4.833702595951957e-06,
      "loss": 0.6112,
      "step": 2702
    },
    {
      "epoch": 0.724373576309795,
      "grad_norm": 0.6757743948208808,
      "learning_rate": 4.833576186335679e-06,
      "loss": 0.5866,
      "step": 2703
    },
    {
      "epoch": 0.7246415650542677,
      "grad_norm": 0.6756402268014987,
      "learning_rate": 4.833449730347225e-06,
      "loss": 0.5675,
      "step": 2704
    },
    {
      "epoch": 0.7249095537987404,
      "grad_norm": 0.658204289947095,
      "learning_rate": 4.833323227989109e-06,
      "loss": 0.5922,
      "step": 2705
    },
    {
      "epoch": 0.7251775425432132,
      "grad_norm": 0.6906013818248596,
      "learning_rate": 4.8331966792638415e-06,
      "loss": 0.5616,
      "step": 2706
    },
    {
      "epoch": 0.725445531287686,
      "grad_norm": 0.7233301611027628,
      "learning_rate": 4.833070084173941e-06,
      "loss": 0.6202,
      "step": 2707
    },
    {
      "epoch": 0.7257135200321586,
      "grad_norm": 0.7061269222493438,
      "learning_rate": 4.832943442721922e-06,
      "loss": 0.5972,
      "step": 2708
    },
    {
      "epoch": 0.7259815087766314,
      "grad_norm": 0.6579314109746552,
      "learning_rate": 4.832816754910299e-06,
      "loss": 0.5508,
      "step": 2709
    },
    {
      "epoch": 0.7262494975211041,
      "grad_norm": 0.6710569756191843,
      "learning_rate": 4.832690020741592e-06,
      "loss": 0.5681,
      "step": 2710
    },
    {
      "epoch": 0.7265174862655769,
      "grad_norm": 0.6751913919327596,
      "learning_rate": 4.832563240218317e-06,
      "loss": 0.5924,
      "step": 2711
    },
    {
      "epoch": 0.7267854750100495,
      "grad_norm": 0.6960108345485668,
      "learning_rate": 4.832436413342997e-06,
      "loss": 0.5703,
      "step": 2712
    },
    {
      "epoch": 0.7270534637545223,
      "grad_norm": 0.6648192692210424,
      "learning_rate": 4.832309540118149e-06,
      "loss": 0.607,
      "step": 2713
    },
    {
      "epoch": 0.727321452498995,
      "grad_norm": 0.6863563735183001,
      "learning_rate": 4.832182620546296e-06,
      "loss": 0.5705,
      "step": 2714
    },
    {
      "epoch": 0.7275894412434678,
      "grad_norm": 0.6546980426228628,
      "learning_rate": 4.832055654629959e-06,
      "loss": 0.5789,
      "step": 2715
    },
    {
      "epoch": 0.7278574299879405,
      "grad_norm": 0.6963156305181395,
      "learning_rate": 4.831928642371662e-06,
      "loss": 0.6157,
      "step": 2716
    },
    {
      "epoch": 0.7281254187324132,
      "grad_norm": 0.6760403932793062,
      "learning_rate": 4.831801583773928e-06,
      "loss": 0.6062,
      "step": 2717
    },
    {
      "epoch": 0.728393407476886,
      "grad_norm": 0.7011744400915794,
      "learning_rate": 4.831674478839281e-06,
      "loss": 0.64,
      "step": 2718
    },
    {
      "epoch": 0.7286613962213587,
      "grad_norm": 0.7019867711983928,
      "learning_rate": 4.831547327570249e-06,
      "loss": 0.5985,
      "step": 2719
    },
    {
      "epoch": 0.7289293849658315,
      "grad_norm": 0.6696874610027009,
      "learning_rate": 4.8314201299693586e-06,
      "loss": 0.5867,
      "step": 2720
    },
    {
      "epoch": 0.7291973737103041,
      "grad_norm": 0.657098355389073,
      "learning_rate": 4.831292886039136e-06,
      "loss": 0.6069,
      "step": 2721
    },
    {
      "epoch": 0.7294653624547769,
      "grad_norm": 0.6745128775106832,
      "learning_rate": 4.8311655957821115e-06,
      "loss": 0.597,
      "step": 2722
    },
    {
      "epoch": 0.7297333511992496,
      "grad_norm": 0.6592591075361169,
      "learning_rate": 4.831038259200812e-06,
      "loss": 0.5826,
      "step": 2723
    },
    {
      "epoch": 0.7300013399437224,
      "grad_norm": 0.6422871882603356,
      "learning_rate": 4.83091087629777e-06,
      "loss": 0.5408,
      "step": 2724
    },
    {
      "epoch": 0.730269328688195,
      "grad_norm": 0.6765472968953306,
      "learning_rate": 4.830783447075516e-06,
      "loss": 0.5743,
      "step": 2725
    },
    {
      "epoch": 0.7305373174326678,
      "grad_norm": 0.6678609759219136,
      "learning_rate": 4.8306559715365825e-06,
      "loss": 0.5942,
      "step": 2726
    },
    {
      "epoch": 0.7308053061771406,
      "grad_norm": 0.6638373114385601,
      "learning_rate": 4.830528449683502e-06,
      "loss": 0.5881,
      "step": 2727
    },
    {
      "epoch": 0.7310732949216133,
      "grad_norm": 0.660082468480251,
      "learning_rate": 4.83040088151881e-06,
      "loss": 0.5732,
      "step": 2728
    },
    {
      "epoch": 0.731341283666086,
      "grad_norm": 0.6591817445507375,
      "learning_rate": 4.830273267045041e-06,
      "loss": 0.6165,
      "step": 2729
    },
    {
      "epoch": 0.7316092724105587,
      "grad_norm": 0.692305568586984,
      "learning_rate": 4.83014560626473e-06,
      "loss": 0.5811,
      "step": 2730
    },
    {
      "epoch": 0.7318772611550315,
      "grad_norm": 0.6544716476176138,
      "learning_rate": 4.830017899180414e-06,
      "loss": 0.6031,
      "step": 2731
    },
    {
      "epoch": 0.7321452498995042,
      "grad_norm": 0.6590809161419735,
      "learning_rate": 4.8298901457946315e-06,
      "loss": 0.5811,
      "step": 2732
    },
    {
      "epoch": 0.732413238643977,
      "grad_norm": 0.6615013392907022,
      "learning_rate": 4.8297623461099215e-06,
      "loss": 0.5764,
      "step": 2733
    },
    {
      "epoch": 0.7326812273884497,
      "grad_norm": 0.647173708264893,
      "learning_rate": 4.829634500128822e-06,
      "loss": 0.5935,
      "step": 2734
    },
    {
      "epoch": 0.7329492161329224,
      "grad_norm": 0.6768373234164147,
      "learning_rate": 4.829506607853875e-06,
      "loss": 0.6018,
      "step": 2735
    },
    {
      "epoch": 0.7332172048773952,
      "grad_norm": 0.6564906732226037,
      "learning_rate": 4.829378669287621e-06,
      "loss": 0.5858,
      "step": 2736
    },
    {
      "epoch": 0.7334851936218679,
      "grad_norm": 0.6682104841419219,
      "learning_rate": 4.829250684432603e-06,
      "loss": 0.5879,
      "step": 2737
    },
    {
      "epoch": 0.7337531823663406,
      "grad_norm": 0.6754116612053565,
      "learning_rate": 4.829122653291365e-06,
      "loss": 0.5973,
      "step": 2738
    },
    {
      "epoch": 0.7340211711108133,
      "grad_norm": 0.6547692182313569,
      "learning_rate": 4.828994575866449e-06,
      "loss": 0.5762,
      "step": 2739
    },
    {
      "epoch": 0.7342891598552861,
      "grad_norm": 0.6792228216816134,
      "learning_rate": 4.828866452160402e-06,
      "loss": 0.5773,
      "step": 2740
    },
    {
      "epoch": 0.7345571485997588,
      "grad_norm": 0.6681899007233951,
      "learning_rate": 4.828738282175769e-06,
      "loss": 0.6014,
      "step": 2741
    },
    {
      "epoch": 0.7348251373442315,
      "grad_norm": 0.7066361469926916,
      "learning_rate": 4.828610065915098e-06,
      "loss": 0.5978,
      "step": 2742
    },
    {
      "epoch": 0.7350931260887043,
      "grad_norm": 0.8066377465921678,
      "learning_rate": 4.828481803380936e-06,
      "loss": 0.5778,
      "step": 2743
    },
    {
      "epoch": 0.735361114833177,
      "grad_norm": 0.6457163564183436,
      "learning_rate": 4.828353494575831e-06,
      "loss": 0.5776,
      "step": 2744
    },
    {
      "epoch": 0.7356291035776498,
      "grad_norm": 0.6707160199200545,
      "learning_rate": 4.828225139502335e-06,
      "loss": 0.6003,
      "step": 2745
    },
    {
      "epoch": 0.7358970923221225,
      "grad_norm": 0.6842293164493326,
      "learning_rate": 4.828096738162997e-06,
      "loss": 0.6215,
      "step": 2746
    },
    {
      "epoch": 0.7361650810665952,
      "grad_norm": 0.7310292777902582,
      "learning_rate": 4.827968290560368e-06,
      "loss": 0.6065,
      "step": 2747
    },
    {
      "epoch": 0.7364330698110679,
      "grad_norm": 0.6632783085679742,
      "learning_rate": 4.827839796697004e-06,
      "loss": 0.6198,
      "step": 2748
    },
    {
      "epoch": 0.7367010585555407,
      "grad_norm": 0.651136136412774,
      "learning_rate": 4.8277112565754536e-06,
      "loss": 0.594,
      "step": 2749
    },
    {
      "epoch": 0.7369690473000134,
      "grad_norm": 0.64702456359802,
      "learning_rate": 4.8275826701982745e-06,
      "loss": 0.5824,
      "step": 2750
    },
    {
      "epoch": 0.7372370360444861,
      "grad_norm": 0.653155069393207,
      "learning_rate": 4.82745403756802e-06,
      "loss": 0.547,
      "step": 2751
    },
    {
      "epoch": 0.7375050247889589,
      "grad_norm": 0.6754012022128203,
      "learning_rate": 4.827325358687247e-06,
      "loss": 0.6049,
      "step": 2752
    },
    {
      "epoch": 0.7377730135334316,
      "grad_norm": 0.6727543722398277,
      "learning_rate": 4.827196633558513e-06,
      "loss": 0.5777,
      "step": 2753
    },
    {
      "epoch": 0.7380410022779044,
      "grad_norm": 0.6747792698227008,
      "learning_rate": 4.827067862184376e-06,
      "loss": 0.6002,
      "step": 2754
    },
    {
      "epoch": 0.738308991022377,
      "grad_norm": 0.6615160048041367,
      "learning_rate": 4.8269390445673936e-06,
      "loss": 0.6093,
      "step": 2755
    },
    {
      "epoch": 0.7385769797668498,
      "grad_norm": 0.661082729720441,
      "learning_rate": 4.826810180710126e-06,
      "loss": 0.5946,
      "step": 2756
    },
    {
      "epoch": 0.7388449685113225,
      "grad_norm": 0.6656734286236924,
      "learning_rate": 4.826681270615136e-06,
      "loss": 0.5943,
      "step": 2757
    },
    {
      "epoch": 0.7391129572557953,
      "grad_norm": 0.6723934967848634,
      "learning_rate": 4.8265523142849835e-06,
      "loss": 0.5863,
      "step": 2758
    },
    {
      "epoch": 0.7393809460002679,
      "grad_norm": 0.6446600782048741,
      "learning_rate": 4.826423311722231e-06,
      "loss": 0.577,
      "step": 2759
    },
    {
      "epoch": 0.7396489347447407,
      "grad_norm": 0.6638260207661811,
      "learning_rate": 4.826294262929442e-06,
      "loss": 0.5517,
      "step": 2760
    },
    {
      "epoch": 0.7399169234892135,
      "grad_norm": 0.6724058532756283,
      "learning_rate": 4.8261651679091815e-06,
      "loss": 0.616,
      "step": 2761
    },
    {
      "epoch": 0.7401849122336862,
      "grad_norm": 0.6535311370214206,
      "learning_rate": 4.826036026664014e-06,
      "loss": 0.564,
      "step": 2762
    },
    {
      "epoch": 0.740452900978159,
      "grad_norm": 0.6494876437477145,
      "learning_rate": 4.825906839196507e-06,
      "loss": 0.5624,
      "step": 2763
    },
    {
      "epoch": 0.7407208897226316,
      "grad_norm": 0.646699458474778,
      "learning_rate": 4.825777605509227e-06,
      "loss": 0.5921,
      "step": 2764
    },
    {
      "epoch": 0.7409888784671044,
      "grad_norm": 0.6491902524981732,
      "learning_rate": 4.8256483256047424e-06,
      "loss": 0.6083,
      "step": 2765
    },
    {
      "epoch": 0.7412568672115771,
      "grad_norm": 0.7010235950090054,
      "learning_rate": 4.825518999485621e-06,
      "loss": 0.6185,
      "step": 2766
    },
    {
      "epoch": 0.7415248559560499,
      "grad_norm": 0.6594583901234069,
      "learning_rate": 4.825389627154435e-06,
      "loss": 0.5723,
      "step": 2767
    },
    {
      "epoch": 0.7417928447005225,
      "grad_norm": 0.6727853896315978,
      "learning_rate": 4.825260208613752e-06,
      "loss": 0.5925,
      "step": 2768
    },
    {
      "epoch": 0.7420608334449953,
      "grad_norm": 0.6508967217640971,
      "learning_rate": 4.825130743866148e-06,
      "loss": 0.569,
      "step": 2769
    },
    {
      "epoch": 0.742328822189468,
      "grad_norm": 0.6590056729657747,
      "learning_rate": 4.8250012329141914e-06,
      "loss": 0.569,
      "step": 2770
    },
    {
      "epoch": 0.7425968109339408,
      "grad_norm": 0.658706773417836,
      "learning_rate": 4.8248716757604586e-06,
      "loss": 0.6147,
      "step": 2771
    },
    {
      "epoch": 0.7428647996784135,
      "grad_norm": 0.6628759689173744,
      "learning_rate": 4.8247420724075235e-06,
      "loss": 0.6009,
      "step": 2772
    },
    {
      "epoch": 0.7431327884228862,
      "grad_norm": 0.6481904103520991,
      "learning_rate": 4.82461242285796e-06,
      "loss": 0.5854,
      "step": 2773
    },
    {
      "epoch": 0.743400777167359,
      "grad_norm": 0.6880314323955756,
      "learning_rate": 4.824482727114348e-06,
      "loss": 0.6201,
      "step": 2774
    },
    {
      "epoch": 0.7436687659118317,
      "grad_norm": 0.6496199376090013,
      "learning_rate": 4.824352985179262e-06,
      "loss": 0.5516,
      "step": 2775
    },
    {
      "epoch": 0.7439367546563045,
      "grad_norm": 0.683118624188379,
      "learning_rate": 4.82422319705528e-06,
      "loss": 0.5711,
      "step": 2776
    },
    {
      "epoch": 0.7442047434007771,
      "grad_norm": 0.6494762476234383,
      "learning_rate": 4.824093362744981e-06,
      "loss": 0.5825,
      "step": 2777
    },
    {
      "epoch": 0.7444727321452499,
      "grad_norm": 0.6505104941996077,
      "learning_rate": 4.823963482250947e-06,
      "loss": 0.5908,
      "step": 2778
    },
    {
      "epoch": 0.7447407208897227,
      "grad_norm": 0.6949049051381081,
      "learning_rate": 4.823833555575759e-06,
      "loss": 0.6061,
      "step": 2779
    },
    {
      "epoch": 0.7450087096341954,
      "grad_norm": 0.6719286340870055,
      "learning_rate": 4.823703582721995e-06,
      "loss": 0.616,
      "step": 2780
    },
    {
      "epoch": 0.7452766983786681,
      "grad_norm": 0.6628128326017232,
      "learning_rate": 4.823573563692243e-06,
      "loss": 0.5897,
      "step": 2781
    },
    {
      "epoch": 0.7455446871231408,
      "grad_norm": 0.6671908467040383,
      "learning_rate": 4.823443498489084e-06,
      "loss": 0.592,
      "step": 2782
    },
    {
      "epoch": 0.7458126758676136,
      "grad_norm": 0.7372958126335851,
      "learning_rate": 4.823313387115102e-06,
      "loss": 0.5612,
      "step": 2783
    },
    {
      "epoch": 0.7460806646120863,
      "grad_norm": 0.644207418486202,
      "learning_rate": 4.823183229572883e-06,
      "loss": 0.5745,
      "step": 2784
    },
    {
      "epoch": 0.746348653356559,
      "grad_norm": 0.6684815281101455,
      "learning_rate": 4.823053025865015e-06,
      "loss": 0.5902,
      "step": 2785
    },
    {
      "epoch": 0.7466166421010317,
      "grad_norm": 0.6725373695158796,
      "learning_rate": 4.822922775994083e-06,
      "loss": 0.5919,
      "step": 2786
    },
    {
      "epoch": 0.7468846308455045,
      "grad_norm": 0.6636818580897049,
      "learning_rate": 4.822792479962678e-06,
      "loss": 0.5712,
      "step": 2787
    },
    {
      "epoch": 0.7471526195899773,
      "grad_norm": 0.6660995945578354,
      "learning_rate": 4.822662137773387e-06,
      "loss": 0.5861,
      "step": 2788
    },
    {
      "epoch": 0.7474206083344499,
      "grad_norm": 0.6504074206053413,
      "learning_rate": 4.8225317494288e-06,
      "loss": 0.5919,
      "step": 2789
    },
    {
      "epoch": 0.7476885970789227,
      "grad_norm": 0.6962052879718275,
      "learning_rate": 4.822401314931509e-06,
      "loss": 0.5649,
      "step": 2790
    },
    {
      "epoch": 0.7479565858233954,
      "grad_norm": 0.688332658166119,
      "learning_rate": 4.822270834284107e-06,
      "loss": 0.5629,
      "step": 2791
    },
    {
      "epoch": 0.7482245745678682,
      "grad_norm": 0.6727766516200722,
      "learning_rate": 4.822140307489185e-06,
      "loss": 0.6208,
      "step": 2792
    },
    {
      "epoch": 0.7484925633123409,
      "grad_norm": 0.6864173274937921,
      "learning_rate": 4.822009734549338e-06,
      "loss": 0.5886,
      "step": 2793
    },
    {
      "epoch": 0.7487605520568136,
      "grad_norm": 0.6746611162168051,
      "learning_rate": 4.821879115467159e-06,
      "loss": 0.6474,
      "step": 2794
    },
    {
      "epoch": 0.7490285408012863,
      "grad_norm": 0.7129422044501444,
      "learning_rate": 4.821748450245246e-06,
      "loss": 0.5632,
      "step": 2795
    },
    {
      "epoch": 0.7492965295457591,
      "grad_norm": 0.6627013580979019,
      "learning_rate": 4.821617738886193e-06,
      "loss": 0.6239,
      "step": 2796
    },
    {
      "epoch": 0.7495645182902319,
      "grad_norm": 0.6550269108469127,
      "learning_rate": 4.8214869813926e-06,
      "loss": 0.5883,
      "step": 2797
    },
    {
      "epoch": 0.7498325070347045,
      "grad_norm": 0.656450304203471,
      "learning_rate": 4.8213561777670644e-06,
      "loss": 0.556,
      "step": 2798
    },
    {
      "epoch": 0.7501004957791773,
      "grad_norm": 0.660689041676969,
      "learning_rate": 4.821225328012185e-06,
      "loss": 0.5793,
      "step": 2799
    },
    {
      "epoch": 0.75036848452365,
      "grad_norm": 0.6480526774146003,
      "learning_rate": 4.821094432130562e-06,
      "loss": 0.5697,
      "step": 2800
    },
    {
      "epoch": 0.7506364732681228,
      "grad_norm": 0.6730295246238743,
      "learning_rate": 4.820963490124796e-06,
      "loss": 0.6046,
      "step": 2801
    },
    {
      "epoch": 0.7509044620125954,
      "grad_norm": 0.6613415330405211,
      "learning_rate": 4.820832501997491e-06,
      "loss": 0.5814,
      "step": 2802
    },
    {
      "epoch": 0.7511724507570682,
      "grad_norm": 0.6596932993180411,
      "learning_rate": 4.8207014677512485e-06,
      "loss": 0.5833,
      "step": 2803
    },
    {
      "epoch": 0.7514404395015409,
      "grad_norm": 0.6583580162703475,
      "learning_rate": 4.820570387388673e-06,
      "loss": 0.5646,
      "step": 2804
    },
    {
      "epoch": 0.7517084282460137,
      "grad_norm": 0.6967754252939334,
      "learning_rate": 4.82043926091237e-06,
      "loss": 0.5995,
      "step": 2805
    },
    {
      "epoch": 0.7519764169904865,
      "grad_norm": 0.64311444646128,
      "learning_rate": 4.820308088324942e-06,
      "loss": 0.602,
      "step": 2806
    },
    {
      "epoch": 0.7522444057349591,
      "grad_norm": 0.6745082219972228,
      "learning_rate": 4.820176869629e-06,
      "loss": 0.6384,
      "step": 2807
    },
    {
      "epoch": 0.7525123944794319,
      "grad_norm": 0.6677584195619719,
      "learning_rate": 4.8200456048271474e-06,
      "loss": 0.5299,
      "step": 2808
    },
    {
      "epoch": 0.7527803832239046,
      "grad_norm": 0.6518952925133784,
      "learning_rate": 4.819914293921996e-06,
      "loss": 0.5582,
      "step": 2809
    },
    {
      "epoch": 0.7530483719683774,
      "grad_norm": 0.6729858063664517,
      "learning_rate": 4.819782936916153e-06,
      "loss": 0.5976,
      "step": 2810
    },
    {
      "epoch": 0.75331636071285,
      "grad_norm": 0.6873488352054379,
      "learning_rate": 4.819651533812229e-06,
      "loss": 0.5969,
      "step": 2811
    },
    {
      "epoch": 0.7535843494573228,
      "grad_norm": 0.6659484810034774,
      "learning_rate": 4.8195200846128374e-06,
      "loss": 0.5922,
      "step": 2812
    },
    {
      "epoch": 0.7538523382017955,
      "grad_norm": 0.6634187288360235,
      "learning_rate": 4.8193885893205875e-06,
      "loss": 0.5848,
      "step": 2813
    },
    {
      "epoch": 0.7541203269462683,
      "grad_norm": 0.6869107421868127,
      "learning_rate": 4.819257047938094e-06,
      "loss": 0.6055,
      "step": 2814
    },
    {
      "epoch": 0.7543883156907409,
      "grad_norm": 0.6765045486976652,
      "learning_rate": 4.81912546046797e-06,
      "loss": 0.6289,
      "step": 2815
    },
    {
      "epoch": 0.7546563044352137,
      "grad_norm": 0.6946090535092946,
      "learning_rate": 4.8189938269128315e-06,
      "loss": 0.599,
      "step": 2816
    },
    {
      "epoch": 0.7549242931796865,
      "grad_norm": 0.6623070955500369,
      "learning_rate": 4.818862147275292e-06,
      "loss": 0.6032,
      "step": 2817
    },
    {
      "epoch": 0.7551922819241592,
      "grad_norm": 0.6501327154409787,
      "learning_rate": 4.8187304215579705e-06,
      "loss": 0.5864,
      "step": 2818
    },
    {
      "epoch": 0.755460270668632,
      "grad_norm": 0.6584740878007221,
      "learning_rate": 4.818598649763484e-06,
      "loss": 0.5956,
      "step": 2819
    },
    {
      "epoch": 0.7557282594131046,
      "grad_norm": 0.6466803553058256,
      "learning_rate": 4.818466831894451e-06,
      "loss": 0.5645,
      "step": 2820
    },
    {
      "epoch": 0.7559962481575774,
      "grad_norm": 0.6464818315387261,
      "learning_rate": 4.818334967953489e-06,
      "loss": 0.5933,
      "step": 2821
    },
    {
      "epoch": 0.7562642369020501,
      "grad_norm": 0.6552166530757398,
      "learning_rate": 4.818203057943222e-06,
      "loss": 0.6066,
      "step": 2822
    },
    {
      "epoch": 0.7565322256465229,
      "grad_norm": 0.6441811134153448,
      "learning_rate": 4.818071101866269e-06,
      "loss": 0.6118,
      "step": 2823
    },
    {
      "epoch": 0.7568002143909955,
      "grad_norm": 0.6492289223925564,
      "learning_rate": 4.817939099725252e-06,
      "loss": 0.5912,
      "step": 2824
    },
    {
      "epoch": 0.7570682031354683,
      "grad_norm": 0.6627140932820433,
      "learning_rate": 4.817807051522796e-06,
      "loss": 0.5998,
      "step": 2825
    },
    {
      "epoch": 0.7573361918799411,
      "grad_norm": 0.6326073446413176,
      "learning_rate": 4.8176749572615224e-06,
      "loss": 0.5858,
      "step": 2826
    },
    {
      "epoch": 0.7576041806244138,
      "grad_norm": 0.671635699949192,
      "learning_rate": 4.817542816944058e-06,
      "loss": 0.5906,
      "step": 2827
    },
    {
      "epoch": 0.7578721693688865,
      "grad_norm": 0.6747263554688916,
      "learning_rate": 4.8174106305730284e-06,
      "loss": 0.6095,
      "step": 2828
    },
    {
      "epoch": 0.7581401581133592,
      "grad_norm": 0.6385044091934784,
      "learning_rate": 4.817278398151059e-06,
      "loss": 0.619,
      "step": 2829
    },
    {
      "epoch": 0.758408146857832,
      "grad_norm": 0.6534171821254928,
      "learning_rate": 4.81714611968078e-06,
      "loss": 0.6077,
      "step": 2830
    },
    {
      "epoch": 0.7586761356023047,
      "grad_norm": 0.647609638186141,
      "learning_rate": 4.817013795164817e-06,
      "loss": 0.6104,
      "step": 2831
    },
    {
      "epoch": 0.7589441243467774,
      "grad_norm": 0.6562223325234862,
      "learning_rate": 4.816881424605803e-06,
      "loss": 0.5492,
      "step": 2832
    },
    {
      "epoch": 0.7592121130912501,
      "grad_norm": 0.6356227777963731,
      "learning_rate": 4.816749008006366e-06,
      "loss": 0.5962,
      "step": 2833
    },
    {
      "epoch": 0.7594801018357229,
      "grad_norm": 0.6761604503795712,
      "learning_rate": 4.816616545369137e-06,
      "loss": 0.6251,
      "step": 2834
    },
    {
      "epoch": 0.7597480905801957,
      "grad_norm": 0.6485958634942411,
      "learning_rate": 4.81648403669675e-06,
      "loss": 0.5707,
      "step": 2835
    },
    {
      "epoch": 0.7600160793246684,
      "grad_norm": 0.6541201459244833,
      "learning_rate": 4.816351481991837e-06,
      "loss": 0.5838,
      "step": 2836
    },
    {
      "epoch": 0.7602840680691411,
      "grad_norm": 0.6643490203769432,
      "learning_rate": 4.816218881257033e-06,
      "loss": 0.5829,
      "step": 2837
    },
    {
      "epoch": 0.7605520568136138,
      "grad_norm": 0.6519514682785608,
      "learning_rate": 4.8160862344949715e-06,
      "loss": 0.6,
      "step": 2838
    },
    {
      "epoch": 0.7608200455580866,
      "grad_norm": 0.6578778232446779,
      "learning_rate": 4.81595354170829e-06,
      "loss": 0.5718,
      "step": 2839
    },
    {
      "epoch": 0.7610880343025593,
      "grad_norm": 0.642711742654275,
      "learning_rate": 4.815820802899625e-06,
      "loss": 0.5681,
      "step": 2840
    },
    {
      "epoch": 0.761356023047032,
      "grad_norm": 0.6660323512089279,
      "learning_rate": 4.815688018071612e-06,
      "loss": 0.6098,
      "step": 2841
    },
    {
      "epoch": 0.7616240117915047,
      "grad_norm": 0.6744747775774028,
      "learning_rate": 4.815555187226893e-06,
      "loss": 0.6317,
      "step": 2842
    },
    {
      "epoch": 0.7618920005359775,
      "grad_norm": 0.6624341768547912,
      "learning_rate": 4.8154223103681065e-06,
      "loss": 0.5816,
      "step": 2843
    },
    {
      "epoch": 0.7621599892804503,
      "grad_norm": 0.6641068379461653,
      "learning_rate": 4.815289387497892e-06,
      "loss": 0.5856,
      "step": 2844
    },
    {
      "epoch": 0.7624279780249229,
      "grad_norm": 0.6902855439536755,
      "learning_rate": 4.815156418618892e-06,
      "loss": 0.5727,
      "step": 2845
    },
    {
      "epoch": 0.7626959667693957,
      "grad_norm": 0.6441102953041197,
      "learning_rate": 4.815023403733748e-06,
      "loss": 0.6029,
      "step": 2846
    },
    {
      "epoch": 0.7629639555138684,
      "grad_norm": 0.6490608045230486,
      "learning_rate": 4.8148903428451044e-06,
      "loss": 0.5972,
      "step": 2847
    },
    {
      "epoch": 0.7632319442583412,
      "grad_norm": 0.6748199081873888,
      "learning_rate": 4.814757235955604e-06,
      "loss": 0.5846,
      "step": 2848
    },
    {
      "epoch": 0.7634999330028139,
      "grad_norm": 0.666858023178861,
      "learning_rate": 4.814624083067892e-06,
      "loss": 0.6057,
      "step": 2849
    },
    {
      "epoch": 0.7637679217472866,
      "grad_norm": 0.6351855825935329,
      "learning_rate": 4.814490884184615e-06,
      "loss": 0.6012,
      "step": 2850
    },
    {
      "epoch": 0.7640359104917593,
      "grad_norm": 0.6591869754107149,
      "learning_rate": 4.81435763930842e-06,
      "loss": 0.6151,
      "step": 2851
    },
    {
      "epoch": 0.7643038992362321,
      "grad_norm": 0.6436033203531742,
      "learning_rate": 4.814224348441955e-06,
      "loss": 0.5625,
      "step": 2852
    },
    {
      "epoch": 0.7645718879807049,
      "grad_norm": 0.6407050574500989,
      "learning_rate": 4.814091011587867e-06,
      "loss": 0.5601,
      "step": 2853
    },
    {
      "epoch": 0.7648398767251775,
      "grad_norm": 0.6489320190541433,
      "learning_rate": 4.813957628748808e-06,
      "loss": 0.6062,
      "step": 2854
    },
    {
      "epoch": 0.7651078654696503,
      "grad_norm": 0.6630981925251811,
      "learning_rate": 4.813824199927426e-06,
      "loss": 0.5959,
      "step": 2855
    },
    {
      "epoch": 0.765375854214123,
      "grad_norm": 0.6754598333161376,
      "learning_rate": 4.813690725126375e-06,
      "loss": 0.6023,
      "step": 2856
    },
    {
      "epoch": 0.7656438429585958,
      "grad_norm": 0.6728064815836541,
      "learning_rate": 4.813557204348306e-06,
      "loss": 0.5745,
      "step": 2857
    },
    {
      "epoch": 0.7659118317030684,
      "grad_norm": 0.6471903267159527,
      "learning_rate": 4.813423637595873e-06,
      "loss": 0.5607,
      "step": 2858
    },
    {
      "epoch": 0.7661798204475412,
      "grad_norm": 0.6666190965465059,
      "learning_rate": 4.813290024871729e-06,
      "loss": 0.5744,
      "step": 2859
    },
    {
      "epoch": 0.7664478091920139,
      "grad_norm": 0.6429128296815976,
      "learning_rate": 4.81315636617853e-06,
      "loss": 0.601,
      "step": 2860
    },
    {
      "epoch": 0.7667157979364867,
      "grad_norm": 0.6767052291579255,
      "learning_rate": 4.8130226615189325e-06,
      "loss": 0.6413,
      "step": 2861
    },
    {
      "epoch": 0.7669837866809593,
      "grad_norm": 0.6601003103252601,
      "learning_rate": 4.8128889108955916e-06,
      "loss": 0.6019,
      "step": 2862
    },
    {
      "epoch": 0.7672517754254321,
      "grad_norm": 0.646745328088378,
      "learning_rate": 4.812755114311167e-06,
      "loss": 0.574,
      "step": 2863
    },
    {
      "epoch": 0.7675197641699049,
      "grad_norm": 0.6706072550853903,
      "learning_rate": 4.812621271768317e-06,
      "loss": 0.5925,
      "step": 2864
    },
    {
      "epoch": 0.7677877529143776,
      "grad_norm": 0.6897662639788775,
      "learning_rate": 4.812487383269701e-06,
      "loss": 0.5674,
      "step": 2865
    },
    {
      "epoch": 0.7680557416588504,
      "grad_norm": 0.6547689936960712,
      "learning_rate": 4.812353448817979e-06,
      "loss": 0.5569,
      "step": 2866
    },
    {
      "epoch": 0.768323730403323,
      "grad_norm": 0.6472148969480424,
      "learning_rate": 4.8122194684158146e-06,
      "loss": 0.6247,
      "step": 2867
    },
    {
      "epoch": 0.7685917191477958,
      "grad_norm": 0.6824874394309414,
      "learning_rate": 4.812085442065868e-06,
      "loss": 0.5677,
      "step": 2868
    },
    {
      "epoch": 0.7688597078922685,
      "grad_norm": 0.6623143706039516,
      "learning_rate": 4.811951369770803e-06,
      "loss": 0.5931,
      "step": 2869
    },
    {
      "epoch": 0.7691276966367413,
      "grad_norm": 0.6764003141688455,
      "learning_rate": 4.811817251533285e-06,
      "loss": 0.5938,
      "step": 2870
    },
    {
      "epoch": 0.769395685381214,
      "grad_norm": 0.6751287923861984,
      "learning_rate": 4.8116830873559785e-06,
      "loss": 0.6009,
      "step": 2871
    },
    {
      "epoch": 0.7696636741256867,
      "grad_norm": 0.6844356314864841,
      "learning_rate": 4.811548877241549e-06,
      "loss": 0.5916,
      "step": 2872
    },
    {
      "epoch": 0.7699316628701595,
      "grad_norm": 0.6518818279637699,
      "learning_rate": 4.811414621192664e-06,
      "loss": 0.6257,
      "step": 2873
    },
    {
      "epoch": 0.7701996516146322,
      "grad_norm": 0.7423683184494823,
      "learning_rate": 4.8112803192119915e-06,
      "loss": 0.5807,
      "step": 2874
    },
    {
      "epoch": 0.7704676403591049,
      "grad_norm": 0.6913359374142037,
      "learning_rate": 4.811145971302201e-06,
      "loss": 0.5722,
      "step": 2875
    },
    {
      "epoch": 0.7707356291035776,
      "grad_norm": 0.6622122047879877,
      "learning_rate": 4.81101157746596e-06,
      "loss": 0.5502,
      "step": 2876
    },
    {
      "epoch": 0.7710036178480504,
      "grad_norm": 0.6853552882414088,
      "learning_rate": 4.8108771377059405e-06,
      "loss": 0.5846,
      "step": 2877
    },
    {
      "epoch": 0.7712716065925231,
      "grad_norm": 0.6733015512354006,
      "learning_rate": 4.810742652024814e-06,
      "loss": 0.5894,
      "step": 2878
    },
    {
      "epoch": 0.7715395953369959,
      "grad_norm": 0.6562362364182271,
      "learning_rate": 4.810608120425254e-06,
      "loss": 0.5908,
      "step": 2879
    },
    {
      "epoch": 0.7718075840814685,
      "grad_norm": 0.6474900122067756,
      "learning_rate": 4.810473542909932e-06,
      "loss": 0.5887,
      "step": 2880
    },
    {
      "epoch": 0.7720755728259413,
      "grad_norm": 0.6795295259692713,
      "learning_rate": 4.810338919481523e-06,
      "loss": 0.5801,
      "step": 2881
    },
    {
      "epoch": 0.7723435615704141,
      "grad_norm": 0.7079604063463245,
      "learning_rate": 4.810204250142703e-06,
      "loss": 0.5827,
      "step": 2882
    },
    {
      "epoch": 0.7726115503148868,
      "grad_norm": 0.6689764672502089,
      "learning_rate": 4.810069534896148e-06,
      "loss": 0.6241,
      "step": 2883
    },
    {
      "epoch": 0.7728795390593595,
      "grad_norm": 0.6474450064607651,
      "learning_rate": 4.809934773744534e-06,
      "loss": 0.6118,
      "step": 2884
    },
    {
      "epoch": 0.7731475278038322,
      "grad_norm": 0.6851703825563384,
      "learning_rate": 4.8097999666905396e-06,
      "loss": 0.6006,
      "step": 2885
    },
    {
      "epoch": 0.773415516548305,
      "grad_norm": 0.6631521739177234,
      "learning_rate": 4.809665113736843e-06,
      "loss": 0.5541,
      "step": 2886
    },
    {
      "epoch": 0.7736835052927777,
      "grad_norm": 0.7268897021841676,
      "learning_rate": 4.809530214886124e-06,
      "loss": 0.558,
      "step": 2887
    },
    {
      "epoch": 0.7739514940372504,
      "grad_norm": 0.6861699724322082,
      "learning_rate": 4.809395270141065e-06,
      "loss": 0.6122,
      "step": 2888
    },
    {
      "epoch": 0.7742194827817231,
      "grad_norm": 0.68666886268849,
      "learning_rate": 4.809260279504346e-06,
      "loss": 0.5622,
      "step": 2889
    },
    {
      "epoch": 0.7744874715261959,
      "grad_norm": 0.655859937436294,
      "learning_rate": 4.809125242978649e-06,
      "loss": 0.6154,
      "step": 2890
    },
    {
      "epoch": 0.7747554602706687,
      "grad_norm": 0.6774307844310289,
      "learning_rate": 4.808990160566659e-06,
      "loss": 0.6125,
      "step": 2891
    },
    {
      "epoch": 0.7750234490151414,
      "grad_norm": 0.7454576265172617,
      "learning_rate": 4.80885503227106e-06,
      "loss": 0.5444,
      "step": 2892
    },
    {
      "epoch": 0.7752914377596141,
      "grad_norm": 0.6666830458241166,
      "learning_rate": 4.808719858094536e-06,
      "loss": 0.5832,
      "step": 2893
    },
    {
      "epoch": 0.7755594265040868,
      "grad_norm": 0.6734770622425453,
      "learning_rate": 4.808584638039774e-06,
      "loss": 0.5506,
      "step": 2894
    },
    {
      "epoch": 0.7758274152485596,
      "grad_norm": 0.6664085801230233,
      "learning_rate": 4.808449372109461e-06,
      "loss": 0.5609,
      "step": 2895
    },
    {
      "epoch": 0.7760954039930323,
      "grad_norm": 0.6532696298590651,
      "learning_rate": 4.808314060306285e-06,
      "loss": 0.5856,
      "step": 2896
    },
    {
      "epoch": 0.776363392737505,
      "grad_norm": 0.6772342356397524,
      "learning_rate": 4.808178702632935e-06,
      "loss": 0.5731,
      "step": 2897
    },
    {
      "epoch": 0.7766313814819777,
      "grad_norm": 0.6384273410366538,
      "learning_rate": 4.8080432990921004e-06,
      "loss": 0.5855,
      "step": 2898
    },
    {
      "epoch": 0.7768993702264505,
      "grad_norm": 0.6557126379316016,
      "learning_rate": 4.807907849686472e-06,
      "loss": 0.5574,
      "step": 2899
    },
    {
      "epoch": 0.7771673589709233,
      "grad_norm": 0.6953010194230796,
      "learning_rate": 4.807772354418742e-06,
      "loss": 0.6082,
      "step": 2900
    },
    {
      "epoch": 0.7774353477153959,
      "grad_norm": 0.6670960982054208,
      "learning_rate": 4.807636813291602e-06,
      "loss": 0.5551,
      "step": 2901
    },
    {
      "epoch": 0.7777033364598687,
      "grad_norm": 0.6946792625429201,
      "learning_rate": 4.807501226307746e-06,
      "loss": 0.578,
      "step": 2902
    },
    {
      "epoch": 0.7779713252043414,
      "grad_norm": 0.6345773836476848,
      "learning_rate": 4.807365593469869e-06,
      "loss": 0.6041,
      "step": 2903
    },
    {
      "epoch": 0.7782393139488142,
      "grad_norm": 0.660853898190136,
      "learning_rate": 4.807229914780665e-06,
      "loss": 0.6051,
      "step": 2904
    },
    {
      "epoch": 0.7785073026932868,
      "grad_norm": 0.6724699346628104,
      "learning_rate": 4.8070941902428295e-06,
      "loss": 0.6105,
      "step": 2905
    },
    {
      "epoch": 0.7787752914377596,
      "grad_norm": 0.6728041135323354,
      "learning_rate": 4.806958419859062e-06,
      "loss": 0.6009,
      "step": 2906
    },
    {
      "epoch": 0.7790432801822323,
      "grad_norm": 0.6520223460008702,
      "learning_rate": 4.806822603632059e-06,
      "loss": 0.6184,
      "step": 2907
    },
    {
      "epoch": 0.7793112689267051,
      "grad_norm": 0.6390301236570929,
      "learning_rate": 4.806686741564519e-06,
      "loss": 0.5495,
      "step": 2908
    },
    {
      "epoch": 0.7795792576711779,
      "grad_norm": 0.6781956173836734,
      "learning_rate": 4.806550833659143e-06,
      "loss": 0.5739,
      "step": 2909
    },
    {
      "epoch": 0.7798472464156505,
      "grad_norm": 0.677938275622111,
      "learning_rate": 4.8064148799186315e-06,
      "loss": 0.5924,
      "step": 2910
    },
    {
      "epoch": 0.7801152351601233,
      "grad_norm": 0.6548505711611879,
      "learning_rate": 4.806278880345685e-06,
      "loss": 0.5416,
      "step": 2911
    },
    {
      "epoch": 0.780383223904596,
      "grad_norm": 0.6582081792378668,
      "learning_rate": 4.8061428349430075e-06,
      "loss": 0.5805,
      "step": 2912
    },
    {
      "epoch": 0.7806512126490688,
      "grad_norm": 0.6592670023494928,
      "learning_rate": 4.8060067437133015e-06,
      "loss": 0.6064,
      "step": 2913
    },
    {
      "epoch": 0.7809192013935414,
      "grad_norm": 0.6573515577432775,
      "learning_rate": 4.805870606659272e-06,
      "loss": 0.5944,
      "step": 2914
    },
    {
      "epoch": 0.7811871901380142,
      "grad_norm": 0.6732626483964955,
      "learning_rate": 4.805734423783624e-06,
      "loss": 0.5972,
      "step": 2915
    },
    {
      "epoch": 0.781455178882487,
      "grad_norm": 0.6776133280535275,
      "learning_rate": 4.805598195089063e-06,
      "loss": 0.62,
      "step": 2916
    },
    {
      "epoch": 0.7817231676269597,
      "grad_norm": 0.6552424678549008,
      "learning_rate": 4.805461920578297e-06,
      "loss": 0.5951,
      "step": 2917
    },
    {
      "epoch": 0.7819911563714323,
      "grad_norm": 0.6662045689394535,
      "learning_rate": 4.805325600254034e-06,
      "loss": 0.6435,
      "step": 2918
    },
    {
      "epoch": 0.7822591451159051,
      "grad_norm": 0.6625001528537887,
      "learning_rate": 4.805189234118982e-06,
      "loss": 0.6084,
      "step": 2919
    },
    {
      "epoch": 0.7825271338603779,
      "grad_norm": 0.6741813972750625,
      "learning_rate": 4.805052822175853e-06,
      "loss": 0.6458,
      "step": 2920
    },
    {
      "epoch": 0.7827951226048506,
      "grad_norm": 0.6498129105160936,
      "learning_rate": 4.804916364427354e-06,
      "loss": 0.5835,
      "step": 2921
    },
    {
      "epoch": 0.7830631113493234,
      "grad_norm": 0.6562329440966447,
      "learning_rate": 4.804779860876201e-06,
      "loss": 0.5944,
      "step": 2922
    },
    {
      "epoch": 0.783331100093796,
      "grad_norm": 0.6409868717112647,
      "learning_rate": 4.8046433115251035e-06,
      "loss": 0.6011,
      "step": 2923
    },
    {
      "epoch": 0.7835990888382688,
      "grad_norm": 0.659841612639022,
      "learning_rate": 4.804506716376775e-06,
      "loss": 0.5919,
      "step": 2924
    },
    {
      "epoch": 0.7838670775827415,
      "grad_norm": 0.63998511804325,
      "learning_rate": 4.8043700754339325e-06,
      "loss": 0.5934,
      "step": 2925
    },
    {
      "epoch": 0.7841350663272143,
      "grad_norm": 0.6520838033205772,
      "learning_rate": 4.804233388699289e-06,
      "loss": 0.591,
      "step": 2926
    },
    {
      "epoch": 0.784403055071687,
      "grad_norm": 0.6671488954201286,
      "learning_rate": 4.804096656175561e-06,
      "loss": 0.6142,
      "step": 2927
    },
    {
      "epoch": 0.7846710438161597,
      "grad_norm": 0.6366958311015696,
      "learning_rate": 4.803959877865466e-06,
      "loss": 0.5759,
      "step": 2928
    },
    {
      "epoch": 0.7849390325606325,
      "grad_norm": 0.685670619046015,
      "learning_rate": 4.8038230537717226e-06,
      "loss": 0.6117,
      "step": 2929
    },
    {
      "epoch": 0.7852070213051052,
      "grad_norm": 0.6482771314171567,
      "learning_rate": 4.803686183897048e-06,
      "loss": 0.5696,
      "step": 2930
    },
    {
      "epoch": 0.7854750100495779,
      "grad_norm": 0.6561604238412759,
      "learning_rate": 4.803549268244164e-06,
      "loss": 0.5994,
      "step": 2931
    },
    {
      "epoch": 0.7857429987940506,
      "grad_norm": 0.648896478897163,
      "learning_rate": 4.80341230681579e-06,
      "loss": 0.5532,
      "step": 2932
    },
    {
      "epoch": 0.7860109875385234,
      "grad_norm": 0.6502928201600394,
      "learning_rate": 4.803275299614648e-06,
      "loss": 0.5632,
      "step": 2933
    },
    {
      "epoch": 0.7862789762829961,
      "grad_norm": 0.6529845813274041,
      "learning_rate": 4.80313824664346e-06,
      "loss": 0.5779,
      "step": 2934
    },
    {
      "epoch": 0.7865469650274688,
      "grad_norm": 0.6693535023871023,
      "learning_rate": 4.803001147904952e-06,
      "loss": 0.5654,
      "step": 2935
    },
    {
      "epoch": 0.7868149537719415,
      "grad_norm": 0.6461994987133466,
      "learning_rate": 4.802864003401845e-06,
      "loss": 0.5805,
      "step": 2936
    },
    {
      "epoch": 0.7870829425164143,
      "grad_norm": 0.652155426480758,
      "learning_rate": 4.8027268131368665e-06,
      "loss": 0.5689,
      "step": 2937
    },
    {
      "epoch": 0.7873509312608871,
      "grad_norm": 0.6563406698208643,
      "learning_rate": 4.802589577112742e-06,
      "loss": 0.585,
      "step": 2938
    },
    {
      "epoch": 0.7876189200053598,
      "grad_norm": 0.6528809822531875,
      "learning_rate": 4.802452295332198e-06,
      "loss": 0.5711,
      "step": 2939
    },
    {
      "epoch": 0.7878869087498325,
      "grad_norm": 0.6595126575925201,
      "learning_rate": 4.802314967797964e-06,
      "loss": 0.5915,
      "step": 2940
    },
    {
      "epoch": 0.7881548974943052,
      "grad_norm": 0.6483804255366084,
      "learning_rate": 4.802177594512768e-06,
      "loss": 0.5796,
      "step": 2941
    },
    {
      "epoch": 0.788422886238778,
      "grad_norm": 0.6671439527376688,
      "learning_rate": 4.80204017547934e-06,
      "loss": 0.5748,
      "step": 2942
    },
    {
      "epoch": 0.7886908749832507,
      "grad_norm": 0.6503696938355544,
      "learning_rate": 4.80190271070041e-06,
      "loss": 0.6013,
      "step": 2943
    },
    {
      "epoch": 0.7889588637277234,
      "grad_norm": 0.6508146060947494,
      "learning_rate": 4.801765200178711e-06,
      "loss": 0.603,
      "step": 2944
    },
    {
      "epoch": 0.7892268524721961,
      "grad_norm": 0.6407972263574927,
      "learning_rate": 4.8016276439169745e-06,
      "loss": 0.5764,
      "step": 2945
    },
    {
      "epoch": 0.7894948412166689,
      "grad_norm": 0.6417499039124136,
      "learning_rate": 4.801490041917935e-06,
      "loss": 0.5792,
      "step": 2946
    },
    {
      "epoch": 0.7897628299611417,
      "grad_norm": 0.6426130573927153,
      "learning_rate": 4.801352394184326e-06,
      "loss": 0.5983,
      "step": 2947
    },
    {
      "epoch": 0.7900308187056143,
      "grad_norm": 0.6568061942585212,
      "learning_rate": 4.801214700718883e-06,
      "loss": 0.5798,
      "step": 2948
    },
    {
      "epoch": 0.7902988074500871,
      "grad_norm": 0.6626171191917994,
      "learning_rate": 4.8010769615243424e-06,
      "loss": 0.57,
      "step": 2949
    },
    {
      "epoch": 0.7905667961945598,
      "grad_norm": 0.6524125480540224,
      "learning_rate": 4.800939176603442e-06,
      "loss": 0.5912,
      "step": 2950
    },
    {
      "epoch": 0.7908347849390326,
      "grad_norm": 0.6706937076471325,
      "learning_rate": 4.8008013459589175e-06,
      "loss": 0.5953,
      "step": 2951
    },
    {
      "epoch": 0.7911027736835053,
      "grad_norm": 0.6583076325985197,
      "learning_rate": 4.800663469593511e-06,
      "loss": 0.5939,
      "step": 2952
    },
    {
      "epoch": 0.791370762427978,
      "grad_norm": 0.6463395839327285,
      "learning_rate": 4.80052554750996e-06,
      "loss": 0.5785,
      "step": 2953
    },
    {
      "epoch": 0.7916387511724507,
      "grad_norm": 0.6463286340664628,
      "learning_rate": 4.800387579711005e-06,
      "loss": 0.6121,
      "step": 2954
    },
    {
      "epoch": 0.7919067399169235,
      "grad_norm": 0.6605281923691958,
      "learning_rate": 4.80024956619939e-06,
      "loss": 0.5617,
      "step": 2955
    },
    {
      "epoch": 0.7921747286613963,
      "grad_norm": 0.6444059631631343,
      "learning_rate": 4.8001115069778555e-06,
      "loss": 0.5972,
      "step": 2956
    },
    {
      "epoch": 0.7924427174058689,
      "grad_norm": 0.6456948213676389,
      "learning_rate": 4.799973402049146e-06,
      "loss": 0.5798,
      "step": 2957
    },
    {
      "epoch": 0.7927107061503417,
      "grad_norm": 0.6466242527058088,
      "learning_rate": 4.799835251416006e-06,
      "loss": 0.5981,
      "step": 2958
    },
    {
      "epoch": 0.7929786948948144,
      "grad_norm": 0.6980233494149324,
      "learning_rate": 4.79969705508118e-06,
      "loss": 0.579,
      "step": 2959
    },
    {
      "epoch": 0.7932466836392872,
      "grad_norm": 0.6852897811350326,
      "learning_rate": 4.7995588130474145e-06,
      "loss": 0.564,
      "step": 2960
    },
    {
      "epoch": 0.7935146723837598,
      "grad_norm": 0.6560603427536871,
      "learning_rate": 4.799420525317456e-06,
      "loss": 0.5716,
      "step": 2961
    },
    {
      "epoch": 0.7937826611282326,
      "grad_norm": 0.6506972321891027,
      "learning_rate": 4.799282191894054e-06,
      "loss": 0.5509,
      "step": 2962
    },
    {
      "epoch": 0.7940506498727053,
      "grad_norm": 0.6644064567671513,
      "learning_rate": 4.799143812779956e-06,
      "loss": 0.5996,
      "step": 2963
    },
    {
      "epoch": 0.7943186386171781,
      "grad_norm": 0.6618734172403267,
      "learning_rate": 4.799005387977913e-06,
      "loss": 0.5942,
      "step": 2964
    },
    {
      "epoch": 0.7945866273616509,
      "grad_norm": 0.6696253540918796,
      "learning_rate": 4.798866917490675e-06,
      "loss": 0.6279,
      "step": 2965
    },
    {
      "epoch": 0.7948546161061235,
      "grad_norm": 0.6705102474407717,
      "learning_rate": 4.798728401320994e-06,
      "loss": 0.576,
      "step": 2966
    },
    {
      "epoch": 0.7951226048505963,
      "grad_norm": 0.6917903760296652,
      "learning_rate": 4.798589839471622e-06,
      "loss": 0.5816,
      "step": 2967
    },
    {
      "epoch": 0.795390593595069,
      "grad_norm": 0.6619908538084645,
      "learning_rate": 4.798451231945313e-06,
      "loss": 0.5841,
      "step": 2968
    },
    {
      "epoch": 0.7956585823395418,
      "grad_norm": 0.6495954827773578,
      "learning_rate": 4.7983125787448214e-06,
      "loss": 0.6074,
      "step": 2969
    },
    {
      "epoch": 0.7959265710840144,
      "grad_norm": 0.6623236951061974,
      "learning_rate": 4.798173879872901e-06,
      "loss": 0.5743,
      "step": 2970
    },
    {
      "epoch": 0.7961945598284872,
      "grad_norm": 0.6553843456782158,
      "learning_rate": 4.798035135332311e-06,
      "loss": 0.6036,
      "step": 2971
    },
    {
      "epoch": 0.79646254857296,
      "grad_norm": 0.673341410661886,
      "learning_rate": 4.7978963451258064e-06,
      "loss": 0.5982,
      "step": 2972
    },
    {
      "epoch": 0.7967305373174327,
      "grad_norm": 0.6431946496777229,
      "learning_rate": 4.797757509256144e-06,
      "loss": 0.5817,
      "step": 2973
    },
    {
      "epoch": 0.7969985260619054,
      "grad_norm": 0.6419846968660678,
      "learning_rate": 4.797618627726086e-06,
      "loss": 0.5895,
      "step": 2974
    },
    {
      "epoch": 0.7972665148063781,
      "grad_norm": 0.7073677529397774,
      "learning_rate": 4.79747970053839e-06,
      "loss": 0.5825,
      "step": 2975
    },
    {
      "epoch": 0.7975345035508509,
      "grad_norm": 0.7571574455320537,
      "learning_rate": 4.797340727695817e-06,
      "loss": 0.5959,
      "step": 2976
    },
    {
      "epoch": 0.7978024922953236,
      "grad_norm": 0.6778431293451058,
      "learning_rate": 4.797201709201129e-06,
      "loss": 0.54,
      "step": 2977
    },
    {
      "epoch": 0.7980704810397963,
      "grad_norm": 0.6597093275518331,
      "learning_rate": 4.797062645057089e-06,
      "loss": 0.572,
      "step": 2978
    },
    {
      "epoch": 0.798338469784269,
      "grad_norm": 0.6797264089017466,
      "learning_rate": 4.796923535266459e-06,
      "loss": 0.6193,
      "step": 2979
    },
    {
      "epoch": 0.7986064585287418,
      "grad_norm": 0.6682857356537149,
      "learning_rate": 4.796784379832005e-06,
      "loss": 0.6047,
      "step": 2980
    },
    {
      "epoch": 0.7988744472732145,
      "grad_norm": 0.6724917512903097,
      "learning_rate": 4.79664517875649e-06,
      "loss": 0.5577,
      "step": 2981
    },
    {
      "epoch": 0.7991424360176873,
      "grad_norm": 0.6429642373848581,
      "learning_rate": 4.7965059320426825e-06,
      "loss": 0.5718,
      "step": 2982
    },
    {
      "epoch": 0.79941042476216,
      "grad_norm": 0.6469650270990893,
      "learning_rate": 4.796366639693348e-06,
      "loss": 0.6045,
      "step": 2983
    },
    {
      "epoch": 0.7996784135066327,
      "grad_norm": 0.64000935186663,
      "learning_rate": 4.796227301711256e-06,
      "loss": 0.5727,
      "step": 2984
    },
    {
      "epoch": 0.7999464022511055,
      "grad_norm": 0.6548400440154241,
      "learning_rate": 4.796087918099174e-06,
      "loss": 0.5894,
      "step": 2985
    },
    {
      "epoch": 0.8002143909955782,
      "grad_norm": 0.6682078241377518,
      "learning_rate": 4.795948488859873e-06,
      "loss": 0.5665,
      "step": 2986
    },
    {
      "epoch": 0.8004823797400509,
      "grad_norm": 0.6601338649723331,
      "learning_rate": 4.795809013996122e-06,
      "loss": 0.6133,
      "step": 2987
    },
    {
      "epoch": 0.8007503684845236,
      "grad_norm": 0.6648035196943359,
      "learning_rate": 4.795669493510694e-06,
      "loss": 0.5773,
      "step": 2988
    },
    {
      "epoch": 0.8010183572289964,
      "grad_norm": 0.6376267108726354,
      "learning_rate": 4.795529927406361e-06,
      "loss": 0.5524,
      "step": 2989
    },
    {
      "epoch": 0.8012863459734691,
      "grad_norm": 0.6698829687784668,
      "learning_rate": 4.795390315685897e-06,
      "loss": 0.6128,
      "step": 2990
    },
    {
      "epoch": 0.8015543347179418,
      "grad_norm": 0.6856212673694193,
      "learning_rate": 4.795250658352076e-06,
      "loss": 0.5893,
      "step": 2991
    },
    {
      "epoch": 0.8018223234624146,
      "grad_norm": 0.6431101203451904,
      "learning_rate": 4.795110955407673e-06,
      "loss": 0.584,
      "step": 2992
    },
    {
      "epoch": 0.8020903122068873,
      "grad_norm": 0.6545053755309579,
      "learning_rate": 4.794971206855464e-06,
      "loss": 0.563,
      "step": 2993
    },
    {
      "epoch": 0.8023583009513601,
      "grad_norm": 0.6535149023090531,
      "learning_rate": 4.7948314126982265e-06,
      "loss": 0.6052,
      "step": 2994
    },
    {
      "epoch": 0.8026262896958328,
      "grad_norm": 0.6568897282782076,
      "learning_rate": 4.794691572938738e-06,
      "loss": 0.593,
      "step": 2995
    },
    {
      "epoch": 0.8028942784403055,
      "grad_norm": 0.6583080012505652,
      "learning_rate": 4.794551687579777e-06,
      "loss": 0.6056,
      "step": 2996
    },
    {
      "epoch": 0.8031622671847782,
      "grad_norm": 0.6473943600814118,
      "learning_rate": 4.794411756624125e-06,
      "loss": 0.6038,
      "step": 2997
    },
    {
      "epoch": 0.803430255929251,
      "grad_norm": 0.662539565969468,
      "learning_rate": 4.794271780074561e-06,
      "loss": 0.5973,
      "step": 2998
    },
    {
      "epoch": 0.8036982446737237,
      "grad_norm": 0.6631687100507824,
      "learning_rate": 4.794131757933867e-06,
      "loss": 0.579,
      "step": 2999
    },
    {
      "epoch": 0.8039662334181964,
      "grad_norm": 0.6453970867157985,
      "learning_rate": 4.793991690204826e-06,
      "loss": 0.5771,
      "step": 3000
    },
    {
      "epoch": 0.8042342221626692,
      "grad_norm": 0.7043280817804779,
      "learning_rate": 4.79385157689022e-06,
      "loss": 0.5804,
      "step": 3001
    },
    {
      "epoch": 0.8045022109071419,
      "grad_norm": 0.6581961680418803,
      "learning_rate": 4.793711417992836e-06,
      "loss": 0.5688,
      "step": 3002
    },
    {
      "epoch": 0.8047701996516147,
      "grad_norm": 0.6693681676973052,
      "learning_rate": 4.793571213515456e-06,
      "loss": 0.5682,
      "step": 3003
    },
    {
      "epoch": 0.8050381883960873,
      "grad_norm": 0.673333163660613,
      "learning_rate": 4.7934309634608676e-06,
      "loss": 0.5934,
      "step": 3004
    },
    {
      "epoch": 0.8053061771405601,
      "grad_norm": 0.6853145009976335,
      "learning_rate": 4.793290667831858e-06,
      "loss": 0.5976,
      "step": 3005
    },
    {
      "epoch": 0.8055741658850328,
      "grad_norm": 0.6511516819439997,
      "learning_rate": 4.7931503266312154e-06,
      "loss": 0.6002,
      "step": 3006
    },
    {
      "epoch": 0.8058421546295056,
      "grad_norm": 0.6485175014285213,
      "learning_rate": 4.793009939861728e-06,
      "loss": 0.5993,
      "step": 3007
    },
    {
      "epoch": 0.8061101433739782,
      "grad_norm": 0.655674710565923,
      "learning_rate": 4.792869507526185e-06,
      "loss": 0.5802,
      "step": 3008
    },
    {
      "epoch": 0.806378132118451,
      "grad_norm": 0.645231378218113,
      "learning_rate": 4.792729029627378e-06,
      "loss": 0.5737,
      "step": 3009
    },
    {
      "epoch": 0.8066461208629238,
      "grad_norm": 0.6482141853734756,
      "learning_rate": 4.792588506168099e-06,
      "loss": 0.5496,
      "step": 3010
    },
    {
      "epoch": 0.8069141096073965,
      "grad_norm": 0.6513888900418324,
      "learning_rate": 4.792447937151138e-06,
      "loss": 0.5744,
      "step": 3011
    },
    {
      "epoch": 0.8071820983518693,
      "grad_norm": 0.6611139017664094,
      "learning_rate": 4.792307322579292e-06,
      "loss": 0.6055,
      "step": 3012
    },
    {
      "epoch": 0.8074500870963419,
      "grad_norm": 0.6556091625244778,
      "learning_rate": 4.792166662455352e-06,
      "loss": 0.5678,
      "step": 3013
    },
    {
      "epoch": 0.8077180758408147,
      "grad_norm": 0.6373769826784949,
      "learning_rate": 4.792025956782114e-06,
      "loss": 0.5772,
      "step": 3014
    },
    {
      "epoch": 0.8079860645852874,
      "grad_norm": 0.734634716482114,
      "learning_rate": 4.791885205562375e-06,
      "loss": 0.6167,
      "step": 3015
    },
    {
      "epoch": 0.8082540533297602,
      "grad_norm": 0.6314328907172989,
      "learning_rate": 4.791744408798932e-06,
      "loss": 0.5702,
      "step": 3016
    },
    {
      "epoch": 0.8085220420742328,
      "grad_norm": 0.6896380912342294,
      "learning_rate": 4.791603566494582e-06,
      "loss": 0.5941,
      "step": 3017
    },
    {
      "epoch": 0.8087900308187056,
      "grad_norm": 0.6638935461328299,
      "learning_rate": 4.791462678652124e-06,
      "loss": 0.5718,
      "step": 3018
    },
    {
      "epoch": 0.8090580195631784,
      "grad_norm": 0.6582239665118582,
      "learning_rate": 4.791321745274358e-06,
      "loss": 0.6067,
      "step": 3019
    },
    {
      "epoch": 0.8093260083076511,
      "grad_norm": 0.6756552444091574,
      "learning_rate": 4.7911807663640834e-06,
      "loss": 0.6008,
      "step": 3020
    },
    {
      "epoch": 0.8095939970521238,
      "grad_norm": 0.695631859171892,
      "learning_rate": 4.791039741924104e-06,
      "loss": 0.5886,
      "step": 3021
    },
    {
      "epoch": 0.8098619857965965,
      "grad_norm": 0.6373777215117694,
      "learning_rate": 4.79089867195722e-06,
      "loss": 0.581,
      "step": 3022
    },
    {
      "epoch": 0.8101299745410693,
      "grad_norm": 0.674865457742749,
      "learning_rate": 4.7907575564662354e-06,
      "loss": 0.6077,
      "step": 3023
    },
    {
      "epoch": 0.810397963285542,
      "grad_norm": 0.662597604720015,
      "learning_rate": 4.790616395453955e-06,
      "loss": 0.568,
      "step": 3024
    },
    {
      "epoch": 0.8106659520300148,
      "grad_norm": 0.6705398427677154,
      "learning_rate": 4.790475188923184e-06,
      "loss": 0.6017,
      "step": 3025
    },
    {
      "epoch": 0.8109339407744874,
      "grad_norm": 0.6636310183456529,
      "learning_rate": 4.790333936876727e-06,
      "loss": 0.6114,
      "step": 3026
    },
    {
      "epoch": 0.8112019295189602,
      "grad_norm": 0.6535875856388607,
      "learning_rate": 4.790192639317392e-06,
      "loss": 0.583,
      "step": 3027
    },
    {
      "epoch": 0.811469918263433,
      "grad_norm": 0.6520803595951317,
      "learning_rate": 4.790051296247987e-06,
      "loss": 0.5765,
      "step": 3028
    },
    {
      "epoch": 0.8117379070079057,
      "grad_norm": 0.6680505985671777,
      "learning_rate": 4.78990990767132e-06,
      "loss": 0.5744,
      "step": 3029
    },
    {
      "epoch": 0.8120058957523784,
      "grad_norm": 0.6629849832521669,
      "learning_rate": 4.789768473590201e-06,
      "loss": 0.5689,
      "step": 3030
    },
    {
      "epoch": 0.8122738844968511,
      "grad_norm": 0.6707275808163337,
      "learning_rate": 4.789626994007442e-06,
      "loss": 0.6057,
      "step": 3031
    },
    {
      "epoch": 0.8125418732413239,
      "grad_norm": 0.6672675083944271,
      "learning_rate": 4.789485468925852e-06,
      "loss": 0.578,
      "step": 3032
    },
    {
      "epoch": 0.8128098619857966,
      "grad_norm": 0.6637475955600209,
      "learning_rate": 4.789343898348245e-06,
      "loss": 0.6134,
      "step": 3033
    },
    {
      "epoch": 0.8130778507302693,
      "grad_norm": 0.6700705749644202,
      "learning_rate": 4.789202282277432e-06,
      "loss": 0.581,
      "step": 3034
    },
    {
      "epoch": 0.813345839474742,
      "grad_norm": 0.7346037610393753,
      "learning_rate": 4.789060620716229e-06,
      "loss": 0.6025,
      "step": 3035
    },
    {
      "epoch": 0.8136138282192148,
      "grad_norm": 0.6653717727668559,
      "learning_rate": 4.788918913667452e-06,
      "loss": 0.5876,
      "step": 3036
    },
    {
      "epoch": 0.8138818169636876,
      "grad_norm": 0.6458162014655767,
      "learning_rate": 4.7887771611339145e-06,
      "loss": 0.5523,
      "step": 3037
    },
    {
      "epoch": 0.8141498057081603,
      "grad_norm": 0.6333285375362169,
      "learning_rate": 4.788635363118435e-06,
      "loss": 0.5708,
      "step": 3038
    },
    {
      "epoch": 0.814417794452633,
      "grad_norm": 0.652478760368634,
      "learning_rate": 4.788493519623831e-06,
      "loss": 0.5977,
      "step": 3039
    },
    {
      "epoch": 0.8146857831971057,
      "grad_norm": 0.6351939362658836,
      "learning_rate": 4.7883516306529225e-06,
      "loss": 0.586,
      "step": 3040
    },
    {
      "epoch": 0.8149537719415785,
      "grad_norm": 0.6362360865421935,
      "learning_rate": 4.788209696208526e-06,
      "loss": 0.5868,
      "step": 3041
    },
    {
      "epoch": 0.8152217606860512,
      "grad_norm": 0.666134862153884,
      "learning_rate": 4.788067716293464e-06,
      "loss": 0.5972,
      "step": 3042
    },
    {
      "epoch": 0.8154897494305239,
      "grad_norm": 0.6518692120046199,
      "learning_rate": 4.787925690910557e-06,
      "loss": 0.5976,
      "step": 3043
    },
    {
      "epoch": 0.8157577381749966,
      "grad_norm": 0.6471819944701265,
      "learning_rate": 4.787783620062629e-06,
      "loss": 0.5695,
      "step": 3044
    },
    {
      "epoch": 0.8160257269194694,
      "grad_norm": 0.6462345971005997,
      "learning_rate": 4.7876415037525015e-06,
      "loss": 0.5779,
      "step": 3045
    },
    {
      "epoch": 0.8162937156639422,
      "grad_norm": 0.6575617859962053,
      "learning_rate": 4.787499341982999e-06,
      "loss": 0.56,
      "step": 3046
    },
    {
      "epoch": 0.8165617044084148,
      "grad_norm": 0.6609334228268408,
      "learning_rate": 4.787357134756947e-06,
      "loss": 0.5802,
      "step": 3047
    },
    {
      "epoch": 0.8168296931528876,
      "grad_norm": 0.6592322538442348,
      "learning_rate": 4.78721488207717e-06,
      "loss": 0.5794,
      "step": 3048
    },
    {
      "epoch": 0.8170976818973603,
      "grad_norm": 0.6769767872868268,
      "learning_rate": 4.787072583946497e-06,
      "loss": 0.6071,
      "step": 3049
    },
    {
      "epoch": 0.8173656706418331,
      "grad_norm": 0.6582484554983433,
      "learning_rate": 4.786930240367754e-06,
      "loss": 0.5867,
      "step": 3050
    },
    {
      "epoch": 0.8176336593863057,
      "grad_norm": 0.654810827773208,
      "learning_rate": 4.78678785134377e-06,
      "loss": 0.574,
      "step": 3051
    },
    {
      "epoch": 0.8179016481307785,
      "grad_norm": 0.6795588923260416,
      "learning_rate": 4.786645416877375e-06,
      "loss": 0.5578,
      "step": 3052
    },
    {
      "epoch": 0.8181696368752512,
      "grad_norm": 0.6556024335937456,
      "learning_rate": 4.786502936971399e-06,
      "loss": 0.5868,
      "step": 3053
    },
    {
      "epoch": 0.818437625619724,
      "grad_norm": 0.6720866547451356,
      "learning_rate": 4.786360411628673e-06,
      "loss": 0.6009,
      "step": 3054
    },
    {
      "epoch": 0.8187056143641968,
      "grad_norm": 0.6487025460697482,
      "learning_rate": 4.78621784085203e-06,
      "loss": 0.5175,
      "step": 3055
    },
    {
      "epoch": 0.8189736031086694,
      "grad_norm": 0.6487432036343007,
      "learning_rate": 4.786075224644302e-06,
      "loss": 0.5782,
      "step": 3056
    },
    {
      "epoch": 0.8192415918531422,
      "grad_norm": 0.6592083417776743,
      "learning_rate": 4.7859325630083245e-06,
      "loss": 0.5804,
      "step": 3057
    },
    {
      "epoch": 0.8195095805976149,
      "grad_norm": 0.7022632338124812,
      "learning_rate": 4.785789855946932e-06,
      "loss": 0.6125,
      "step": 3058
    },
    {
      "epoch": 0.8197775693420877,
      "grad_norm": 0.6528922512513946,
      "learning_rate": 4.785647103462959e-06,
      "loss": 0.5361,
      "step": 3059
    },
    {
      "epoch": 0.8200455580865603,
      "grad_norm": 0.660010570097635,
      "learning_rate": 4.785504305559244e-06,
      "loss": 0.6225,
      "step": 3060
    },
    {
      "epoch": 0.8203135468310331,
      "grad_norm": 0.6606484270822474,
      "learning_rate": 4.785361462238624e-06,
      "loss": 0.5957,
      "step": 3061
    },
    {
      "epoch": 0.8205815355755058,
      "grad_norm": 0.6615058412831081,
      "learning_rate": 4.785218573503938e-06,
      "loss": 0.5595,
      "step": 3062
    },
    {
      "epoch": 0.8208495243199786,
      "grad_norm": 0.6671024097166335,
      "learning_rate": 4.7850756393580234e-06,
      "loss": 0.6119,
      "step": 3063
    },
    {
      "epoch": 0.8211175130644512,
      "grad_norm": 0.6897919899858682,
      "learning_rate": 4.784932659803723e-06,
      "loss": 0.6244,
      "step": 3064
    },
    {
      "epoch": 0.821385501808924,
      "grad_norm": 0.641997547162022,
      "learning_rate": 4.7847896348438775e-06,
      "loss": 0.5831,
      "step": 3065
    },
    {
      "epoch": 0.8216534905533968,
      "grad_norm": 0.6630935449691221,
      "learning_rate": 4.784646564481328e-06,
      "loss": 0.6086,
      "step": 3066
    },
    {
      "epoch": 0.8219214792978695,
      "grad_norm": 0.6076689918441524,
      "learning_rate": 4.784503448718918e-06,
      "loss": 0.5115,
      "step": 3067
    },
    {
      "epoch": 0.8221894680423423,
      "grad_norm": 0.6440198592105716,
      "learning_rate": 4.784360287559492e-06,
      "loss": 0.5608,
      "step": 3068
    },
    {
      "epoch": 0.8224574567868149,
      "grad_norm": 0.6520223332662822,
      "learning_rate": 4.784217081005894e-06,
      "loss": 0.5246,
      "step": 3069
    },
    {
      "epoch": 0.8227254455312877,
      "grad_norm": 0.6614603623247449,
      "learning_rate": 4.7840738290609714e-06,
      "loss": 0.5902,
      "step": 3070
    },
    {
      "epoch": 0.8229934342757604,
      "grad_norm": 0.6798050720865745,
      "learning_rate": 4.783930531727568e-06,
      "loss": 0.5865,
      "step": 3071
    },
    {
      "epoch": 0.8232614230202332,
      "grad_norm": 0.6628935566460417,
      "learning_rate": 4.783787189008535e-06,
      "loss": 0.6246,
      "step": 3072
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.6493490942336831,
      "learning_rate": 4.783643800906718e-06,
      "loss": 0.5781,
      "step": 3073
    },
    {
      "epoch": 0.8237974005091786,
      "grad_norm": 0.6479266716914118,
      "learning_rate": 4.783500367424968e-06,
      "loss": 0.5593,
      "step": 3074
    },
    {
      "epoch": 0.8240653892536514,
      "grad_norm": 0.6542400843052544,
      "learning_rate": 4.783356888566134e-06,
      "loss": 0.5867,
      "step": 3075
    },
    {
      "epoch": 0.8243333779981241,
      "grad_norm": 0.6722009467255908,
      "learning_rate": 4.783213364333068e-06,
      "loss": 0.5694,
      "step": 3076
    },
    {
      "epoch": 0.8246013667425968,
      "grad_norm": 0.6845192646854604,
      "learning_rate": 4.7830697947286215e-06,
      "loss": 0.5807,
      "step": 3077
    },
    {
      "epoch": 0.8248693554870695,
      "grad_norm": 0.6571874495519858,
      "learning_rate": 4.782926179755648e-06,
      "loss": 0.5477,
      "step": 3078
    },
    {
      "epoch": 0.8251373442315423,
      "grad_norm": 0.658340613454381,
      "learning_rate": 4.782782519417002e-06,
      "loss": 0.5843,
      "step": 3079
    },
    {
      "epoch": 0.825405332976015,
      "grad_norm": 0.6910353850591002,
      "learning_rate": 4.782638813715537e-06,
      "loss": 0.5552,
      "step": 3080
    },
    {
      "epoch": 0.8256733217204877,
      "grad_norm": 0.64919745622588,
      "learning_rate": 4.782495062654108e-06,
      "loss": 0.577,
      "step": 3081
    },
    {
      "epoch": 0.8259413104649604,
      "grad_norm": 0.6555984403689648,
      "learning_rate": 4.7823512662355745e-06,
      "loss": 0.5479,
      "step": 3082
    },
    {
      "epoch": 0.8262092992094332,
      "grad_norm": 0.6588216287308839,
      "learning_rate": 4.782207424462791e-06,
      "loss": 0.5585,
      "step": 3083
    },
    {
      "epoch": 0.826477287953906,
      "grad_norm": 0.6604418381576681,
      "learning_rate": 4.782063537338617e-06,
      "loss": 0.5999,
      "step": 3084
    },
    {
      "epoch": 0.8267452766983787,
      "grad_norm": 0.6903300615386141,
      "learning_rate": 4.781919604865914e-06,
      "loss": 0.6141,
      "step": 3085
    },
    {
      "epoch": 0.8270132654428514,
      "grad_norm": 0.655110574755576,
      "learning_rate": 4.781775627047538e-06,
      "loss": 0.5476,
      "step": 3086
    },
    {
      "epoch": 0.8272812541873241,
      "grad_norm": 0.6508091330957595,
      "learning_rate": 4.781631603886353e-06,
      "loss": 0.6149,
      "step": 3087
    },
    {
      "epoch": 0.8275492429317969,
      "grad_norm": 0.6427198016994086,
      "learning_rate": 4.78148753538522e-06,
      "loss": 0.5641,
      "step": 3088
    },
    {
      "epoch": 0.8278172316762696,
      "grad_norm": 0.6545717759733436,
      "learning_rate": 4.781343421547001e-06,
      "loss": 0.5937,
      "step": 3089
    },
    {
      "epoch": 0.8280852204207423,
      "grad_norm": 0.7249891167497441,
      "learning_rate": 4.781199262374563e-06,
      "loss": 0.6126,
      "step": 3090
    },
    {
      "epoch": 0.828353209165215,
      "grad_norm": 0.6641732380946422,
      "learning_rate": 4.781055057870767e-06,
      "loss": 0.5988,
      "step": 3091
    },
    {
      "epoch": 0.8286211979096878,
      "grad_norm": 0.6561534399640059,
      "learning_rate": 4.78091080803848e-06,
      "loss": 0.571,
      "step": 3092
    },
    {
      "epoch": 0.8288891866541606,
      "grad_norm": 0.6550296039910594,
      "learning_rate": 4.78076651288057e-06,
      "loss": 0.5698,
      "step": 3093
    },
    {
      "epoch": 0.8291571753986332,
      "grad_norm": 0.662216502189422,
      "learning_rate": 4.780622172399902e-06,
      "loss": 0.6005,
      "step": 3094
    },
    {
      "epoch": 0.829425164143106,
      "grad_norm": 0.6649222201755896,
      "learning_rate": 4.780477786599346e-06,
      "loss": 0.6101,
      "step": 3095
    },
    {
      "epoch": 0.8296931528875787,
      "grad_norm": 0.6847759327918028,
      "learning_rate": 4.780333355481769e-06,
      "loss": 0.5716,
      "step": 3096
    },
    {
      "epoch": 0.8299611416320515,
      "grad_norm": 0.6738285160433493,
      "learning_rate": 4.780188879050044e-06,
      "loss": 0.5844,
      "step": 3097
    },
    {
      "epoch": 0.8302291303765242,
      "grad_norm": 0.6670528874521068,
      "learning_rate": 4.78004435730704e-06,
      "loss": 0.5851,
      "step": 3098
    },
    {
      "epoch": 0.8304971191209969,
      "grad_norm": 0.6496727558420994,
      "learning_rate": 4.7798997902556295e-06,
      "loss": 0.5746,
      "step": 3099
    },
    {
      "epoch": 0.8307651078654696,
      "grad_norm": 0.6842359269812276,
      "learning_rate": 4.779755177898686e-06,
      "loss": 0.6062,
      "step": 3100
    },
    {
      "epoch": 0.8310330966099424,
      "grad_norm": 0.665853022476175,
      "learning_rate": 4.779610520239081e-06,
      "loss": 0.6024,
      "step": 3101
    },
    {
      "epoch": 0.8313010853544152,
      "grad_norm": 0.6390409626842742,
      "learning_rate": 4.779465817279692e-06,
      "loss": 0.5533,
      "step": 3102
    },
    {
      "epoch": 0.8315690740988878,
      "grad_norm": 0.6499930028933891,
      "learning_rate": 4.779321069023393e-06,
      "loss": 0.5796,
      "step": 3103
    },
    {
      "epoch": 0.8318370628433606,
      "grad_norm": 0.6909685863214112,
      "learning_rate": 4.77917627547306e-06,
      "loss": 0.5915,
      "step": 3104
    },
    {
      "epoch": 0.8321050515878333,
      "grad_norm": 0.6685122172854053,
      "learning_rate": 4.77903143663157e-06,
      "loss": 0.5466,
      "step": 3105
    },
    {
      "epoch": 0.8323730403323061,
      "grad_norm": 0.673850635570106,
      "learning_rate": 4.778886552501803e-06,
      "loss": 0.5434,
      "step": 3106
    },
    {
      "epoch": 0.8326410290767787,
      "grad_norm": 0.7031969404492439,
      "learning_rate": 4.778741623086637e-06,
      "loss": 0.5699,
      "step": 3107
    },
    {
      "epoch": 0.8329090178212515,
      "grad_norm": 0.670106316768374,
      "learning_rate": 4.778596648388952e-06,
      "loss": 0.5907,
      "step": 3108
    },
    {
      "epoch": 0.8331770065657242,
      "grad_norm": 0.6771246530714944,
      "learning_rate": 4.778451628411629e-06,
      "loss": 0.5662,
      "step": 3109
    },
    {
      "epoch": 0.833444995310197,
      "grad_norm": 0.6633131769777094,
      "learning_rate": 4.778306563157549e-06,
      "loss": 0.6164,
      "step": 3110
    },
    {
      "epoch": 0.8337129840546698,
      "grad_norm": 0.6657744480168081,
      "learning_rate": 4.778161452629596e-06,
      "loss": 0.5871,
      "step": 3111
    },
    {
      "epoch": 0.8339809727991424,
      "grad_norm": 0.6636601140125372,
      "learning_rate": 4.7780162968306535e-06,
      "loss": 0.6017,
      "step": 3112
    },
    {
      "epoch": 0.8342489615436152,
      "grad_norm": 0.6680282294101109,
      "learning_rate": 4.777871095763605e-06,
      "loss": 0.6129,
      "step": 3113
    },
    {
      "epoch": 0.8345169502880879,
      "grad_norm": 0.7050851505892064,
      "learning_rate": 4.777725849431336e-06,
      "loss": 0.5694,
      "step": 3114
    },
    {
      "epoch": 0.8347849390325607,
      "grad_norm": 0.6287968713988534,
      "learning_rate": 4.777580557836734e-06,
      "loss": 0.5689,
      "step": 3115
    },
    {
      "epoch": 0.8350529277770333,
      "grad_norm": 0.6586857820036668,
      "learning_rate": 4.777435220982685e-06,
      "loss": 0.5713,
      "step": 3116
    },
    {
      "epoch": 0.8353209165215061,
      "grad_norm": 0.6837995379948701,
      "learning_rate": 4.777289838872078e-06,
      "loss": 0.5842,
      "step": 3117
    },
    {
      "epoch": 0.8355889052659788,
      "grad_norm": 0.6721460590664821,
      "learning_rate": 4.777144411507801e-06,
      "loss": 0.5762,
      "step": 3118
    },
    {
      "epoch": 0.8358568940104516,
      "grad_norm": 0.6395882973019483,
      "learning_rate": 4.776998938892745e-06,
      "loss": 0.5577,
      "step": 3119
    },
    {
      "epoch": 0.8361248827549242,
      "grad_norm": 0.6400963597401125,
      "learning_rate": 4.7768534210298e-06,
      "loss": 0.598,
      "step": 3120
    },
    {
      "epoch": 0.836392871499397,
      "grad_norm": 0.642515144860881,
      "learning_rate": 4.776707857921858e-06,
      "loss": 0.5709,
      "step": 3121
    },
    {
      "epoch": 0.8366608602438698,
      "grad_norm": 0.6603162423935492,
      "learning_rate": 4.776562249571811e-06,
      "loss": 0.6045,
      "step": 3122
    },
    {
      "epoch": 0.8369288489883425,
      "grad_norm": 0.6487136652969808,
      "learning_rate": 4.776416595982553e-06,
      "loss": 0.574,
      "step": 3123
    },
    {
      "epoch": 0.8371968377328152,
      "grad_norm": 0.667384389524996,
      "learning_rate": 4.776270897156978e-06,
      "loss": 0.5883,
      "step": 3124
    },
    {
      "epoch": 0.8374648264772879,
      "grad_norm": 0.6579715920634281,
      "learning_rate": 4.776125153097982e-06,
      "loss": 0.5757,
      "step": 3125
    },
    {
      "epoch": 0.8377328152217607,
      "grad_norm": 0.6540748887972132,
      "learning_rate": 4.775979363808461e-06,
      "loss": 0.5612,
      "step": 3126
    },
    {
      "epoch": 0.8380008039662334,
      "grad_norm": 0.6619911628962946,
      "learning_rate": 4.7758335292913125e-06,
      "loss": 0.5799,
      "step": 3127
    },
    {
      "epoch": 0.8382687927107062,
      "grad_norm": 0.6676241311441824,
      "learning_rate": 4.775687649549433e-06,
      "loss": 0.5965,
      "step": 3128
    },
    {
      "epoch": 0.8385367814551788,
      "grad_norm": 0.6704240792245163,
      "learning_rate": 4.775541724585723e-06,
      "loss": 0.5623,
      "step": 3129
    },
    {
      "epoch": 0.8388047701996516,
      "grad_norm": 0.6548427899656942,
      "learning_rate": 4.775395754403081e-06,
      "loss": 0.6046,
      "step": 3130
    },
    {
      "epoch": 0.8390727589441244,
      "grad_norm": 0.6677781741925785,
      "learning_rate": 4.775249739004408e-06,
      "loss": 0.5981,
      "step": 3131
    },
    {
      "epoch": 0.8393407476885971,
      "grad_norm": 0.7058846740100233,
      "learning_rate": 4.775103678392606e-06,
      "loss": 0.5986,
      "step": 3132
    },
    {
      "epoch": 0.8396087364330698,
      "grad_norm": 0.6644140447264388,
      "learning_rate": 4.774957572570577e-06,
      "loss": 0.5707,
      "step": 3133
    },
    {
      "epoch": 0.8398767251775425,
      "grad_norm": 0.6531062385961287,
      "learning_rate": 4.7748114215412255e-06,
      "loss": 0.5776,
      "step": 3134
    },
    {
      "epoch": 0.8401447139220153,
      "grad_norm": 0.6735514651225047,
      "learning_rate": 4.774665225307454e-06,
      "loss": 0.5977,
      "step": 3135
    },
    {
      "epoch": 0.840412702666488,
      "grad_norm": 0.653129968247748,
      "learning_rate": 4.774518983872169e-06,
      "loss": 0.6064,
      "step": 3136
    },
    {
      "epoch": 0.8406806914109607,
      "grad_norm": 0.6511675223450772,
      "learning_rate": 4.774372697238276e-06,
      "loss": 0.5529,
      "step": 3137
    },
    {
      "epoch": 0.8409486801554334,
      "grad_norm": 0.6561300425153519,
      "learning_rate": 4.774226365408683e-06,
      "loss": 0.5748,
      "step": 3138
    },
    {
      "epoch": 0.8412166688999062,
      "grad_norm": 0.6742017491595041,
      "learning_rate": 4.7740799883862966e-06,
      "loss": 0.6057,
      "step": 3139
    },
    {
      "epoch": 0.841484657644379,
      "grad_norm": 0.6468662481500759,
      "learning_rate": 4.773933566174026e-06,
      "loss": 0.5924,
      "step": 3140
    },
    {
      "epoch": 0.8417526463888517,
      "grad_norm": 0.6638153117261949,
      "learning_rate": 4.7737870987747805e-06,
      "loss": 0.6119,
      "step": 3141
    },
    {
      "epoch": 0.8420206351333244,
      "grad_norm": 0.7361718999026284,
      "learning_rate": 4.77364058619147e-06,
      "loss": 0.5562,
      "step": 3142
    },
    {
      "epoch": 0.8422886238777971,
      "grad_norm": 0.668978908390312,
      "learning_rate": 4.773494028427009e-06,
      "loss": 0.5665,
      "step": 3143
    },
    {
      "epoch": 0.8425566126222699,
      "grad_norm": 0.6687412536875977,
      "learning_rate": 4.7733474254843056e-06,
      "loss": 0.5819,
      "step": 3144
    },
    {
      "epoch": 0.8428246013667426,
      "grad_norm": 0.6558180617217609,
      "learning_rate": 4.773200777366277e-06,
      "loss": 0.5905,
      "step": 3145
    },
    {
      "epoch": 0.8430925901112153,
      "grad_norm": 0.646586682294032,
      "learning_rate": 4.773054084075836e-06,
      "loss": 0.5889,
      "step": 3146
    },
    {
      "epoch": 0.843360578855688,
      "grad_norm": 0.6767777748808228,
      "learning_rate": 4.772907345615896e-06,
      "loss": 0.5775,
      "step": 3147
    },
    {
      "epoch": 0.8436285676001608,
      "grad_norm": 0.6660283621218177,
      "learning_rate": 4.772760561989374e-06,
      "loss": 0.5787,
      "step": 3148
    },
    {
      "epoch": 0.8438965563446336,
      "grad_norm": 0.6825137039725411,
      "learning_rate": 4.772613733199187e-06,
      "loss": 0.5994,
      "step": 3149
    },
    {
      "epoch": 0.8441645450891062,
      "grad_norm": 0.6838419825959201,
      "learning_rate": 4.772466859248253e-06,
      "loss": 0.5932,
      "step": 3150
    },
    {
      "epoch": 0.844432533833579,
      "grad_norm": 0.6827245142953945,
      "learning_rate": 4.772319940139492e-06,
      "loss": 0.5768,
      "step": 3151
    },
    {
      "epoch": 0.8447005225780517,
      "grad_norm": 0.6911158944966432,
      "learning_rate": 4.772172975875821e-06,
      "loss": 0.5634,
      "step": 3152
    },
    {
      "epoch": 0.8449685113225245,
      "grad_norm": 0.6943344911805317,
      "learning_rate": 4.7720259664601606e-06,
      "loss": 0.6222,
      "step": 3153
    },
    {
      "epoch": 0.8452365000669971,
      "grad_norm": 0.6559846598233787,
      "learning_rate": 4.7718789118954335e-06,
      "loss": 0.604,
      "step": 3154
    },
    {
      "epoch": 0.8455044888114699,
      "grad_norm": 0.650772953163498,
      "learning_rate": 4.771731812184561e-06,
      "loss": 0.599,
      "step": 3155
    },
    {
      "epoch": 0.8457724775559426,
      "grad_norm": 0.6733255402984325,
      "learning_rate": 4.771584667330466e-06,
      "loss": 0.6211,
      "step": 3156
    },
    {
      "epoch": 0.8460404663004154,
      "grad_norm": 0.6471948026119477,
      "learning_rate": 4.771437477336075e-06,
      "loss": 0.5935,
      "step": 3157
    },
    {
      "epoch": 0.8463084550448882,
      "grad_norm": 0.6593155405874451,
      "learning_rate": 4.77129024220431e-06,
      "loss": 0.5924,
      "step": 3158
    },
    {
      "epoch": 0.8465764437893608,
      "grad_norm": 0.6731987624982129,
      "learning_rate": 4.771142961938097e-06,
      "loss": 0.5532,
      "step": 3159
    },
    {
      "epoch": 0.8468444325338336,
      "grad_norm": 0.6187349021836394,
      "learning_rate": 4.770995636540364e-06,
      "loss": 0.5812,
      "step": 3160
    },
    {
      "epoch": 0.8471124212783063,
      "grad_norm": 0.6641227576469878,
      "learning_rate": 4.770848266014038e-06,
      "loss": 0.5449,
      "step": 3161
    },
    {
      "epoch": 0.8473804100227791,
      "grad_norm": 0.6363089927989328,
      "learning_rate": 4.770700850362049e-06,
      "loss": 0.5443,
      "step": 3162
    },
    {
      "epoch": 0.8476483987672517,
      "grad_norm": 0.6674816514611182,
      "learning_rate": 4.770553389587324e-06,
      "loss": 0.574,
      "step": 3163
    },
    {
      "epoch": 0.8479163875117245,
      "grad_norm": 0.6627860838283095,
      "learning_rate": 4.7704058836927945e-06,
      "loss": 0.5598,
      "step": 3164
    },
    {
      "epoch": 0.8481843762561972,
      "grad_norm": 0.6612800181072016,
      "learning_rate": 4.770258332681391e-06,
      "loss": 0.5707,
      "step": 3165
    },
    {
      "epoch": 0.84845236500067,
      "grad_norm": 0.6437683443652636,
      "learning_rate": 4.770110736556047e-06,
      "loss": 0.5978,
      "step": 3166
    },
    {
      "epoch": 0.8487203537451427,
      "grad_norm": 0.653547786239741,
      "learning_rate": 4.7699630953196945e-06,
      "loss": 0.605,
      "step": 3167
    },
    {
      "epoch": 0.8489883424896154,
      "grad_norm": 0.6555657895659907,
      "learning_rate": 4.7698154089752674e-06,
      "loss": 0.6081,
      "step": 3168
    },
    {
      "epoch": 0.8492563312340882,
      "grad_norm": 0.650801718904471,
      "learning_rate": 4.769667677525701e-06,
      "loss": 0.552,
      "step": 3169
    },
    {
      "epoch": 0.8495243199785609,
      "grad_norm": 0.6427768045556459,
      "learning_rate": 4.76951990097393e-06,
      "loss": 0.5607,
      "step": 3170
    },
    {
      "epoch": 0.8497923087230337,
      "grad_norm": 0.6416210564847686,
      "learning_rate": 4.769372079322891e-06,
      "loss": 0.5632,
      "step": 3171
    },
    {
      "epoch": 0.8500602974675063,
      "grad_norm": 0.6604105528405225,
      "learning_rate": 4.769224212575523e-06,
      "loss": 0.6191,
      "step": 3172
    },
    {
      "epoch": 0.8503282862119791,
      "grad_norm": 0.6442722847074478,
      "learning_rate": 4.769076300734763e-06,
      "loss": 0.5513,
      "step": 3173
    },
    {
      "epoch": 0.8505962749564518,
      "grad_norm": 0.6527687349521738,
      "learning_rate": 4.768928343803551e-06,
      "loss": 0.5749,
      "step": 3174
    },
    {
      "epoch": 0.8508642637009246,
      "grad_norm": 0.6471362784781073,
      "learning_rate": 4.768780341784826e-06,
      "loss": 0.571,
      "step": 3175
    },
    {
      "epoch": 0.8511322524453973,
      "grad_norm": 0.6655980218680901,
      "learning_rate": 4.76863229468153e-06,
      "loss": 0.6023,
      "step": 3176
    },
    {
      "epoch": 0.85140024118987,
      "grad_norm": 0.6775812262733332,
      "learning_rate": 4.768484202496606e-06,
      "loss": 0.6133,
      "step": 3177
    },
    {
      "epoch": 0.8516682299343428,
      "grad_norm": 0.6649740391552023,
      "learning_rate": 4.768336065232995e-06,
      "loss": 0.6222,
      "step": 3178
    },
    {
      "epoch": 0.8519362186788155,
      "grad_norm": 0.6683684291072788,
      "learning_rate": 4.768187882893641e-06,
      "loss": 0.6153,
      "step": 3179
    },
    {
      "epoch": 0.8522042074232882,
      "grad_norm": 0.6927092912862322,
      "learning_rate": 4.7680396554814886e-06,
      "loss": 0.576,
      "step": 3180
    },
    {
      "epoch": 0.8524721961677609,
      "grad_norm": 0.6370384297827594,
      "learning_rate": 4.767891382999485e-06,
      "loss": 0.6007,
      "step": 3181
    },
    {
      "epoch": 0.8527401849122337,
      "grad_norm": 0.6381453512867714,
      "learning_rate": 4.767743065450575e-06,
      "loss": 0.6054,
      "step": 3182
    },
    {
      "epoch": 0.8530081736567064,
      "grad_norm": 0.6415380666065423,
      "learning_rate": 4.767594702837707e-06,
      "loss": 0.5554,
      "step": 3183
    },
    {
      "epoch": 0.8532761624011792,
      "grad_norm": 0.6584905364760156,
      "learning_rate": 4.7674462951638265e-06,
      "loss": 0.5926,
      "step": 3184
    },
    {
      "epoch": 0.8535441511456519,
      "grad_norm": 0.6716012579872028,
      "learning_rate": 4.7672978424318865e-06,
      "loss": 0.612,
      "step": 3185
    },
    {
      "epoch": 0.8538121398901246,
      "grad_norm": 0.6668158551990547,
      "learning_rate": 4.767149344644835e-06,
      "loss": 0.6015,
      "step": 3186
    },
    {
      "epoch": 0.8540801286345974,
      "grad_norm": 0.6373927446245135,
      "learning_rate": 4.7670008018056225e-06,
      "loss": 0.5645,
      "step": 3187
    },
    {
      "epoch": 0.8543481173790701,
      "grad_norm": 0.6521891641931692,
      "learning_rate": 4.766852213917201e-06,
      "loss": 0.5679,
      "step": 3188
    },
    {
      "epoch": 0.8546161061235428,
      "grad_norm": 0.6413862184416523,
      "learning_rate": 4.766703580982525e-06,
      "loss": 0.5885,
      "step": 3189
    },
    {
      "epoch": 0.8548840948680155,
      "grad_norm": 0.6377099364493513,
      "learning_rate": 4.766554903004546e-06,
      "loss": 0.5991,
      "step": 3190
    },
    {
      "epoch": 0.8551520836124883,
      "grad_norm": 0.64108477506779,
      "learning_rate": 4.7664061799862185e-06,
      "loss": 0.5793,
      "step": 3191
    },
    {
      "epoch": 0.855420072356961,
      "grad_norm": 0.6483555379800019,
      "learning_rate": 4.7662574119305e-06,
      "loss": 0.5293,
      "step": 3192
    },
    {
      "epoch": 0.8556880611014337,
      "grad_norm": 0.6346428167372958,
      "learning_rate": 4.766108598840344e-06,
      "loss": 0.5542,
      "step": 3193
    },
    {
      "epoch": 0.8559560498459065,
      "grad_norm": 0.6522721451659285,
      "learning_rate": 4.765959740718709e-06,
      "loss": 0.5907,
      "step": 3194
    },
    {
      "epoch": 0.8562240385903792,
      "grad_norm": 0.642055893726123,
      "learning_rate": 4.7658108375685535e-06,
      "loss": 0.5573,
      "step": 3195
    },
    {
      "epoch": 0.856492027334852,
      "grad_norm": 0.6616473599651914,
      "learning_rate": 4.765661889392836e-06,
      "loss": 0.5836,
      "step": 3196
    },
    {
      "epoch": 0.8567600160793246,
      "grad_norm": 0.6590155702410356,
      "learning_rate": 4.765512896194517e-06,
      "loss": 0.5788,
      "step": 3197
    },
    {
      "epoch": 0.8570280048237974,
      "grad_norm": 0.6434439940034746,
      "learning_rate": 4.765363857976556e-06,
      "loss": 0.5636,
      "step": 3198
    },
    {
      "epoch": 0.8572959935682701,
      "grad_norm": 0.6497906675086813,
      "learning_rate": 4.7652147747419155e-06,
      "loss": 0.5909,
      "step": 3199
    },
    {
      "epoch": 0.8575639823127429,
      "grad_norm": 0.6365976481261912,
      "learning_rate": 4.765065646493557e-06,
      "loss": 0.5763,
      "step": 3200
    },
    {
      "epoch": 0.8578319710572156,
      "grad_norm": 0.6464067729779847,
      "learning_rate": 4.764916473234446e-06,
      "loss": 0.5858,
      "step": 3201
    },
    {
      "epoch": 0.8580999598016883,
      "grad_norm": 0.6496988102296476,
      "learning_rate": 4.764767254967544e-06,
      "loss": 0.5814,
      "step": 3202
    },
    {
      "epoch": 0.858367948546161,
      "grad_norm": 0.6914580476689312,
      "learning_rate": 4.764617991695819e-06,
      "loss": 0.6232,
      "step": 3203
    },
    {
      "epoch": 0.8586359372906338,
      "grad_norm": 0.6526463272137439,
      "learning_rate": 4.764468683422236e-06,
      "loss": 0.5984,
      "step": 3204
    },
    {
      "epoch": 0.8589039260351066,
      "grad_norm": 0.6577814127761742,
      "learning_rate": 4.764319330149761e-06,
      "loss": 0.56,
      "step": 3205
    },
    {
      "epoch": 0.8591719147795792,
      "grad_norm": 0.6589034572796547,
      "learning_rate": 4.7641699318813635e-06,
      "loss": 0.6039,
      "step": 3206
    },
    {
      "epoch": 0.859439903524052,
      "grad_norm": 0.6499401191906315,
      "learning_rate": 4.764020488620012e-06,
      "loss": 0.5693,
      "step": 3207
    },
    {
      "epoch": 0.8597078922685247,
      "grad_norm": 0.6459459494425217,
      "learning_rate": 4.763871000368676e-06,
      "loss": 0.5913,
      "step": 3208
    },
    {
      "epoch": 0.8599758810129975,
      "grad_norm": 0.6473766713478994,
      "learning_rate": 4.763721467130325e-06,
      "loss": 0.583,
      "step": 3209
    },
    {
      "epoch": 0.8602438697574701,
      "grad_norm": 0.6476459218600354,
      "learning_rate": 4.763571888907932e-06,
      "loss": 0.5952,
      "step": 3210
    },
    {
      "epoch": 0.8605118585019429,
      "grad_norm": 0.6362372400135851,
      "learning_rate": 4.763422265704469e-06,
      "loss": 0.5353,
      "step": 3211
    },
    {
      "epoch": 0.8607798472464157,
      "grad_norm": 0.6322601181608958,
      "learning_rate": 4.763272597522908e-06,
      "loss": 0.5417,
      "step": 3212
    },
    {
      "epoch": 0.8610478359908884,
      "grad_norm": 0.6640205804714657,
      "learning_rate": 4.7631228843662255e-06,
      "loss": 0.551,
      "step": 3213
    },
    {
      "epoch": 0.8613158247353612,
      "grad_norm": 0.6672010075281846,
      "learning_rate": 4.762973126237395e-06,
      "loss": 0.6033,
      "step": 3214
    },
    {
      "epoch": 0.8615838134798338,
      "grad_norm": 0.6584035082555381,
      "learning_rate": 4.7628233231393915e-06,
      "loss": 0.6112,
      "step": 3215
    },
    {
      "epoch": 0.8618518022243066,
      "grad_norm": 0.6509843784473753,
      "learning_rate": 4.762673475075194e-06,
      "loss": 0.5882,
      "step": 3216
    },
    {
      "epoch": 0.8621197909687793,
      "grad_norm": 0.6498852119322758,
      "learning_rate": 4.76252358204778e-06,
      "loss": 0.5741,
      "step": 3217
    },
    {
      "epoch": 0.8623877797132521,
      "grad_norm": 0.6636574648730721,
      "learning_rate": 4.762373644060127e-06,
      "loss": 0.5894,
      "step": 3218
    },
    {
      "epoch": 0.8626557684577247,
      "grad_norm": 0.6731304593579279,
      "learning_rate": 4.762223661115214e-06,
      "loss": 0.5946,
      "step": 3219
    },
    {
      "epoch": 0.8629237572021975,
      "grad_norm": 0.6461207080468656,
      "learning_rate": 4.762073633216023e-06,
      "loss": 0.59,
      "step": 3220
    },
    {
      "epoch": 0.8631917459466703,
      "grad_norm": 0.6511799167596549,
      "learning_rate": 4.761923560365535e-06,
      "loss": 0.578,
      "step": 3221
    },
    {
      "epoch": 0.863459734691143,
      "grad_norm": 0.6588403179600775,
      "learning_rate": 4.761773442566732e-06,
      "loss": 0.5924,
      "step": 3222
    },
    {
      "epoch": 0.8637277234356157,
      "grad_norm": 0.6688215202208388,
      "learning_rate": 4.761623279822597e-06,
      "loss": 0.594,
      "step": 3223
    },
    {
      "epoch": 0.8639957121800884,
      "grad_norm": 0.6441413472349149,
      "learning_rate": 4.761473072136114e-06,
      "loss": 0.5672,
      "step": 3224
    },
    {
      "epoch": 0.8642637009245612,
      "grad_norm": 0.6298553848944114,
      "learning_rate": 4.761322819510268e-06,
      "loss": 0.5534,
      "step": 3225
    },
    {
      "epoch": 0.8645316896690339,
      "grad_norm": 0.6447316909739723,
      "learning_rate": 4.761172521948044e-06,
      "loss": 0.564,
      "step": 3226
    },
    {
      "epoch": 0.8647996784135066,
      "grad_norm": 0.6360367102675497,
      "learning_rate": 4.761022179452431e-06,
      "loss": 0.5842,
      "step": 3227
    },
    {
      "epoch": 0.8650676671579793,
      "grad_norm": 0.6535154839999836,
      "learning_rate": 4.760871792026413e-06,
      "loss": 0.5865,
      "step": 3228
    },
    {
      "epoch": 0.8653356559024521,
      "grad_norm": 0.6924166297799622,
      "learning_rate": 4.7607213596729805e-06,
      "loss": 0.5696,
      "step": 3229
    },
    {
      "epoch": 0.8656036446469249,
      "grad_norm": 0.6458111158815054,
      "learning_rate": 4.760570882395124e-06,
      "loss": 0.5842,
      "step": 3230
    },
    {
      "epoch": 0.8658716333913976,
      "grad_norm": 0.6377037617971301,
      "learning_rate": 4.760420360195831e-06,
      "loss": 0.5422,
      "step": 3231
    },
    {
      "epoch": 0.8661396221358703,
      "grad_norm": 0.6413730180171819,
      "learning_rate": 4.760269793078095e-06,
      "loss": 0.5488,
      "step": 3232
    },
    {
      "epoch": 0.866407610880343,
      "grad_norm": 0.6601897056682376,
      "learning_rate": 4.760119181044907e-06,
      "loss": 0.5693,
      "step": 3233
    },
    {
      "epoch": 0.8666755996248158,
      "grad_norm": 0.6532551985032985,
      "learning_rate": 4.75996852409926e-06,
      "loss": 0.5881,
      "step": 3234
    },
    {
      "epoch": 0.8669435883692885,
      "grad_norm": 0.6460459405147907,
      "learning_rate": 4.759817822244147e-06,
      "loss": 0.5532,
      "step": 3235
    },
    {
      "epoch": 0.8672115771137612,
      "grad_norm": 0.6593858736242975,
      "learning_rate": 4.7596670754825644e-06,
      "loss": 0.592,
      "step": 3236
    },
    {
      "epoch": 0.8674795658582339,
      "grad_norm": 0.627852384336917,
      "learning_rate": 4.759516283817508e-06,
      "loss": 0.5751,
      "step": 3237
    },
    {
      "epoch": 0.8677475546027067,
      "grad_norm": 0.657710422810941,
      "learning_rate": 4.759365447251971e-06,
      "loss": 0.5915,
      "step": 3238
    },
    {
      "epoch": 0.8680155433471795,
      "grad_norm": 0.6568054570262605,
      "learning_rate": 4.759214565788955e-06,
      "loss": 0.5789,
      "step": 3239
    },
    {
      "epoch": 0.8682835320916521,
      "grad_norm": 0.6822924282579698,
      "learning_rate": 4.7590636394314555e-06,
      "loss": 0.5609,
      "step": 3240
    },
    {
      "epoch": 0.8685515208361249,
      "grad_norm": 0.6764691713679462,
      "learning_rate": 4.758912668182473e-06,
      "loss": 0.6249,
      "step": 3241
    },
    {
      "epoch": 0.8688195095805976,
      "grad_norm": 0.6392049156664723,
      "learning_rate": 4.758761652045006e-06,
      "loss": 0.5745,
      "step": 3242
    },
    {
      "epoch": 0.8690874983250704,
      "grad_norm": 0.6436706024527438,
      "learning_rate": 4.758610591022057e-06,
      "loss": 0.5874,
      "step": 3243
    },
    {
      "epoch": 0.8693554870695431,
      "grad_norm": 0.6582398234024065,
      "learning_rate": 4.758459485116627e-06,
      "loss": 0.5607,
      "step": 3244
    },
    {
      "epoch": 0.8696234758140158,
      "grad_norm": 0.6628877513726351,
      "learning_rate": 4.758308334331719e-06,
      "loss": 0.5926,
      "step": 3245
    },
    {
      "epoch": 0.8698914645584885,
      "grad_norm": 0.6671932483862738,
      "learning_rate": 4.758157138670337e-06,
      "loss": 0.5656,
      "step": 3246
    },
    {
      "epoch": 0.8701594533029613,
      "grad_norm": 0.7001816391005472,
      "learning_rate": 4.758005898135486e-06,
      "loss": 0.5699,
      "step": 3247
    },
    {
      "epoch": 0.870427442047434,
      "grad_norm": 0.6757985306742152,
      "learning_rate": 4.757854612730169e-06,
      "loss": 0.5731,
      "step": 3248
    },
    {
      "epoch": 0.8706954307919067,
      "grad_norm": 0.6453075019323944,
      "learning_rate": 4.757703282457395e-06,
      "loss": 0.5951,
      "step": 3249
    },
    {
      "epoch": 0.8709634195363795,
      "grad_norm": 0.6661556265087716,
      "learning_rate": 4.75755190732017e-06,
      "loss": 0.601,
      "step": 3250
    },
    {
      "epoch": 0.8712314082808522,
      "grad_norm": 0.6547906122969417,
      "learning_rate": 4.757400487321502e-06,
      "loss": 0.5607,
      "step": 3251
    },
    {
      "epoch": 0.871499397025325,
      "grad_norm": 0.6812939856507824,
      "learning_rate": 4.7572490224644e-06,
      "loss": 0.5775,
      "step": 3252
    },
    {
      "epoch": 0.8717673857697976,
      "grad_norm": 0.6663191919060963,
      "learning_rate": 4.757097512751875e-06,
      "loss": 0.5892,
      "step": 3253
    },
    {
      "epoch": 0.8720353745142704,
      "grad_norm": 0.6458770706444957,
      "learning_rate": 4.756945958186936e-06,
      "loss": 0.5711,
      "step": 3254
    },
    {
      "epoch": 0.8723033632587431,
      "grad_norm": 0.660507465655499,
      "learning_rate": 4.756794358772595e-06,
      "loss": 0.5962,
      "step": 3255
    },
    {
      "epoch": 0.8725713520032159,
      "grad_norm": 0.6662907215356184,
      "learning_rate": 4.7566427145118655e-06,
      "loss": 0.5926,
      "step": 3256
    },
    {
      "epoch": 0.8728393407476887,
      "grad_norm": 0.6834883380218512,
      "learning_rate": 4.756491025407761e-06,
      "loss": 0.6048,
      "step": 3257
    },
    {
      "epoch": 0.8731073294921613,
      "grad_norm": 0.6572004140967049,
      "learning_rate": 4.756339291463295e-06,
      "loss": 0.5766,
      "step": 3258
    },
    {
      "epoch": 0.8733753182366341,
      "grad_norm": 0.6413214404216485,
      "learning_rate": 4.756187512681482e-06,
      "loss": 0.5605,
      "step": 3259
    },
    {
      "epoch": 0.8736433069811068,
      "grad_norm": 0.6452073451962534,
      "learning_rate": 4.756035689065339e-06,
      "loss": 0.5747,
      "step": 3260
    },
    {
      "epoch": 0.8739112957255796,
      "grad_norm": 0.6852004942528124,
      "learning_rate": 4.755883820617884e-06,
      "loss": 0.5668,
      "step": 3261
    },
    {
      "epoch": 0.8741792844700522,
      "grad_norm": 0.6839703879193435,
      "learning_rate": 4.755731907342133e-06,
      "loss": 0.6089,
      "step": 3262
    },
    {
      "epoch": 0.874447273214525,
      "grad_norm": 0.657281551176564,
      "learning_rate": 4.755579949241107e-06,
      "loss": 0.5935,
      "step": 3263
    },
    {
      "epoch": 0.8747152619589977,
      "grad_norm": 0.6599236807141726,
      "learning_rate": 4.7554279463178235e-06,
      "loss": 0.5761,
      "step": 3264
    },
    {
      "epoch": 0.8749832507034705,
      "grad_norm": 0.6656217090550969,
      "learning_rate": 4.755275898575304e-06,
      "loss": 0.5737,
      "step": 3265
    },
    {
      "epoch": 0.8752512394479431,
      "grad_norm": 0.687075182793911,
      "learning_rate": 4.75512380601657e-06,
      "loss": 0.5703,
      "step": 3266
    },
    {
      "epoch": 0.8755192281924159,
      "grad_norm": 0.6481278605841635,
      "learning_rate": 4.754971668644645e-06,
      "loss": 0.5799,
      "step": 3267
    },
    {
      "epoch": 0.8757872169368887,
      "grad_norm": 0.6745504280090725,
      "learning_rate": 4.75481948646255e-06,
      "loss": 0.5856,
      "step": 3268
    },
    {
      "epoch": 0.8760552056813614,
      "grad_norm": 0.6676614049163286,
      "learning_rate": 4.754667259473309e-06,
      "loss": 0.5938,
      "step": 3269
    },
    {
      "epoch": 0.8763231944258341,
      "grad_norm": 0.6517542899582469,
      "learning_rate": 4.754514987679949e-06,
      "loss": 0.5569,
      "step": 3270
    },
    {
      "epoch": 0.8765911831703068,
      "grad_norm": 0.628481318529479,
      "learning_rate": 4.754362671085495e-06,
      "loss": 0.5547,
      "step": 3271
    },
    {
      "epoch": 0.8768591719147796,
      "grad_norm": 0.6854083609839099,
      "learning_rate": 4.754210309692974e-06,
      "loss": 0.6279,
      "step": 3272
    },
    {
      "epoch": 0.8771271606592523,
      "grad_norm": 0.6685746728100495,
      "learning_rate": 4.754057903505414e-06,
      "loss": 0.569,
      "step": 3273
    },
    {
      "epoch": 0.8773951494037251,
      "grad_norm": 0.6393562408633156,
      "learning_rate": 4.7539054525258425e-06,
      "loss": 0.5635,
      "step": 3274
    },
    {
      "epoch": 0.8776631381481977,
      "grad_norm": 0.6435530485933503,
      "learning_rate": 4.753752956757289e-06,
      "loss": 0.5858,
      "step": 3275
    },
    {
      "epoch": 0.8779311268926705,
      "grad_norm": 0.6506318512297781,
      "learning_rate": 4.753600416202784e-06,
      "loss": 0.5867,
      "step": 3276
    },
    {
      "epoch": 0.8781991156371433,
      "grad_norm": 0.7586247773691235,
      "learning_rate": 4.75344783086536e-06,
      "loss": 0.5736,
      "step": 3277
    },
    {
      "epoch": 0.878467104381616,
      "grad_norm": 0.6865617466373237,
      "learning_rate": 4.753295200748048e-06,
      "loss": 0.6164,
      "step": 3278
    },
    {
      "epoch": 0.8787350931260887,
      "grad_norm": 0.6855320975687336,
      "learning_rate": 4.753142525853882e-06,
      "loss": 0.615,
      "step": 3279
    },
    {
      "epoch": 0.8790030818705614,
      "grad_norm": 0.6500096703363435,
      "learning_rate": 4.752989806185895e-06,
      "loss": 0.5593,
      "step": 3280
    },
    {
      "epoch": 0.8792710706150342,
      "grad_norm": 0.6577725489641091,
      "learning_rate": 4.752837041747122e-06,
      "loss": 0.5911,
      "step": 3281
    },
    {
      "epoch": 0.8795390593595069,
      "grad_norm": 0.6899659944861188,
      "learning_rate": 4.752684232540597e-06,
      "loss": 0.5494,
      "step": 3282
    },
    {
      "epoch": 0.8798070481039796,
      "grad_norm": 0.6518423637426826,
      "learning_rate": 4.75253137856936e-06,
      "loss": 0.5914,
      "step": 3283
    },
    {
      "epoch": 0.8800750368484523,
      "grad_norm": 0.6639030216004308,
      "learning_rate": 4.752378479836447e-06,
      "loss": 0.5695,
      "step": 3284
    },
    {
      "epoch": 0.8803430255929251,
      "grad_norm": 0.6796131013370093,
      "learning_rate": 4.752225536344894e-06,
      "loss": 0.5749,
      "step": 3285
    },
    {
      "epoch": 0.8806110143373979,
      "grad_norm": 0.6693166125823026,
      "learning_rate": 4.752072548097745e-06,
      "loss": 0.5917,
      "step": 3286
    },
    {
      "epoch": 0.8808790030818706,
      "grad_norm": 0.6561358932154493,
      "learning_rate": 4.7519195150980355e-06,
      "loss": 0.5834,
      "step": 3287
    },
    {
      "epoch": 0.8811469918263433,
      "grad_norm": 0.6648878598497896,
      "learning_rate": 4.751766437348809e-06,
      "loss": 0.6077,
      "step": 3288
    },
    {
      "epoch": 0.881414980570816,
      "grad_norm": 0.6636801616108946,
      "learning_rate": 4.751613314853107e-06,
      "loss": 0.556,
      "step": 3289
    },
    {
      "epoch": 0.8816829693152888,
      "grad_norm": 0.7162759251249186,
      "learning_rate": 4.751460147613973e-06,
      "loss": 0.5546,
      "step": 3290
    },
    {
      "epoch": 0.8819509580597615,
      "grad_norm": 0.7763158581732347,
      "learning_rate": 4.75130693563445e-06,
      "loss": 0.5994,
      "step": 3291
    },
    {
      "epoch": 0.8822189468042342,
      "grad_norm": 0.6912561657772665,
      "learning_rate": 4.751153678917582e-06,
      "loss": 0.5865,
      "step": 3292
    },
    {
      "epoch": 0.8824869355487069,
      "grad_norm": 0.6938567850527845,
      "learning_rate": 4.751000377466414e-06,
      "loss": 0.5697,
      "step": 3293
    },
    {
      "epoch": 0.8827549242931797,
      "grad_norm": 0.6742947424341534,
      "learning_rate": 4.750847031283995e-06,
      "loss": 0.5814,
      "step": 3294
    },
    {
      "epoch": 0.8830229130376525,
      "grad_norm": 0.653234058647415,
      "learning_rate": 4.7506936403733695e-06,
      "loss": 0.5716,
      "step": 3295
    },
    {
      "epoch": 0.8832909017821251,
      "grad_norm": 0.6411016938387866,
      "learning_rate": 4.750540204737587e-06,
      "loss": 0.5799,
      "step": 3296
    },
    {
      "epoch": 0.8835588905265979,
      "grad_norm": 0.6743685617295563,
      "learning_rate": 4.750386724379696e-06,
      "loss": 0.596,
      "step": 3297
    },
    {
      "epoch": 0.8838268792710706,
      "grad_norm": 0.6894799490310402,
      "learning_rate": 4.750233199302748e-06,
      "loss": 0.5577,
      "step": 3298
    },
    {
      "epoch": 0.8840948680155434,
      "grad_norm": 0.6599603828683883,
      "learning_rate": 4.750079629509792e-06,
      "loss": 0.5954,
      "step": 3299
    },
    {
      "epoch": 0.884362856760016,
      "grad_norm": 0.6674426758066785,
      "learning_rate": 4.7499260150038795e-06,
      "loss": 0.5806,
      "step": 3300
    },
    {
      "epoch": 0.8846308455044888,
      "grad_norm": 0.6563495489630458,
      "learning_rate": 4.749772355788065e-06,
      "loss": 0.5986,
      "step": 3301
    },
    {
      "epoch": 0.8848988342489615,
      "grad_norm": 0.6522528280976891,
      "learning_rate": 4.749618651865399e-06,
      "loss": 0.5795,
      "step": 3302
    },
    {
      "epoch": 0.8851668229934343,
      "grad_norm": 0.6730635369955903,
      "learning_rate": 4.749464903238938e-06,
      "loss": 0.5732,
      "step": 3303
    },
    {
      "epoch": 0.8854348117379071,
      "grad_norm": 0.6701741012532996,
      "learning_rate": 4.749311109911739e-06,
      "loss": 0.5944,
      "step": 3304
    },
    {
      "epoch": 0.8857028004823797,
      "grad_norm": 0.6393251904014069,
      "learning_rate": 4.7491572718868536e-06,
      "loss": 0.6034,
      "step": 3305
    },
    {
      "epoch": 0.8859707892268525,
      "grad_norm": 0.6533985180990393,
      "learning_rate": 4.7490033891673425e-06,
      "loss": 0.5711,
      "step": 3306
    },
    {
      "epoch": 0.8862387779713252,
      "grad_norm": 0.6551238221841935,
      "learning_rate": 4.748849461756262e-06,
      "loss": 0.6029,
      "step": 3307
    },
    {
      "epoch": 0.886506766715798,
      "grad_norm": 0.6618290074594522,
      "learning_rate": 4.748695489656671e-06,
      "loss": 0.6475,
      "step": 3308
    },
    {
      "epoch": 0.8867747554602706,
      "grad_norm": 0.6476929081087859,
      "learning_rate": 4.748541472871629e-06,
      "loss": 0.5834,
      "step": 3309
    },
    {
      "epoch": 0.8870427442047434,
      "grad_norm": 0.638359503352522,
      "learning_rate": 4.7483874114041985e-06,
      "loss": 0.5869,
      "step": 3310
    },
    {
      "epoch": 0.8873107329492161,
      "grad_norm": 0.6730449033904835,
      "learning_rate": 4.7482333052574384e-06,
      "loss": 0.6159,
      "step": 3311
    },
    {
      "epoch": 0.8875787216936889,
      "grad_norm": 0.6659256290783242,
      "learning_rate": 4.748079154434413e-06,
      "loss": 0.549,
      "step": 3312
    },
    {
      "epoch": 0.8878467104381615,
      "grad_norm": 0.6846736486927558,
      "learning_rate": 4.747924958938184e-06,
      "loss": 0.5869,
      "step": 3313
    },
    {
      "epoch": 0.8881146991826343,
      "grad_norm": 0.673170728953228,
      "learning_rate": 4.747770718771817e-06,
      "loss": 0.6062,
      "step": 3314
    },
    {
      "epoch": 0.8883826879271071,
      "grad_norm": 0.6558515913987606,
      "learning_rate": 4.747616433938375e-06,
      "loss": 0.5747,
      "step": 3315
    },
    {
      "epoch": 0.8886506766715798,
      "grad_norm": 0.6581079054537934,
      "learning_rate": 4.747462104440926e-06,
      "loss": 0.5898,
      "step": 3316
    },
    {
      "epoch": 0.8889186654160526,
      "grad_norm": 0.662477943890675,
      "learning_rate": 4.747307730282536e-06,
      "loss": 0.5781,
      "step": 3317
    },
    {
      "epoch": 0.8891866541605252,
      "grad_norm": 0.6397880865071216,
      "learning_rate": 4.747153311466272e-06,
      "loss": 0.555,
      "step": 3318
    },
    {
      "epoch": 0.889454642904998,
      "grad_norm": 0.6686927367500722,
      "learning_rate": 4.746998847995204e-06,
      "loss": 0.5836,
      "step": 3319
    },
    {
      "epoch": 0.8897226316494707,
      "grad_norm": 0.6682528507004605,
      "learning_rate": 4.7468443398724e-06,
      "loss": 0.6153,
      "step": 3320
    },
    {
      "epoch": 0.8899906203939435,
      "grad_norm": 0.6574187783938742,
      "learning_rate": 4.746689787100931e-06,
      "loss": 0.6033,
      "step": 3321
    },
    {
      "epoch": 0.8902586091384161,
      "grad_norm": 0.6544936948866459,
      "learning_rate": 4.746535189683869e-06,
      "loss": 0.5727,
      "step": 3322
    },
    {
      "epoch": 0.8905265978828889,
      "grad_norm": 0.64461718224856,
      "learning_rate": 4.746380547624284e-06,
      "loss": 0.5606,
      "step": 3323
    },
    {
      "epoch": 0.8907945866273617,
      "grad_norm": 0.6537369218834893,
      "learning_rate": 4.746225860925251e-06,
      "loss": 0.5565,
      "step": 3324
    },
    {
      "epoch": 0.8910625753718344,
      "grad_norm": 0.6826630919413882,
      "learning_rate": 4.746071129589844e-06,
      "loss": 0.5722,
      "step": 3325
    },
    {
      "epoch": 0.8913305641163071,
      "grad_norm": 0.6638967808445948,
      "learning_rate": 4.745916353621136e-06,
      "loss": 0.6016,
      "step": 3326
    },
    {
      "epoch": 0.8915985528607798,
      "grad_norm": 0.6532526064717987,
      "learning_rate": 4.745761533022204e-06,
      "loss": 0.5559,
      "step": 3327
    },
    {
      "epoch": 0.8918665416052526,
      "grad_norm": 0.6474912021204854,
      "learning_rate": 4.745606667796124e-06,
      "loss": 0.5296,
      "step": 3328
    },
    {
      "epoch": 0.8921345303497253,
      "grad_norm": 0.6492318992977568,
      "learning_rate": 4.745451757945975e-06,
      "loss": 0.5579,
      "step": 3329
    },
    {
      "epoch": 0.8924025190941981,
      "grad_norm": 0.6684382920866259,
      "learning_rate": 4.745296803474832e-06,
      "loss": 0.6147,
      "step": 3330
    },
    {
      "epoch": 0.8926705078386707,
      "grad_norm": 0.6797513076172523,
      "learning_rate": 4.745141804385777e-06,
      "loss": 0.5675,
      "step": 3331
    },
    {
      "epoch": 0.8929384965831435,
      "grad_norm": 0.6538866166015569,
      "learning_rate": 4.74498676068189e-06,
      "loss": 0.5788,
      "step": 3332
    },
    {
      "epoch": 0.8932064853276163,
      "grad_norm": 0.6761650107425454,
      "learning_rate": 4.74483167236625e-06,
      "loss": 0.6034,
      "step": 3333
    },
    {
      "epoch": 0.893474474072089,
      "grad_norm": 0.6577622394425608,
      "learning_rate": 4.744676539441941e-06,
      "loss": 0.5655,
      "step": 3334
    },
    {
      "epoch": 0.8937424628165617,
      "grad_norm": 0.6557942037197131,
      "learning_rate": 4.744521361912045e-06,
      "loss": 0.5817,
      "step": 3335
    },
    {
      "epoch": 0.8940104515610344,
      "grad_norm": 0.6601391146225761,
      "learning_rate": 4.7443661397796455e-06,
      "loss": 0.5648,
      "step": 3336
    },
    {
      "epoch": 0.8942784403055072,
      "grad_norm": 0.6435160855979146,
      "learning_rate": 4.744210873047826e-06,
      "loss": 0.5655,
      "step": 3337
    },
    {
      "epoch": 0.8945464290499799,
      "grad_norm": 0.6498297077579243,
      "learning_rate": 4.744055561719674e-06,
      "loss": 0.5768,
      "step": 3338
    },
    {
      "epoch": 0.8948144177944526,
      "grad_norm": 0.6597075642112427,
      "learning_rate": 4.743900205798275e-06,
      "loss": 0.6005,
      "step": 3339
    },
    {
      "epoch": 0.8950824065389253,
      "grad_norm": 0.6702018625242018,
      "learning_rate": 4.743744805286715e-06,
      "loss": 0.5689,
      "step": 3340
    },
    {
      "epoch": 0.8953503952833981,
      "grad_norm": 0.666584593437903,
      "learning_rate": 4.743589360188085e-06,
      "loss": 0.6504,
      "step": 3341
    },
    {
      "epoch": 0.8956183840278709,
      "grad_norm": 0.6534301366203433,
      "learning_rate": 4.743433870505471e-06,
      "loss": 0.5675,
      "step": 3342
    },
    {
      "epoch": 0.8958863727723435,
      "grad_norm": 0.6519381601150565,
      "learning_rate": 4.743278336241964e-06,
      "loss": 0.5911,
      "step": 3343
    },
    {
      "epoch": 0.8961543615168163,
      "grad_norm": 0.6291951088570248,
      "learning_rate": 4.743122757400653e-06,
      "loss": 0.5836,
      "step": 3344
    },
    {
      "epoch": 0.896422350261289,
      "grad_norm": 0.6720994493726508,
      "learning_rate": 4.742967133984633e-06,
      "loss": 0.6032,
      "step": 3345
    },
    {
      "epoch": 0.8966903390057618,
      "grad_norm": 0.665355072146895,
      "learning_rate": 4.742811465996995e-06,
      "loss": 0.5702,
      "step": 3346
    },
    {
      "epoch": 0.8969583277502345,
      "grad_norm": 0.6620210542041498,
      "learning_rate": 4.742655753440831e-06,
      "loss": 0.5788,
      "step": 3347
    },
    {
      "epoch": 0.8972263164947072,
      "grad_norm": 0.6363959326009667,
      "learning_rate": 4.742499996319236e-06,
      "loss": 0.5483,
      "step": 3348
    },
    {
      "epoch": 0.89749430523918,
      "grad_norm": 0.6656056425936615,
      "learning_rate": 4.742344194635307e-06,
      "loss": 0.5805,
      "step": 3349
    },
    {
      "epoch": 0.8977622939836527,
      "grad_norm": 0.6389187332768715,
      "learning_rate": 4.742188348392138e-06,
      "loss": 0.5613,
      "step": 3350
    },
    {
      "epoch": 0.8980302827281255,
      "grad_norm": 0.6502971231603512,
      "learning_rate": 4.7420324575928265e-06,
      "loss": 0.5909,
      "step": 3351
    },
    {
      "epoch": 0.8982982714725981,
      "grad_norm": 0.6543674731196273,
      "learning_rate": 4.7418765222404706e-06,
      "loss": 0.5839,
      "step": 3352
    },
    {
      "epoch": 0.8985662602170709,
      "grad_norm": 0.655143909076731,
      "learning_rate": 4.741720542338169e-06,
      "loss": 0.547,
      "step": 3353
    },
    {
      "epoch": 0.8988342489615436,
      "grad_norm": 0.6523943330299161,
      "learning_rate": 4.74156451788902e-06,
      "loss": 0.6,
      "step": 3354
    },
    {
      "epoch": 0.8991022377060164,
      "grad_norm": 0.6450834403872607,
      "learning_rate": 4.741408448896126e-06,
      "loss": 0.558,
      "step": 3355
    },
    {
      "epoch": 0.899370226450489,
      "grad_norm": 0.6386076018037562,
      "learning_rate": 4.741252335362588e-06,
      "loss": 0.5632,
      "step": 3356
    },
    {
      "epoch": 0.8996382151949618,
      "grad_norm": 0.6317200567581133,
      "learning_rate": 4.741096177291507e-06,
      "loss": 0.5356,
      "step": 3357
    },
    {
      "epoch": 0.8999062039394345,
      "grad_norm": 0.6691918530157192,
      "learning_rate": 4.740939974685987e-06,
      "loss": 0.6136,
      "step": 3358
    },
    {
      "epoch": 0.9001741926839073,
      "grad_norm": 0.6462032170141451,
      "learning_rate": 4.740783727549132e-06,
      "loss": 0.5675,
      "step": 3359
    },
    {
      "epoch": 0.9004421814283801,
      "grad_norm": 0.6383780999270949,
      "learning_rate": 4.740627435884046e-06,
      "loss": 0.5852,
      "step": 3360
    },
    {
      "epoch": 0.9007101701728527,
      "grad_norm": 0.6359473203981045,
      "learning_rate": 4.740471099693836e-06,
      "loss": 0.5551,
      "step": 3361
    },
    {
      "epoch": 0.9009781589173255,
      "grad_norm": 0.6492867131609878,
      "learning_rate": 4.740314718981609e-06,
      "loss": 0.5816,
      "step": 3362
    },
    {
      "epoch": 0.9012461476617982,
      "grad_norm": 0.6496275761759395,
      "learning_rate": 4.7401582937504705e-06,
      "loss": 0.596,
      "step": 3363
    },
    {
      "epoch": 0.901514136406271,
      "grad_norm": 0.6615719623647749,
      "learning_rate": 4.7400018240035315e-06,
      "loss": 0.5745,
      "step": 3364
    },
    {
      "epoch": 0.9017821251507436,
      "grad_norm": 0.6636988276946751,
      "learning_rate": 4.7398453097439e-06,
      "loss": 0.5995,
      "step": 3365
    },
    {
      "epoch": 0.9020501138952164,
      "grad_norm": 0.6659416339050986,
      "learning_rate": 4.739688750974686e-06,
      "loss": 0.5897,
      "step": 3366
    },
    {
      "epoch": 0.9023181026396891,
      "grad_norm": 0.6380438833859048,
      "learning_rate": 4.739532147699e-06,
      "loss": 0.5335,
      "step": 3367
    },
    {
      "epoch": 0.9025860913841619,
      "grad_norm": 0.6658803561786877,
      "learning_rate": 4.739375499919956e-06,
      "loss": 0.5805,
      "step": 3368
    },
    {
      "epoch": 0.9028540801286346,
      "grad_norm": 0.6636718141010041,
      "learning_rate": 4.739218807640666e-06,
      "loss": 0.5594,
      "step": 3369
    },
    {
      "epoch": 0.9031220688731073,
      "grad_norm": 0.6460820970481297,
      "learning_rate": 4.739062070864241e-06,
      "loss": 0.5353,
      "step": 3370
    },
    {
      "epoch": 0.9033900576175801,
      "grad_norm": 0.6733403098387296,
      "learning_rate": 4.7389052895938e-06,
      "loss": 0.6046,
      "step": 3371
    },
    {
      "epoch": 0.9036580463620528,
      "grad_norm": 0.6569561295697207,
      "learning_rate": 4.7387484638324566e-06,
      "loss": 0.5835,
      "step": 3372
    },
    {
      "epoch": 0.9039260351065255,
      "grad_norm": 0.6497716188824916,
      "learning_rate": 4.738591593583327e-06,
      "loss": 0.5501,
      "step": 3373
    },
    {
      "epoch": 0.9041940238509982,
      "grad_norm": 0.629937798702561,
      "learning_rate": 4.738434678849528e-06,
      "loss": 0.5729,
      "step": 3374
    },
    {
      "epoch": 0.904462012595471,
      "grad_norm": 0.6362707504495142,
      "learning_rate": 4.73827771963418e-06,
      "loss": 0.5798,
      "step": 3375
    },
    {
      "epoch": 0.9047300013399437,
      "grad_norm": 0.6414020479546952,
      "learning_rate": 4.738120715940398e-06,
      "loss": 0.5689,
      "step": 3376
    },
    {
      "epoch": 0.9049979900844165,
      "grad_norm": 0.6337174429876339,
      "learning_rate": 4.737963667771307e-06,
      "loss": 0.5802,
      "step": 3377
    },
    {
      "epoch": 0.9052659788288892,
      "grad_norm": 0.6490964259867523,
      "learning_rate": 4.737806575130024e-06,
      "loss": 0.6072,
      "step": 3378
    },
    {
      "epoch": 0.9055339675733619,
      "grad_norm": 0.6548193856640449,
      "learning_rate": 4.737649438019672e-06,
      "loss": 0.588,
      "step": 3379
    },
    {
      "epoch": 0.9058019563178347,
      "grad_norm": 0.6504004101548762,
      "learning_rate": 4.737492256443373e-06,
      "loss": 0.5968,
      "step": 3380
    },
    {
      "epoch": 0.9060699450623074,
      "grad_norm": 0.6356717397755209,
      "learning_rate": 4.737335030404252e-06,
      "loss": 0.5725,
      "step": 3381
    },
    {
      "epoch": 0.9063379338067801,
      "grad_norm": 0.6389443897812183,
      "learning_rate": 4.737177759905432e-06,
      "loss": 0.5802,
      "step": 3382
    },
    {
      "epoch": 0.9066059225512528,
      "grad_norm": 0.6773700655262276,
      "learning_rate": 4.737020444950039e-06,
      "loss": 0.5786,
      "step": 3383
    },
    {
      "epoch": 0.9068739112957256,
      "grad_norm": 0.6618887285025437,
      "learning_rate": 4.7368630855411975e-06,
      "loss": 0.5926,
      "step": 3384
    },
    {
      "epoch": 0.9071419000401983,
      "grad_norm": 0.6360310923428607,
      "learning_rate": 4.7367056816820365e-06,
      "loss": 0.5748,
      "step": 3385
    },
    {
      "epoch": 0.907409888784671,
      "grad_norm": 0.6448336957488253,
      "learning_rate": 4.736548233375683e-06,
      "loss": 0.5917,
      "step": 3386
    },
    {
      "epoch": 0.9076778775291438,
      "grad_norm": 0.620764997389957,
      "learning_rate": 4.736390740625265e-06,
      "loss": 0.5429,
      "step": 3387
    },
    {
      "epoch": 0.9079458662736165,
      "grad_norm": 0.6610191373450296,
      "learning_rate": 4.7362332034339146e-06,
      "loss": 0.5766,
      "step": 3388
    },
    {
      "epoch": 0.9082138550180893,
      "grad_norm": 0.6574591225410856,
      "learning_rate": 4.73607562180476e-06,
      "loss": 0.5668,
      "step": 3389
    },
    {
      "epoch": 0.908481843762562,
      "grad_norm": 0.6554581010796001,
      "learning_rate": 4.735917995740933e-06,
      "loss": 0.5843,
      "step": 3390
    },
    {
      "epoch": 0.9087498325070347,
      "grad_norm": 0.6642808088104696,
      "learning_rate": 4.735760325245566e-06,
      "loss": 0.5981,
      "step": 3391
    },
    {
      "epoch": 0.9090178212515074,
      "grad_norm": 0.6365027629886802,
      "learning_rate": 4.735602610321794e-06,
      "loss": 0.5558,
      "step": 3392
    },
    {
      "epoch": 0.9092858099959802,
      "grad_norm": 0.6539090982999031,
      "learning_rate": 4.735444850972748e-06,
      "loss": 0.5761,
      "step": 3393
    },
    {
      "epoch": 0.909553798740453,
      "grad_norm": 0.6489724596781357,
      "learning_rate": 4.7352870472015645e-06,
      "loss": 0.5724,
      "step": 3394
    },
    {
      "epoch": 0.9098217874849256,
      "grad_norm": 0.6314427951730932,
      "learning_rate": 4.735129199011379e-06,
      "loss": 0.5762,
      "step": 3395
    },
    {
      "epoch": 0.9100897762293984,
      "grad_norm": 0.642866012165145,
      "learning_rate": 4.73497130640533e-06,
      "loss": 0.5628,
      "step": 3396
    },
    {
      "epoch": 0.9103577649738711,
      "grad_norm": 0.6564760308998532,
      "learning_rate": 4.7348133693865525e-06,
      "loss": 0.5932,
      "step": 3397
    },
    {
      "epoch": 0.9106257537183439,
      "grad_norm": 0.6565341884999675,
      "learning_rate": 4.734655387958187e-06,
      "loss": 0.5948,
      "step": 3398
    },
    {
      "epoch": 0.9108937424628165,
      "grad_norm": 0.6418004989653503,
      "learning_rate": 4.734497362123371e-06,
      "loss": 0.5519,
      "step": 3399
    },
    {
      "epoch": 0.9111617312072893,
      "grad_norm": 0.6263741947309276,
      "learning_rate": 4.734339291885246e-06,
      "loss": 0.5538,
      "step": 3400
    },
    {
      "epoch": 0.911429719951762,
      "grad_norm": 0.658335361733373,
      "learning_rate": 4.734181177246953e-06,
      "loss": 0.5739,
      "step": 3401
    },
    {
      "epoch": 0.9116977086962348,
      "grad_norm": 0.6515699239316548,
      "learning_rate": 4.734023018211633e-06,
      "loss": 0.5946,
      "step": 3402
    },
    {
      "epoch": 0.9119656974407075,
      "grad_norm": 0.6493929561853011,
      "learning_rate": 4.733864814782431e-06,
      "loss": 0.5905,
      "step": 3403
    },
    {
      "epoch": 0.9122336861851802,
      "grad_norm": 0.6400824840673427,
      "learning_rate": 4.733706566962489e-06,
      "loss": 0.5318,
      "step": 3404
    },
    {
      "epoch": 0.912501674929653,
      "grad_norm": 0.6404371106645456,
      "learning_rate": 4.733548274754952e-06,
      "loss": 0.573,
      "step": 3405
    },
    {
      "epoch": 0.9127696636741257,
      "grad_norm": 0.6473991867412168,
      "learning_rate": 4.733389938162966e-06,
      "loss": 0.5861,
      "step": 3406
    },
    {
      "epoch": 0.9130376524185985,
      "grad_norm": 0.6538082667431149,
      "learning_rate": 4.733231557189677e-06,
      "loss": 0.595,
      "step": 3407
    },
    {
      "epoch": 0.9133056411630711,
      "grad_norm": 0.6660559629896472,
      "learning_rate": 4.733073131838232e-06,
      "loss": 0.6015,
      "step": 3408
    },
    {
      "epoch": 0.9135736299075439,
      "grad_norm": 0.6618456422758199,
      "learning_rate": 4.73291466211178e-06,
      "loss": 0.5843,
      "step": 3409
    },
    {
      "epoch": 0.9138416186520166,
      "grad_norm": 0.6164503014067026,
      "learning_rate": 4.732756148013469e-06,
      "loss": 0.5865,
      "step": 3410
    },
    {
      "epoch": 0.9141096073964894,
      "grad_norm": 0.647033716952279,
      "learning_rate": 4.7325975895464504e-06,
      "loss": 0.5689,
      "step": 3411
    },
    {
      "epoch": 0.914377596140962,
      "grad_norm": 0.6496168146950265,
      "learning_rate": 4.732438986713874e-06,
      "loss": 0.5512,
      "step": 3412
    },
    {
      "epoch": 0.9146455848854348,
      "grad_norm": 0.6325320728758925,
      "learning_rate": 4.732280339518892e-06,
      "loss": 0.5824,
      "step": 3413
    },
    {
      "epoch": 0.9149135736299076,
      "grad_norm": 0.6810076454845677,
      "learning_rate": 4.732121647964656e-06,
      "loss": 0.5638,
      "step": 3414
    },
    {
      "epoch": 0.9151815623743803,
      "grad_norm": 0.6494153281789957,
      "learning_rate": 4.731962912054321e-06,
      "loss": 0.5882,
      "step": 3415
    },
    {
      "epoch": 0.915449551118853,
      "grad_norm": 0.6390360547113809,
      "learning_rate": 4.73180413179104e-06,
      "loss": 0.5796,
      "step": 3416
    },
    {
      "epoch": 0.9157175398633257,
      "grad_norm": 0.6446320887727869,
      "learning_rate": 4.731645307177969e-06,
      "loss": 0.5883,
      "step": 3417
    },
    {
      "epoch": 0.9159855286077985,
      "grad_norm": 0.6724712586294336,
      "learning_rate": 4.731486438218264e-06,
      "loss": 0.604,
      "step": 3418
    },
    {
      "epoch": 0.9162535173522712,
      "grad_norm": 0.6639660896319671,
      "learning_rate": 4.731327524915082e-06,
      "loss": 0.5773,
      "step": 3419
    },
    {
      "epoch": 0.916521506096744,
      "grad_norm": 0.6215693705137899,
      "learning_rate": 4.7311685672715805e-06,
      "loss": 0.5779,
      "step": 3420
    },
    {
      "epoch": 0.9167894948412166,
      "grad_norm": 0.6570517187515561,
      "learning_rate": 4.731009565290919e-06,
      "loss": 0.5855,
      "step": 3421
    },
    {
      "epoch": 0.9170574835856894,
      "grad_norm": 0.6691522988388598,
      "learning_rate": 4.7308505189762565e-06,
      "loss": 0.5648,
      "step": 3422
    },
    {
      "epoch": 0.9173254723301622,
      "grad_norm": 0.6596585013705025,
      "learning_rate": 4.730691428330754e-06,
      "loss": 0.6077,
      "step": 3423
    },
    {
      "epoch": 0.9175934610746349,
      "grad_norm": 0.6614029586723783,
      "learning_rate": 4.730532293357572e-06,
      "loss": 0.5793,
      "step": 3424
    },
    {
      "epoch": 0.9178614498191076,
      "grad_norm": 0.6582392936431979,
      "learning_rate": 4.730373114059874e-06,
      "loss": 0.6096,
      "step": 3425
    },
    {
      "epoch": 0.9181294385635803,
      "grad_norm": 0.6600790335692477,
      "learning_rate": 4.7302138904408224e-06,
      "loss": 0.5978,
      "step": 3426
    },
    {
      "epoch": 0.9183974273080531,
      "grad_norm": 0.6440354861684854,
      "learning_rate": 4.730054622503581e-06,
      "loss": 0.5944,
      "step": 3427
    },
    {
      "epoch": 0.9186654160525258,
      "grad_norm": 0.6590216290171245,
      "learning_rate": 4.729895310251317e-06,
      "loss": 0.5474,
      "step": 3428
    },
    {
      "epoch": 0.9189334047969985,
      "grad_norm": 0.6516903403385634,
      "learning_rate": 4.729735953687192e-06,
      "loss": 0.5843,
      "step": 3429
    },
    {
      "epoch": 0.9192013935414712,
      "grad_norm": 0.6536405681563027,
      "learning_rate": 4.7295765528143765e-06,
      "loss": 0.5529,
      "step": 3430
    },
    {
      "epoch": 0.919469382285944,
      "grad_norm": 0.6491715675014015,
      "learning_rate": 4.7294171076360365e-06,
      "loss": 0.5565,
      "step": 3431
    },
    {
      "epoch": 0.9197373710304168,
      "grad_norm": 0.6543024650930254,
      "learning_rate": 4.729257618155341e-06,
      "loss": 0.5927,
      "step": 3432
    },
    {
      "epoch": 0.9200053597748895,
      "grad_norm": 0.6682423062433424,
      "learning_rate": 4.729098084375457e-06,
      "loss": 0.5999,
      "step": 3433
    },
    {
      "epoch": 0.9202733485193622,
      "grad_norm": 0.6491558863560826,
      "learning_rate": 4.7289385062995586e-06,
      "loss": 0.5483,
      "step": 3434
    },
    {
      "epoch": 0.9205413372638349,
      "grad_norm": 0.653050884613408,
      "learning_rate": 4.728778883930815e-06,
      "loss": 0.5527,
      "step": 3435
    },
    {
      "epoch": 0.9208093260083077,
      "grad_norm": 0.6544952161676165,
      "learning_rate": 4.728619217272397e-06,
      "loss": 0.5896,
      "step": 3436
    },
    {
      "epoch": 0.9210773147527804,
      "grad_norm": 0.6716935705656689,
      "learning_rate": 4.728459506327478e-06,
      "loss": 0.5957,
      "step": 3437
    },
    {
      "epoch": 0.9213453034972531,
      "grad_norm": 0.6745879902873367,
      "learning_rate": 4.728299751099234e-06,
      "loss": 0.6249,
      "step": 3438
    },
    {
      "epoch": 0.9216132922417258,
      "grad_norm": 0.6672618855970893,
      "learning_rate": 4.728139951590837e-06,
      "loss": 0.5983,
      "step": 3439
    },
    {
      "epoch": 0.9218812809861986,
      "grad_norm": 0.6526559516531001,
      "learning_rate": 4.727980107805463e-06,
      "loss": 0.5928,
      "step": 3440
    },
    {
      "epoch": 0.9221492697306714,
      "grad_norm": 0.6505741823248506,
      "learning_rate": 4.72782021974629e-06,
      "loss": 0.5877,
      "step": 3441
    },
    {
      "epoch": 0.922417258475144,
      "grad_norm": 0.6554574931820256,
      "learning_rate": 4.727660287416493e-06,
      "loss": 0.5612,
      "step": 3442
    },
    {
      "epoch": 0.9226852472196168,
      "grad_norm": 0.6460824076476511,
      "learning_rate": 4.727500310819252e-06,
      "loss": 0.5817,
      "step": 3443
    },
    {
      "epoch": 0.9229532359640895,
      "grad_norm": 0.6417735418322913,
      "learning_rate": 4.727340289957744e-06,
      "loss": 0.5761,
      "step": 3444
    },
    {
      "epoch": 0.9232212247085623,
      "grad_norm": 0.6478320726238821,
      "learning_rate": 4.7271802248351504e-06,
      "loss": 0.5901,
      "step": 3445
    },
    {
      "epoch": 0.9234892134530349,
      "grad_norm": 0.6474122128896038,
      "learning_rate": 4.727020115454652e-06,
      "loss": 0.594,
      "step": 3446
    },
    {
      "epoch": 0.9237572021975077,
      "grad_norm": 0.6525736767439271,
      "learning_rate": 4.72685996181943e-06,
      "loss": 0.5776,
      "step": 3447
    },
    {
      "epoch": 0.9240251909419804,
      "grad_norm": 0.6461985751960613,
      "learning_rate": 4.726699763932666e-06,
      "loss": 0.5951,
      "step": 3448
    },
    {
      "epoch": 0.9242931796864532,
      "grad_norm": 0.6422819011858776,
      "learning_rate": 4.726539521797546e-06,
      "loss": 0.6046,
      "step": 3449
    },
    {
      "epoch": 0.924561168430926,
      "grad_norm": 0.6593790800130451,
      "learning_rate": 4.726379235417251e-06,
      "loss": 0.5701,
      "step": 3450
    },
    {
      "epoch": 0.9248291571753986,
      "grad_norm": 0.6452134331584533,
      "learning_rate": 4.726218904794969e-06,
      "loss": 0.5871,
      "step": 3451
    },
    {
      "epoch": 0.9250971459198714,
      "grad_norm": 0.6437121323659916,
      "learning_rate": 4.7260585299338845e-06,
      "loss": 0.5828,
      "step": 3452
    },
    {
      "epoch": 0.9253651346643441,
      "grad_norm": 0.6837249517724922,
      "learning_rate": 4.725898110837185e-06,
      "loss": 0.5424,
      "step": 3453
    },
    {
      "epoch": 0.9256331234088169,
      "grad_norm": 0.6069528183770303,
      "learning_rate": 4.725737647508058e-06,
      "loss": 0.5498,
      "step": 3454
    },
    {
      "epoch": 0.9259011121532895,
      "grad_norm": 0.6615830503209451,
      "learning_rate": 4.725577139949693e-06,
      "loss": 0.5581,
      "step": 3455
    },
    {
      "epoch": 0.9261691008977623,
      "grad_norm": 0.6518773385691436,
      "learning_rate": 4.725416588165278e-06,
      "loss": 0.5548,
      "step": 3456
    },
    {
      "epoch": 0.926437089642235,
      "grad_norm": 0.6414461875138879,
      "learning_rate": 4.725255992158005e-06,
      "loss": 0.5883,
      "step": 3457
    },
    {
      "epoch": 0.9267050783867078,
      "grad_norm": 0.6783130299233177,
      "learning_rate": 4.725095351931064e-06,
      "loss": 0.6211,
      "step": 3458
    },
    {
      "epoch": 0.9269730671311804,
      "grad_norm": 0.6855578219941352,
      "learning_rate": 4.724934667487647e-06,
      "loss": 0.5917,
      "step": 3459
    },
    {
      "epoch": 0.9272410558756532,
      "grad_norm": 0.672019921958756,
      "learning_rate": 4.72477393883095e-06,
      "loss": 0.5871,
      "step": 3460
    },
    {
      "epoch": 0.927509044620126,
      "grad_norm": 0.6640884404219292,
      "learning_rate": 4.724613165964163e-06,
      "loss": 0.5948,
      "step": 3461
    },
    {
      "epoch": 0.9277770333645987,
      "grad_norm": 0.6720767307114507,
      "learning_rate": 4.724452348890484e-06,
      "loss": 0.5831,
      "step": 3462
    },
    {
      "epoch": 0.9280450221090715,
      "grad_norm": 0.6687956873848233,
      "learning_rate": 4.724291487613106e-06,
      "loss": 0.612,
      "step": 3463
    },
    {
      "epoch": 0.9283130108535441,
      "grad_norm": 0.676585861501015,
      "learning_rate": 4.7241305821352275e-06,
      "loss": 0.5975,
      "step": 3464
    },
    {
      "epoch": 0.9285809995980169,
      "grad_norm": 0.6591735872790031,
      "learning_rate": 4.7239696324600456e-06,
      "loss": 0.5682,
      "step": 3465
    },
    {
      "epoch": 0.9288489883424896,
      "grad_norm": 0.6644835957355981,
      "learning_rate": 4.723808638590759e-06,
      "loss": 0.5832,
      "step": 3466
    },
    {
      "epoch": 0.9291169770869624,
      "grad_norm": 0.6501376649486361,
      "learning_rate": 4.723647600530567e-06,
      "loss": 0.5636,
      "step": 3467
    },
    {
      "epoch": 0.929384965831435,
      "grad_norm": 0.8270762238045251,
      "learning_rate": 4.723486518282667e-06,
      "loss": 0.5724,
      "step": 3468
    },
    {
      "epoch": 0.9296529545759078,
      "grad_norm": 0.6971939773345143,
      "learning_rate": 4.723325391850264e-06,
      "loss": 0.5611,
      "step": 3469
    },
    {
      "epoch": 0.9299209433203806,
      "grad_norm": 0.6627584598052714,
      "learning_rate": 4.723164221236557e-06,
      "loss": 0.5676,
      "step": 3470
    },
    {
      "epoch": 0.9301889320648533,
      "grad_norm": 0.6310555848539339,
      "learning_rate": 4.72300300644475e-06,
      "loss": 0.5682,
      "step": 3471
    },
    {
      "epoch": 0.930456920809326,
      "grad_norm": 0.6424925486927668,
      "learning_rate": 4.722841747478047e-06,
      "loss": 0.5682,
      "step": 3472
    },
    {
      "epoch": 0.9307249095537987,
      "grad_norm": 0.6547270098197885,
      "learning_rate": 4.72268044433965e-06,
      "loss": 0.5887,
      "step": 3473
    },
    {
      "epoch": 0.9309928982982715,
      "grad_norm": 0.6990412671552086,
      "learning_rate": 4.722519097032768e-06,
      "loss": 0.6008,
      "step": 3474
    },
    {
      "epoch": 0.9312608870427442,
      "grad_norm": 0.650104009361929,
      "learning_rate": 4.722357705560605e-06,
      "loss": 0.5552,
      "step": 3475
    },
    {
      "epoch": 0.931528875787217,
      "grad_norm": 0.673937297883845,
      "learning_rate": 4.722196269926368e-06,
      "loss": 0.5832,
      "step": 3476
    },
    {
      "epoch": 0.9317968645316896,
      "grad_norm": 0.6521802856330762,
      "learning_rate": 4.7220347901332654e-06,
      "loss": 0.5748,
      "step": 3477
    },
    {
      "epoch": 0.9320648532761624,
      "grad_norm": 0.669611913321045,
      "learning_rate": 4.7218732661845065e-06,
      "loss": 0.5773,
      "step": 3478
    },
    {
      "epoch": 0.9323328420206352,
      "grad_norm": 0.6480943657745494,
      "learning_rate": 4.721711698083301e-06,
      "loss": 0.5912,
      "step": 3479
    },
    {
      "epoch": 0.9326008307651079,
      "grad_norm": 0.6667028181713447,
      "learning_rate": 4.72155008583286e-06,
      "loss": 0.5993,
      "step": 3480
    },
    {
      "epoch": 0.9328688195095806,
      "grad_norm": 0.6493631095667144,
      "learning_rate": 4.721388429436394e-06,
      "loss": 0.5991,
      "step": 3481
    },
    {
      "epoch": 0.9331368082540533,
      "grad_norm": 0.6439727672721103,
      "learning_rate": 4.721226728897116e-06,
      "loss": 0.5593,
      "step": 3482
    },
    {
      "epoch": 0.9334047969985261,
      "grad_norm": 0.6578631897710628,
      "learning_rate": 4.721064984218238e-06,
      "loss": 0.5642,
      "step": 3483
    },
    {
      "epoch": 0.9336727857429988,
      "grad_norm": 0.6649416094958109,
      "learning_rate": 4.720903195402976e-06,
      "loss": 0.5685,
      "step": 3484
    },
    {
      "epoch": 0.9339407744874715,
      "grad_norm": 0.6430837702973909,
      "learning_rate": 4.720741362454544e-06,
      "loss": 0.5938,
      "step": 3485
    },
    {
      "epoch": 0.9342087632319442,
      "grad_norm": 0.6362372330564003,
      "learning_rate": 4.720579485376157e-06,
      "loss": 0.5673,
      "step": 3486
    },
    {
      "epoch": 0.934476751976417,
      "grad_norm": 0.6399584158875357,
      "learning_rate": 4.720417564171034e-06,
      "loss": 0.5836,
      "step": 3487
    },
    {
      "epoch": 0.9347447407208898,
      "grad_norm": 0.6402502152362434,
      "learning_rate": 4.720255598842392e-06,
      "loss": 0.5863,
      "step": 3488
    },
    {
      "epoch": 0.9350127294653624,
      "grad_norm": 0.6186967553982875,
      "learning_rate": 4.720093589393448e-06,
      "loss": 0.5897,
      "step": 3489
    },
    {
      "epoch": 0.9352807182098352,
      "grad_norm": 0.6401605740086946,
      "learning_rate": 4.7199315358274225e-06,
      "loss": 0.5897,
      "step": 3490
    },
    {
      "epoch": 0.9355487069543079,
      "grad_norm": 0.6437929878405332,
      "learning_rate": 4.719769438147537e-06,
      "loss": 0.618,
      "step": 3491
    },
    {
      "epoch": 0.9358166956987807,
      "grad_norm": 0.6957734487273594,
      "learning_rate": 4.71960729635701e-06,
      "loss": 0.5705,
      "step": 3492
    },
    {
      "epoch": 0.9360846844432534,
      "grad_norm": 0.636159911861215,
      "learning_rate": 4.719445110459065e-06,
      "loss": 0.5906,
      "step": 3493
    },
    {
      "epoch": 0.9363526731877261,
      "grad_norm": 0.6349092973402326,
      "learning_rate": 4.719282880456926e-06,
      "loss": 0.5491,
      "step": 3494
    },
    {
      "epoch": 0.9366206619321988,
      "grad_norm": 0.6697843225509369,
      "learning_rate": 4.719120606353814e-06,
      "loss": 0.6133,
      "step": 3495
    },
    {
      "epoch": 0.9368886506766716,
      "grad_norm": 0.6494001521760704,
      "learning_rate": 4.718958288152957e-06,
      "loss": 0.5843,
      "step": 3496
    },
    {
      "epoch": 0.9371566394211444,
      "grad_norm": 0.6682447023594194,
      "learning_rate": 4.718795925857578e-06,
      "loss": 0.6214,
      "step": 3497
    },
    {
      "epoch": 0.937424628165617,
      "grad_norm": 0.6652942606466947,
      "learning_rate": 4.718633519470904e-06,
      "loss": 0.5701,
      "step": 3498
    },
    {
      "epoch": 0.9376926169100898,
      "grad_norm": 0.6399134443163886,
      "learning_rate": 4.718471068996163e-06,
      "loss": 0.6111,
      "step": 3499
    },
    {
      "epoch": 0.9379606056545625,
      "grad_norm": 0.6290028786924972,
      "learning_rate": 4.718308574436582e-06,
      "loss": 0.551,
      "step": 3500
    },
    {
      "epoch": 0.9382285943990353,
      "grad_norm": 0.6237869025466269,
      "learning_rate": 4.718146035795392e-06,
      "loss": 0.5476,
      "step": 3501
    },
    {
      "epoch": 0.9384965831435079,
      "grad_norm": 0.6411731512738869,
      "learning_rate": 4.717983453075821e-06,
      "loss": 0.589,
      "step": 3502
    },
    {
      "epoch": 0.9387645718879807,
      "grad_norm": 0.6491234463668287,
      "learning_rate": 4.7178208262811e-06,
      "loss": 0.5839,
      "step": 3503
    },
    {
      "epoch": 0.9390325606324534,
      "grad_norm": 0.6407121124953173,
      "learning_rate": 4.717658155414462e-06,
      "loss": 0.5703,
      "step": 3504
    },
    {
      "epoch": 0.9393005493769262,
      "grad_norm": 0.6322394064393162,
      "learning_rate": 4.717495440479138e-06,
      "loss": 0.5489,
      "step": 3505
    },
    {
      "epoch": 0.939568538121399,
      "grad_norm": 0.6328748586094857,
      "learning_rate": 4.717332681478363e-06,
      "loss": 0.5863,
      "step": 3506
    },
    {
      "epoch": 0.9398365268658716,
      "grad_norm": 0.6357168743998336,
      "learning_rate": 4.717169878415369e-06,
      "loss": 0.563,
      "step": 3507
    },
    {
      "epoch": 0.9401045156103444,
      "grad_norm": 0.6647994899362336,
      "learning_rate": 4.717007031293395e-06,
      "loss": 0.5753,
      "step": 3508
    },
    {
      "epoch": 0.9403725043548171,
      "grad_norm": 0.6413509812707774,
      "learning_rate": 4.7168441401156724e-06,
      "loss": 0.5612,
      "step": 3509
    },
    {
      "epoch": 0.9406404930992899,
      "grad_norm": 0.611284228907506,
      "learning_rate": 4.716681204885442e-06,
      "loss": 0.5408,
      "step": 3510
    },
    {
      "epoch": 0.9409084818437625,
      "grad_norm": 0.6446105266228809,
      "learning_rate": 4.716518225605939e-06,
      "loss": 0.6038,
      "step": 3511
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.6303019187319486,
      "learning_rate": 4.716355202280403e-06,
      "loss": 0.5578,
      "step": 3512
    },
    {
      "epoch": 0.941444459332708,
      "grad_norm": 0.6511067498346684,
      "learning_rate": 4.716192134912074e-06,
      "loss": 0.5977,
      "step": 3513
    },
    {
      "epoch": 0.9417124480771808,
      "grad_norm": 0.6561121660180187,
      "learning_rate": 4.716029023504193e-06,
      "loss": 0.6045,
      "step": 3514
    },
    {
      "epoch": 0.9419804368216534,
      "grad_norm": 0.6435404303809539,
      "learning_rate": 4.715865868059999e-06,
      "loss": 0.6019,
      "step": 3515
    },
    {
      "epoch": 0.9422484255661262,
      "grad_norm": 0.6543825573991033,
      "learning_rate": 4.715702668582735e-06,
      "loss": 0.5682,
      "step": 3516
    },
    {
      "epoch": 0.942516414310599,
      "grad_norm": 0.6651868792395932,
      "learning_rate": 4.715539425075646e-06,
      "loss": 0.5876,
      "step": 3517
    },
    {
      "epoch": 0.9427844030550717,
      "grad_norm": 0.6583808312292776,
      "learning_rate": 4.715376137541975e-06,
      "loss": 0.5713,
      "step": 3518
    },
    {
      "epoch": 0.9430523917995444,
      "grad_norm": 0.6362023692064257,
      "learning_rate": 4.7152128059849654e-06,
      "loss": 0.5946,
      "step": 3519
    },
    {
      "epoch": 0.9433203805440171,
      "grad_norm": 0.6609281565092965,
      "learning_rate": 4.715049430407864e-06,
      "loss": 0.5515,
      "step": 3520
    },
    {
      "epoch": 0.9435883692884899,
      "grad_norm": 0.6428624034624197,
      "learning_rate": 4.714886010813918e-06,
      "loss": 0.568,
      "step": 3521
    },
    {
      "epoch": 0.9438563580329626,
      "grad_norm": 0.649795627994689,
      "learning_rate": 4.714722547206372e-06,
      "loss": 0.5763,
      "step": 3522
    },
    {
      "epoch": 0.9441243467774354,
      "grad_norm": 0.6831910569418175,
      "learning_rate": 4.714559039588477e-06,
      "loss": 0.5771,
      "step": 3523
    },
    {
      "epoch": 0.944392335521908,
      "grad_norm": 0.6645600744488169,
      "learning_rate": 4.7143954879634826e-06,
      "loss": 0.5658,
      "step": 3524
    },
    {
      "epoch": 0.9446603242663808,
      "grad_norm": 0.631458072246413,
      "learning_rate": 4.714231892334636e-06,
      "loss": 0.5684,
      "step": 3525
    },
    {
      "epoch": 0.9449283130108536,
      "grad_norm": 0.6370145291673889,
      "learning_rate": 4.714068252705192e-06,
      "loss": 0.5837,
      "step": 3526
    },
    {
      "epoch": 0.9451963017553263,
      "grad_norm": 0.6364817989850866,
      "learning_rate": 4.713904569078399e-06,
      "loss": 0.5826,
      "step": 3527
    },
    {
      "epoch": 0.945464290499799,
      "grad_norm": 0.6181602602988938,
      "learning_rate": 4.713740841457511e-06,
      "loss": 0.5597,
      "step": 3528
    },
    {
      "epoch": 0.9457322792442717,
      "grad_norm": 0.6619116202460178,
      "learning_rate": 4.713577069845782e-06,
      "loss": 0.57,
      "step": 3529
    },
    {
      "epoch": 0.9460002679887445,
      "grad_norm": 0.6347534458363195,
      "learning_rate": 4.713413254246465e-06,
      "loss": 0.5949,
      "step": 3530
    },
    {
      "epoch": 0.9462682567332172,
      "grad_norm": 0.6460172102387315,
      "learning_rate": 4.713249394662816e-06,
      "loss": 0.552,
      "step": 3531
    },
    {
      "epoch": 0.9465362454776899,
      "grad_norm": 0.6468390265039428,
      "learning_rate": 4.713085491098093e-06,
      "loss": 0.5791,
      "step": 3532
    },
    {
      "epoch": 0.9468042342221626,
      "grad_norm": 0.654232724407092,
      "learning_rate": 4.7129215435555496e-06,
      "loss": 0.5736,
      "step": 3533
    },
    {
      "epoch": 0.9470722229666354,
      "grad_norm": 0.6598036812980641,
      "learning_rate": 4.712757552038447e-06,
      "loss": 0.6279,
      "step": 3534
    },
    {
      "epoch": 0.9473402117111082,
      "grad_norm": 0.6420875087135753,
      "learning_rate": 4.712593516550042e-06,
      "loss": 0.5823,
      "step": 3535
    },
    {
      "epoch": 0.9476082004555809,
      "grad_norm": 0.6567650672476831,
      "learning_rate": 4.712429437093595e-06,
      "loss": 0.559,
      "step": 3536
    },
    {
      "epoch": 0.9478761892000536,
      "grad_norm": 0.6955605355665977,
      "learning_rate": 4.712265313672366e-06,
      "loss": 0.5635,
      "step": 3537
    },
    {
      "epoch": 0.9481441779445263,
      "grad_norm": 0.6654555027727724,
      "learning_rate": 4.712101146289617e-06,
      "loss": 0.5752,
      "step": 3538
    },
    {
      "epoch": 0.9484121666889991,
      "grad_norm": 0.664261452442603,
      "learning_rate": 4.71193693494861e-06,
      "loss": 0.6088,
      "step": 3539
    },
    {
      "epoch": 0.9486801554334718,
      "grad_norm": 0.6560504350295401,
      "learning_rate": 4.7117726796526084e-06,
      "loss": 0.5763,
      "step": 3540
    },
    {
      "epoch": 0.9489481441779445,
      "grad_norm": 0.6562171625062716,
      "learning_rate": 4.711608380404876e-06,
      "loss": 0.5748,
      "step": 3541
    },
    {
      "epoch": 0.9492161329224172,
      "grad_norm": 0.6357455157811097,
      "learning_rate": 4.7114440372086775e-06,
      "loss": 0.5621,
      "step": 3542
    },
    {
      "epoch": 0.94948412166689,
      "grad_norm": 0.6567251476308007,
      "learning_rate": 4.711279650067279e-06,
      "loss": 0.5662,
      "step": 3543
    },
    {
      "epoch": 0.9497521104113628,
      "grad_norm": 0.6423733838463973,
      "learning_rate": 4.711115218983947e-06,
      "loss": 0.5652,
      "step": 3544
    },
    {
      "epoch": 0.9500200991558354,
      "grad_norm": 0.6964129441273931,
      "learning_rate": 4.71095074396195e-06,
      "loss": 0.5598,
      "step": 3545
    },
    {
      "epoch": 0.9502880879003082,
      "grad_norm": 0.6605253888457552,
      "learning_rate": 4.710786225004555e-06,
      "loss": 0.5761,
      "step": 3546
    },
    {
      "epoch": 0.9505560766447809,
      "grad_norm": 0.6618967188772595,
      "learning_rate": 4.710621662115031e-06,
      "loss": 0.5762,
      "step": 3547
    },
    {
      "epoch": 0.9508240653892537,
      "grad_norm": 0.6695254084423258,
      "learning_rate": 4.710457055296649e-06,
      "loss": 0.591,
      "step": 3548
    },
    {
      "epoch": 0.9510920541337264,
      "grad_norm": 0.6648529925349119,
      "learning_rate": 4.710292404552681e-06,
      "loss": 0.5911,
      "step": 3549
    },
    {
      "epoch": 0.9513600428781991,
      "grad_norm": 0.6440632421355821,
      "learning_rate": 4.710127709886398e-06,
      "loss": 0.5768,
      "step": 3550
    },
    {
      "epoch": 0.9516280316226718,
      "grad_norm": 0.6536405699792122,
      "learning_rate": 4.709962971301071e-06,
      "loss": 0.6014,
      "step": 3551
    },
    {
      "epoch": 0.9518960203671446,
      "grad_norm": 0.6531706192950796,
      "learning_rate": 4.709798188799977e-06,
      "loss": 0.6148,
      "step": 3552
    },
    {
      "epoch": 0.9521640091116174,
      "grad_norm": 0.7132203981660442,
      "learning_rate": 4.709633362386388e-06,
      "loss": 0.5752,
      "step": 3553
    },
    {
      "epoch": 0.95243199785609,
      "grad_norm": 0.6547113973861152,
      "learning_rate": 4.70946849206358e-06,
      "loss": 0.5907,
      "step": 3554
    },
    {
      "epoch": 0.9526999866005628,
      "grad_norm": 0.7145941726750553,
      "learning_rate": 4.70930357783483e-06,
      "loss": 0.5756,
      "step": 3555
    },
    {
      "epoch": 0.9529679753450355,
      "grad_norm": 0.6568202207496506,
      "learning_rate": 4.709138619703414e-06,
      "loss": 0.6064,
      "step": 3556
    },
    {
      "epoch": 0.9532359640895083,
      "grad_norm": 0.665369394774644,
      "learning_rate": 4.708973617672611e-06,
      "loss": 0.5778,
      "step": 3557
    },
    {
      "epoch": 0.9535039528339809,
      "grad_norm": 0.6520603461022358,
      "learning_rate": 4.708808571745699e-06,
      "loss": 0.5938,
      "step": 3558
    },
    {
      "epoch": 0.9537719415784537,
      "grad_norm": 0.6787392046781323,
      "learning_rate": 4.708643481925958e-06,
      "loss": 0.5756,
      "step": 3559
    },
    {
      "epoch": 0.9540399303229264,
      "grad_norm": 0.7105812558633876,
      "learning_rate": 4.70847834821667e-06,
      "loss": 0.5512,
      "step": 3560
    },
    {
      "epoch": 0.9543079190673992,
      "grad_norm": 0.6497894826349736,
      "learning_rate": 4.7083131706211144e-06,
      "loss": 0.6206,
      "step": 3561
    },
    {
      "epoch": 0.9545759078118718,
      "grad_norm": 0.647377208803838,
      "learning_rate": 4.708147949142574e-06,
      "loss": 0.6192,
      "step": 3562
    },
    {
      "epoch": 0.9548438965563446,
      "grad_norm": 0.647721839226532,
      "learning_rate": 4.7079826837843335e-06,
      "loss": 0.5681,
      "step": 3563
    },
    {
      "epoch": 0.9551118853008174,
      "grad_norm": 0.6513079756752616,
      "learning_rate": 4.707817374549676e-06,
      "loss": 0.5372,
      "step": 3564
    },
    {
      "epoch": 0.9553798740452901,
      "grad_norm": 0.6602330012647872,
      "learning_rate": 4.707652021441885e-06,
      "loss": 0.5523,
      "step": 3565
    },
    {
      "epoch": 0.9556478627897629,
      "grad_norm": 0.6385404914293289,
      "learning_rate": 4.7074866244642494e-06,
      "loss": 0.5539,
      "step": 3566
    },
    {
      "epoch": 0.9559158515342355,
      "grad_norm": 0.6664873775775892,
      "learning_rate": 4.707321183620054e-06,
      "loss": 0.5562,
      "step": 3567
    },
    {
      "epoch": 0.9561838402787083,
      "grad_norm": 0.6591473386717521,
      "learning_rate": 4.707155698912587e-06,
      "loss": 0.5866,
      "step": 3568
    },
    {
      "epoch": 0.956451829023181,
      "grad_norm": 0.6982630396731848,
      "learning_rate": 4.706990170345136e-06,
      "loss": 0.5614,
      "step": 3569
    },
    {
      "epoch": 0.9567198177676538,
      "grad_norm": 0.6595050437505207,
      "learning_rate": 4.7068245979209905e-06,
      "loss": 0.5853,
      "step": 3570
    },
    {
      "epoch": 0.9569878065121264,
      "grad_norm": 0.6536226063011584,
      "learning_rate": 4.706658981643442e-06,
      "loss": 0.5951,
      "step": 3571
    },
    {
      "epoch": 0.9572557952565992,
      "grad_norm": 0.646236718357016,
      "learning_rate": 4.706493321515781e-06,
      "loss": 0.5787,
      "step": 3572
    },
    {
      "epoch": 0.957523784001072,
      "grad_norm": 0.6726097332650802,
      "learning_rate": 4.706327617541299e-06,
      "loss": 0.6173,
      "step": 3573
    },
    {
      "epoch": 0.9577917727455447,
      "grad_norm": 0.6556417772532419,
      "learning_rate": 4.706161869723288e-06,
      "loss": 0.5715,
      "step": 3574
    },
    {
      "epoch": 0.9580597614900174,
      "grad_norm": 0.6508752533588633,
      "learning_rate": 4.7059960780650435e-06,
      "loss": 0.5792,
      "step": 3575
    },
    {
      "epoch": 0.9583277502344901,
      "grad_norm": 0.6579566799825168,
      "learning_rate": 4.705830242569859e-06,
      "loss": 0.57,
      "step": 3576
    },
    {
      "epoch": 0.9585957389789629,
      "grad_norm": 0.6603562317824893,
      "learning_rate": 4.70566436324103e-06,
      "loss": 0.5709,
      "step": 3577
    },
    {
      "epoch": 0.9588637277234356,
      "grad_norm": 0.6486659221194284,
      "learning_rate": 4.705498440081853e-06,
      "loss": 0.5944,
      "step": 3578
    },
    {
      "epoch": 0.9591317164679084,
      "grad_norm": 0.6890795891625732,
      "learning_rate": 4.705332473095627e-06,
      "loss": 0.6041,
      "step": 3579
    },
    {
      "epoch": 0.959399705212381,
      "grad_norm": 0.6324690482694049,
      "learning_rate": 4.705166462285647e-06,
      "loss": 0.5645,
      "step": 3580
    },
    {
      "epoch": 0.9596676939568538,
      "grad_norm": 0.6288371868635207,
      "learning_rate": 4.705000407655212e-06,
      "loss": 0.569,
      "step": 3581
    },
    {
      "epoch": 0.9599356827013266,
      "grad_norm": 0.6955432805792463,
      "learning_rate": 4.704834309207624e-06,
      "loss": 0.5717,
      "step": 3582
    },
    {
      "epoch": 0.9602036714457993,
      "grad_norm": 0.6962942635099264,
      "learning_rate": 4.704668166946183e-06,
      "loss": 0.5948,
      "step": 3583
    },
    {
      "epoch": 0.960471660190272,
      "grad_norm": 0.6402762898168255,
      "learning_rate": 4.70450198087419e-06,
      "loss": 0.5445,
      "step": 3584
    },
    {
      "epoch": 0.9607396489347447,
      "grad_norm": 0.6570821026668283,
      "learning_rate": 4.704335750994947e-06,
      "loss": 0.5298,
      "step": 3585
    },
    {
      "epoch": 0.9610076376792175,
      "grad_norm": 0.6782944001042529,
      "learning_rate": 4.704169477311758e-06,
      "loss": 0.5881,
      "step": 3586
    },
    {
      "epoch": 0.9612756264236902,
      "grad_norm": 0.6568325574663612,
      "learning_rate": 4.704003159827928e-06,
      "loss": 0.5728,
      "step": 3587
    },
    {
      "epoch": 0.9615436151681629,
      "grad_norm": 0.6437134423798472,
      "learning_rate": 4.703836798546759e-06,
      "loss": 0.567,
      "step": 3588
    },
    {
      "epoch": 0.9618116039126356,
      "grad_norm": 0.6487887625562613,
      "learning_rate": 4.70367039347156e-06,
      "loss": 0.5876,
      "step": 3589
    },
    {
      "epoch": 0.9620795926571084,
      "grad_norm": 0.6764693507394768,
      "learning_rate": 4.703503944605637e-06,
      "loss": 0.6039,
      "step": 3590
    },
    {
      "epoch": 0.9623475814015812,
      "grad_norm": 0.6717183308923813,
      "learning_rate": 4.703337451952297e-06,
      "loss": 0.5761,
      "step": 3591
    },
    {
      "epoch": 0.9626155701460538,
      "grad_norm": 0.6698796892083738,
      "learning_rate": 4.703170915514848e-06,
      "loss": 0.5946,
      "step": 3592
    },
    {
      "epoch": 0.9628835588905266,
      "grad_norm": 0.6559368170949526,
      "learning_rate": 4.703004335296601e-06,
      "loss": 0.5927,
      "step": 3593
    },
    {
      "epoch": 0.9631515476349993,
      "grad_norm": 0.6576608139199022,
      "learning_rate": 4.702837711300865e-06,
      "loss": 0.5975,
      "step": 3594
    },
    {
      "epoch": 0.9634195363794721,
      "grad_norm": 0.6324637342137288,
      "learning_rate": 4.7026710435309516e-06,
      "loss": 0.5889,
      "step": 3595
    },
    {
      "epoch": 0.9636875251239448,
      "grad_norm": 0.6276224215206759,
      "learning_rate": 4.702504331990172e-06,
      "loss": 0.5495,
      "step": 3596
    },
    {
      "epoch": 0.9639555138684175,
      "grad_norm": 0.6493761955286033,
      "learning_rate": 4.70233757668184e-06,
      "loss": 0.5885,
      "step": 3597
    },
    {
      "epoch": 0.9642235026128902,
      "grad_norm": 0.6594041138516836,
      "learning_rate": 4.70217077760927e-06,
      "loss": 0.604,
      "step": 3598
    },
    {
      "epoch": 0.964491491357363,
      "grad_norm": 0.6214305201606131,
      "learning_rate": 4.702003934775775e-06,
      "loss": 0.5582,
      "step": 3599
    },
    {
      "epoch": 0.9647594801018358,
      "grad_norm": 0.6296585415620911,
      "learning_rate": 4.70183704818467e-06,
      "loss": 0.5398,
      "step": 3600
    },
    {
      "epoch": 0.9650274688463084,
      "grad_norm": 0.6499001240401445,
      "learning_rate": 4.7016701178392734e-06,
      "loss": 0.5776,
      "step": 3601
    },
    {
      "epoch": 0.9652954575907812,
      "grad_norm": 0.6281359499667094,
      "learning_rate": 4.701503143742901e-06,
      "loss": 0.5479,
      "step": 3602
    },
    {
      "epoch": 0.9655634463352539,
      "grad_norm": 0.6638634339579238,
      "learning_rate": 4.701336125898871e-06,
      "loss": 0.5853,
      "step": 3603
    },
    {
      "epoch": 0.9658314350797267,
      "grad_norm": 0.6236839108519305,
      "learning_rate": 4.701169064310503e-06,
      "loss": 0.5516,
      "step": 3604
    },
    {
      "epoch": 0.9660994238241993,
      "grad_norm": 0.6728076324597139,
      "learning_rate": 4.701001958981116e-06,
      "loss": 0.5992,
      "step": 3605
    },
    {
      "epoch": 0.9663674125686721,
      "grad_norm": 0.6556360369984029,
      "learning_rate": 4.700834809914032e-06,
      "loss": 0.5995,
      "step": 3606
    },
    {
      "epoch": 0.9666354013131448,
      "grad_norm": 0.656297572650713,
      "learning_rate": 4.70066761711257e-06,
      "loss": 0.5836,
      "step": 3607
    },
    {
      "epoch": 0.9669033900576176,
      "grad_norm": 0.6494586794436598,
      "learning_rate": 4.700500380580055e-06,
      "loss": 0.6145,
      "step": 3608
    },
    {
      "epoch": 0.9671713788020904,
      "grad_norm": 0.6504169835346458,
      "learning_rate": 4.700333100319809e-06,
      "loss": 0.562,
      "step": 3609
    },
    {
      "epoch": 0.967439367546563,
      "grad_norm": 0.6370385407643261,
      "learning_rate": 4.700165776335156e-06,
      "loss": 0.5583,
      "step": 3610
    },
    {
      "epoch": 0.9677073562910358,
      "grad_norm": 0.6519884791282987,
      "learning_rate": 4.699998408629422e-06,
      "loss": 0.607,
      "step": 3611
    },
    {
      "epoch": 0.9679753450355085,
      "grad_norm": 0.6477745209375867,
      "learning_rate": 4.699830997205932e-06,
      "loss": 0.5813,
      "step": 3612
    },
    {
      "epoch": 0.9682433337799813,
      "grad_norm": 0.647358068044171,
      "learning_rate": 4.6996635420680135e-06,
      "loss": 0.5899,
      "step": 3613
    },
    {
      "epoch": 0.9685113225244539,
      "grad_norm": 0.6571576321869307,
      "learning_rate": 4.6994960432189925e-06,
      "loss": 0.561,
      "step": 3614
    },
    {
      "epoch": 0.9687793112689267,
      "grad_norm": 0.6329202785035706,
      "learning_rate": 4.6993285006622e-06,
      "loss": 0.5641,
      "step": 3615
    },
    {
      "epoch": 0.9690473000133994,
      "grad_norm": 0.6555508353377388,
      "learning_rate": 4.6991609144009635e-06,
      "loss": 0.6002,
      "step": 3616
    },
    {
      "epoch": 0.9693152887578722,
      "grad_norm": 0.6521413438360688,
      "learning_rate": 4.698993284438613e-06,
      "loss": 0.5848,
      "step": 3617
    },
    {
      "epoch": 0.9695832775023449,
      "grad_norm": 0.6498490247536117,
      "learning_rate": 4.698825610778481e-06,
      "loss": 0.5996,
      "step": 3618
    },
    {
      "epoch": 0.9698512662468176,
      "grad_norm": 0.6492372739353627,
      "learning_rate": 4.698657893423899e-06,
      "loss": 0.5586,
      "step": 3619
    },
    {
      "epoch": 0.9701192549912904,
      "grad_norm": 0.6378289576785083,
      "learning_rate": 4.6984901323781996e-06,
      "loss": 0.595,
      "step": 3620
    },
    {
      "epoch": 0.9703872437357631,
      "grad_norm": 0.6453618813965615,
      "learning_rate": 4.698322327644716e-06,
      "loss": 0.5561,
      "step": 3621
    },
    {
      "epoch": 0.9706552324802359,
      "grad_norm": 0.6318134540658313,
      "learning_rate": 4.6981544792267845e-06,
      "loss": 0.5617,
      "step": 3622
    },
    {
      "epoch": 0.9709232212247085,
      "grad_norm": 0.7168013331348658,
      "learning_rate": 4.697986587127738e-06,
      "loss": 0.5637,
      "step": 3623
    },
    {
      "epoch": 0.9711912099691813,
      "grad_norm": 0.6332542171700722,
      "learning_rate": 4.697818651350915e-06,
      "loss": 0.6085,
      "step": 3624
    },
    {
      "epoch": 0.971459198713654,
      "grad_norm": 0.6367001133305797,
      "learning_rate": 4.697650671899651e-06,
      "loss": 0.5733,
      "step": 3625
    },
    {
      "epoch": 0.9717271874581268,
      "grad_norm": 0.63473818146096,
      "learning_rate": 4.697482648777286e-06,
      "loss": 0.5731,
      "step": 3626
    },
    {
      "epoch": 0.9719951762025995,
      "grad_norm": 0.6502327597298099,
      "learning_rate": 4.697314581987157e-06,
      "loss": 0.5913,
      "step": 3627
    },
    {
      "epoch": 0.9722631649470722,
      "grad_norm": 0.6554131011106271,
      "learning_rate": 4.697146471532604e-06,
      "loss": 0.5782,
      "step": 3628
    },
    {
      "epoch": 0.972531153691545,
      "grad_norm": 0.6492984154166548,
      "learning_rate": 4.696978317416969e-06,
      "loss": 0.5873,
      "step": 3629
    },
    {
      "epoch": 0.9727991424360177,
      "grad_norm": 0.6244987991532543,
      "learning_rate": 4.696810119643592e-06,
      "loss": 0.5453,
      "step": 3630
    },
    {
      "epoch": 0.9730671311804904,
      "grad_norm": 0.649159479186248,
      "learning_rate": 4.696641878215818e-06,
      "loss": 0.5663,
      "step": 3631
    },
    {
      "epoch": 0.9733351199249631,
      "grad_norm": 0.6546716549347812,
      "learning_rate": 4.696473593136986e-06,
      "loss": 0.5701,
      "step": 3632
    },
    {
      "epoch": 0.9736031086694359,
      "grad_norm": 0.6623182427672153,
      "learning_rate": 4.696305264410443e-06,
      "loss": 0.549,
      "step": 3633
    },
    {
      "epoch": 0.9738710974139086,
      "grad_norm": 0.6456930660482872,
      "learning_rate": 4.696136892039535e-06,
      "loss": 0.5818,
      "step": 3634
    },
    {
      "epoch": 0.9741390861583813,
      "grad_norm": 0.6556592259980485,
      "learning_rate": 4.695968476027604e-06,
      "loss": 0.5838,
      "step": 3635
    },
    {
      "epoch": 0.974407074902854,
      "grad_norm": 0.6745412945782988,
      "learning_rate": 4.695800016378e-06,
      "loss": 0.5739,
      "step": 3636
    },
    {
      "epoch": 0.9746750636473268,
      "grad_norm": 0.7116178716088697,
      "learning_rate": 4.69563151309407e-06,
      "loss": 0.5432,
      "step": 3637
    },
    {
      "epoch": 0.9749430523917996,
      "grad_norm": 0.6292854150824467,
      "learning_rate": 4.695462966179162e-06,
      "loss": 0.5698,
      "step": 3638
    },
    {
      "epoch": 0.9752110411362723,
      "grad_norm": 0.6484967253891765,
      "learning_rate": 4.695294375636624e-06,
      "loss": 0.5763,
      "step": 3639
    },
    {
      "epoch": 0.975479029880745,
      "grad_norm": 0.7204885322641233,
      "learning_rate": 4.69512574146981e-06,
      "loss": 0.5991,
      "step": 3640
    },
    {
      "epoch": 0.9757470186252177,
      "grad_norm": 0.7433698673609244,
      "learning_rate": 4.694957063682066e-06,
      "loss": 0.5904,
      "step": 3641
    },
    {
      "epoch": 0.9760150073696905,
      "grad_norm": 0.6816053171443545,
      "learning_rate": 4.6947883422767475e-06,
      "loss": 0.5805,
      "step": 3642
    },
    {
      "epoch": 0.9762829961141632,
      "grad_norm": 0.6824124551719561,
      "learning_rate": 4.694619577257207e-06,
      "loss": 0.5529,
      "step": 3643
    },
    {
      "epoch": 0.9765509848586359,
      "grad_norm": 0.6711404840074447,
      "learning_rate": 4.694450768626796e-06,
      "loss": 0.5563,
      "step": 3644
    },
    {
      "epoch": 0.9768189736031087,
      "grad_norm": 0.6666167929978988,
      "learning_rate": 4.694281916388871e-06,
      "loss": 0.5538,
      "step": 3645
    },
    {
      "epoch": 0.9770869623475814,
      "grad_norm": 0.6618271684277123,
      "learning_rate": 4.694113020546788e-06,
      "loss": 0.6124,
      "step": 3646
    },
    {
      "epoch": 0.9773549510920542,
      "grad_norm": 0.6756244121398133,
      "learning_rate": 4.693944081103901e-06,
      "loss": 0.6119,
      "step": 3647
    },
    {
      "epoch": 0.9776229398365268,
      "grad_norm": 0.6579517234437223,
      "learning_rate": 4.693775098063568e-06,
      "loss": 0.5939,
      "step": 3648
    },
    {
      "epoch": 0.9778909285809996,
      "grad_norm": 0.6398023296023155,
      "learning_rate": 4.693606071429147e-06,
      "loss": 0.5647,
      "step": 3649
    },
    {
      "epoch": 0.9781589173254723,
      "grad_norm": 0.6405373765892327,
      "learning_rate": 4.693437001203997e-06,
      "loss": 0.5695,
      "step": 3650
    },
    {
      "epoch": 0.9784269060699451,
      "grad_norm": 0.6336071875375705,
      "learning_rate": 4.693267887391479e-06,
      "loss": 0.5669,
      "step": 3651
    },
    {
      "epoch": 0.9786948948144178,
      "grad_norm": 0.6674678141420214,
      "learning_rate": 4.693098729994951e-06,
      "loss": 0.594,
      "step": 3652
    },
    {
      "epoch": 0.9789628835588905,
      "grad_norm": 0.6681678780530913,
      "learning_rate": 4.692929529017777e-06,
      "loss": 0.5645,
      "step": 3653
    },
    {
      "epoch": 0.9792308723033633,
      "grad_norm": 0.6743657856883705,
      "learning_rate": 4.692760284463317e-06,
      "loss": 0.5936,
      "step": 3654
    },
    {
      "epoch": 0.979498861047836,
      "grad_norm": 0.6596156183491241,
      "learning_rate": 4.692590996334936e-06,
      "loss": 0.5858,
      "step": 3655
    },
    {
      "epoch": 0.9797668497923088,
      "grad_norm": 0.6383699777358531,
      "learning_rate": 4.692421664635998e-06,
      "loss": 0.5441,
      "step": 3656
    },
    {
      "epoch": 0.9800348385367814,
      "grad_norm": 0.6440554383188176,
      "learning_rate": 4.692252289369865e-06,
      "loss": 0.5687,
      "step": 3657
    },
    {
      "epoch": 0.9803028272812542,
      "grad_norm": 0.6539967503504834,
      "learning_rate": 4.692082870539907e-06,
      "loss": 0.5815,
      "step": 3658
    },
    {
      "epoch": 0.9805708160257269,
      "grad_norm": 0.6399353465410915,
      "learning_rate": 4.691913408149488e-06,
      "loss": 0.597,
      "step": 3659
    },
    {
      "epoch": 0.9808388047701997,
      "grad_norm": 0.6263977110881341,
      "learning_rate": 4.691743902201976e-06,
      "loss": 0.5914,
      "step": 3660
    },
    {
      "epoch": 0.9811067935146723,
      "grad_norm": 0.6482229236602403,
      "learning_rate": 4.69157435270074e-06,
      "loss": 0.6064,
      "step": 3661
    },
    {
      "epoch": 0.9813747822591451,
      "grad_norm": 0.6568494520869423,
      "learning_rate": 4.691404759649148e-06,
      "loss": 0.575,
      "step": 3662
    },
    {
      "epoch": 0.9816427710036179,
      "grad_norm": 0.6350581537660648,
      "learning_rate": 4.691235123050572e-06,
      "loss": 0.5421,
      "step": 3663
    },
    {
      "epoch": 0.9819107597480906,
      "grad_norm": 0.6773109753056816,
      "learning_rate": 4.69106544290838e-06,
      "loss": 0.5994,
      "step": 3664
    },
    {
      "epoch": 0.9821787484925633,
      "grad_norm": 0.647093485207133,
      "learning_rate": 4.690895719225947e-06,
      "loss": 0.5853,
      "step": 3665
    },
    {
      "epoch": 0.982446737237036,
      "grad_norm": 0.6706848057554287,
      "learning_rate": 4.690725952006645e-06,
      "loss": 0.5912,
      "step": 3666
    },
    {
      "epoch": 0.9827147259815088,
      "grad_norm": 0.6430755213039979,
      "learning_rate": 4.690556141253846e-06,
      "loss": 0.5521,
      "step": 3667
    },
    {
      "epoch": 0.9829827147259815,
      "grad_norm": 0.6469169438187203,
      "learning_rate": 4.6903862869709255e-06,
      "loss": 0.5661,
      "step": 3668
    },
    {
      "epoch": 0.9832507034704543,
      "grad_norm": 0.65964757544449,
      "learning_rate": 4.690216389161259e-06,
      "loss": 0.607,
      "step": 3669
    },
    {
      "epoch": 0.9835186922149269,
      "grad_norm": 0.6685604183557473,
      "learning_rate": 4.6900464478282225e-06,
      "loss": 0.5721,
      "step": 3670
    },
    {
      "epoch": 0.9837866809593997,
      "grad_norm": 0.6710679917523679,
      "learning_rate": 4.6898764629751926e-06,
      "loss": 0.5737,
      "step": 3671
    },
    {
      "epoch": 0.9840546697038725,
      "grad_norm": 0.6512572580212656,
      "learning_rate": 4.689706434605547e-06,
      "loss": 0.5821,
      "step": 3672
    },
    {
      "epoch": 0.9843226584483452,
      "grad_norm": 0.6703606553161069,
      "learning_rate": 4.689536362722665e-06,
      "loss": 0.5581,
      "step": 3673
    },
    {
      "epoch": 0.9845906471928179,
      "grad_norm": 0.6358536582062997,
      "learning_rate": 4.689366247329927e-06,
      "loss": 0.5599,
      "step": 3674
    },
    {
      "epoch": 0.9848586359372906,
      "grad_norm": 0.6678847713531305,
      "learning_rate": 4.689196088430712e-06,
      "loss": 0.5912,
      "step": 3675
    },
    {
      "epoch": 0.9851266246817634,
      "grad_norm": 0.665928201593799,
      "learning_rate": 4.689025886028402e-06,
      "loss": 0.5981,
      "step": 3676
    },
    {
      "epoch": 0.9853946134262361,
      "grad_norm": 0.6521422283074183,
      "learning_rate": 4.68885564012638e-06,
      "loss": 0.5989,
      "step": 3677
    },
    {
      "epoch": 0.9856626021707088,
      "grad_norm": 0.6480716663973936,
      "learning_rate": 4.6886853507280275e-06,
      "loss": 0.6264,
      "step": 3678
    },
    {
      "epoch": 0.9859305909151815,
      "grad_norm": 0.6392060846485915,
      "learning_rate": 4.688515017836729e-06,
      "loss": 0.5487,
      "step": 3679
    },
    {
      "epoch": 0.9861985796596543,
      "grad_norm": 0.641700381206584,
      "learning_rate": 4.68834464145587e-06,
      "loss": 0.5435,
      "step": 3680
    },
    {
      "epoch": 0.986466568404127,
      "grad_norm": 0.6757960098176891,
      "learning_rate": 4.688174221588836e-06,
      "loss": 0.5996,
      "step": 3681
    },
    {
      "epoch": 0.9867345571485998,
      "grad_norm": 0.6544824147072539,
      "learning_rate": 4.6880037582390125e-06,
      "loss": 0.5807,
      "step": 3682
    },
    {
      "epoch": 0.9870025458930725,
      "grad_norm": 0.6645040542978521,
      "learning_rate": 4.687833251409788e-06,
      "loss": 0.5477,
      "step": 3683
    },
    {
      "epoch": 0.9872705346375452,
      "grad_norm": 0.6580972573568555,
      "learning_rate": 4.68766270110455e-06,
      "loss": 0.5987,
      "step": 3684
    },
    {
      "epoch": 0.987538523382018,
      "grad_norm": 0.6443733358940973,
      "learning_rate": 4.687492107326689e-06,
      "loss": 0.5781,
      "step": 3685
    },
    {
      "epoch": 0.9878065121264907,
      "grad_norm": 0.6578451277515922,
      "learning_rate": 4.687321470079593e-06,
      "loss": 0.5743,
      "step": 3686
    },
    {
      "epoch": 0.9880745008709634,
      "grad_norm": 0.6310330152546723,
      "learning_rate": 4.687150789366654e-06,
      "loss": 0.586,
      "step": 3687
    },
    {
      "epoch": 0.9883424896154361,
      "grad_norm": 0.6569763169930007,
      "learning_rate": 4.6869800651912645e-06,
      "loss": 0.5679,
      "step": 3688
    },
    {
      "epoch": 0.9886104783599089,
      "grad_norm": 0.6272993452321952,
      "learning_rate": 4.686809297556816e-06,
      "loss": 0.5928,
      "step": 3689
    },
    {
      "epoch": 0.9888784671043817,
      "grad_norm": 0.6457758889259962,
      "learning_rate": 4.686638486466701e-06,
      "loss": 0.5604,
      "step": 3690
    },
    {
      "epoch": 0.9891464558488543,
      "grad_norm": 0.65726906596196,
      "learning_rate": 4.686467631924316e-06,
      "loss": 0.5588,
      "step": 3691
    },
    {
      "epoch": 0.9894144445933271,
      "grad_norm": 0.6572543512690429,
      "learning_rate": 4.6862967339330545e-06,
      "loss": 0.5994,
      "step": 3692
    },
    {
      "epoch": 0.9896824333377998,
      "grad_norm": 0.6585801976046233,
      "learning_rate": 4.686125792496313e-06,
      "loss": 0.6017,
      "step": 3693
    },
    {
      "epoch": 0.9899504220822726,
      "grad_norm": 0.6495915800652504,
      "learning_rate": 4.685954807617489e-06,
      "loss": 0.5475,
      "step": 3694
    },
    {
      "epoch": 0.9902184108267453,
      "grad_norm": 0.6248193170628848,
      "learning_rate": 4.685783779299981e-06,
      "loss": 0.5536,
      "step": 3695
    },
    {
      "epoch": 0.990486399571218,
      "grad_norm": 0.685951233787337,
      "learning_rate": 4.685612707547184e-06,
      "loss": 0.5422,
      "step": 3696
    },
    {
      "epoch": 0.9907543883156907,
      "grad_norm": 0.6681921576312514,
      "learning_rate": 4.6854415923625015e-06,
      "loss": 0.5952,
      "step": 3697
    },
    {
      "epoch": 0.9910223770601635,
      "grad_norm": 0.645187276300403,
      "learning_rate": 4.685270433749332e-06,
      "loss": 0.5805,
      "step": 3698
    },
    {
      "epoch": 0.9912903658046363,
      "grad_norm": 0.6576249033142075,
      "learning_rate": 4.685099231711077e-06,
      "loss": 0.587,
      "step": 3699
    },
    {
      "epoch": 0.9915583545491089,
      "grad_norm": 0.6668794399735911,
      "learning_rate": 4.684927986251139e-06,
      "loss": 0.5628,
      "step": 3700
    },
    {
      "epoch": 0.9918263432935817,
      "grad_norm": 0.6774691700552868,
      "learning_rate": 4.684756697372919e-06,
      "loss": 0.5944,
      "step": 3701
    },
    {
      "epoch": 0.9920943320380544,
      "grad_norm": 0.6473423822963896,
      "learning_rate": 4.6845853650798246e-06,
      "loss": 0.5736,
      "step": 3702
    },
    {
      "epoch": 0.9923623207825272,
      "grad_norm": 0.6372520467760491,
      "learning_rate": 4.684413989375256e-06,
      "loss": 0.5635,
      "step": 3703
    },
    {
      "epoch": 0.9926303095269998,
      "grad_norm": 0.6622151047390362,
      "learning_rate": 4.684242570262623e-06,
      "loss": 0.583,
      "step": 3704
    },
    {
      "epoch": 0.9928982982714726,
      "grad_norm": 0.6735708012836784,
      "learning_rate": 4.684071107745328e-06,
      "loss": 0.5837,
      "step": 3705
    },
    {
      "epoch": 0.9931662870159453,
      "grad_norm": 0.6619030088338704,
      "learning_rate": 4.683899601826781e-06,
      "loss": 0.594,
      "step": 3706
    },
    {
      "epoch": 0.9934342757604181,
      "grad_norm": 0.6607686042526517,
      "learning_rate": 4.683728052510389e-06,
      "loss": 0.5856,
      "step": 3707
    },
    {
      "epoch": 0.9937022645048907,
      "grad_norm": 0.6434927804652357,
      "learning_rate": 4.683556459799562e-06,
      "loss": 0.5674,
      "step": 3708
    },
    {
      "epoch": 0.9939702532493635,
      "grad_norm": 0.6470230644923146,
      "learning_rate": 4.683384823697708e-06,
      "loss": 0.598,
      "step": 3709
    },
    {
      "epoch": 0.9942382419938363,
      "grad_norm": 0.6348000719576417,
      "learning_rate": 4.683213144208241e-06,
      "loss": 0.6061,
      "step": 3710
    },
    {
      "epoch": 0.994506230738309,
      "grad_norm": 0.6459143221723709,
      "learning_rate": 4.683041421334568e-06,
      "loss": 0.5328,
      "step": 3711
    },
    {
      "epoch": 0.9947742194827818,
      "grad_norm": 0.6712801576552944,
      "learning_rate": 4.682869655080105e-06,
      "loss": 0.5694,
      "step": 3712
    },
    {
      "epoch": 0.9950422082272544,
      "grad_norm": 0.6490070275887392,
      "learning_rate": 4.682697845448264e-06,
      "loss": 0.6021,
      "step": 3713
    },
    {
      "epoch": 0.9953101969717272,
      "grad_norm": 0.6584845066632391,
      "learning_rate": 4.682525992442459e-06,
      "loss": 0.5962,
      "step": 3714
    },
    {
      "epoch": 0.9955781857161999,
      "grad_norm": 0.6499705650399213,
      "learning_rate": 4.682354096066105e-06,
      "loss": 0.5766,
      "step": 3715
    },
    {
      "epoch": 0.9958461744606727,
      "grad_norm": 0.6846270505321216,
      "learning_rate": 4.682182156322619e-06,
      "loss": 0.5806,
      "step": 3716
    },
    {
      "epoch": 0.9961141632051453,
      "grad_norm": 0.6632082715001447,
      "learning_rate": 4.682010173215416e-06,
      "loss": 0.5606,
      "step": 3717
    },
    {
      "epoch": 0.9963821519496181,
      "grad_norm": 0.6622121707683738,
      "learning_rate": 4.681838146747915e-06,
      "loss": 0.5851,
      "step": 3718
    },
    {
      "epoch": 0.9966501406940909,
      "grad_norm": 0.6760299386858977,
      "learning_rate": 4.681666076923534e-06,
      "loss": 0.5538,
      "step": 3719
    },
    {
      "epoch": 0.9969181294385636,
      "grad_norm": 0.6658600429147093,
      "learning_rate": 4.6814939637456924e-06,
      "loss": 0.5603,
      "step": 3720
    },
    {
      "epoch": 0.9971861181830363,
      "grad_norm": 0.6870636117191197,
      "learning_rate": 4.68132180721781e-06,
      "loss": 0.5728,
      "step": 3721
    },
    {
      "epoch": 0.997454106927509,
      "grad_norm": 0.643322693752352,
      "learning_rate": 4.681149607343308e-06,
      "loss": 0.5753,
      "step": 3722
    },
    {
      "epoch": 0.9977220956719818,
      "grad_norm": 0.6411870463095163,
      "learning_rate": 4.680977364125608e-06,
      "loss": 0.5733,
      "step": 3723
    },
    {
      "epoch": 0.9979900844164545,
      "grad_norm": 0.6493745274510955,
      "learning_rate": 4.6808050775681345e-06,
      "loss": 0.5649,
      "step": 3724
    },
    {
      "epoch": 0.9982580731609273,
      "grad_norm": 0.6565952027812364,
      "learning_rate": 4.6806327476743085e-06,
      "loss": 0.6267,
      "step": 3725
    },
    {
      "epoch": 0.9985260619053999,
      "grad_norm": 0.644517545343833,
      "learning_rate": 4.680460374447556e-06,
      "loss": 0.574,
      "step": 3726
    },
    {
      "epoch": 0.9987940506498727,
      "grad_norm": 0.6367375241073371,
      "learning_rate": 4.680287957891303e-06,
      "loss": 0.5574,
      "step": 3727
    },
    {
      "epoch": 0.9990620393943455,
      "grad_norm": 0.6607007677607291,
      "learning_rate": 4.680115498008975e-06,
      "loss": 0.5576,
      "step": 3728
    },
    {
      "epoch": 0.9993300281388182,
      "grad_norm": 0.6414658044111372,
      "learning_rate": 4.6799429948039975e-06,
      "loss": 0.5264,
      "step": 3729
    },
    {
      "epoch": 0.9995980168832909,
      "grad_norm": 0.6391822614818372,
      "learning_rate": 4.679770448279801e-06,
      "loss": 0.5744,
      "step": 3730
    },
    {
      "epoch": 0.9998660056277636,
      "grad_norm": 0.649682318251365,
      "learning_rate": 4.679597858439813e-06,
      "loss": 0.5715,
      "step": 3731
    }
  ],
  "logging_steps": 1,
  "max_steps": 22386,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 3731,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 384443130839040.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}