{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 903,
  "global_step": 3611,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00027693159789531985,
      "grad_norm": 24.09097671508789,
      "learning_rate": 4.000000000000001e-06,
      "loss": 52.9022,
      "step": 1
    },
    {
      "epoch": 0.00027693159789531985,
      "eval_loss": 6.285526275634766,
      "eval_runtime": 810.5492,
      "eval_samples_per_second": 7.504,
      "eval_steps_per_second": 1.877,
      "step": 1
    },
    {
      "epoch": 0.0005538631957906397,
      "grad_norm": 26.264270782470703,
      "learning_rate": 8.000000000000001e-06,
      "loss": 50.5719,
      "step": 2
    },
    {
      "epoch": 0.0008307947936859596,
      "grad_norm": 24.111114501953125,
      "learning_rate": 1.2e-05,
      "loss": 51.6265,
      "step": 3
    },
    {
      "epoch": 0.0011077263915812794,
      "grad_norm": 23.482765197753906,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 48.2097,
      "step": 4
    },
    {
      "epoch": 0.0013846579894765993,
      "grad_norm": 23.02947235107422,
      "learning_rate": 2e-05,
      "loss": 50.7815,
      "step": 5
    },
    {
      "epoch": 0.001661589587371919,
      "grad_norm": 21.88750648498535,
      "learning_rate": 2.4e-05,
      "loss": 51.859,
      "step": 6
    },
    {
      "epoch": 0.001938521185267239,
      "grad_norm": 23.62565803527832,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 52.6551,
      "step": 7
    },
    {
      "epoch": 0.002215452783162559,
      "grad_norm": 28.487407684326172,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 46.7997,
      "step": 8
    },
    {
      "epoch": 0.0024923843810578787,
      "grad_norm": 30.373289108276367,
      "learning_rate": 3.6e-05,
      "loss": 46.1472,
      "step": 9
    },
    {
      "epoch": 0.0027693159789531985,
      "grad_norm": 27.043048858642578,
      "learning_rate": 4e-05,
      "loss": 51.6249,
      "step": 10
    },
    {
      "epoch": 0.0030462475768485184,
      "grad_norm": 32.95696258544922,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 39.4519,
      "step": 11
    },
    {
      "epoch": 0.003323179174743838,
      "grad_norm": 26.835208892822266,
      "learning_rate": 4.8e-05,
      "loss": 42.0588,
      "step": 12
    },
    {
      "epoch": 0.003600110772639158,
      "grad_norm": 29.825239181518555,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 43.2945,
      "step": 13
    },
    {
      "epoch": 0.003877042370534478,
      "grad_norm": 29.57949447631836,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 42.6722,
      "step": 14
    },
    {
      "epoch": 0.004153973968429798,
      "grad_norm": 35.65092849731445,
      "learning_rate": 6e-05,
      "loss": 46.1249,
      "step": 15
    },
    {
      "epoch": 0.004430905566325118,
      "grad_norm": 33.953556060791016,
      "learning_rate": 6.400000000000001e-05,
      "loss": 38.0391,
      "step": 16
    },
    {
      "epoch": 0.004707837164220438,
      "grad_norm": 36.93070983886719,
      "learning_rate": 6.800000000000001e-05,
      "loss": 40.2576,
      "step": 17
    },
    {
      "epoch": 0.004984768762115757,
      "grad_norm": 37.30794143676758,
      "learning_rate": 7.2e-05,
      "loss": 35.7331,
      "step": 18
    },
    {
      "epoch": 0.005261700360011078,
      "grad_norm": 35.6324577331543,
      "learning_rate": 7.6e-05,
      "loss": 33.1261,
      "step": 19
    },
    {
      "epoch": 0.005538631957906397,
      "grad_norm": 37.69329833984375,
      "learning_rate": 8e-05,
      "loss": 34.4139,
      "step": 20
    },
    {
      "epoch": 0.005815563555801717,
      "grad_norm": 39.204811096191406,
      "learning_rate": 8.4e-05,
      "loss": 25.1644,
      "step": 21
    },
    {
      "epoch": 0.006092495153697037,
      "grad_norm": 77.47728729248047,
      "learning_rate": 8.800000000000001e-05,
      "loss": 24.5212,
      "step": 22
    },
    {
      "epoch": 0.006369426751592357,
      "grad_norm": 35.82951354980469,
      "learning_rate": 9.200000000000001e-05,
      "loss": 23.2199,
      "step": 23
    },
    {
      "epoch": 0.006646358349487676,
      "grad_norm": 37.444766998291016,
      "learning_rate": 9.6e-05,
      "loss": 22.59,
      "step": 24
    },
    {
      "epoch": 0.006923289947382997,
      "grad_norm": 42.77293014526367,
      "learning_rate": 0.0001,
      "loss": 24.103,
      "step": 25
    },
    {
      "epoch": 0.007200221545278316,
      "grad_norm": 34.63228988647461,
      "learning_rate": 0.00010400000000000001,
      "loss": 21.5396,
      "step": 26
    },
    {
      "epoch": 0.007477153143173636,
      "grad_norm": 33.11638259887695,
      "learning_rate": 0.00010800000000000001,
      "loss": 18.6901,
      "step": 27
    },
    {
      "epoch": 0.007754084741068956,
      "grad_norm": 25.6596736907959,
      "learning_rate": 0.00011200000000000001,
      "loss": 16.7677,
      "step": 28
    },
    {
      "epoch": 0.008031016338964275,
      "grad_norm": 26.76543426513672,
      "learning_rate": 0.000116,
      "loss": 15.0301,
      "step": 29
    },
    {
      "epoch": 0.008307947936859596,
      "grad_norm": 27.160602569580078,
      "learning_rate": 0.00012,
      "loss": 12.1681,
      "step": 30
    },
    {
      "epoch": 0.008584879534754916,
      "grad_norm": 27.516878128051758,
      "learning_rate": 0.000124,
      "loss": 11.442,
      "step": 31
    },
    {
      "epoch": 0.008861811132650235,
      "grad_norm": 31.20770263671875,
      "learning_rate": 0.00012800000000000002,
      "loss": 7.6272,
      "step": 32
    },
    {
      "epoch": 0.009138742730545555,
      "grad_norm": 20.336532592773438,
      "learning_rate": 0.000132,
      "loss": 6.3337,
      "step": 33
    },
    {
      "epoch": 0.009415674328440876,
      "grad_norm": 26.660675048828125,
      "learning_rate": 0.00013600000000000003,
      "loss": 4.5571,
      "step": 34
    },
    {
      "epoch": 0.009692605926336195,
      "grad_norm": 21.37610626220703,
      "learning_rate": 0.00014,
      "loss": 3.659,
      "step": 35
    },
    {
      "epoch": 0.009969537524231515,
      "grad_norm": 32.23893356323242,
      "learning_rate": 0.000144,
      "loss": 5.3739,
      "step": 36
    },
    {
      "epoch": 0.010246469122126834,
      "grad_norm": 24.701919555664062,
      "learning_rate": 0.000148,
      "loss": 3.8149,
      "step": 37
    },
    {
      "epoch": 0.010523400720022155,
      "grad_norm": 22.42169189453125,
      "learning_rate": 0.000152,
      "loss": 3.7777,
      "step": 38
    },
    {
      "epoch": 0.010800332317917475,
      "grad_norm": 16.242782592773438,
      "learning_rate": 0.00015600000000000002,
      "loss": 3.3434,
      "step": 39
    },
    {
      "epoch": 0.011077263915812794,
      "grad_norm": 12.208995819091797,
      "learning_rate": 0.00016,
      "loss": 2.9402,
      "step": 40
    },
    {
      "epoch": 0.011354195513708113,
      "grad_norm": 20.679357528686523,
      "learning_rate": 0.000164,
      "loss": 3.7932,
      "step": 41
    },
    {
      "epoch": 0.011631127111603435,
      "grad_norm": 15.438067436218262,
      "learning_rate": 0.000168,
      "loss": 4.0106,
      "step": 42
    },
    {
      "epoch": 0.011908058709498754,
      "grad_norm": 18.89388656616211,
      "learning_rate": 0.000172,
      "loss": 3.9148,
      "step": 43
    },
    {
      "epoch": 0.012184990307394073,
      "grad_norm": 10.173064231872559,
      "learning_rate": 0.00017600000000000002,
      "loss": 2.9731,
      "step": 44
    },
    {
      "epoch": 0.012461921905289393,
      "grad_norm": 12.305123329162598,
      "learning_rate": 0.00018,
      "loss": 3.1847,
      "step": 45
    },
    {
      "epoch": 0.012738853503184714,
      "grad_norm": 10.65303897857666,
      "learning_rate": 0.00018400000000000003,
      "loss": 3.2399,
      "step": 46
    },
    {
      "epoch": 0.013015785101080033,
      "grad_norm": 12.510346412658691,
      "learning_rate": 0.000188,
      "loss": 3.1796,
      "step": 47
    },
    {
      "epoch": 0.013292716698975353,
      "grad_norm": 17.95038604736328,
      "learning_rate": 0.000192,
      "loss": 3.3959,
      "step": 48
    },
    {
      "epoch": 0.013569648296870672,
      "grad_norm": 7.766382217407227,
      "learning_rate": 0.000196,
      "loss": 3.1379,
      "step": 49
    },
    {
      "epoch": 0.013846579894765993,
      "grad_norm": 18.828136444091797,
      "learning_rate": 0.0002,
      "loss": 4.2675,
      "step": 50
    },
    {
      "epoch": 0.014123511492661313,
      "grad_norm": 8.210630416870117,
      "learning_rate": 0.0001999999610842169,
      "loss": 2.8736,
      "step": 51
    },
    {
      "epoch": 0.014400443090556632,
      "grad_norm": 6.19714879989624,
      "learning_rate": 0.00019999984433689786,
      "loss": 2.9424,
      "step": 52
    },
    {
      "epoch": 0.014677374688451952,
      "grad_norm": 8.041394233703613,
      "learning_rate": 0.0001999996497581338,
      "loss": 3.0541,
      "step": 53
    },
    {
      "epoch": 0.014954306286347273,
      "grad_norm": 8.640958786010742,
      "learning_rate": 0.00019999937734807612,
      "loss": 3.2354,
      "step": 54
    },
    {
      "epoch": 0.015231237884242592,
      "grad_norm": 6.604707717895508,
      "learning_rate": 0.00019999902710693683,
      "loss": 3.5656,
      "step": 55
    },
    {
      "epoch": 0.015508169482137912,
      "grad_norm": 5.787939548492432,
      "learning_rate": 0.00019999859903498856,
      "loss": 2.9423,
      "step": 56
    },
    {
      "epoch": 0.015785101080033233,
      "grad_norm": 5.111888408660889,
      "learning_rate": 0.00019999809313256446,
      "loss": 2.926,
      "step": 57
    },
    {
      "epoch": 0.01606203267792855,
      "grad_norm": 5.36300802230835,
      "learning_rate": 0.0001999975094000583,
      "loss": 2.9573,
      "step": 58
    },
    {
      "epoch": 0.01633896427582387,
      "grad_norm": 5.115900039672852,
      "learning_rate": 0.00019999684783792443,
      "loss": 2.8367,
      "step": 59
    },
    {
      "epoch": 0.016615895873719193,
      "grad_norm": 6.6548075675964355,
      "learning_rate": 0.00019999610844667772,
      "loss": 2.8389,
      "step": 60
    },
    {
      "epoch": 0.01689282747161451,
      "grad_norm": 3.7862396240234375,
      "learning_rate": 0.00019999529122689363,
      "loss": 2.7203,
      "step": 61
    },
    {
      "epoch": 0.01716975906950983,
      "grad_norm": 4.995016098022461,
      "learning_rate": 0.00019999439617920825,
      "loss": 3.084,
      "step": 62
    },
    {
      "epoch": 0.01744669066740515,
      "grad_norm": 7.811636447906494,
      "learning_rate": 0.00019999342330431823,
      "loss": 3.8168,
      "step": 63
    },
    {
      "epoch": 0.01772362226530047,
      "grad_norm": 8.833688735961914,
      "learning_rate": 0.00019999237260298072,
      "loss": 3.2426,
      "step": 64
    },
    {
      "epoch": 0.01800055386319579,
      "grad_norm": 7.167825222015381,
      "learning_rate": 0.00019999124407601353,
      "loss": 3.4158,
      "step": 65
    },
    {
      "epoch": 0.01827748546109111,
      "grad_norm": 9.22640609741211,
      "learning_rate": 0.000199990037724295,
      "loss": 3.6612,
      "step": 66
    },
    {
      "epoch": 0.01855441705898643,
      "grad_norm": 7.312063217163086,
      "learning_rate": 0.00019998875354876406,
      "loss": 3.4862,
      "step": 67
    },
    {
      "epoch": 0.01883134865688175,
      "grad_norm": 7.195916652679443,
      "learning_rate": 0.0001999873915504202,
      "loss": 3.1429,
      "step": 68
    },
    {
      "epoch": 0.01910828025477707,
      "grad_norm": 5.664394855499268,
      "learning_rate": 0.00019998595173032347,
      "loss": 2.7398,
      "step": 69
    },
    {
      "epoch": 0.01938521185267239,
      "grad_norm": 8.402835845947266,
      "learning_rate": 0.00019998443408959454,
      "loss": 2.8071,
      "step": 70
    },
    {
      "epoch": 0.019662143450567708,
      "grad_norm": 6.3753790855407715,
      "learning_rate": 0.00019998283862941458,
      "loss": 2.9036,
      "step": 71
    },
    {
      "epoch": 0.01993907504846303,
      "grad_norm": 6.662612438201904,
      "learning_rate": 0.0001999811653510254,
      "loss": 3.0893,
      "step": 72
    },
    {
      "epoch": 0.02021600664635835,
      "grad_norm": 7.037993907928467,
      "learning_rate": 0.00019997941425572928,
      "loss": 3.5855,
      "step": 73
    },
    {
      "epoch": 0.020492938244253668,
      "grad_norm": 4.904363632202148,
      "learning_rate": 0.00019997758534488915,
      "loss": 3.2274,
      "step": 74
    },
    {
      "epoch": 0.02076986984214899,
      "grad_norm": 9.387070655822754,
      "learning_rate": 0.0001999756786199285,
      "loss": 4.6668,
      "step": 75
    },
    {
      "epoch": 0.02104680144004431,
      "grad_norm": 2.9691836833953857,
      "learning_rate": 0.00019997369408233134,
      "loss": 3.1311,
      "step": 76
    },
    {
      "epoch": 0.021323733037939628,
      "grad_norm": 4.634636878967285,
      "learning_rate": 0.00019997163173364233,
      "loss": 3.2597,
      "step": 77
    },
    {
      "epoch": 0.02160066463583495,
      "grad_norm": 5.3558759689331055,
      "learning_rate": 0.00019996949157546654,
      "loss": 3.3337,
      "step": 78
    },
    {
      "epoch": 0.021877596233730267,
      "grad_norm": 4.282164096832275,
      "learning_rate": 0.00019996727360946972,
      "loss": 3.2675,
      "step": 79
    },
    {
      "epoch": 0.022154527831625588,
      "grad_norm": 4.271265506744385,
      "learning_rate": 0.00019996497783737817,
      "loss": 3.049,
      "step": 80
    },
    {
      "epoch": 0.02243145942952091,
      "grad_norm": 3.7361857891082764,
      "learning_rate": 0.0001999626042609787,
      "loss": 2.9496,
      "step": 81
    },
    {
      "epoch": 0.022708391027416227,
      "grad_norm": 6.172094345092773,
      "learning_rate": 0.00019996015288211872,
      "loss": 2.8293,
      "step": 82
    },
    {
      "epoch": 0.022985322625311548,
      "grad_norm": 4.567731857299805,
      "learning_rate": 0.00019995762370270614,
      "loss": 3.11,
      "step": 83
    },
    {
      "epoch": 0.02326225422320687,
      "grad_norm": 9.828329086303711,
      "learning_rate": 0.00019995501672470951,
      "loss": 3.1826,
      "step": 84
    },
    {
      "epoch": 0.023539185821102187,
      "grad_norm": 7.35584831237793,
      "learning_rate": 0.00019995233195015783,
      "loss": 3.1093,
      "step": 85
    },
    {
      "epoch": 0.023816117418997508,
      "grad_norm": 6.213433265686035,
      "learning_rate": 0.00019994956938114075,
      "loss": 3.0147,
      "step": 86
    },
    {
      "epoch": 0.024093049016892826,
      "grad_norm": 7.855972766876221,
      "learning_rate": 0.0001999467290198084,
      "loss": 3.8945,
      "step": 87
    },
    {
      "epoch": 0.024369980614788147,
      "grad_norm": 6.687296390533447,
      "learning_rate": 0.00019994381086837148,
      "loss": 3.1185,
      "step": 88
    },
    {
      "epoch": 0.024646912212683468,
      "grad_norm": 5.172341823577881,
      "learning_rate": 0.00019994081492910124,
      "loss": 3.0799,
      "step": 89
    },
    {
      "epoch": 0.024923843810578786,
      "grad_norm": 4.361382484436035,
      "learning_rate": 0.00019993774120432946,
      "loss": 2.8538,
      "step": 90
    },
    {
      "epoch": 0.025200775408474107,
      "grad_norm": 2.921813726425171,
      "learning_rate": 0.00019993458969644844,
      "loss": 3.1521,
      "step": 91
    },
    {
      "epoch": 0.025477707006369428,
      "grad_norm": 5.353600025177002,
      "learning_rate": 0.0001999313604079111,
      "loss": 3.3836,
      "step": 92
    },
    {
      "epoch": 0.025754638604264746,
      "grad_norm": 24.217308044433594,
      "learning_rate": 0.00019992805334123078,
      "loss": 3.0601,
      "step": 93
    },
    {
      "epoch": 0.026031570202160067,
      "grad_norm": 15.985791206359863,
      "learning_rate": 0.0001999246684989815,
      "loss": 3.2496,
      "step": 94
    },
    {
      "epoch": 0.026308501800055388,
      "grad_norm": 5.215633869171143,
      "learning_rate": 0.00019992120588379773,
      "loss": 3.2689,
      "step": 95
    },
    {
      "epoch": 0.026585433397950706,
      "grad_norm": 5.993437767028809,
      "learning_rate": 0.00019991766549837438,
      "loss": 3.0728,
      "step": 96
    },
    {
      "epoch": 0.026862364995846027,
      "grad_norm": 8.502284049987793,
      "learning_rate": 0.00019991404734546708,
      "loss": 3.1705,
      "step": 97
    },
    {
      "epoch": 0.027139296593741345,
      "grad_norm": 8.067936897277832,
      "learning_rate": 0.00019991035142789186,
      "loss": 2.809,
      "step": 98
    },
    {
      "epoch": 0.027416228191636666,
      "grad_norm": 4.59942626953125,
      "learning_rate": 0.00019990657774852534,
      "loss": 2.972,
      "step": 99
    },
    {
      "epoch": 0.027693159789531987,
      "grad_norm": 11.113409042358398,
      "learning_rate": 0.0001999027263103046,
      "loss": 4.6655,
      "step": 100
    },
    {
      "epoch": 0.027970091387427305,
      "grad_norm": 7.234707832336426,
      "learning_rate": 0.00019989879711622725,
      "loss": 3.2355,
      "step": 101
    },
    {
      "epoch": 0.028247022985322626,
      "grad_norm": 2.6375296115875244,
      "learning_rate": 0.00019989479016935152,
      "loss": 2.6118,
      "step": 102
    },
    {
      "epoch": 0.028523954583217947,
      "grad_norm": 3.2083632946014404,
      "learning_rate": 0.00019989070547279605,
      "loss": 3.017,
      "step": 103
    },
    {
      "epoch": 0.028800886181113265,
      "grad_norm": 3.726649522781372,
      "learning_rate": 0.00019988654302974,
      "loss": 3.1657,
      "step": 104
    },
    {
      "epoch": 0.029077817779008586,
      "grad_norm": 3.17037296295166,
      "learning_rate": 0.00019988230284342308,
      "loss": 2.8929,
      "step": 105
    },
    {
      "epoch": 0.029354749376903903,
      "grad_norm": 3.549494504928589,
      "learning_rate": 0.00019987798491714548,
      "loss": 2.9011,
      "step": 106
    },
    {
      "epoch": 0.029631680974799224,
      "grad_norm": 3.223787784576416,
      "learning_rate": 0.0001998735892542679,
      "loss": 3.0911,
      "step": 107
    },
    {
      "epoch": 0.029908612572694546,
      "grad_norm": 10.136894226074219,
      "learning_rate": 0.0001998691158582116,
      "loss": 3.3907,
      "step": 108
    },
    {
      "epoch": 0.030185544170589863,
      "grad_norm": 4.363787651062012,
      "learning_rate": 0.00019986456473245826,
      "loss": 3.1103,
      "step": 109
    },
    {
      "epoch": 0.030462475768485184,
      "grad_norm": 3.338764190673828,
      "learning_rate": 0.00019985993588055013,
      "loss": 2.8275,
      "step": 110
    },
    {
      "epoch": 0.030739407366380506,
      "grad_norm": 4.156650543212891,
      "learning_rate": 0.00019985522930608985,
      "loss": 3.1066,
      "step": 111
    },
    {
      "epoch": 0.031016338964275823,
      "grad_norm": 3.8609375953674316,
      "learning_rate": 0.0001998504450127407,
      "loss": 2.9508,
      "step": 112
    },
    {
      "epoch": 0.03129327056217114,
      "grad_norm": 18.497459411621094,
      "learning_rate": 0.00019984558300422628,
      "loss": 3.165,
      "step": 113
    },
    {
      "epoch": 0.031570202160066466,
      "grad_norm": 3.3081798553466797,
      "learning_rate": 0.00019984064328433084,
      "loss": 2.9892,
      "step": 114
    },
    {
      "epoch": 0.03184713375796178,
      "grad_norm": 5.705508232116699,
      "learning_rate": 0.000199835625856899,
      "loss": 3.1684,
      "step": 115
    },
    {
      "epoch": 0.0321240653558571,
      "grad_norm": 5.110528469085693,
      "learning_rate": 0.00019983053072583596,
      "loss": 3.1865,
      "step": 116
    },
    {
      "epoch": 0.032400996953752426,
      "grad_norm": 5.05802059173584,
      "learning_rate": 0.0001998253578951073,
      "loss": 2.9766,
      "step": 117
    },
    {
      "epoch": 0.03267792855164774,
      "grad_norm": 7.192070484161377,
      "learning_rate": 0.00019982010736873908,
      "loss": 3.2792,
      "step": 118
    },
    {
      "epoch": 0.03295486014954306,
      "grad_norm": 3.830031156539917,
      "learning_rate": 0.00019981477915081793,
      "loss": 2.721,
      "step": 119
    },
    {
      "epoch": 0.033231791747438386,
      "grad_norm": 3.6124696731567383,
      "learning_rate": 0.00019980937324549084,
      "loss": 2.6529,
      "step": 120
    },
    {
      "epoch": 0.0335087233453337,
      "grad_norm": 3.6310627460479736,
      "learning_rate": 0.00019980388965696534,
      "loss": 2.9436,
      "step": 121
    },
    {
      "epoch": 0.03378565494322902,
      "grad_norm": 6.552757740020752,
      "learning_rate": 0.00019979832838950938,
      "loss": 3.1585,
      "step": 122
    },
    {
      "epoch": 0.034062586541124346,
      "grad_norm": 3.328778028488159,
      "learning_rate": 0.00019979268944745137,
      "loss": 2.8573,
      "step": 123
    },
    {
      "epoch": 0.03433951813901966,
      "grad_norm": 4.673979759216309,
      "learning_rate": 0.00019978697283518023,
      "loss": 3.4607,
      "step": 124
    },
    {
      "epoch": 0.03461644973691498,
      "grad_norm": 5.156442165374756,
      "learning_rate": 0.00019978117855714526,
      "loss": 3.9396,
      "step": 125
    },
    {
      "epoch": 0.0348933813348103,
      "grad_norm": 20.98853874206543,
      "learning_rate": 0.00019977530661785618,
      "loss": 3.2191,
      "step": 126
    },
    {
      "epoch": 0.03517031293270562,
      "grad_norm": 2.979680299758911,
      "learning_rate": 0.0001997693570218833,
      "loss": 3.0521,
      "step": 127
    },
    {
      "epoch": 0.03544724453060094,
      "grad_norm": 3.1123874187469482,
      "learning_rate": 0.00019976332977385727,
      "loss": 2.9523,
      "step": 128
    },
    {
      "epoch": 0.03572417612849626,
      "grad_norm": 4.905653953552246,
      "learning_rate": 0.00019975722487846918,
      "loss": 3.0467,
      "step": 129
    },
    {
      "epoch": 0.03600110772639158,
      "grad_norm": 2.7707560062408447,
      "learning_rate": 0.00019975104234047053,
      "loss": 3.1212,
      "step": 130
    },
    {
      "epoch": 0.0362780393242869,
      "grad_norm": 3.286872148513794,
      "learning_rate": 0.00019974478216467335,
      "loss": 3.0796,
      "step": 131
    },
    {
      "epoch": 0.03655497092218222,
      "grad_norm": 4.147380352020264,
      "learning_rate": 0.00019973844435594996,
      "loss": 3.1645,
      "step": 132
    },
    {
      "epoch": 0.03683190252007754,
      "grad_norm": 3.7690138816833496,
      "learning_rate": 0.00019973202891923324,
      "loss": 2.9335,
      "step": 133
    },
    {
      "epoch": 0.03710883411797286,
      "grad_norm": 3.537510633468628,
      "learning_rate": 0.0001997255358595164,
      "loss": 3.0739,
      "step": 134
    },
    {
      "epoch": 0.03738576571586818,
      "grad_norm": 2.9214696884155273,
      "learning_rate": 0.0001997189651818531,
      "loss": 2.9609,
      "step": 135
    },
    {
      "epoch": 0.0376626973137635,
      "grad_norm": 4.733245372772217,
      "learning_rate": 0.00019971231689135737,
      "loss": 2.8016,
      "step": 136
    },
    {
      "epoch": 0.03793962891165882,
      "grad_norm": 6.125110149383545,
      "learning_rate": 0.0001997055909932037,
      "loss": 3.333,
      "step": 137
    },
    {
      "epoch": 0.03821656050955414,
      "grad_norm": 4.021219730377197,
      "learning_rate": 0.00019969878749262696,
      "loss": 3.4326,
      "step": 138
    },
    {
      "epoch": 0.03849349210744946,
      "grad_norm": 6.539845943450928,
      "learning_rate": 0.00019969190639492244,
      "loss": 3.0441,
      "step": 139
    },
    {
      "epoch": 0.03877042370534478,
      "grad_norm": 3.1108386516571045,
      "learning_rate": 0.00019968494770544577,
      "loss": 2.7918,
      "step": 140
    },
    {
      "epoch": 0.0390473553032401,
      "grad_norm": 2.605855703353882,
      "learning_rate": 0.00019967791142961304,
      "loss": 2.6988,
      "step": 141
    },
    {
      "epoch": 0.039324286901135416,
      "grad_norm": 2.4704034328460693,
      "learning_rate": 0.00019967079757290067,
      "loss": 2.9355,
      "step": 142
    },
    {
      "epoch": 0.03960121849903074,
      "grad_norm": 5.313930511474609,
      "learning_rate": 0.00019966360614084549,
      "loss": 2.9783,
      "step": 143
    },
    {
      "epoch": 0.03987815009692606,
      "grad_norm": 4.905053615570068,
      "learning_rate": 0.00019965633713904472,
      "loss": 2.7292,
      "step": 144
    },
    {
      "epoch": 0.040155081694821376,
      "grad_norm": 6.81724214553833,
      "learning_rate": 0.00019964899057315591,
      "loss": 3.4628,
      "step": 145
    },
    {
      "epoch": 0.0404320132927167,
      "grad_norm": 3.764613151550293,
      "learning_rate": 0.00019964156644889706,
      "loss": 2.9632,
      "step": 146
    },
    {
      "epoch": 0.04070894489061202,
      "grad_norm": 3.9981393814086914,
      "learning_rate": 0.00019963406477204644,
      "loss": 2.6181,
      "step": 147
    },
    {
      "epoch": 0.040985876488507336,
      "grad_norm": 26.09824562072754,
      "learning_rate": 0.0001996264855484427,
      "loss": 3.2855,
      "step": 148
    },
    {
      "epoch": 0.04126280808640266,
      "grad_norm": 3.9404141902923584,
      "learning_rate": 0.00019961882878398492,
      "loss": 2.9334,
      "step": 149
    },
    {
      "epoch": 0.04153973968429798,
      "grad_norm": 8.878583908081055,
      "learning_rate": 0.00019961109448463248,
      "loss": 4.5557,
      "step": 150
    },
    {
      "epoch": 0.041816671282193296,
      "grad_norm": 21.1249942779541,
      "learning_rate": 0.00019960328265640506,
      "loss": 3.4292,
      "step": 151
    },
    {
      "epoch": 0.04209360288008862,
      "grad_norm": 3.0727407932281494,
      "learning_rate": 0.00019959539330538273,
      "loss": 2.7764,
      "step": 152
    },
    {
      "epoch": 0.04237053447798394,
      "grad_norm": 16.201549530029297,
      "learning_rate": 0.00019958742643770593,
      "loss": 3.2883,
      "step": 153
    },
    {
      "epoch": 0.042647466075879256,
      "grad_norm": 11.132794380187988,
      "learning_rate": 0.0001995793820595754,
      "loss": 3.4929,
      "step": 154
    },
    {
      "epoch": 0.04292439767377458,
      "grad_norm": 3.189621686935425,
      "learning_rate": 0.0001995712601772522,
      "loss": 3.0601,
      "step": 155
    },
    {
      "epoch": 0.0432013292716699,
      "grad_norm": 5.149763107299805,
      "learning_rate": 0.0001995630607970577,
      "loss": 3.0358,
      "step": 156
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 3.466712474822998,
      "learning_rate": 0.00019955478392537362,
      "loss": 2.9031,
      "step": 157
    },
    {
      "epoch": 0.043755192467460534,
      "grad_norm": 1.9142956733703613,
      "learning_rate": 0.00019954642956864196,
      "loss": 2.8122,
      "step": 158
    },
    {
      "epoch": 0.04403212406535586,
      "grad_norm": 2.5402872562408447,
      "learning_rate": 0.00019953799773336507,
      "loss": 2.8495,
      "step": 159
    },
    {
      "epoch": 0.044309055663251176,
      "grad_norm": 3.6350300312042236,
      "learning_rate": 0.00019952948842610554,
      "loss": 3.1537,
      "step": 160
    },
    {
      "epoch": 0.044585987261146494,
      "grad_norm": 4.203430652618408,
      "learning_rate": 0.00019952090165348637,
      "loss": 2.7296,
      "step": 161
    },
    {
      "epoch": 0.04486291885904182,
      "grad_norm": 3.4311680793762207,
      "learning_rate": 0.0001995122374221907,
      "loss": 3.0541,
      "step": 162
    },
    {
      "epoch": 0.045139850456937136,
      "grad_norm": 3.866399049758911,
      "learning_rate": 0.00019950349573896213,
      "loss": 3.0068,
      "step": 163
    },
    {
      "epoch": 0.045416782054832454,
      "grad_norm": 3.7808001041412354,
      "learning_rate": 0.00019949467661060433,
      "loss": 3.0596,
      "step": 164
    },
    {
      "epoch": 0.04569371365272778,
      "grad_norm": 4.888076305389404,
      "learning_rate": 0.00019948578004398148,
      "loss": 3.3471,
      "step": 165
    },
    {
      "epoch": 0.045970645250623096,
      "grad_norm": 3.015810251235962,
      "learning_rate": 0.00019947680604601783,
      "loss": 2.868,
      "step": 166
    },
    {
      "epoch": 0.046247576848518414,
      "grad_norm": 2.8251118659973145,
      "learning_rate": 0.00019946775462369805,
      "loss": 2.7384,
      "step": 167
    },
    {
      "epoch": 0.04652450844641374,
      "grad_norm": 4.449611186981201,
      "learning_rate": 0.00019945862578406697,
      "loss": 3.2198,
      "step": 168
    },
    {
      "epoch": 0.046801440044309056,
      "grad_norm": 5.678244590759277,
      "learning_rate": 0.00019944941953422968,
      "loss": 3.2162,
      "step": 169
    },
    {
      "epoch": 0.047078371642204374,
      "grad_norm": 8.139033317565918,
      "learning_rate": 0.0001994401358813516,
      "loss": 2.859,
      "step": 170
    },
    {
      "epoch": 0.0473553032400997,
      "grad_norm": 6.555064678192139,
      "learning_rate": 0.00019943077483265833,
      "loss": 3.0907,
      "step": 171
    },
    {
      "epoch": 0.047632234837995016,
      "grad_norm": 4.014291763305664,
      "learning_rate": 0.00019942133639543573,
      "loss": 2.679,
      "step": 172
    },
    {
      "epoch": 0.047909166435890334,
      "grad_norm": 3.2168991565704346,
      "learning_rate": 0.0001994118205770298,
      "loss": 2.7681,
      "step": 173
    },
    {
      "epoch": 0.04818609803378565,
      "grad_norm": 7.394007205963135,
      "learning_rate": 0.000199402227384847,
      "loss": 3.5997,
      "step": 174
    },
    {
      "epoch": 0.048463029631680976,
      "grad_norm": 16.315059661865234,
      "learning_rate": 0.00019939255682635372,
      "loss": 4.6607,
      "step": 175
    },
    {
      "epoch": 0.048739961229576294,
      "grad_norm": 7.469921588897705,
      "learning_rate": 0.0001993828089090768,
      "loss": 3.3174,
      "step": 176
    },
    {
      "epoch": 0.04901689282747161,
      "grad_norm": 3.1955409049987793,
      "learning_rate": 0.00019937298364060312,
      "loss": 2.6456,
      "step": 177
    },
    {
      "epoch": 0.049293824425366936,
      "grad_norm": 6.934082508087158,
      "learning_rate": 0.00019936308102857993,
      "loss": 2.7126,
      "step": 178
    },
    {
      "epoch": 0.049570756023262254,
      "grad_norm": 2.4637463092803955,
      "learning_rate": 0.00019935310108071453,
      "loss": 2.608,
      "step": 179
    },
    {
      "epoch": 0.04984768762115757,
      "grad_norm": 2.381906747817993,
      "learning_rate": 0.0001993430438047745,
      "loss": 2.8141,
      "step": 180
    },
    {
      "epoch": 0.050124619219052896,
      "grad_norm": 12.824419975280762,
      "learning_rate": 0.00019933290920858752,
      "loss": 2.8778,
      "step": 181
    },
    {
      "epoch": 0.050401550816948214,
      "grad_norm": 2.076671838760376,
      "learning_rate": 0.00019932269730004153,
      "loss": 2.8058,
      "step": 182
    },
    {
      "epoch": 0.05067848241484353,
      "grad_norm": 2.2687172889709473,
      "learning_rate": 0.00019931240808708468,
      "loss": 2.9901,
      "step": 183
    },
    {
      "epoch": 0.050955414012738856,
      "grad_norm": 13.4617919921875,
      "learning_rate": 0.00019930204157772515,
      "loss": 2.9649,
      "step": 184
    },
    {
      "epoch": 0.051232345610634174,
      "grad_norm": 1.9112941026687622,
      "learning_rate": 0.00019929159778003138,
      "loss": 2.6419,
      "step": 185
    },
    {
      "epoch": 0.05150927720852949,
      "grad_norm": 3.7725117206573486,
      "learning_rate": 0.00019928107670213194,
      "loss": 2.993,
      "step": 186
    },
    {
      "epoch": 0.051786208806424816,
      "grad_norm": 4.537817478179932,
      "learning_rate": 0.00019927047835221554,
      "loss": 2.8517,
      "step": 187
    },
    {
      "epoch": 0.052063140404320134,
      "grad_norm": 2.7799930572509766,
      "learning_rate": 0.00019925980273853105,
      "loss": 2.9868,
      "step": 188
    },
    {
      "epoch": 0.05234007200221545,
      "grad_norm": 3.7512753009796143,
      "learning_rate": 0.00019924904986938754,
      "loss": 2.9711,
      "step": 189
    },
    {
      "epoch": 0.052617003600110776,
      "grad_norm": 3.921067714691162,
      "learning_rate": 0.00019923821975315403,
      "loss": 3.1084,
      "step": 190
    },
    {
      "epoch": 0.052893935198006094,
      "grad_norm": 6.35083532333374,
      "learning_rate": 0.00019922731239825978,
      "loss": 3.294,
      "step": 191
    },
    {
      "epoch": 0.05317086679590141,
      "grad_norm": 8.527310371398926,
      "learning_rate": 0.0001992163278131942,
      "loss": 2.9271,
      "step": 192
    },
    {
      "epoch": 0.05344779839379673,
      "grad_norm": 4.5866193771362305,
      "learning_rate": 0.00019920526600650674,
      "loss": 3.0789,
      "step": 193
    },
    {
      "epoch": 0.053724729991692054,
      "grad_norm": 2.8476200103759766,
      "learning_rate": 0.000199194126986807,
      "loss": 2.9648,
      "step": 194
    },
    {
      "epoch": 0.05400166158958737,
      "grad_norm": 4.191553592681885,
      "learning_rate": 0.00019918291076276466,
      "loss": 3.2052,
      "step": 195
    },
    {
      "epoch": 0.05427859318748269,
      "grad_norm": 12.183512687683105,
      "learning_rate": 0.00019917161734310943,
      "loss": 3.8112,
      "step": 196
    },
    {
      "epoch": 0.054555524785378014,
      "grad_norm": 3.692249059677124,
      "learning_rate": 0.0001991602467366312,
      "loss": 2.8357,
      "step": 197
    },
    {
      "epoch": 0.05483245638327333,
      "grad_norm": 6.3173394203186035,
      "learning_rate": 0.00019914879895217987,
      "loss": 3.1846,
      "step": 198
    },
    {
      "epoch": 0.05510938798116865,
      "grad_norm": 6.148683547973633,
      "learning_rate": 0.00019913727399866545,
      "loss": 2.8057,
      "step": 199
    },
    {
      "epoch": 0.055386319579063974,
      "grad_norm": 3.915633201599121,
      "learning_rate": 0.00019912567188505795,
      "loss": 3.6289,
      "step": 200
    },
    {
      "epoch": 0.05566325117695929,
      "grad_norm": 2.9523115158081055,
      "learning_rate": 0.00019911399262038753,
      "loss": 2.9399,
      "step": 201
    },
    {
      "epoch": 0.05594018277485461,
      "grad_norm": 2.9612538814544678,
      "learning_rate": 0.0001991022362137443,
      "loss": 2.9523,
      "step": 202
    },
    {
      "epoch": 0.056217114372749934,
      "grad_norm": 3.0619959831237793,
      "learning_rate": 0.0001990904026742785,
      "loss": 3.1371,
      "step": 203
    },
    {
      "epoch": 0.05649404597064525,
      "grad_norm": 2.0102360248565674,
      "learning_rate": 0.00019907849201120033,
      "loss": 3.1148,
      "step": 204
    },
    {
      "epoch": 0.05677097756854057,
      "grad_norm": 3.8012619018554688,
      "learning_rate": 0.00019906650423378005,
      "loss": 2.9833,
      "step": 205
    },
    {
      "epoch": 0.057047909166435894,
      "grad_norm": 3.3233330249786377,
      "learning_rate": 0.00019905443935134791,
      "loss": 2.5441,
      "step": 206
    },
    {
      "epoch": 0.05732484076433121,
      "grad_norm": 3.41413950920105,
      "learning_rate": 0.00019904229737329425,
      "loss": 3.0065,
      "step": 207
    },
    {
      "epoch": 0.05760177236222653,
      "grad_norm": 3.2364377975463867,
      "learning_rate": 0.00019903007830906935,
      "loss": 3.0149,
      "step": 208
    },
    {
      "epoch": 0.05787870396012185,
      "grad_norm": 2.9118869304656982,
      "learning_rate": 0.00019901778216818345,
      "loss": 2.8871,
      "step": 209
    },
    {
      "epoch": 0.05815563555801717,
      "grad_norm": 1.9055243730545044,
      "learning_rate": 0.00019900540896020688,
      "loss": 2.6669,
      "step": 210
    },
    {
      "epoch": 0.05843256715591249,
      "grad_norm": 2.51979923248291,
      "learning_rate": 0.0001989929586947699,
      "loss": 2.903,
      "step": 211
    },
    {
      "epoch": 0.05870949875380781,
      "grad_norm": 6.010900497436523,
      "learning_rate": 0.00019898043138156268,
      "loss": 3.3382,
      "step": 212
    },
    {
      "epoch": 0.05898643035170313,
      "grad_norm": 3.1433265209198,
      "learning_rate": 0.00019896782703033555,
      "loss": 2.7796,
      "step": 213
    },
    {
      "epoch": 0.05926336194959845,
      "grad_norm": 3.7899844646453857,
      "learning_rate": 0.00019895514565089855,
      "loss": 3.1045,
      "step": 214
    },
    {
      "epoch": 0.05954029354749377,
      "grad_norm": 4.300248146057129,
      "learning_rate": 0.00019894238725312184,
      "loss": 2.8547,
      "step": 215
    },
    {
      "epoch": 0.05981722514538909,
      "grad_norm": 4.351460933685303,
      "learning_rate": 0.0001989295518469355,
      "loss": 3.1198,
      "step": 216
    },
    {
      "epoch": 0.06009415674328441,
      "grad_norm": 14.191424369812012,
      "learning_rate": 0.0001989166394423295,
      "loss": 2.9771,
      "step": 217
    },
    {
      "epoch": 0.06037108834117973,
      "grad_norm": 8.077985763549805,
      "learning_rate": 0.00019890365004935379,
      "loss": 2.4881,
      "step": 218
    },
    {
      "epoch": 0.06064801993907505,
      "grad_norm": 5.447228908538818,
      "learning_rate": 0.00019889058367811822,
      "loss": 2.7146,
      "step": 219
    },
    {
      "epoch": 0.06092495153697037,
      "grad_norm": 2.0953845977783203,
      "learning_rate": 0.00019887744033879253,
      "loss": 2.6072,
      "step": 220
    },
    {
      "epoch": 0.06120188313486569,
      "grad_norm": 3.406092643737793,
      "learning_rate": 0.0001988642200416064,
      "loss": 2.8637,
      "step": 221
    },
    {
      "epoch": 0.06147881473276101,
      "grad_norm": 4.163158893585205,
      "learning_rate": 0.00019885092279684936,
      "loss": 2.4997,
      "step": 222
    },
    {
      "epoch": 0.06175574633065633,
      "grad_norm": 3.912731647491455,
      "learning_rate": 0.00019883754861487095,
      "loss": 2.8006,
      "step": 223
    },
    {
      "epoch": 0.06203267792855165,
      "grad_norm": 4.682294845581055,
      "learning_rate": 0.0001988240975060804,
      "loss": 2.9797,
      "step": 224
    },
    {
      "epoch": 0.062309609526446964,
      "grad_norm": 7.165354251861572,
      "learning_rate": 0.000198810569480947,
      "loss": 3.9649,
      "step": 225
    },
    {
      "epoch": 0.06258654112434228,
      "grad_norm": 1.584733247756958,
      "learning_rate": 0.00019879696454999977,
      "loss": 2.7378,
      "step": 226
    },
    {
      "epoch": 0.0628634727222376,
      "grad_norm": 3.165452241897583,
      "learning_rate": 0.00019878328272382768,
      "loss": 2.8733,
      "step": 227
    },
    {
      "epoch": 0.06314040432013293,
      "grad_norm": 2.215864419937134,
      "learning_rate": 0.0001987695240130795,
      "loss": 2.8075,
      "step": 228
    },
    {
      "epoch": 0.06341733591802824,
      "grad_norm": 2.2456507682800293,
      "learning_rate": 0.00019875568842846382,
      "loss": 2.7353,
      "step": 229
    },
    {
      "epoch": 0.06369426751592357,
      "grad_norm": 3.5446226596832275,
      "learning_rate": 0.0001987417759807491,
      "loss": 2.8128,
      "step": 230
    },
    {
      "epoch": 0.06397119911381889,
      "grad_norm": 2.351714611053467,
      "learning_rate": 0.0001987277866807637,
      "loss": 2.6989,
      "step": 231
    },
    {
      "epoch": 0.0642481307117142,
      "grad_norm": 1.9342094659805298,
      "learning_rate": 0.00019871372053939558,
      "loss": 2.8764,
      "step": 232
    },
    {
      "epoch": 0.06452506230960953,
      "grad_norm": 2.296945571899414,
      "learning_rate": 0.00019869957756759273,
      "loss": 2.7458,
      "step": 233
    },
    {
      "epoch": 0.06480199390750485,
      "grad_norm": 4.668132781982422,
      "learning_rate": 0.0001986853577763628,
      "loss": 3.1173,
      "step": 234
    },
    {
      "epoch": 0.06507892550540016,
      "grad_norm": 7.3507256507873535,
      "learning_rate": 0.0001986710611767733,
      "loss": 3.1764,
      "step": 235
    },
    {
      "epoch": 0.06535585710329549,
      "grad_norm": 7.586774826049805,
      "learning_rate": 0.00019865668777995147,
      "loss": 3.2926,
      "step": 236
    },
    {
      "epoch": 0.06563278870119081,
      "grad_norm": 7.294936656951904,
      "learning_rate": 0.0001986422375970844,
      "loss": 3.0098,
      "step": 237
    },
    {
      "epoch": 0.06590972029908612,
      "grad_norm": 3.664259195327759,
      "learning_rate": 0.00019862771063941886,
      "loss": 2.5577,
      "step": 238
    },
    {
      "epoch": 0.06618665189698145,
      "grad_norm": 1.538918375968933,
      "learning_rate": 0.00019861310691826143,
      "loss": 2.7598,
      "step": 239
    },
    {
      "epoch": 0.06646358349487677,
      "grad_norm": 5.194944381713867,
      "learning_rate": 0.00019859842644497837,
      "loss": 3.0205,
      "step": 240
    },
    {
      "epoch": 0.06674051509277208,
      "grad_norm": 3.615199327468872,
      "learning_rate": 0.00019858366923099574,
      "loss": 2.8842,
      "step": 241
    },
    {
      "epoch": 0.0670174466906674,
      "grad_norm": 2.8806331157684326,
      "learning_rate": 0.00019856883528779934,
      "loss": 2.8442,
      "step": 242
    },
    {
      "epoch": 0.06729437828856273,
      "grad_norm": 3.9716224670410156,
      "learning_rate": 0.00019855392462693466,
      "loss": 2.935,
      "step": 243
    },
    {
      "epoch": 0.06757130988645804,
      "grad_norm": 3.705488920211792,
      "learning_rate": 0.00019853893726000683,
      "loss": 3.1247,
      "step": 244
    },
    {
      "epoch": 0.06784824148435337,
      "grad_norm": 4.608498573303223,
      "learning_rate": 0.00019852387319868085,
      "loss": 2.9861,
      "step": 245
    },
    {
      "epoch": 0.06812517308224869,
      "grad_norm": 3.4963815212249756,
      "learning_rate": 0.00019850873245468126,
      "loss": 2.5856,
      "step": 246
    },
    {
      "epoch": 0.068402104680144,
      "grad_norm": 2.8581809997558594,
      "learning_rate": 0.00019849351503979233,
      "loss": 2.8976,
      "step": 247
    },
    {
      "epoch": 0.06867903627803933,
      "grad_norm": 4.240769863128662,
      "learning_rate": 0.00019847822096585803,
      "loss": 2.8831,
      "step": 248
    },
    {
      "epoch": 0.06895596787593464,
      "grad_norm": 13.28221321105957,
      "learning_rate": 0.00019846285024478202,
      "loss": 3.1482,
      "step": 249
    },
    {
      "epoch": 0.06923289947382996,
      "grad_norm": 11.003414154052734,
      "learning_rate": 0.00019844740288852748,
      "loss": 4.2067,
      "step": 250
    },
    {
      "epoch": 0.06950983107172529,
      "grad_norm": 2.4025285243988037,
      "learning_rate": 0.00019843187890911739,
      "loss": 2.8043,
      "step": 251
    },
    {
      "epoch": 0.0697867626696206,
      "grad_norm": 3.831378936767578,
      "learning_rate": 0.0001984162783186343,
      "loss": 3.1313,
      "step": 252
    },
    {
      "epoch": 0.07006369426751592,
      "grad_norm": 4.178756237030029,
      "learning_rate": 0.0001984006011292204,
      "loss": 3.4844,
      "step": 253
    },
    {
      "epoch": 0.07034062586541125,
      "grad_norm": 2.4836204051971436,
      "learning_rate": 0.00019838484735307748,
      "loss": 3.1151,
      "step": 254
    },
    {
      "epoch": 0.07061755746330656,
      "grad_norm": 3.50260591506958,
      "learning_rate": 0.00019836901700246695,
      "loss": 2.7573,
      "step": 255
    },
    {
      "epoch": 0.07089448906120188,
      "grad_norm": 2.2057502269744873,
      "learning_rate": 0.0001983531100897098,
      "loss": 2.7927,
      "step": 256
    },
    {
      "epoch": 0.0711714206590972,
      "grad_norm": 3.3176257610321045,
      "learning_rate": 0.0001983371266271867,
      "loss": 3.1583,
      "step": 257
    },
    {
      "epoch": 0.07144835225699252,
      "grad_norm": 43.778228759765625,
      "learning_rate": 0.00019832106662733774,
      "loss": 3.1222,
      "step": 258
    },
    {
      "epoch": 0.07172528385488784,
      "grad_norm": 2.3279812335968018,
      "learning_rate": 0.0001983049301026627,
      "loss": 2.7614,
      "step": 259
    },
    {
      "epoch": 0.07200221545278317,
      "grad_norm": 2.8496177196502686,
      "learning_rate": 0.00019828871706572095,
      "loss": 2.7267,
      "step": 260
    },
    {
      "epoch": 0.07227914705067848,
      "grad_norm": 2.8000481128692627,
      "learning_rate": 0.00019827242752913125,
      "loss": 3.2881,
      "step": 261
    },
    {
      "epoch": 0.0725560786485738,
      "grad_norm": 3.695194721221924,
      "learning_rate": 0.0001982560615055721,
      "loss": 2.9656,
      "step": 262
    },
    {
      "epoch": 0.07283301024646913,
      "grad_norm": 4.314358711242676,
      "learning_rate": 0.00019823961900778135,
      "loss": 2.8443,
      "step": 263
    },
    {
      "epoch": 0.07310994184436444,
      "grad_norm": 3.7146267890930176,
      "learning_rate": 0.00019822310004855652,
      "loss": 3.1693,
      "step": 264
    },
    {
      "epoch": 0.07338687344225976,
      "grad_norm": 3.826491117477417,
      "learning_rate": 0.0001982065046407545,
      "loss": 3.0863,
      "step": 265
    },
    {
      "epoch": 0.07366380504015509,
      "grad_norm": 4.198593616485596,
      "learning_rate": 0.0001981898327972918,
      "loss": 2.846,
      "step": 266
    },
    {
      "epoch": 0.0739407366380504,
      "grad_norm": 3.1046805381774902,
      "learning_rate": 0.0001981730845311444,
      "loss": 2.6716,
      "step": 267
    },
    {
      "epoch": 0.07421766823594572,
      "grad_norm": 2.8134427070617676,
      "learning_rate": 0.00019815625985534772,
      "loss": 2.8223,
      "step": 268
    },
    {
      "epoch": 0.07449459983384105,
      "grad_norm": 2.7822787761688232,
      "learning_rate": 0.00019813935878299662,
      "loss": 2.6138,
      "step": 269
    },
    {
      "epoch": 0.07477153143173636,
      "grad_norm": 4.507596969604492,
      "learning_rate": 0.00019812238132724556,
      "loss": 2.6792,
      "step": 270
    },
    {
      "epoch": 0.07504846302963168,
      "grad_norm": 3.206634283065796,
      "learning_rate": 0.0001981053275013083,
      "loss": 2.8331,
      "step": 271
    },
    {
      "epoch": 0.075325394627527,
      "grad_norm": 1.985059380531311,
      "learning_rate": 0.00019808819731845809,
      "loss": 2.4029,
      "step": 272
    },
    {
      "epoch": 0.07560232622542232,
      "grad_norm": 4.91208028793335,
      "learning_rate": 0.00019807099079202764,
      "loss": 2.7546,
      "step": 273
    },
    {
      "epoch": 0.07587925782331764,
      "grad_norm": 6.044703960418701,
      "learning_rate": 0.0001980537079354091,
      "loss": 3.9813,
      "step": 274
    },
    {
      "epoch": 0.07615618942121297,
      "grad_norm": 7.671187400817871,
      "learning_rate": 0.0001980363487620539,
      "loss": 4.2854,
      "step": 275
    },
    {
      "epoch": 0.07643312101910828,
      "grad_norm": 4.184301853179932,
      "learning_rate": 0.00019801891328547302,
      "loss": 3.0334,
      "step": 276
    },
    {
      "epoch": 0.0767100526170036,
      "grad_norm": 3.0849671363830566,
      "learning_rate": 0.00019800140151923675,
      "loss": 2.56,
      "step": 277
    },
    {
      "epoch": 0.07698698421489893,
      "grad_norm": 2.949519395828247,
      "learning_rate": 0.0001979838134769748,
      "loss": 3.0188,
      "step": 278
    },
    {
      "epoch": 0.07726391581279424,
      "grad_norm": 2.7396950721740723,
      "learning_rate": 0.00019796614917237616,
      "loss": 2.8281,
      "step": 279
    },
    {
      "epoch": 0.07754084741068956,
      "grad_norm": 1.4398963451385498,
      "learning_rate": 0.00019794840861918925,
      "loss": 2.6675,
      "step": 280
    },
    {
      "epoch": 0.07781777900858489,
      "grad_norm": 1.690563440322876,
      "learning_rate": 0.00019793059183122186,
      "loss": 2.9267,
      "step": 281
    },
    {
      "epoch": 0.0780947106064802,
      "grad_norm": 2.1529276371002197,
      "learning_rate": 0.00019791269882234102,
      "loss": 2.9394,
      "step": 282
    },
    {
      "epoch": 0.07837164220437552,
      "grad_norm": 2.6295652389526367,
      "learning_rate": 0.0001978947296064732,
      "loss": 2.9206,
      "step": 283
    },
    {
      "epoch": 0.07864857380227083,
      "grad_norm": 2.713244676589966,
      "learning_rate": 0.00019787668419760408,
      "loss": 2.7909,
      "step": 284
    },
    {
      "epoch": 0.07892550540016616,
      "grad_norm": 2.071441173553467,
      "learning_rate": 0.0001978585626097787,
      "loss": 2.9111,
      "step": 285
    },
    {
      "epoch": 0.07920243699806148,
      "grad_norm": 2.382073163986206,
      "learning_rate": 0.00019784036485710136,
      "loss": 2.6643,
      "step": 286
    },
    {
      "epoch": 0.07947936859595679,
      "grad_norm": 2.6257436275482178,
      "learning_rate": 0.00019782209095373565,
      "loss": 2.7276,
      "step": 287
    },
    {
      "epoch": 0.07975630019385212,
      "grad_norm": 2.6829757690429688,
      "learning_rate": 0.00019780374091390448,
      "loss": 2.3726,
      "step": 288
    },
    {
      "epoch": 0.08003323179174744,
      "grad_norm": 4.82999849319458,
      "learning_rate": 0.00019778531475188996,
      "loss": 3.1286,
      "step": 289
    },
    {
      "epoch": 0.08031016338964275,
      "grad_norm": 4.4267401695251465,
      "learning_rate": 0.0001977668124820334,
      "loss": 3.0422,
      "step": 290
    },
    {
      "epoch": 0.08058709498753808,
      "grad_norm": 6.372005462646484,
      "learning_rate": 0.00019774823411873547,
      "loss": 3.2041,
      "step": 291
    },
    {
      "epoch": 0.0808640265854334,
      "grad_norm": 3.2059247493743896,
      "learning_rate": 0.000197729579676456,
      "loss": 3.0659,
      "step": 292
    },
    {
      "epoch": 0.08114095818332871,
      "grad_norm": 4.272333145141602,
      "learning_rate": 0.000197710849169714,
      "loss": 3.1635,
      "step": 293
    },
    {
      "epoch": 0.08141788978122404,
      "grad_norm": 3.1921799182891846,
      "learning_rate": 0.00019769204261308774,
      "loss": 2.8431,
      "step": 294
    },
    {
      "epoch": 0.08169482137911936,
      "grad_norm": 2.311687469482422,
      "learning_rate": 0.00019767316002121467,
      "loss": 2.7492,
      "step": 295
    },
    {
      "epoch": 0.08197175297701467,
      "grad_norm": 5.720884799957275,
      "learning_rate": 0.00019765420140879135,
      "loss": 3.0525,
      "step": 296
    },
    {
      "epoch": 0.08224868457491,
      "grad_norm": 4.252990245819092,
      "learning_rate": 0.00019763516679057364,
      "loss": 2.785,
      "step": 297
    },
    {
      "epoch": 0.08252561617280532,
      "grad_norm": 3.961559772491455,
      "learning_rate": 0.00019761605618137643,
      "loss": 3.0525,
      "step": 298
    },
    {
      "epoch": 0.08280254777070063,
      "grad_norm": 5.757613182067871,
      "learning_rate": 0.00019759686959607383,
      "loss": 3.1933,
      "step": 299
    },
    {
      "epoch": 0.08307947936859596,
      "grad_norm": 6.132602691650391,
      "learning_rate": 0.00019757760704959903,
      "loss": 3.8208,
      "step": 300
    },
    {
      "epoch": 0.08335641096649128,
      "grad_norm": 4.055561065673828,
      "learning_rate": 0.00019755826855694442,
      "loss": 2.9217,
      "step": 301
    },
    {
      "epoch": 0.08363334256438659,
      "grad_norm": 3.2529594898223877,
      "learning_rate": 0.00019753885413316142,
      "loss": 3.1784,
      "step": 302
    },
    {
      "epoch": 0.08391027416228192,
      "grad_norm": 2.9904534816741943,
      "learning_rate": 0.0001975193637933606,
      "loss": 2.9318,
      "step": 303
    },
    {
      "epoch": 0.08418720576017724,
      "grad_norm": 3.231020927429199,
      "learning_rate": 0.00019749979755271155,
      "loss": 3.032,
      "step": 304
    },
    {
      "epoch": 0.08446413735807255,
      "grad_norm": 1.5338724851608276,
      "learning_rate": 0.00019748015542644302,
      "loss": 2.8338,
      "step": 305
    },
    {
      "epoch": 0.08474106895596788,
      "grad_norm": 2.9697089195251465,
      "learning_rate": 0.0001974604374298428,
      "loss": 2.7904,
      "step": 306
    },
    {
      "epoch": 0.0850180005538632,
      "grad_norm": 1.9385228157043457,
      "learning_rate": 0.00019744064357825767,
      "loss": 2.7301,
      "step": 307
    },
    {
      "epoch": 0.08529493215175851,
      "grad_norm": 2.643333911895752,
      "learning_rate": 0.00019742077388709353,
      "loss": 2.6874,
      "step": 308
    },
    {
      "epoch": 0.08557186374965384,
      "grad_norm": 2.934795379638672,
      "learning_rate": 0.00019740082837181526,
      "loss": 3.0575,
      "step": 309
    },
    {
      "epoch": 0.08584879534754916,
      "grad_norm": 2.615273952484131,
      "learning_rate": 0.00019738080704794677,
      "loss": 2.8407,
      "step": 310
    },
    {
      "epoch": 0.08612572694544447,
      "grad_norm": 2.9731595516204834,
      "learning_rate": 0.00019736070993107093,
      "loss": 2.8984,
      "step": 311
    },
    {
      "epoch": 0.0864026585433398,
      "grad_norm": 2.208294153213501,
      "learning_rate": 0.00019734053703682972,
      "loss": 2.7479,
      "step": 312
    },
    {
      "epoch": 0.08667959014123512,
      "grad_norm": 4.021787643432617,
      "learning_rate": 0.00019732028838092397,
      "loss": 3.1724,
      "step": 313
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 3.455578327178955,
      "learning_rate": 0.00019729996397911356,
      "loss": 3.1545,
      "step": 314
    },
    {
      "epoch": 0.08723345333702576,
      "grad_norm": 3.4853105545043945,
      "learning_rate": 0.00019727956384721723,
      "loss": 3.0364,
      "step": 315
    },
    {
      "epoch": 0.08751038493492107,
      "grad_norm": 1.7653639316558838,
      "learning_rate": 0.00019725908800111275,
      "loss": 2.8587,
      "step": 316
    },
    {
      "epoch": 0.08778731653281639,
      "grad_norm": 3.85060715675354,
      "learning_rate": 0.0001972385364567368,
      "loss": 2.8053,
      "step": 317
    },
    {
      "epoch": 0.08806424813071172,
      "grad_norm": 3.2734038829803467,
      "learning_rate": 0.00019721790923008499,
      "loss": 2.9927,
      "step": 318
    },
    {
      "epoch": 0.08834117972860703,
      "grad_norm": 3.687723159790039,
      "learning_rate": 0.00019719720633721178,
      "loss": 2.7902,
      "step": 319
    },
    {
      "epoch": 0.08861811132650235,
      "grad_norm": 3.100510358810425,
      "learning_rate": 0.00019717642779423057,
      "loss": 2.9339,
      "step": 320
    },
    {
      "epoch": 0.08889504292439768,
      "grad_norm": 2.6783623695373535,
      "learning_rate": 0.00019715557361731358,
      "loss": 2.6569,
      "step": 321
    },
    {
      "epoch": 0.08917197452229299,
      "grad_norm": 4.09005880355835,
      "learning_rate": 0.00019713464382269203,
      "loss": 2.9551,
      "step": 322
    },
    {
      "epoch": 0.08944890612018831,
      "grad_norm": 3.5520811080932617,
      "learning_rate": 0.00019711363842665587,
      "loss": 2.8597,
      "step": 323
    },
    {
      "epoch": 0.08972583771808364,
      "grad_norm": 4.7246479988098145,
      "learning_rate": 0.00019709255744555389,
      "loss": 2.9633,
      "step": 324
    },
    {
      "epoch": 0.09000276931597895,
      "grad_norm": 11.66451644897461,
      "learning_rate": 0.00019707140089579377,
      "loss": 4.6186,
      "step": 325
    },
    {
      "epoch": 0.09027970091387427,
      "grad_norm": 3.7271742820739746,
      "learning_rate": 0.00019705016879384201,
      "loss": 2.6191,
      "step": 326
    },
    {
      "epoch": 0.0905566325117696,
      "grad_norm": 4.173647880554199,
      "learning_rate": 0.00019702886115622388,
      "loss": 2.7914,
      "step": 327
    },
    {
      "epoch": 0.09083356410966491,
      "grad_norm": 2.0531363487243652,
      "learning_rate": 0.00019700747799952342,
      "loss": 2.8658,
      "step": 328
    },
    {
      "epoch": 0.09111049570756023,
      "grad_norm": 2.2616448402404785,
      "learning_rate": 0.0001969860193403835,
      "loss": 2.5976,
      "step": 329
    },
    {
      "epoch": 0.09138742730545556,
      "grad_norm": 2.0705738067626953,
      "learning_rate": 0.0001969644851955057,
      "loss": 2.6588,
      "step": 330
    },
    {
      "epoch": 0.09166435890335087,
      "grad_norm": 2.060875177383423,
      "learning_rate": 0.00019694287558165042,
      "loss": 2.6176,
      "step": 331
    },
    {
      "epoch": 0.09194129050124619,
      "grad_norm": 2.033216953277588,
      "learning_rate": 0.00019692119051563676,
      "loss": 2.7997,
      "step": 332
    },
    {
      "epoch": 0.09221822209914152,
      "grad_norm": 1.9967268705368042,
      "learning_rate": 0.00019689943001434252,
      "loss": 2.5866,
      "step": 333
    },
    {
      "epoch": 0.09249515369703683,
      "grad_norm": 2.241673707962036,
      "learning_rate": 0.00019687759409470426,
      "loss": 2.7305,
      "step": 334
    },
    {
      "epoch": 0.09277208529493215,
      "grad_norm": 3.0635623931884766,
      "learning_rate": 0.0001968556827737172,
      "loss": 3.0122,
      "step": 335
    },
    {
      "epoch": 0.09304901689282748,
      "grad_norm": 2.500697374343872,
      "learning_rate": 0.00019683369606843528,
      "loss": 2.7502,
      "step": 336
    },
    {
      "epoch": 0.09332594849072279,
      "grad_norm": 2.314696788787842,
      "learning_rate": 0.0001968116339959711,
      "loss": 3.0437,
      "step": 337
    },
    {
      "epoch": 0.09360288008861811,
      "grad_norm": 2.801692247390747,
      "learning_rate": 0.00019678949657349587,
      "loss": 2.8603,
      "step": 338
    },
    {
      "epoch": 0.09387981168651344,
      "grad_norm": 2.075801134109497,
      "learning_rate": 0.00019676728381823956,
      "loss": 2.9299,
      "step": 339
    },
    {
      "epoch": 0.09415674328440875,
      "grad_norm": 1.7948641777038574,
      "learning_rate": 0.00019674499574749067,
      "loss": 2.646,
      "step": 340
    },
    {
      "epoch": 0.09443367488230407,
      "grad_norm": 2.968350887298584,
      "learning_rate": 0.00019672263237859638,
      "loss": 2.7008,
      "step": 341
    },
    {
      "epoch": 0.0947106064801994,
      "grad_norm": 4.761722087860107,
      "learning_rate": 0.0001967001937289624,
      "loss": 3.13,
      "step": 342
    },
    {
      "epoch": 0.09498753807809471,
      "grad_norm": 3.11910343170166,
      "learning_rate": 0.00019667767981605314,
      "loss": 2.8321,
      "step": 343
    },
    {
      "epoch": 0.09526446967599003,
      "grad_norm": 4.676784515380859,
      "learning_rate": 0.00019665509065739149,
      "loss": 3.0559,
      "step": 344
    },
    {
      "epoch": 0.09554140127388536,
      "grad_norm": 2.8339955806732178,
      "learning_rate": 0.00019663242627055897,
      "loss": 2.7198,
      "step": 345
    },
    {
      "epoch": 0.09581833287178067,
      "grad_norm": 3.6793911457061768,
      "learning_rate": 0.00019660968667319561,
      "loss": 2.7427,
      "step": 346
    },
    {
      "epoch": 0.09609526446967599,
      "grad_norm": 1.839174747467041,
      "learning_rate": 0.000196586871883,
      "loss": 2.5297,
      "step": 347
    },
    {
      "epoch": 0.0963721960675713,
      "grad_norm": 3.2524006366729736,
      "learning_rate": 0.00019656398191772928,
      "loss": 2.6822,
      "step": 348
    },
    {
      "epoch": 0.09664912766546663,
      "grad_norm": 4.079854488372803,
      "learning_rate": 0.000196541016795199,
      "loss": 3.0894,
      "step": 349
    },
    {
      "epoch": 0.09692605926336195,
      "grad_norm": 5.71146297454834,
      "learning_rate": 0.00019651797653328333,
      "loss": 3.7383,
      "step": 350
    },
    {
      "epoch": 0.09720299086125726,
      "grad_norm": 2.944096088409424,
      "learning_rate": 0.00019649486114991485,
      "loss": 2.6763,
      "step": 351
    },
    {
      "epoch": 0.09747992245915259,
      "grad_norm": 3.183020830154419,
      "learning_rate": 0.0001964716706630846,
      "loss": 2.8159,
      "step": 352
    },
    {
      "epoch": 0.09775685405704791,
      "grad_norm": 2.9949750900268555,
      "learning_rate": 0.00019644840509084216,
      "loss": 3.2972,
      "step": 353
    },
    {
      "epoch": 0.09803378565494322,
      "grad_norm": 2.2809813022613525,
      "learning_rate": 0.00019642506445129545,
      "loss": 2.8126,
      "step": 354
    },
    {
      "epoch": 0.09831071725283855,
      "grad_norm": 1.4152350425720215,
      "learning_rate": 0.00019640164876261085,
      "loss": 2.6471,
      "step": 355
    },
    {
      "epoch": 0.09858764885073387,
      "grad_norm": 2.2027335166931152,
      "learning_rate": 0.00019637815804301315,
      "loss": 2.818,
      "step": 356
    },
    {
      "epoch": 0.09886458044862918,
      "grad_norm": 4.638129711151123,
      "learning_rate": 0.00019635459231078558,
      "loss": 2.8069,
      "step": 357
    },
    {
      "epoch": 0.09914151204652451,
      "grad_norm": 5.524168491363525,
      "learning_rate": 0.00019633095158426967,
      "loss": 3.0269,
      "step": 358
    },
    {
      "epoch": 0.09941844364441983,
      "grad_norm": 5.6533708572387695,
      "learning_rate": 0.00019630723588186545,
      "loss": 3.221,
      "step": 359
    },
    {
      "epoch": 0.09969537524231514,
      "grad_norm": 5.9534077644348145,
      "learning_rate": 0.00019628344522203115,
      "loss": 3.6454,
      "step": 360
    },
    {
      "epoch": 0.09997230684021047,
      "grad_norm": 2.3247969150543213,
      "learning_rate": 0.00019625957962328343,
      "loss": 3.2326,
      "step": 361
    },
    {
      "epoch": 0.10024923843810579,
      "grad_norm": 2.465853452682495,
      "learning_rate": 0.00019623563910419725,
      "loss": 2.7036,
      "step": 362
    },
    {
      "epoch": 0.1005261700360011,
      "grad_norm": 2.02891206741333,
      "learning_rate": 0.0001962116236834059,
      "loss": 2.7621,
      "step": 363
    },
    {
      "epoch": 0.10080310163389643,
      "grad_norm": 2.185584306716919,
      "learning_rate": 0.000196187533379601,
      "loss": 2.9568,
      "step": 364
    },
    {
      "epoch": 0.10108003323179175,
      "grad_norm": 1.705273985862732,
      "learning_rate": 0.00019616336821153235,
      "loss": 2.6296,
      "step": 365
    },
    {
      "epoch": 0.10135696482968706,
      "grad_norm": 3.615710973739624,
      "learning_rate": 0.0001961391281980081,
      "loss": 3.4777,
      "step": 366
    },
    {
      "epoch": 0.10163389642758239,
      "grad_norm": 5.6888957023620605,
      "learning_rate": 0.00019611481335789463,
      "loss": 3.3798,
      "step": 367
    },
    {
      "epoch": 0.10191082802547771,
      "grad_norm": 2.6681437492370605,
      "learning_rate": 0.00019609042371011655,
      "loss": 3.1311,
      "step": 368
    },
    {
      "epoch": 0.10218775962337302,
      "grad_norm": 2.4236390590667725,
      "learning_rate": 0.00019606595927365675,
      "loss": 2.7262,
      "step": 369
    },
    {
      "epoch": 0.10246469122126835,
      "grad_norm": 2.5280492305755615,
      "learning_rate": 0.00019604142006755625,
      "loss": 2.7093,
      "step": 370
    },
    {
      "epoch": 0.10274162281916367,
      "grad_norm": 2.658708333969116,
      "learning_rate": 0.0001960168061109143,
      "loss": 3.0635,
      "step": 371
    },
    {
      "epoch": 0.10301855441705898,
      "grad_norm": 6.890584468841553,
      "learning_rate": 0.0001959921174228883,
      "loss": 3.1091,
      "step": 372
    },
    {
      "epoch": 0.10329548601495431,
      "grad_norm": 3.1579232215881348,
      "learning_rate": 0.00019596735402269388,
      "loss": 3.1025,
      "step": 373
    },
    {
      "epoch": 0.10357241761284963,
      "grad_norm": 3.9812111854553223,
      "learning_rate": 0.00019594251592960479,
      "loss": 2.9105,
      "step": 374
    },
    {
      "epoch": 0.10384934921074494,
      "grad_norm": 6.3323187828063965,
      "learning_rate": 0.00019591760316295292,
      "loss": 4.0633,
      "step": 375
    },
    {
      "epoch": 0.10412628080864027,
      "grad_norm": 22.30817985534668,
      "learning_rate": 0.00019589261574212818,
      "loss": 3.3351,
      "step": 376
    },
    {
      "epoch": 0.10440321240653559,
      "grad_norm": 5.525020122528076,
      "learning_rate": 0.00019586755368657878,
      "loss": 2.7819,
      "step": 377
    },
    {
      "epoch": 0.1046801440044309,
      "grad_norm": 3.135986328125,
      "learning_rate": 0.0001958424170158108,
      "loss": 2.7154,
      "step": 378
    },
    {
      "epoch": 0.10495707560232623,
      "grad_norm": 5.768637180328369,
      "learning_rate": 0.0001958172057493886,
      "loss": 3.3564,
      "step": 379
    },
    {
      "epoch": 0.10523400720022155,
      "grad_norm": 2.8191492557525635,
      "learning_rate": 0.0001957919199069345,
      "loss": 3.0567,
      "step": 380
    },
    {
      "epoch": 0.10551093879811686,
      "grad_norm": 2.9175446033477783,
      "learning_rate": 0.0001957665595081288,
      "loss": 2.9957,
      "step": 381
    },
    {
      "epoch": 0.10578787039601219,
      "grad_norm": 2.604487180709839,
      "learning_rate": 0.00019574112457270993,
      "loss": 2.7528,
      "step": 382
    },
    {
      "epoch": 0.1060648019939075,
      "grad_norm": 2.066667079925537,
      "learning_rate": 0.00019571561512047428,
      "loss": 3.0062,
      "step": 383
    },
    {
      "epoch": 0.10634173359180282,
      "grad_norm": 3.283092975616455,
      "learning_rate": 0.0001956900311712763,
      "loss": 2.9717,
      "step": 384
    },
    {
      "epoch": 0.10661866518969815,
      "grad_norm": 1.97118079662323,
      "learning_rate": 0.00019566437274502833,
      "loss": 2.6684,
      "step": 385
    },
    {
      "epoch": 0.10689559678759346,
      "grad_norm": 1.9094443321228027,
      "learning_rate": 0.00019563863986170077,
      "loss": 2.8776,
      "step": 386
    },
    {
      "epoch": 0.10717252838548878,
      "grad_norm": 2.3373727798461914,
      "learning_rate": 0.0001956128325413219,
      "loss": 2.6434,
      "step": 387
    },
    {
      "epoch": 0.10744945998338411,
      "grad_norm": 5.657153129577637,
      "learning_rate": 0.00019558695080397796,
      "loss": 2.9741,
      "step": 388
    },
    {
      "epoch": 0.10772639158127942,
      "grad_norm": 3.7068936824798584,
      "learning_rate": 0.0001955609946698131,
      "loss": 2.8159,
      "step": 389
    },
    {
      "epoch": 0.10800332317917474,
      "grad_norm": 2.457514524459839,
      "learning_rate": 0.00019553496415902945,
      "loss": 3.0508,
      "step": 390
    },
    {
      "epoch": 0.10828025477707007,
      "grad_norm": 3.0572688579559326,
      "learning_rate": 0.00019550885929188686,
      "loss": 2.9603,
      "step": 391
    },
    {
      "epoch": 0.10855718637496538,
      "grad_norm": 3.693903684616089,
      "learning_rate": 0.00019548268008870329,
      "loss": 2.9386,
      "step": 392
    },
    {
      "epoch": 0.1088341179728607,
      "grad_norm": 5.157990455627441,
      "learning_rate": 0.00019545642656985428,
      "loss": 2.8769,
      "step": 393
    },
    {
      "epoch": 0.10911104957075603,
      "grad_norm": 4.02393913269043,
      "learning_rate": 0.00019543009875577346,
      "loss": 3.116,
      "step": 394
    },
    {
      "epoch": 0.10938798116865134,
      "grad_norm": 2.50396728515625,
      "learning_rate": 0.00019540369666695213,
      "loss": 2.7121,
      "step": 395
    },
    {
      "epoch": 0.10966491276654666,
      "grad_norm": 2.8286375999450684,
      "learning_rate": 0.00019537722032393945,
      "loss": 2.7944,
      "step": 396
    },
    {
      "epoch": 0.10994184436444199,
      "grad_norm": 2.6441237926483154,
      "learning_rate": 0.0001953506697473424,
      "loss": 2.7884,
      "step": 397
    },
    {
      "epoch": 0.1102187759623373,
      "grad_norm": 1.8171393871307373,
      "learning_rate": 0.0001953240449578257,
      "loss": 2.9018,
      "step": 398
    },
    {
      "epoch": 0.11049570756023262,
      "grad_norm": 5.175047874450684,
      "learning_rate": 0.0001952973459761118,
      "loss": 3.4427,
      "step": 399
    },
    {
      "epoch": 0.11077263915812795,
      "grad_norm": 7.9577860832214355,
      "learning_rate": 0.00019527057282298102,
      "loss": 4.3943,
      "step": 400
    },
    {
      "epoch": 0.11104957075602326,
      "grad_norm": 3.050788402557373,
      "learning_rate": 0.00019524372551927126,
      "loss": 2.8492,
      "step": 401
    },
    {
      "epoch": 0.11132650235391858,
      "grad_norm": 2.6500747203826904,
      "learning_rate": 0.0001952168040858782,
      "loss": 3.1055,
      "step": 402
    },
    {
      "epoch": 0.11160343395181391,
      "grad_norm": 2.206071615219116,
      "learning_rate": 0.00019518980854375521,
      "loss": 2.9417,
      "step": 403
    },
    {
      "epoch": 0.11188036554970922,
      "grad_norm": 2.746514081954956,
      "learning_rate": 0.0001951627389139134,
      "loss": 2.7294,
      "step": 404
    },
    {
      "epoch": 0.11215729714760454,
      "grad_norm": 2.190720319747925,
      "learning_rate": 0.00019513559521742142,
      "loss": 2.7132,
      "step": 405
    },
    {
      "epoch": 0.11243422874549987,
      "grad_norm": 1.882227897644043,
      "learning_rate": 0.00019510837747540566,
      "loss": 2.7519,
      "step": 406
    },
    {
      "epoch": 0.11271116034339518,
      "grad_norm": 3.664625644683838,
      "learning_rate": 0.00019508108570905013,
      "loss": 2.953,
      "step": 407
    },
    {
      "epoch": 0.1129880919412905,
      "grad_norm": 2.6911027431488037,
      "learning_rate": 0.0001950537199395964,
      "loss": 2.7795,
      "step": 408
    },
    {
      "epoch": 0.11326502353918583,
      "grad_norm": 2.9484004974365234,
      "learning_rate": 0.00019502628018834372,
      "loss": 3.057,
      "step": 409
    },
    {
      "epoch": 0.11354195513708114,
      "grad_norm": 2.3461923599243164,
      "learning_rate": 0.00019499876647664885,
      "loss": 2.8541,
      "step": 410
    },
    {
      "epoch": 0.11381888673497646,
      "grad_norm": 2.187255620956421,
      "learning_rate": 0.00019497117882592618,
      "loss": 2.6141,
      "step": 411
    },
    {
      "epoch": 0.11409581833287179,
      "grad_norm": 2.4059369564056396,
      "learning_rate": 0.00019494351725764752,
      "loss": 2.8761,
      "step": 412
    },
    {
      "epoch": 0.1143727499307671,
      "grad_norm": 4.065573692321777,
      "learning_rate": 0.00019491578179334243,
      "loss": 3.5152,
      "step": 413
    },
    {
      "epoch": 0.11464968152866242,
      "grad_norm": 2.2918248176574707,
      "learning_rate": 0.00019488797245459773,
      "loss": 2.8289,
      "step": 414
    },
    {
      "epoch": 0.11492661312655773,
      "grad_norm": 2.248016834259033,
      "learning_rate": 0.00019486008926305798,
      "loss": 2.8703,
      "step": 415
    },
    {
      "epoch": 0.11520354472445306,
      "grad_norm": 3.0664451122283936,
      "learning_rate": 0.000194832132240425,
      "loss": 2.7859,
      "step": 416
    },
    {
      "epoch": 0.11548047632234838,
      "grad_norm": 3.346709966659546,
      "learning_rate": 0.00019480410140845826,
      "loss": 2.8393,
      "step": 417
    },
    {
      "epoch": 0.1157574079202437,
      "grad_norm": 3.573076009750366,
      "learning_rate": 0.00019477599678897455,
      "loss": 3.0581,
      "step": 418
    },
    {
      "epoch": 0.11603433951813902,
      "grad_norm": 1.8886693716049194,
      "learning_rate": 0.00019474781840384816,
      "loss": 2.6688,
      "step": 419
    },
    {
      "epoch": 0.11631127111603434,
      "grad_norm": 3.444385051727295,
      "learning_rate": 0.00019471956627501076,
      "loss": 2.8626,
      "step": 420
    },
    {
      "epoch": 0.11658820271392965,
      "grad_norm": 2.1609256267547607,
      "learning_rate": 0.00019469124042445138,
      "loss": 2.8123,
      "step": 421
    },
    {
      "epoch": 0.11686513431182498,
      "grad_norm": 3.179786443710327,
      "learning_rate": 0.00019466284087421654,
      "loss": 2.8059,
      "step": 422
    },
    {
      "epoch": 0.1171420659097203,
      "grad_norm": 2.9083259105682373,
      "learning_rate": 0.00019463436764641007,
      "loss": 2.6698,
      "step": 423
    },
    {
      "epoch": 0.11741899750761561,
      "grad_norm": 6.696518898010254,
      "learning_rate": 0.00019460582076319302,
      "loss": 3.1366,
      "step": 424
    },
    {
      "epoch": 0.11769592910551094,
      "grad_norm": 4.36393928527832,
      "learning_rate": 0.000194577200246784,
      "loss": 3.3773,
      "step": 425
    },
    {
      "epoch": 0.11797286070340626,
      "grad_norm": 3.4995036125183105,
      "learning_rate": 0.00019454850611945872,
      "loss": 2.8375,
      "step": 426
    },
    {
      "epoch": 0.11824979230130157,
      "grad_norm": 2.442023277282715,
      "learning_rate": 0.0001945197384035503,
      "loss": 2.6869,
      "step": 427
    },
    {
      "epoch": 0.1185267238991969,
      "grad_norm": 3.9752767086029053,
      "learning_rate": 0.0001944908971214491,
      "loss": 3.051,
      "step": 428
    },
    {
      "epoch": 0.11880365549709222,
      "grad_norm": 2.235665798187256,
      "learning_rate": 0.00019446198229560276,
      "loss": 3.1079,
      "step": 429
    },
    {
      "epoch": 0.11908058709498753,
      "grad_norm": 1.734859824180603,
      "learning_rate": 0.00019443299394851616,
      "loss": 2.6128,
      "step": 430
    },
    {
      "epoch": 0.11935751869288286,
      "grad_norm": 2.163276433944702,
      "learning_rate": 0.0001944039321027513,
      "loss": 2.8891,
      "step": 431
    },
    {
      "epoch": 0.11963445029077818,
      "grad_norm": 2.1512417793273926,
      "learning_rate": 0.00019437479678092754,
      "loss": 3.2543,
      "step": 432
    },
    {
      "epoch": 0.1199113818886735,
      "grad_norm": 2.4014272689819336,
      "learning_rate": 0.0001943455880057213,
      "loss": 2.9538,
      "step": 433
    },
    {
      "epoch": 0.12018831348656882,
      "grad_norm": 2.0314533710479736,
      "learning_rate": 0.00019431630579986632,
      "loss": 2.7067,
      "step": 434
    },
    {
      "epoch": 0.12046524508446414,
      "grad_norm": 3.206453323364258,
      "learning_rate": 0.00019428695018615334,
      "loss": 3.1961,
      "step": 435
    },
    {
      "epoch": 0.12074217668235945,
      "grad_norm": 1.3990651369094849,
      "learning_rate": 0.00019425752118743027,
      "loss": 2.9877,
      "step": 436
    },
    {
      "epoch": 0.12101910828025478,
      "grad_norm": 2.4721100330352783,
      "learning_rate": 0.0001942280188266022,
      "loss": 3.0155,
      "step": 437
    },
    {
      "epoch": 0.1212960398781501,
      "grad_norm": 1.831649661064148,
      "learning_rate": 0.00019419844312663128,
      "loss": 2.8572,
      "step": 438
    },
    {
      "epoch": 0.12157297147604541,
      "grad_norm": 2.862865924835205,
      "learning_rate": 0.00019416879411053673,
      "loss": 2.9184,
      "step": 439
    },
    {
      "epoch": 0.12184990307394074,
      "grad_norm": 3.795285224914551,
      "learning_rate": 0.00019413907180139482,
      "loss": 3.9096,
      "step": 440
    },
    {
      "epoch": 0.12212683467183606,
      "grad_norm": 2.3346879482269287,
      "learning_rate": 0.00019410927622233893,
      "loss": 2.732,
      "step": 441
    },
    {
      "epoch": 0.12240376626973137,
      "grad_norm": 6.56414270401001,
      "learning_rate": 0.0001940794073965594,
      "loss": 2.9084,
      "step": 442
    },
    {
      "epoch": 0.1226806978676267,
      "grad_norm": 3.394613742828369,
      "learning_rate": 0.00019404946534730364,
      "loss": 3.0567,
      "step": 443
    },
    {
      "epoch": 0.12295762946552202,
      "grad_norm": 2.5969808101654053,
      "learning_rate": 0.00019401945009787594,
      "loss": 2.7193,
      "step": 444
    },
    {
      "epoch": 0.12323456106341733,
      "grad_norm": 2.7926831245422363,
      "learning_rate": 0.00019398936167163772,
      "loss": 2.4651,
      "step": 445
    },
    {
      "epoch": 0.12351149266131266,
      "grad_norm": 3.630769729614258,
      "learning_rate": 0.00019395920009200723,
      "loss": 2.8965,
      "step": 446
    },
    {
      "epoch": 0.12378842425920798,
      "grad_norm": 1.7607187032699585,
      "learning_rate": 0.0001939289653824597,
      "loss": 2.5714,
      "step": 447
    },
    {
      "epoch": 0.1240653558571033,
      "grad_norm": 4.170645236968994,
      "learning_rate": 0.00019389865756652732,
      "loss": 2.9125,
      "step": 448
    },
    {
      "epoch": 0.12434228745499862,
      "grad_norm": 5.252502918243408,
      "learning_rate": 0.0001938682766677991,
      "loss": 3.238,
      "step": 449
    },
    {
      "epoch": 0.12461921905289393,
      "grad_norm": 9.120099067687988,
      "learning_rate": 0.00019383782270992098,
      "loss": 4.1508,
      "step": 450
    },
    {
      "epoch": 0.12489615065078925,
      "grad_norm": 2.3955490589141846,
      "learning_rate": 0.00019380729571659574,
      "loss": 3.0882,
      "step": 451
    },
    {
      "epoch": 0.12517308224868456,
      "grad_norm": 2.8864545822143555,
      "learning_rate": 0.000193776695711583,
      "loss": 2.94,
      "step": 452
    },
    {
      "epoch": 0.1254500138465799,
      "grad_norm": 4.40331506729126,
      "learning_rate": 0.00019374602271869925,
      "loss": 3.0988,
      "step": 453
    },
    {
      "epoch": 0.1257269454444752,
      "grad_norm": 2.07773494720459,
      "learning_rate": 0.00019371527676181777,
      "loss": 2.8605,
      "step": 454
    },
    {
      "epoch": 0.12600387704237054,
      "grad_norm": 2.9056828022003174,
      "learning_rate": 0.00019368445786486862,
      "loss": 3.1449,
      "step": 455
    },
    {
      "epoch": 0.12628080864026586,
      "grad_norm": 2.314566135406494,
      "learning_rate": 0.00019365356605183863,
      "loss": 2.7058,
      "step": 456
    },
    {
      "epoch": 0.1265577402381612,
      "grad_norm": 3.2491626739501953,
      "learning_rate": 0.00019362260134677135,
      "loss": 2.8069,
      "step": 457
    },
    {
      "epoch": 0.12683467183605648,
      "grad_norm": 3.122917890548706,
      "learning_rate": 0.00019359156377376713,
      "loss": 2.9945,
      "step": 458
    },
    {
      "epoch": 0.1271116034339518,
      "grad_norm": 2.2590036392211914,
      "learning_rate": 0.00019356045335698296,
      "loss": 2.707,
      "step": 459
    },
    {
      "epoch": 0.12738853503184713,
      "grad_norm": 3.042433023452759,
      "learning_rate": 0.0001935292701206326,
      "loss": 2.9189,
      "step": 460
    },
    {
      "epoch": 0.12766546662974246,
      "grad_norm": 2.72328519821167,
      "learning_rate": 0.00019349801408898647,
      "loss": 3.2472,
      "step": 461
    },
    {
      "epoch": 0.12794239822763778,
      "grad_norm": 2.6480560302734375,
      "learning_rate": 0.00019346668528637156,
      "loss": 2.9503,
      "step": 462
    },
    {
      "epoch": 0.1282193298255331,
      "grad_norm": 3.1094019412994385,
      "learning_rate": 0.0001934352837371716,
      "loss": 3.017,
      "step": 463
    },
    {
      "epoch": 0.1284962614234284,
      "grad_norm": 5.252288341522217,
      "learning_rate": 0.00019340380946582695,
      "loss": 3.051,
      "step": 464
    },
    {
      "epoch": 0.12877319302132373,
      "grad_norm": 2.4221389293670654,
      "learning_rate": 0.0001933722624968345,
      "loss": 2.9024,
      "step": 465
    },
    {
      "epoch": 0.12905012461921905,
      "grad_norm": 2.4375619888305664,
      "learning_rate": 0.00019334064285474772,
      "loss": 2.7439,
      "step": 466
    },
    {
      "epoch": 0.12932705621711438,
      "grad_norm": 5.243897914886475,
      "learning_rate": 0.00019330895056417671,
      "loss": 3.147,
      "step": 467
    },
    {
      "epoch": 0.1296039878150097,
      "grad_norm": 3.8889989852905273,
      "learning_rate": 0.00019327718564978805,
      "loss": 2.8308,
      "step": 468
    },
    {
      "epoch": 0.129880919412905,
      "grad_norm": 3.177091360092163,
      "learning_rate": 0.00019324534813630487,
      "loss": 2.7943,
      "step": 469
    },
    {
      "epoch": 0.13015785101080032,
      "grad_norm": 3.306352376937866,
      "learning_rate": 0.00019321343804850683,
      "loss": 2.539,
      "step": 470
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 1.4211779832839966,
      "learning_rate": 0.00019318145541123005,
      "loss": 2.9397,
      "step": 471
    },
    {
      "epoch": 0.13071171420659097,
      "grad_norm": 2.220968723297119,
      "learning_rate": 0.0001931494002493671,
      "loss": 2.9446,
      "step": 472
    },
    {
      "epoch": 0.1309886458044863,
      "grad_norm": 4.130578994750977,
      "learning_rate": 0.00019311727258786702,
      "loss": 3.1133,
      "step": 473
    },
    {
      "epoch": 0.13126557740238162,
      "grad_norm": 3.1133687496185303,
      "learning_rate": 0.00019308507245173527,
      "loss": 3.1137,
      "step": 474
    },
    {
      "epoch": 0.13154250900027692,
      "grad_norm": 8.326321601867676,
      "learning_rate": 0.00019305279986603374,
      "loss": 4.4815,
      "step": 475
    },
    {
      "epoch": 0.13181944059817224,
      "grad_norm": 1.7736440896987915,
      "learning_rate": 0.00019302045485588068,
      "loss": 2.7976,
      "step": 476
    },
    {
      "epoch": 0.13209637219606757,
      "grad_norm": 2.2228281497955322,
      "learning_rate": 0.00019298803744645067,
      "loss": 2.8211,
      "step": 477
    },
    {
      "epoch": 0.1323733037939629,
      "grad_norm": 2.290677070617676,
      "learning_rate": 0.00019295554766297476,
      "loss": 2.7382,
      "step": 478
    },
    {
      "epoch": 0.13265023539185822,
      "grad_norm": 2.2529354095458984,
      "learning_rate": 0.0001929229855307402,
      "loss": 3.0548,
      "step": 479
    },
    {
      "epoch": 0.13292716698975354,
      "grad_norm": 1.641726016998291,
      "learning_rate": 0.00019289035107509066,
      "loss": 2.9073,
      "step": 480
    },
    {
      "epoch": 0.13320409858764884,
      "grad_norm": 2.6638753414154053,
      "learning_rate": 0.00019285764432142602,
      "loss": 2.8904,
      "step": 481
    },
    {
      "epoch": 0.13348103018554416,
      "grad_norm": 2.0153493881225586,
      "learning_rate": 0.00019282486529520243,
      "loss": 2.813,
      "step": 482
    },
    {
      "epoch": 0.1337579617834395,
      "grad_norm": 2.089580535888672,
      "learning_rate": 0.00019279201402193236,
      "loss": 3.0678,
      "step": 483
    },
    {
      "epoch": 0.1340348933813348,
      "grad_norm": 5.025186061859131,
      "learning_rate": 0.00019275909052718447,
      "loss": 2.6991,
      "step": 484
    },
    {
      "epoch": 0.13431182497923014,
      "grad_norm": 2.2681591510772705,
      "learning_rate": 0.00019272609483658362,
      "loss": 2.9733,
      "step": 485
    },
    {
      "epoch": 0.13458875657712546,
      "grad_norm": 3.0130293369293213,
      "learning_rate": 0.00019269302697581087,
      "loss": 2.8882,
      "step": 486
    },
    {
      "epoch": 0.13486568817502076,
      "grad_norm": 2.0153720378875732,
      "learning_rate": 0.00019265988697060346,
      "loss": 2.6695,
      "step": 487
    },
    {
      "epoch": 0.13514261977291608,
      "grad_norm": 3.0560152530670166,
      "learning_rate": 0.00019262667484675475,
      "loss": 2.8081,
      "step": 488
    },
    {
      "epoch": 0.1354195513708114,
      "grad_norm": 2.948871374130249,
      "learning_rate": 0.00019259339063011432,
      "loss": 2.9432,
      "step": 489
    },
    {
      "epoch": 0.13569648296870673,
      "grad_norm": 1.9918317794799805,
      "learning_rate": 0.00019256003434658773,
      "loss": 2.5663,
      "step": 490
    },
    {
      "epoch": 0.13597341456660206,
      "grad_norm": 2.153956651687622,
      "learning_rate": 0.00019252660602213674,
      "loss": 2.8127,
      "step": 491
    },
    {
      "epoch": 0.13625034616449738,
      "grad_norm": 2.756939649581909,
      "learning_rate": 0.00019249310568277909,
      "loss": 2.7582,
      "step": 492
    },
    {
      "epoch": 0.13652727776239268,
      "grad_norm": 2.7311301231384277,
      "learning_rate": 0.0001924595333545887,
      "loss": 2.3999,
      "step": 493
    },
    {
      "epoch": 0.136804209360288,
      "grad_norm": 3.7439494132995605,
      "learning_rate": 0.00019242588906369536,
      "loss": 2.8778,
      "step": 494
    },
    {
      "epoch": 0.13708114095818333,
      "grad_norm": 2.8746988773345947,
      "learning_rate": 0.00019239217283628497,
      "loss": 2.7263,
      "step": 495
    },
    {
      "epoch": 0.13735807255607865,
      "grad_norm": 1.9332902431488037,
      "learning_rate": 0.00019235838469859942,
      "loss": 2.7315,
      "step": 496
    },
    {
      "epoch": 0.13763500415397398,
      "grad_norm": 2.4865143299102783,
      "learning_rate": 0.00019232452467693657,
      "loss": 2.6248,
      "step": 497
    },
    {
      "epoch": 0.13791193575186927,
      "grad_norm": 2.1645824909210205,
      "learning_rate": 0.00019229059279765012,
      "loss": 2.7823,
      "step": 498
    },
    {
      "epoch": 0.1381888673497646,
      "grad_norm": 3.73016619682312,
      "learning_rate": 0.00019225658908714983,
      "loss": 3.0859,
      "step": 499
    },
    {
      "epoch": 0.13846579894765992,
      "grad_norm": 9.9523344039917,
      "learning_rate": 0.00019222251357190132,
      "loss": 4.6159,
      "step": 500
    },
    {
      "epoch": 0.13874273054555525,
      "grad_norm": 2.717419385910034,
      "learning_rate": 0.0001921883662784261,
      "loss": 2.9515,
      "step": 501
    },
    {
      "epoch": 0.13901966214345057,
      "grad_norm": 2.07205867767334,
      "learning_rate": 0.00019215414723330158,
      "loss": 2.764,
      "step": 502
    },
    {
      "epoch": 0.1392965937413459,
      "grad_norm": 2.078087091445923,
      "learning_rate": 0.00019211985646316093,
      "loss": 2.9914,
      "step": 503
    },
    {
      "epoch": 0.1395735253392412,
      "grad_norm": 1.61542546749115,
      "learning_rate": 0.00019208549399469318,
      "loss": 2.5703,
      "step": 504
    },
    {
      "epoch": 0.13985045693713652,
      "grad_norm": 1.77694833278656,
      "learning_rate": 0.0001920510598546432,
      "loss": 2.773,
      "step": 505
    },
    {
      "epoch": 0.14012738853503184,
      "grad_norm": 1.663738489151001,
      "learning_rate": 0.00019201655406981164,
      "loss": 2.6654,
      "step": 506
    },
    {
      "epoch": 0.14040432013292717,
      "grad_norm": 1.5260848999023438,
      "learning_rate": 0.00019198197666705488,
      "loss": 2.6962,
      "step": 507
    },
    {
      "epoch": 0.1406812517308225,
      "grad_norm": 1.298994541168213,
      "learning_rate": 0.00019194732767328505,
      "loss": 2.9068,
      "step": 508
    },
    {
      "epoch": 0.14095818332871782,
      "grad_norm": 1.8828246593475342,
      "learning_rate": 0.00019191260711547001,
      "loss": 3.1734,
      "step": 509
    },
    {
      "epoch": 0.14123511492661311,
      "grad_norm": 2.350924491882324,
      "learning_rate": 0.00019187781502063328,
      "loss": 2.9027,
      "step": 510
    },
    {
      "epoch": 0.14151204652450844,
      "grad_norm": 2.7922251224517822,
      "learning_rate": 0.00019184295141585415,
      "loss": 2.9525,
      "step": 511
    },
    {
      "epoch": 0.14178897812240376,
      "grad_norm": 2.086071491241455,
      "learning_rate": 0.00019180801632826748,
      "loss": 2.7488,
      "step": 512
    },
    {
      "epoch": 0.1420659097202991,
      "grad_norm": 4.257290840148926,
      "learning_rate": 0.00019177300978506377,
      "loss": 3.0051,
      "step": 513
    },
    {
      "epoch": 0.1423428413181944,
      "grad_norm": 4.1589508056640625,
      "learning_rate": 0.0001917379318134892,
      "loss": 3.1527,
      "step": 514
    },
    {
      "epoch": 0.14261977291608974,
      "grad_norm": 2.253479242324829,
      "learning_rate": 0.0001917027824408455,
      "loss": 2.9246,
      "step": 515
    },
    {
      "epoch": 0.14289670451398503,
      "grad_norm": 2.4339916706085205,
      "learning_rate": 0.00019166756169448993,
      "loss": 2.8065,
      "step": 516
    },
    {
      "epoch": 0.14317363611188036,
      "grad_norm": 1.9720395803451538,
      "learning_rate": 0.00019163226960183542,
      "loss": 3.0954,
      "step": 517
    },
    {
      "epoch": 0.14345056770977568,
      "grad_norm": 2.3299455642700195,
      "learning_rate": 0.00019159690619035034,
      "loss": 2.7443,
      "step": 518
    },
    {
      "epoch": 0.143727499307671,
      "grad_norm": 2.715243101119995,
      "learning_rate": 0.00019156147148755855,
      "loss": 2.7845,
      "step": 519
    },
    {
      "epoch": 0.14400443090556633,
      "grad_norm": 1.9602571725845337,
      "learning_rate": 0.00019152596552103948,
      "loss": 2.6434,
      "step": 520
    },
    {
      "epoch": 0.14428136250346166,
      "grad_norm": 1.381770372390747,
      "learning_rate": 0.00019149038831842792,
      "loss": 2.8227,
      "step": 521
    },
    {
      "epoch": 0.14455829410135695,
      "grad_norm": 3.3014719486236572,
      "learning_rate": 0.0001914547399074142,
      "loss": 3.0247,
      "step": 522
    },
    {
      "epoch": 0.14483522569925228,
      "grad_norm": 2.787161111831665,
      "learning_rate": 0.0001914190203157441,
      "loss": 3.0022,
      "step": 523
    },
    {
      "epoch": 0.1451121572971476,
      "grad_norm": 5.41040563583374,
      "learning_rate": 0.0001913832295712186,
      "loss": 3.4196,
      "step": 524
    },
    {
      "epoch": 0.14538908889504293,
      "grad_norm": 7.442314624786377,
      "learning_rate": 0.0001913473677016943,
      "loss": 4.4476,
      "step": 525
    },
    {
      "epoch": 0.14566602049293825,
      "grad_norm": 1.430985689163208,
      "learning_rate": 0.00019131143473508303,
      "loss": 2.5798,
      "step": 526
    },
    {
      "epoch": 0.14594295209083358,
      "grad_norm": 1.5545196533203125,
      "learning_rate": 0.000191275430699352,
      "loss": 2.9568,
      "step": 527
    },
    {
      "epoch": 0.14621988368872887,
      "grad_norm": 2.2793681621551514,
      "learning_rate": 0.00019123935562252365,
      "loss": 2.9475,
      "step": 528
    },
    {
      "epoch": 0.1464968152866242,
      "grad_norm": 1.8129557371139526,
      "learning_rate": 0.00019120320953267586,
      "loss": 2.8559,
      "step": 529
    },
    {
      "epoch": 0.14677374688451952,
      "grad_norm": 1.2468783855438232,
      "learning_rate": 0.0001911669924579416,
      "loss": 2.596,
      "step": 530
    },
    {
      "epoch": 0.14705067848241485,
      "grad_norm": 2.115652084350586,
      "learning_rate": 0.00019113070442650928,
      "loss": 2.6525,
      "step": 531
    },
    {
      "epoch": 0.14732761008031017,
      "grad_norm": 2.7340188026428223,
      "learning_rate": 0.00019109434546662243,
      "loss": 2.6955,
      "step": 532
    },
    {
      "epoch": 0.14760454167820547,
      "grad_norm": 1.7179347276687622,
      "learning_rate": 0.00019105791560657974,
      "loss": 3.0466,
      "step": 533
    },
    {
      "epoch": 0.1478814732761008,
      "grad_norm": 1.8532159328460693,
      "learning_rate": 0.0001910214148747352,
      "loss": 2.9043,
      "step": 534
    },
    {
      "epoch": 0.14815840487399612,
      "grad_norm": 2.033792734146118,
      "learning_rate": 0.00019098484329949786,
      "loss": 2.8517,
      "step": 535
    },
    {
      "epoch": 0.14843533647189144,
      "grad_norm": 1.300536036491394,
      "learning_rate": 0.00019094820090933195,
      "loss": 2.9107,
      "step": 536
    },
    {
      "epoch": 0.14871226806978677,
      "grad_norm": 3.2628746032714844,
      "learning_rate": 0.00019091148773275687,
      "loss": 2.829,
      "step": 537
    },
    {
      "epoch": 0.1489891996676821,
      "grad_norm": 1.9415562152862549,
      "learning_rate": 0.00019087470379834702,
      "loss": 2.6483,
      "step": 538
    },
    {
      "epoch": 0.1492661312655774,
      "grad_norm": 3.041368007659912,
      "learning_rate": 0.0001908378491347319,
      "loss": 3.1466,
      "step": 539
    },
    {
      "epoch": 0.14954306286347271,
      "grad_norm": 5.678069114685059,
      "learning_rate": 0.00019080092377059607,
      "loss": 2.8426,
      "step": 540
    },
    {
      "epoch": 0.14981999446136804,
      "grad_norm": 3.12460994720459,
      "learning_rate": 0.00019076392773467916,
      "loss": 2.9311,
      "step": 541
    },
    {
      "epoch": 0.15009692605926336,
      "grad_norm": 2.576343536376953,
      "learning_rate": 0.00019072686105577576,
      "loss": 2.8172,
      "step": 542
    },
    {
      "epoch": 0.1503738576571587,
      "grad_norm": 3.4362986087799072,
      "learning_rate": 0.0001906897237627354,
      "loss": 3.1724,
      "step": 543
    },
    {
      "epoch": 0.150650789255054,
      "grad_norm": 1.4936844110488892,
      "learning_rate": 0.00019065251588446265,
      "loss": 2.6596,
      "step": 544
    },
    {
      "epoch": 0.1509277208529493,
      "grad_norm": 1.8611481189727783,
      "learning_rate": 0.00019061523744991698,
      "loss": 2.7357,
      "step": 545
    },
    {
      "epoch": 0.15120465245084463,
      "grad_norm": 2.7158865928649902,
      "learning_rate": 0.00019057788848811277,
      "loss": 3.0875,
      "step": 546
    },
    {
      "epoch": 0.15148158404873996,
      "grad_norm": 2.2609493732452393,
      "learning_rate": 0.00019054046902811932,
      "loss": 2.7561,
      "step": 547
    },
    {
      "epoch": 0.15175851564663528,
      "grad_norm": 2.341095209121704,
      "learning_rate": 0.00019050297909906075,
      "loss": 2.8444,
      "step": 548
    },
    {
      "epoch": 0.1520354472445306,
      "grad_norm": 3.354525327682495,
      "learning_rate": 0.0001904654187301161,
      "loss": 3.2922,
      "step": 549
    },
    {
      "epoch": 0.15231237884242593,
      "grad_norm": 4.269205570220947,
      "learning_rate": 0.0001904277879505192,
      "loss": 3.528,
      "step": 550
    },
    {
      "epoch": 0.15258931044032123,
      "grad_norm": 3.570624589920044,
      "learning_rate": 0.00019039008678955864,
      "loss": 3.1134,
      "step": 551
    },
    {
      "epoch": 0.15286624203821655,
      "grad_norm": 2.4145169258117676,
      "learning_rate": 0.00019035231527657782,
      "loss": 2.7928,
      "step": 552
    },
    {
      "epoch": 0.15314317363611188,
      "grad_norm": 1.96170175075531,
      "learning_rate": 0.0001903144734409749,
      "loss": 2.8577,
      "step": 553
    },
    {
      "epoch": 0.1534201052340072,
      "grad_norm": 1.7383220195770264,
      "learning_rate": 0.0001902765613122028,
      "loss": 2.7327,
      "step": 554
    },
    {
      "epoch": 0.15369703683190253,
      "grad_norm": 2.1005501747131348,
      "learning_rate": 0.0001902385789197691,
      "loss": 2.948,
      "step": 555
    },
    {
      "epoch": 0.15397396842979785,
      "grad_norm": 2.3060429096221924,
      "learning_rate": 0.0001902005262932361,
      "loss": 2.7076,
      "step": 556
    },
    {
      "epoch": 0.15425090002769315,
      "grad_norm": 2.3842906951904297,
      "learning_rate": 0.00019016240346222079,
      "loss": 2.6827,
      "step": 557
    },
    {
      "epoch": 0.15452783162558847,
      "grad_norm": 2.296273708343506,
      "learning_rate": 0.00019012421045639473,
      "loss": 2.811,
      "step": 558
    },
    {
      "epoch": 0.1548047632234838,
      "grad_norm": 2.9906840324401855,
      "learning_rate": 0.0001900859473054841,
      "loss": 2.5892,
      "step": 559
    },
    {
      "epoch": 0.15508169482137912,
      "grad_norm": 1.905619502067566,
      "learning_rate": 0.00019004761403926978,
      "loss": 3.0653,
      "step": 560
    },
    {
      "epoch": 0.15535862641927445,
      "grad_norm": 3.003749370574951,
      "learning_rate": 0.00019000921068758708,
      "loss": 3.0541,
      "step": 561
    },
    {
      "epoch": 0.15563555801716977,
      "grad_norm": 3.384355068206787,
      "learning_rate": 0.000189970737280326,
      "loss": 3.0699,
      "step": 562
    },
    {
      "epoch": 0.15591248961506507,
      "grad_norm": 1.637637972831726,
      "learning_rate": 0.00018993219384743097,
      "loss": 2.6486,
      "step": 563
    },
    {
      "epoch": 0.1561894212129604,
      "grad_norm": 3.134511947631836,
      "learning_rate": 0.00018989358041890094,
      "loss": 3.2672,
      "step": 564
    },
    {
      "epoch": 0.15646635281085572,
      "grad_norm": 18.818422317504883,
      "learning_rate": 0.00018985489702478934,
      "loss": 2.9634,
      "step": 565
    },
    {
      "epoch": 0.15674328440875104,
      "grad_norm": 2.783501386642456,
      "learning_rate": 0.00018981614369520405,
      "loss": 2.824,
      "step": 566
    },
    {
      "epoch": 0.15702021600664637,
      "grad_norm": 3.167635440826416,
      "learning_rate": 0.00018977732046030746,
      "loss": 2.6968,
      "step": 567
    },
    {
      "epoch": 0.15729714760454167,
      "grad_norm": 3.8831708431243896,
      "learning_rate": 0.00018973842735031622,
      "loss": 2.8965,
      "step": 568
    },
    {
      "epoch": 0.157574079202437,
      "grad_norm": 3.5270354747772217,
      "learning_rate": 0.00018969946439550148,
      "loss": 2.5994,
      "step": 569
    },
    {
      "epoch": 0.15785101080033231,
      "grad_norm": 4.177906036376953,
      "learning_rate": 0.00018966043162618873,
      "loss": 2.9521,
      "step": 570
    },
    {
      "epoch": 0.15812794239822764,
      "grad_norm": 2.400700807571411,
      "learning_rate": 0.00018962132907275776,
      "loss": 2.9044,
      "step": 571
    },
    {
      "epoch": 0.15840487399612296,
      "grad_norm": 3.270749568939209,
      "learning_rate": 0.00018958215676564275,
      "loss": 3.0326,
      "step": 572
    },
    {
      "epoch": 0.1586818055940183,
      "grad_norm": 3.104177713394165,
      "learning_rate": 0.00018954291473533206,
      "loss": 2.8663,
      "step": 573
    },
    {
      "epoch": 0.15895873719191358,
      "grad_norm": 3.4584054946899414,
      "learning_rate": 0.0001895036030123684,
      "loss": 2.9584,
      "step": 574
    },
    {
      "epoch": 0.1592356687898089,
      "grad_norm": 4.127494812011719,
      "learning_rate": 0.00018946422162734872,
      "loss": 3.607,
      "step": 575
    },
    {
      "epoch": 0.15951260038770423,
      "grad_norm": 1.8092525005340576,
      "learning_rate": 0.00018942477061092413,
      "loss": 2.6754,
      "step": 576
    },
    {
      "epoch": 0.15978953198559956,
      "grad_norm": 2.801778554916382,
      "learning_rate": 0.0001893852499938,
      "loss": 2.9081,
      "step": 577
    },
    {
      "epoch": 0.16006646358349488,
      "grad_norm": 2.5070884227752686,
      "learning_rate": 0.00018934565980673583,
      "loss": 3.2574,
      "step": 578
    },
    {
      "epoch": 0.1603433951813902,
      "grad_norm": 1.5043915510177612,
      "learning_rate": 0.0001893060000805453,
      "loss": 3.0069,
      "step": 579
    },
    {
      "epoch": 0.1606203267792855,
      "grad_norm": 2.809873580932617,
      "learning_rate": 0.0001892662708460962,
      "loss": 2.6756,
      "step": 580
    },
    {
      "epoch": 0.16089725837718083,
      "grad_norm": 2.6917896270751953,
      "learning_rate": 0.0001892264721343104,
      "loss": 2.8422,
      "step": 581
    },
    {
      "epoch": 0.16117418997507615,
      "grad_norm": 2.8489990234375,
      "learning_rate": 0.00018918660397616388,
      "loss": 2.8668,
      "step": 582
    },
    {
      "epoch": 0.16145112157297148,
      "grad_norm": 1.7083117961883545,
      "learning_rate": 0.0001891466664026866,
      "loss": 2.8933,
      "step": 583
    },
    {
      "epoch": 0.1617280531708668,
      "grad_norm": 2.7621779441833496,
      "learning_rate": 0.00018910665944496264,
      "loss": 3.0861,
      "step": 584
    },
    {
      "epoch": 0.16200498476876213,
      "grad_norm": 1.6034456491470337,
      "learning_rate": 0.00018906658313413006,
      "loss": 3.081,
      "step": 585
    },
    {
      "epoch": 0.16228191636665742,
      "grad_norm": 2.1977531909942627,
      "learning_rate": 0.00018902643750138084,
      "loss": 2.7185,
      "step": 586
    },
    {
      "epoch": 0.16255884796455275,
      "grad_norm": 2.6708223819732666,
      "learning_rate": 0.00018898622257796098,
      "loss": 2.8751,
      "step": 587
    },
    {
      "epoch": 0.16283577956244807,
      "grad_norm": 2.5136728286743164,
      "learning_rate": 0.0001889459383951704,
      "loss": 3.0376,
      "step": 588
    },
    {
      "epoch": 0.1631127111603434,
      "grad_norm": 2.253354549407959,
      "learning_rate": 0.00018890558498436282,
      "loss": 2.9694,
      "step": 589
    },
    {
      "epoch": 0.16338964275823872,
      "grad_norm": 2.0030698776245117,
      "learning_rate": 0.00018886516237694607,
      "loss": 2.9248,
      "step": 590
    },
    {
      "epoch": 0.16366657435613405,
      "grad_norm": 3.905414342880249,
      "learning_rate": 0.0001888246706043816,
      "loss": 3.0461,
      "step": 591
    },
    {
      "epoch": 0.16394350595402934,
      "grad_norm": 3.2316644191741943,
      "learning_rate": 0.00018878410969818484,
      "loss": 2.9037,
      "step": 592
    },
    {
      "epoch": 0.16422043755192467,
      "grad_norm": 3.1154937744140625,
      "learning_rate": 0.0001887434796899249,
      "loss": 2.8276,
      "step": 593
    },
    {
      "epoch": 0.16449736914982,
      "grad_norm": 2.8231682777404785,
      "learning_rate": 0.00018870278061122484,
      "loss": 2.9152,
      "step": 594
    },
    {
      "epoch": 0.16477430074771532,
      "grad_norm": 2.0204408168792725,
      "learning_rate": 0.00018866201249376137,
      "loss": 2.7603,
      "step": 595
    },
    {
      "epoch": 0.16505123234561064,
      "grad_norm": 2.0140509605407715,
      "learning_rate": 0.00018862117536926496,
      "loss": 2.5941,
      "step": 596
    },
    {
      "epoch": 0.16532816394350594,
      "grad_norm": 3.0981736183166504,
      "learning_rate": 0.00018858026926951973,
      "loss": 2.9203,
      "step": 597
    },
    {
      "epoch": 0.16560509554140126,
      "grad_norm": 3.244689464569092,
      "learning_rate": 0.0001885392942263636,
      "loss": 2.7306,
      "step": 598
    },
    {
      "epoch": 0.1658820271392966,
      "grad_norm": 5.177836894989014,
      "learning_rate": 0.00018849825027168803,
      "loss": 3.3335,
      "step": 599
    },
    {
      "epoch": 0.16615895873719191,
      "grad_norm": 5.458366870880127,
      "learning_rate": 0.00018845713743743822,
      "loss": 3.7187,
      "step": 600
    },
    {
      "epoch": 0.16643589033508724,
      "grad_norm": 2.1460788249969482,
      "learning_rate": 0.00018841595575561294,
      "loss": 2.7101,
      "step": 601
    },
    {
      "epoch": 0.16671282193298256,
      "grad_norm": 1.636173963546753,
      "learning_rate": 0.00018837470525826452,
      "loss": 2.7277,
      "step": 602
    },
    {
      "epoch": 0.16698975353087786,
      "grad_norm": 3.3788509368896484,
      "learning_rate": 0.00018833338597749882,
      "loss": 3.2524,
      "step": 603
    },
    {
      "epoch": 0.16726668512877318,
      "grad_norm": 3.1274211406707764,
      "learning_rate": 0.00018829199794547535,
      "loss": 3.0485,
      "step": 604
    },
    {
      "epoch": 0.1675436167266685,
      "grad_norm": 3.001844644546509,
      "learning_rate": 0.00018825054119440705,
      "loss": 2.8739,
      "step": 605
    },
    {
      "epoch": 0.16782054832456383,
      "grad_norm": 2.3479843139648438,
      "learning_rate": 0.00018820901575656034,
      "loss": 2.9218,
      "step": 606
    },
    {
      "epoch": 0.16809747992245916,
      "grad_norm": 3.5614748001098633,
      "learning_rate": 0.00018816742166425515,
      "loss": 3.1383,
      "step": 607
    },
    {
      "epoch": 0.16837441152035448,
      "grad_norm": 2.6405105590820312,
      "learning_rate": 0.00018812575894986474,
      "loss": 2.7645,
      "step": 608
    },
    {
      "epoch": 0.16865134311824978,
      "grad_norm": 3.3058698177337646,
      "learning_rate": 0.00018808402764581596,
      "loss": 2.9221,
      "step": 609
    },
    {
      "epoch": 0.1689282747161451,
      "grad_norm": 3.376081705093384,
      "learning_rate": 0.00018804222778458885,
      "loss": 2.8451,
      "step": 610
    },
    {
      "epoch": 0.16920520631404043,
      "grad_norm": 2.419116497039795,
      "learning_rate": 0.00018800035939871697,
      "loss": 3.0378,
      "step": 611
    },
    {
      "epoch": 0.16948213791193575,
      "grad_norm": 3.627453327178955,
      "learning_rate": 0.00018795842252078705,
      "loss": 2.9525,
      "step": 612
    },
    {
      "epoch": 0.16975906950983108,
      "grad_norm": 2.9806618690490723,
      "learning_rate": 0.0001879164171834393,
      "loss": 3.2352,
      "step": 613
    },
    {
      "epoch": 0.1700360011077264,
      "grad_norm": 3.6521472930908203,
      "learning_rate": 0.0001878743434193671,
      "loss": 3.1307,
      "step": 614
    },
    {
      "epoch": 0.1703129327056217,
      "grad_norm": 2.553753137588501,
      "learning_rate": 0.0001878322012613171,
      "loss": 2.8143,
      "step": 615
    },
    {
      "epoch": 0.17058986430351702,
      "grad_norm": 3.0864615440368652,
      "learning_rate": 0.00018778999074208925,
      "loss": 2.8784,
      "step": 616
    },
    {
      "epoch": 0.17086679590141235,
      "grad_norm": 2.491281509399414,
      "learning_rate": 0.00018774771189453658,
      "loss": 2.7147,
      "step": 617
    },
    {
      "epoch": 0.17114372749930767,
      "grad_norm": 3.605449914932251,
      "learning_rate": 0.0001877053647515655,
      "loss": 3.0916,
      "step": 618
    },
    {
      "epoch": 0.171420659097203,
      "grad_norm": 2.559891939163208,
      "learning_rate": 0.00018766294934613535,
      "loss": 2.5182,
      "step": 619
    },
    {
      "epoch": 0.17169759069509832,
      "grad_norm": 2.9058518409729004,
      "learning_rate": 0.00018762046571125872,
      "loss": 2.8692,
      "step": 620
    },
    {
      "epoch": 0.17197452229299362,
      "grad_norm": 2.3772425651550293,
      "learning_rate": 0.00018757791388000135,
      "loss": 2.7498,
      "step": 621
    },
    {
      "epoch": 0.17225145389088894,
      "grad_norm": 2.9739794731140137,
      "learning_rate": 0.00018753529388548194,
      "loss": 2.9448,
      "step": 622
    },
    {
      "epoch": 0.17252838548878427,
      "grad_norm": 4.4551215171813965,
      "learning_rate": 0.00018749260576087228,
      "loss": 3.3577,
      "step": 623
    },
    {
      "epoch": 0.1728053170866796,
      "grad_norm": 2.452760934829712,
      "learning_rate": 0.00018744984953939726,
      "loss": 3.1046,
      "step": 624
    },
    {
      "epoch": 0.17308224868457492,
      "grad_norm": 4.560608386993408,
      "learning_rate": 0.00018740702525433468,
      "loss": 3.9237,
      "step": 625
    },
    {
      "epoch": 0.17335918028247024,
      "grad_norm": 1.832899808883667,
      "learning_rate": 0.0001873641329390154,
      "loss": 2.9019,
      "step": 626
    },
    {
      "epoch": 0.17363611188036554,
      "grad_norm": 3.598841428756714,
      "learning_rate": 0.00018732117262682313,
      "loss": 3.0795,
      "step": 627
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 2.3927810192108154,
      "learning_rate": 0.00018727814435119458,
      "loss": 2.8281,
      "step": 628
    },
    {
      "epoch": 0.1741899750761562,
      "grad_norm": 1.6871562004089355,
      "learning_rate": 0.0001872350481456193,
      "loss": 2.709,
      "step": 629
    },
    {
      "epoch": 0.17446690667405151,
      "grad_norm": 1.5506254434585571,
      "learning_rate": 0.0001871918840436398,
      "loss": 2.4343,
      "step": 630
    },
    {
      "epoch": 0.17474383827194684,
      "grad_norm": 1.5275046825408936,
      "learning_rate": 0.0001871486520788513,
      "loss": 2.6755,
      "step": 631
    },
    {
      "epoch": 0.17502076986984214,
      "grad_norm": 1.5423846244812012,
      "learning_rate": 0.000187105352284902,
      "loss": 2.5921,
      "step": 632
    },
    {
      "epoch": 0.17529770146773746,
      "grad_norm": 1.8302452564239502,
      "learning_rate": 0.0001870619846954927,
      "loss": 3.0291,
      "step": 633
    },
    {
      "epoch": 0.17557463306563278,
      "grad_norm": 2.1343085765838623,
      "learning_rate": 0.0001870185493443772,
      "loss": 2.7965,
      "step": 634
    },
    {
      "epoch": 0.1758515646635281,
      "grad_norm": 2.0958521366119385,
      "learning_rate": 0.00018697504626536185,
      "loss": 2.6242,
      "step": 635
    },
    {
      "epoch": 0.17612849626142343,
      "grad_norm": 1.856468915939331,
      "learning_rate": 0.00018693147549230576,
      "loss": 2.7097,
      "step": 636
    },
    {
      "epoch": 0.17640542785931876,
      "grad_norm": 1.8141541481018066,
      "learning_rate": 0.00018688783705912075,
      "loss": 2.7367,
      "step": 637
    },
    {
      "epoch": 0.17668235945721406,
      "grad_norm": 2.124817132949829,
      "learning_rate": 0.00018684413099977136,
      "loss": 2.8737,
      "step": 638
    },
    {
      "epoch": 0.17695929105510938,
      "grad_norm": 3.436328649520874,
      "learning_rate": 0.0001868003573482746,
      "loss": 3.1817,
      "step": 639
    },
    {
      "epoch": 0.1772362226530047,
      "grad_norm": 1.8416308164596558,
      "learning_rate": 0.0001867565161387003,
      "loss": 3.2015,
      "step": 640
    },
    {
      "epoch": 0.17751315425090003,
      "grad_norm": 3.4383628368377686,
      "learning_rate": 0.00018671260740517066,
      "loss": 2.7034,
      "step": 641
    },
    {
      "epoch": 0.17779008584879535,
      "grad_norm": 3.328667163848877,
      "learning_rate": 0.00018666863118186057,
      "loss": 3.1435,
      "step": 642
    },
    {
      "epoch": 0.17806701744669068,
      "grad_norm": 2.535160541534424,
      "learning_rate": 0.00018662458750299743,
      "loss": 2.6899,
      "step": 643
    },
    {
      "epoch": 0.17834394904458598,
      "grad_norm": 3.590573787689209,
      "learning_rate": 0.0001865804764028611,
      "loss": 2.64,
      "step": 644
    },
    {
      "epoch": 0.1786208806424813,
      "grad_norm": 2.0103766918182373,
      "learning_rate": 0.000186536297915784,
      "loss": 2.9379,
      "step": 645
    },
    {
      "epoch": 0.17889781224037662,
      "grad_norm": 1.850217342376709,
      "learning_rate": 0.00018649205207615084,
      "loss": 2.9077,
      "step": 646
    },
    {
      "epoch": 0.17917474383827195,
      "grad_norm": 3.9196999073028564,
      "learning_rate": 0.00018644773891839892,
      "loss": 3.0329,
      "step": 647
    },
    {
      "epoch": 0.17945167543616727,
      "grad_norm": 4.188611030578613,
      "learning_rate": 0.00018640335847701787,
      "loss": 2.9961,
      "step": 648
    },
    {
      "epoch": 0.1797286070340626,
      "grad_norm": 3.173583745956421,
      "learning_rate": 0.0001863589107865496,
      "loss": 3.0233,
      "step": 649
    },
    {
      "epoch": 0.1800055386319579,
      "grad_norm": 4.154677391052246,
      "learning_rate": 0.00018631439588158856,
      "loss": 3.8583,
      "step": 650
    },
    {
      "epoch": 0.18028247022985322,
      "grad_norm": 2.180216073989868,
      "learning_rate": 0.00018626981379678132,
      "loss": 2.9147,
      "step": 651
    },
    {
      "epoch": 0.18055940182774854,
      "grad_norm": 1.7871785163879395,
      "learning_rate": 0.00018622516456682684,
      "loss": 2.7395,
      "step": 652
    },
    {
      "epoch": 0.18083633342564387,
      "grad_norm": 1.6536489725112915,
      "learning_rate": 0.00018618044822647632,
      "loss": 2.7182,
      "step": 653
    },
    {
      "epoch": 0.1811132650235392,
      "grad_norm": 2.316704273223877,
      "learning_rate": 0.00018613566481053315,
      "loss": 2.7115,
      "step": 654
    },
    {
      "epoch": 0.18139019662143452,
      "grad_norm": 1.456023931503296,
      "learning_rate": 0.00018609081435385302,
      "loss": 2.9228,
      "step": 655
    },
    {
      "epoch": 0.18166712821932982,
      "grad_norm": 2.3114585876464844,
      "learning_rate": 0.00018604589689134372,
      "loss": 2.5995,
      "step": 656
    },
    {
      "epoch": 0.18194405981722514,
      "grad_norm": 1.963617205619812,
      "learning_rate": 0.0001860009124579652,
      "loss": 2.5466,
      "step": 657
    },
    {
      "epoch": 0.18222099141512046,
      "grad_norm": 1.4744776487350464,
      "learning_rate": 0.00018595586108872953,
      "loss": 2.7022,
      "step": 658
    },
    {
      "epoch": 0.1824979230130158,
      "grad_norm": 2.431297540664673,
      "learning_rate": 0.00018591074281870099,
      "loss": 2.7521,
      "step": 659
    },
    {
      "epoch": 0.1827748546109111,
      "grad_norm": 2.815333843231201,
      "learning_rate": 0.00018586555768299571,
      "loss": 2.9846,
      "step": 660
    },
    {
      "epoch": 0.18305178620880644,
      "grad_norm": 2.565122365951538,
      "learning_rate": 0.00018582030571678208,
      "loss": 2.789,
      "step": 661
    },
    {
      "epoch": 0.18332871780670174,
      "grad_norm": 2.2769696712493896,
      "learning_rate": 0.0001857749869552804,
      "loss": 2.9159,
      "step": 662
    },
    {
      "epoch": 0.18360564940459706,
      "grad_norm": 2.010282039642334,
      "learning_rate": 0.0001857296014337629,
      "loss": 2.7838,
      "step": 663
    },
    {
      "epoch": 0.18388258100249238,
      "grad_norm": 3.214738368988037,
      "learning_rate": 0.00018568414918755397,
      "loss": 3.1515,
      "step": 664
    },
    {
      "epoch": 0.1841595126003877,
      "grad_norm": 4.236786365509033,
      "learning_rate": 0.0001856386302520297,
      "loss": 3.2493,
      "step": 665
    },
    {
      "epoch": 0.18443644419828303,
      "grad_norm": 2.729323625564575,
      "learning_rate": 0.00018559304466261823,
      "loss": 2.7869,
      "step": 666
    },
    {
      "epoch": 0.18471337579617833,
      "grad_norm": 2.8418619632720947,
      "learning_rate": 0.0001855473924547995,
      "loss": 2.7512,
      "step": 667
    },
    {
      "epoch": 0.18499030739407366,
      "grad_norm": 2.5104856491088867,
      "learning_rate": 0.00018550167366410543,
      "loss": 2.7103,
      "step": 668
    },
    {
      "epoch": 0.18526723899196898,
      "grad_norm": 2.7290761470794678,
      "learning_rate": 0.00018545588832611956,
      "loss": 2.9332,
      "step": 669
    },
    {
      "epoch": 0.1855441705898643,
      "grad_norm": 2.2149202823638916,
      "learning_rate": 0.00018541003647647744,
      "loss": 2.5608,
      "step": 670
    },
    {
      "epoch": 0.18582110218775963,
      "grad_norm": 2.0544304847717285,
      "learning_rate": 0.0001853641181508662,
      "loss": 2.775,
      "step": 671
    },
    {
      "epoch": 0.18609803378565495,
      "grad_norm": 2.9898555278778076,
      "learning_rate": 0.0001853181333850248,
      "loss": 3.0448,
      "step": 672
    },
    {
      "epoch": 0.18637496538355025,
      "grad_norm": 2.771472215652466,
      "learning_rate": 0.00018527208221474396,
      "loss": 2.99,
      "step": 673
    },
    {
      "epoch": 0.18665189698144558,
      "grad_norm": 2.860121726989746,
      "learning_rate": 0.00018522596467586598,
      "loss": 2.9894,
      "step": 674
    },
    {
      "epoch": 0.1869288285793409,
      "grad_norm": 3.368523120880127,
      "learning_rate": 0.00018517978080428486,
      "loss": 3.1418,
      "step": 675
    },
    {
      "epoch": 0.18720576017723622,
      "grad_norm": 1.9053947925567627,
      "learning_rate": 0.00018513353063594627,
      "loss": 2.9306,
      "step": 676
    },
    {
      "epoch": 0.18748269177513155,
      "grad_norm": 3.7781260013580322,
      "learning_rate": 0.00018508721420684742,
      "loss": 3.1047,
      "step": 677
    },
    {
      "epoch": 0.18775962337302687,
      "grad_norm": 2.74504017829895,
      "learning_rate": 0.0001850408315530371,
      "loss": 2.5796,
      "step": 678
    },
    {
      "epoch": 0.18803655497092217,
      "grad_norm": 3.8274240493774414,
      "learning_rate": 0.00018499438271061568,
      "loss": 3.0927,
      "step": 679
    },
    {
      "epoch": 0.1883134865688175,
      "grad_norm": 2.835418701171875,
      "learning_rate": 0.00018494786771573496,
      "loss": 3.0944,
      "step": 680
    },
    {
      "epoch": 0.18859041816671282,
      "grad_norm": 1.6601074934005737,
      "learning_rate": 0.00018490128660459835,
      "loss": 2.8352,
      "step": 681
    },
    {
      "epoch": 0.18886734976460814,
      "grad_norm": 1.8874528408050537,
      "learning_rate": 0.00018485463941346066,
      "loss": 2.9458,
      "step": 682
    },
    {
      "epoch": 0.18914428136250347,
      "grad_norm": 2.7685647010803223,
      "learning_rate": 0.00018480792617862814,
      "loss": 2.9203,
      "step": 683
    },
    {
      "epoch": 0.1894212129603988,
      "grad_norm": 2.1622962951660156,
      "learning_rate": 0.0001847611469364584,
      "loss": 2.8604,
      "step": 684
    },
    {
      "epoch": 0.1896981445582941,
      "grad_norm": 2.5799427032470703,
      "learning_rate": 0.0001847143017233604,
      "loss": 2.7162,
      "step": 685
    },
    {
      "epoch": 0.18997507615618942,
      "grad_norm": 1.5721246004104614,
      "learning_rate": 0.00018466739057579462,
      "loss": 3.0661,
      "step": 686
    },
    {
      "epoch": 0.19025200775408474,
      "grad_norm": 2.5846004486083984,
      "learning_rate": 0.00018462041353027266,
      "loss": 3.0392,
      "step": 687
    },
    {
      "epoch": 0.19052893935198006,
      "grad_norm": 2.5718348026275635,
      "learning_rate": 0.00018457337062335752,
      "loss": 2.6367,
      "step": 688
    },
    {
      "epoch": 0.1908058709498754,
      "grad_norm": 2.7954459190368652,
      "learning_rate": 0.00018452626189166345,
      "loss": 3.039,
      "step": 689
    },
    {
      "epoch": 0.1910828025477707,
      "grad_norm": 1.9670569896697998,
      "learning_rate": 0.00018447908737185585,
      "loss": 2.9294,
      "step": 690
    },
    {
      "epoch": 0.191359734145666,
      "grad_norm": 2.261406421661377,
      "learning_rate": 0.00018443184710065146,
      "loss": 2.8604,
      "step": 691
    },
    {
      "epoch": 0.19163666574356134,
      "grad_norm": 3.316488742828369,
      "learning_rate": 0.00018438454111481808,
      "loss": 2.6537,
      "step": 692
    },
    {
      "epoch": 0.19191359734145666,
      "grad_norm": 3.1909279823303223,
      "learning_rate": 0.00018433716945117474,
      "loss": 2.8554,
      "step": 693
    },
    {
      "epoch": 0.19219052893935198,
      "grad_norm": 1.9768377542495728,
      "learning_rate": 0.0001842897321465915,
      "loss": 2.7122,
      "step": 694
    },
    {
      "epoch": 0.1924674605372473,
      "grad_norm": 3.4397292137145996,
      "learning_rate": 0.00018424222923798958,
      "loss": 2.492,
      "step": 695
    },
    {
      "epoch": 0.1927443921351426,
      "grad_norm": 1.6669995784759521,
      "learning_rate": 0.0001841946607623412,
      "loss": 2.6234,
      "step": 696
    },
    {
      "epoch": 0.19302132373303793,
      "grad_norm": 2.8505983352661133,
      "learning_rate": 0.00018414702675666975,
      "loss": 2.7902,
      "step": 697
    },
    {
      "epoch": 0.19329825533093326,
      "grad_norm": 3.3573968410491943,
      "learning_rate": 0.0001840993272580494,
      "loss": 2.9162,
      "step": 698
    },
    {
      "epoch": 0.19357518692882858,
      "grad_norm": 6.864968299865723,
      "learning_rate": 0.0001840515623036055,
      "loss": 3.9297,
      "step": 699
    },
    {
      "epoch": 0.1938521185267239,
      "grad_norm": 9.950575828552246,
      "learning_rate": 0.00018400373193051424,
      "loss": 4.8462,
      "step": 700
    },
    {
      "epoch": 0.19412905012461923,
      "grad_norm": 1.9874237775802612,
      "learning_rate": 0.0001839558361760027,
      "loss": 2.9369,
      "step": 701
    },
    {
      "epoch": 0.19440598172251453,
      "grad_norm": 2.468550443649292,
      "learning_rate": 0.000183907875077349,
      "loss": 2.9755,
      "step": 702
    },
    {
      "epoch": 0.19468291332040985,
      "grad_norm": 1.8362083435058594,
      "learning_rate": 0.00018385984867188193,
      "loss": 2.7249,
      "step": 703
    },
    {
      "epoch": 0.19495984491830518,
      "grad_norm": 3.345048189163208,
      "learning_rate": 0.0001838117569969812,
      "loss": 3.008,
      "step": 704
    },
    {
      "epoch": 0.1952367765162005,
      "grad_norm": 2.1463112831115723,
      "learning_rate": 0.00018376360009007736,
      "loss": 2.7318,
      "step": 705
    },
    {
      "epoch": 0.19551370811409582,
      "grad_norm": 1.7631304264068604,
      "learning_rate": 0.00018371537798865162,
      "loss": 2.7002,
      "step": 706
    },
    {
      "epoch": 0.19579063971199115,
      "grad_norm": 2.6849911212921143,
      "learning_rate": 0.00018366709073023606,
      "loss": 2.9509,
      "step": 707
    },
    {
      "epoch": 0.19606757130988645,
      "grad_norm": 2.5786337852478027,
      "learning_rate": 0.0001836187383524134,
      "loss": 3.0554,
      "step": 708
    },
    {
      "epoch": 0.19634450290778177,
      "grad_norm": 12.970073699951172,
      "learning_rate": 0.00018357032089281702,
      "loss": 2.7985,
      "step": 709
    },
    {
      "epoch": 0.1966214345056771,
      "grad_norm": 2.3600611686706543,
      "learning_rate": 0.00018352183838913098,
      "loss": 2.8667,
      "step": 710
    },
    {
      "epoch": 0.19689836610357242,
      "grad_norm": 2.256000518798828,
      "learning_rate": 0.00018347329087909002,
      "loss": 2.8481,
      "step": 711
    },
    {
      "epoch": 0.19717529770146774,
      "grad_norm": 2.0184097290039062,
      "learning_rate": 0.00018342467840047939,
      "loss": 3.0471,
      "step": 712
    },
    {
      "epoch": 0.19745222929936307,
      "grad_norm": 3.4170594215393066,
      "learning_rate": 0.00018337600099113495,
      "loss": 2.9388,
      "step": 713
    },
    {
      "epoch": 0.19772916089725837,
      "grad_norm": 2.170293092727661,
      "learning_rate": 0.00018332725868894313,
      "loss": 2.9068,
      "step": 714
    },
    {
      "epoch": 0.1980060924951537,
      "grad_norm": 2.9727580547332764,
      "learning_rate": 0.00018327845153184077,
      "loss": 3.1206,
      "step": 715
    },
    {
      "epoch": 0.19828302409304902,
      "grad_norm": 5.886566638946533,
      "learning_rate": 0.00018322957955781526,
      "loss": 3.09,
      "step": 716
    },
    {
      "epoch": 0.19855995569094434,
      "grad_norm": 11.844672203063965,
      "learning_rate": 0.0001831806428049044,
      "loss": 2.7557,
      "step": 717
    },
    {
      "epoch": 0.19883688728883966,
      "grad_norm": 3.298863172531128,
      "learning_rate": 0.00018313164131119654,
      "loss": 2.6253,
      "step": 718
    },
    {
      "epoch": 0.199113818886735,
      "grad_norm": 18.608060836791992,
      "learning_rate": 0.00018308257511483018,
      "loss": 3.1732,
      "step": 719
    },
    {
      "epoch": 0.19939075048463029,
      "grad_norm": 2.1521401405334473,
      "learning_rate": 0.00018303344425399436,
      "loss": 2.6386,
      "step": 720
    },
    {
      "epoch": 0.1996676820825256,
      "grad_norm": 2.3799684047698975,
      "learning_rate": 0.00018298424876692844,
      "loss": 2.761,
      "step": 721
    },
    {
      "epoch": 0.19994461368042094,
      "grad_norm": 3.3127503395080566,
      "learning_rate": 0.00018293498869192197,
      "loss": 2.8244,
      "step": 722
    },
    {
      "epoch": 0.20022154527831626,
      "grad_norm": 2.539456844329834,
      "learning_rate": 0.00018288566406731484,
      "loss": 2.4344,
      "step": 723
    },
    {
      "epoch": 0.20049847687621158,
      "grad_norm": 2.227818727493286,
      "learning_rate": 0.00018283627493149721,
      "loss": 2.816,
      "step": 724
    },
    {
      "epoch": 0.2007754084741069,
      "grad_norm": 6.158308029174805,
      "learning_rate": 0.00018278682132290944,
      "loss": 4.2033,
      "step": 725
    },
    {
      "epoch": 0.2010523400720022,
      "grad_norm": 2.4608898162841797,
      "learning_rate": 0.000182737303280042,
      "loss": 2.7077,
      "step": 726
    },
    {
      "epoch": 0.20132927166989753,
      "grad_norm": 3.126945734024048,
      "learning_rate": 0.00018268772084143562,
      "loss": 3.1508,
      "step": 727
    },
    {
      "epoch": 0.20160620326779286,
      "grad_norm": 1.4004101753234863,
      "learning_rate": 0.000182638074045681,
      "loss": 2.542,
      "step": 728
    },
    {
      "epoch": 0.20188313486568818,
      "grad_norm": 6.214250564575195,
      "learning_rate": 0.00018258836293141907,
      "loss": 2.8699,
      "step": 729
    },
    {
      "epoch": 0.2021600664635835,
      "grad_norm": 2.3528897762298584,
      "learning_rate": 0.0001825385875373408,
      "loss": 2.7283,
      "step": 730
    },
    {
      "epoch": 0.2024369980614788,
      "grad_norm": 3.6349143981933594,
      "learning_rate": 0.00018248874790218706,
      "loss": 3.2385,
      "step": 731
    },
    {
      "epoch": 0.20271392965937413,
      "grad_norm": 2.3201217651367188,
      "learning_rate": 0.0001824388440647489,
      "loss": 2.6714,
      "step": 732
    },
    {
      "epoch": 0.20299086125726945,
      "grad_norm": 2.7398152351379395,
      "learning_rate": 0.00018238887606386727,
      "loss": 2.8433,
      "step": 733
    },
    {
      "epoch": 0.20326779285516478,
      "grad_norm": 3.190858840942383,
      "learning_rate": 0.000182338843938433,
      "loss": 2.8483,
      "step": 734
    },
    {
      "epoch": 0.2035447244530601,
      "grad_norm": 4.260762691497803,
      "learning_rate": 0.0001822887477273869,
      "loss": 3.0106,
      "step": 735
    },
    {
      "epoch": 0.20382165605095542,
      "grad_norm": 1.6161547899246216,
      "learning_rate": 0.0001822385874697196,
      "loss": 2.6834,
      "step": 736
    },
    {
      "epoch": 0.20409858764885072,
      "grad_norm": 3.6264712810516357,
      "learning_rate": 0.0001821883632044717,
      "loss": 2.6701,
      "step": 737
    },
    {
      "epoch": 0.20437551924674605,
      "grad_norm": 6.149306297302246,
      "learning_rate": 0.00018213807497073346,
      "loss": 2.8069,
      "step": 738
    },
    {
      "epoch": 0.20465245084464137,
      "grad_norm": 3.0085160732269287,
      "learning_rate": 0.000182087722807645,
      "loss": 2.7487,
      "step": 739
    },
    {
      "epoch": 0.2049293824425367,
      "grad_norm": 3.4432058334350586,
      "learning_rate": 0.0001820373067543962,
      "loss": 3.0368,
      "step": 740
    },
    {
      "epoch": 0.20520631404043202,
      "grad_norm": 4.097099304199219,
      "learning_rate": 0.0001819868268502267,
      "loss": 3.002,
      "step": 741
    },
    {
      "epoch": 0.20548324563832734,
      "grad_norm": 3.168367624282837,
      "learning_rate": 0.00018193628313442576,
      "loss": 2.9108,
      "step": 742
    },
    {
      "epoch": 0.20576017723622264,
      "grad_norm": 3.36014986038208,
      "learning_rate": 0.00018188567564633238,
      "loss": 2.5995,
      "step": 743
    },
    {
      "epoch": 0.20603710883411797,
      "grad_norm": 2.911134719848633,
      "learning_rate": 0.00018183500442533514,
      "loss": 2.7216,
      "step": 744
    },
    {
      "epoch": 0.2063140404320133,
      "grad_norm": 2.1692538261413574,
      "learning_rate": 0.00018178426951087224,
      "loss": 2.9378,
      "step": 745
    },
    {
      "epoch": 0.20659097202990861,
      "grad_norm": 2.408252000808716,
      "learning_rate": 0.00018173347094243146,
      "loss": 2.9515,
      "step": 746
    },
    {
      "epoch": 0.20686790362780394,
      "grad_norm": 2.2805047035217285,
      "learning_rate": 0.00018168260875955013,
      "loss": 2.5644,
      "step": 747
    },
    {
      "epoch": 0.20714483522569926,
      "grad_norm": 3.825239658355713,
      "learning_rate": 0.00018163168300181507,
      "loss": 3.0186,
      "step": 748
    },
    {
      "epoch": 0.20742176682359456,
      "grad_norm": 6.781472682952881,
      "learning_rate": 0.00018158069370886266,
      "loss": 3.1843,
      "step": 749
    },
    {
      "epoch": 0.20769869842148989,
      "grad_norm": 6.31098747253418,
      "learning_rate": 0.00018152964092037855,
      "loss": 3.6861,
      "step": 750
    },
    {
      "epoch": 0.2079756300193852,
      "grad_norm": 5.607875823974609,
      "learning_rate": 0.00018147852467609803,
      "loss": 3.0506,
      "step": 751
    },
    {
      "epoch": 0.20825256161728053,
      "grad_norm": 2.671266794204712,
      "learning_rate": 0.0001814273450158056,
      "loss": 2.6357,
      "step": 752
    },
    {
      "epoch": 0.20852949321517586,
      "grad_norm": 2.7472641468048096,
      "learning_rate": 0.00018137610197933523,
      "loss": 3.0306,
      "step": 753
    },
    {
      "epoch": 0.20880642481307118,
      "grad_norm": 1.4980647563934326,
      "learning_rate": 0.0001813247956065702,
      "loss": 2.8052,
      "step": 754
    },
    {
      "epoch": 0.20908335641096648,
      "grad_norm": 2.9251890182495117,
      "learning_rate": 0.00018127342593744297,
      "loss": 3.1088,
      "step": 755
    },
    {
      "epoch": 0.2093602880088618,
      "grad_norm": 2.2277612686157227,
      "learning_rate": 0.00018122199301193548,
      "loss": 2.9202,
      "step": 756
    },
    {
      "epoch": 0.20963721960675713,
      "grad_norm": 2.4915273189544678,
      "learning_rate": 0.0001811704968700787,
      "loss": 3.0571,
      "step": 757
    },
    {
      "epoch": 0.20991415120465245,
      "grad_norm": 2.2147316932678223,
      "learning_rate": 0.0001811189375519529,
      "loss": 2.7547,
      "step": 758
    },
    {
      "epoch": 0.21019108280254778,
      "grad_norm": 1.6362735033035278,
      "learning_rate": 0.00018106731509768753,
      "loss": 2.7775,
      "step": 759
    },
    {
      "epoch": 0.2104680144004431,
      "grad_norm": 2.174014091491699,
      "learning_rate": 0.00018101562954746112,
      "loss": 2.7198,
      "step": 760
    },
    {
      "epoch": 0.2107449459983384,
      "grad_norm": 3.81546688079834,
      "learning_rate": 0.00018096388094150138,
      "loss": 2.9381,
      "step": 761
    },
    {
      "epoch": 0.21102187759623373,
      "grad_norm": 2.7602310180664062,
      "learning_rate": 0.00018091206932008502,
      "loss": 2.796,
      "step": 762
    },
    {
      "epoch": 0.21129880919412905,
      "grad_norm": 3.135514736175537,
      "learning_rate": 0.00018086019472353786,
      "loss": 2.7376,
      "step": 763
    },
    {
      "epoch": 0.21157574079202437,
      "grad_norm": 14.77597427368164,
      "learning_rate": 0.00018080825719223468,
      "loss": 2.8534,
      "step": 764
    },
    {
      "epoch": 0.2118526723899197,
      "grad_norm": 3.037137269973755,
      "learning_rate": 0.00018075625676659932,
      "loss": 3.238,
      "step": 765
    },
    {
      "epoch": 0.212129603987815,
      "grad_norm": 4.679478168487549,
      "learning_rate": 0.00018070419348710452,
      "loss": 3.0818,
      "step": 766
    },
    {
      "epoch": 0.21240653558571032,
      "grad_norm": 3.6584432125091553,
      "learning_rate": 0.00018065206739427192,
      "loss": 2.8912,
      "step": 767
    },
    {
      "epoch": 0.21268346718360565,
      "grad_norm": 7.0400166511535645,
      "learning_rate": 0.00018059987852867208,
      "loss": 3.1298,
      "step": 768
    },
    {
      "epoch": 0.21296039878150097,
      "grad_norm": 1.9162108898162842,
      "learning_rate": 0.00018054762693092444,
      "loss": 3.2406,
      "step": 769
    },
    {
      "epoch": 0.2132373303793963,
      "grad_norm": 2.555687189102173,
      "learning_rate": 0.00018049531264169723,
      "loss": 2.4545,
      "step": 770
    },
    {
      "epoch": 0.21351426197729162,
      "grad_norm": 2.5481579303741455,
      "learning_rate": 0.0001804429357017074,
      "loss": 2.9135,
      "step": 771
    },
    {
      "epoch": 0.21379119357518692,
      "grad_norm": 2.2174980640411377,
      "learning_rate": 0.0001803904961517209,
      "loss": 2.6785,
      "step": 772
    },
    {
      "epoch": 0.21406812517308224,
      "grad_norm": 2.345719814300537,
      "learning_rate": 0.00018033799403255208,
      "loss": 2.976,
      "step": 773
    },
    {
      "epoch": 0.21434505677097757,
      "grad_norm": 2.987143039703369,
      "learning_rate": 0.00018028542938506426,
      "loss": 3.103,
      "step": 774
    },
    {
      "epoch": 0.2146219883688729,
      "grad_norm": 5.510335922241211,
      "learning_rate": 0.00018023280225016932,
      "loss": 3.8119,
      "step": 775
    },
    {
      "epoch": 0.21489891996676821,
      "grad_norm": 1.4958183765411377,
      "learning_rate": 0.0001801801126688278,
      "loss": 2.6603,
      "step": 776
    },
    {
      "epoch": 0.21517585156466354,
      "grad_norm": 2.7116470336914062,
      "learning_rate": 0.0001801273606820488,
      "loss": 2.9782,
      "step": 777
    },
    {
      "epoch": 0.21545278316255884,
      "grad_norm": 1.8536920547485352,
      "learning_rate": 0.00018007454633089,
      "loss": 2.929,
      "step": 778
    },
    {
      "epoch": 0.21572971476045416,
      "grad_norm": 2.6260993480682373,
      "learning_rate": 0.0001800216696564576,
      "loss": 3.168,
      "step": 779
    },
    {
      "epoch": 0.21600664635834949,
      "grad_norm": 1.8232653141021729,
      "learning_rate": 0.00017996873069990647,
      "loss": 2.7345,
      "step": 780
    },
    {
      "epoch": 0.2162835779562448,
      "grad_norm": 2.8388843536376953,
      "learning_rate": 0.00017991572950243977,
      "loss": 2.6816,
      "step": 781
    },
    {
      "epoch": 0.21656050955414013,
      "grad_norm": 2.5359623432159424,
      "learning_rate": 0.0001798626661053091,
      "loss": 2.8546,
      "step": 782
    },
    {
      "epoch": 0.21683744115203546,
      "grad_norm": 2.4679372310638428,
      "learning_rate": 0.0001798095405498146,
      "loss": 2.8639,
      "step": 783
    },
    {
      "epoch": 0.21711437274993076,
      "grad_norm": 48.03520584106445,
      "learning_rate": 0.00017975635287730473,
      "loss": 2.8886,
      "step": 784
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 1.9362636804580688,
      "learning_rate": 0.00017970310312917623,
      "loss": 2.8487,
      "step": 785
    },
    {
      "epoch": 0.2176682359457214,
      "grad_norm": 1.8206771612167358,
      "learning_rate": 0.00017964979134687427,
      "loss": 2.8701,
      "step": 786
    },
    {
      "epoch": 0.21794516754361673,
      "grad_norm": 1.7026715278625488,
      "learning_rate": 0.00017959641757189223,
      "loss": 2.7412,
      "step": 787
    },
    {
      "epoch": 0.21822209914151205,
      "grad_norm": 1.8954119682312012,
      "learning_rate": 0.00017954298184577172,
      "loss": 2.9769,
      "step": 788
    },
    {
      "epoch": 0.21849903073940738,
      "grad_norm": 2.224543571472168,
      "learning_rate": 0.00017948948421010264,
      "loss": 2.9588,
      "step": 789
    },
    {
      "epoch": 0.21877596233730268,
      "grad_norm": 3.9762964248657227,
      "learning_rate": 0.00017943592470652303,
      "loss": 2.9722,
      "step": 790
    },
    {
      "epoch": 0.219052893935198,
      "grad_norm": 1.5863521099090576,
      "learning_rate": 0.00017938230337671908,
      "loss": 2.8169,
      "step": 791
    },
    {
      "epoch": 0.21932982553309333,
      "grad_norm": 1.8167140483856201,
      "learning_rate": 0.00017932862026242514,
      "loss": 2.6676,
      "step": 792
    },
    {
      "epoch": 0.21960675713098865,
      "grad_norm": 3.5854008197784424,
      "learning_rate": 0.00017927487540542357,
      "loss": 2.8992,
      "step": 793
    },
    {
      "epoch": 0.21988368872888397,
      "grad_norm": 2.874563694000244,
      "learning_rate": 0.00017922106884754488,
      "loss": 2.8493,
      "step": 794
    },
    {
      "epoch": 0.2201606203267793,
      "grad_norm": 3.0138399600982666,
      "learning_rate": 0.00017916720063066753,
      "loss": 2.931,
      "step": 795
    },
    {
      "epoch": 0.2204375519246746,
      "grad_norm": 1.9415334463119507,
      "learning_rate": 0.00017911327079671803,
      "loss": 2.7398,
      "step": 796
    },
    {
      "epoch": 0.22071448352256992,
      "grad_norm": 2.626664400100708,
      "learning_rate": 0.00017905927938767076,
      "loss": 2.8844,
      "step": 797
    },
    {
      "epoch": 0.22099141512046525,
      "grad_norm": 2.3523318767547607,
      "learning_rate": 0.00017900522644554812,
      "loss": 2.6385,
      "step": 798
    },
    {
      "epoch": 0.22126834671836057,
      "grad_norm": 3.625702142715454,
      "learning_rate": 0.0001789511120124203,
      "loss": 3.1983,
      "step": 799
    },
    {
      "epoch": 0.2215452783162559,
      "grad_norm": 6.103048324584961,
      "learning_rate": 0.00017889693613040554,
      "loss": 3.9482,
      "step": 800
    },
    {
      "epoch": 0.2218222099141512,
      "grad_norm": 2.523581027984619,
      "learning_rate": 0.00017884269884166967,
      "loss": 2.9803,
      "step": 801
    },
    {
      "epoch": 0.22209914151204652,
      "grad_norm": 1.7089117765426636,
      "learning_rate": 0.00017878840018842643,
      "loss": 2.8673,
      "step": 802
    },
    {
      "epoch": 0.22237607310994184,
      "grad_norm": 1.8813852071762085,
      "learning_rate": 0.0001787340402129374,
      "loss": 2.8749,
      "step": 803
    },
    {
      "epoch": 0.22265300470783717,
      "grad_norm": 1.4631633758544922,
      "learning_rate": 0.00017867961895751163,
      "loss": 2.9742,
      "step": 804
    },
    {
      "epoch": 0.2229299363057325,
      "grad_norm": 2.468569040298462,
      "learning_rate": 0.0001786251364645062,
      "loss": 2.7563,
      "step": 805
    },
    {
      "epoch": 0.22320686790362781,
      "grad_norm": 3.15724515914917,
      "learning_rate": 0.00017857059277632563,
      "loss": 2.9215,
      "step": 806
    },
    {
      "epoch": 0.2234837995015231,
      "grad_norm": 2.501721143722534,
      "learning_rate": 0.00017851598793542208,
      "loss": 2.7974,
      "step": 807
    },
    {
      "epoch": 0.22376073109941844,
      "grad_norm": 1.597951889038086,
      "learning_rate": 0.00017846132198429544,
      "loss": 2.8209,
      "step": 808
    },
    {
      "epoch": 0.22403766269731376,
      "grad_norm": 2.3806540966033936,
      "learning_rate": 0.00017840659496549298,
      "loss": 2.8041,
      "step": 809
    },
    {
      "epoch": 0.22431459429520909,
      "grad_norm": 1.3796292543411255,
      "learning_rate": 0.00017835180692160968,
      "loss": 2.7755,
      "step": 810
    },
    {
      "epoch": 0.2245915258931044,
      "grad_norm": 2.67073392868042,
      "learning_rate": 0.0001782969578952879,
      "loss": 2.9294,
      "step": 811
    },
    {
      "epoch": 0.22486845749099973,
      "grad_norm": 1.391492486000061,
      "learning_rate": 0.0001782420479292175,
      "loss": 2.7277,
      "step": 812
    },
    {
      "epoch": 0.22514538908889503,
      "grad_norm": 3.6642210483551025,
      "learning_rate": 0.00017818707706613576,
      "loss": 3.2001,
      "step": 813
    },
    {
      "epoch": 0.22542232068679036,
      "grad_norm": 2.9600884914398193,
      "learning_rate": 0.00017813204534882738,
      "loss": 3.121,
      "step": 814
    },
    {
      "epoch": 0.22569925228468568,
      "grad_norm": 2.7705891132354736,
      "learning_rate": 0.00017807695282012436,
      "loss": 2.9563,
      "step": 815
    },
    {
      "epoch": 0.225976183882581,
      "grad_norm": 1.836990475654602,
      "learning_rate": 0.00017802179952290613,
      "loss": 2.6764,
      "step": 816
    },
    {
      "epoch": 0.22625311548047633,
      "grad_norm": 3.026313066482544,
      "learning_rate": 0.00017796658550009933,
      "loss": 2.9005,
      "step": 817
    },
    {
      "epoch": 0.22653004707837165,
      "grad_norm": 2.420330047607422,
      "learning_rate": 0.00017791131079467794,
      "loss": 2.7694,
      "step": 818
    },
    {
      "epoch": 0.22680697867626695,
      "grad_norm": 3.7374370098114014,
      "learning_rate": 0.0001778559754496631,
      "loss": 2.9352,
      "step": 819
    },
    {
      "epoch": 0.22708391027416228,
      "grad_norm": 2.7599310874938965,
      "learning_rate": 0.00017780057950812317,
      "loss": 2.5512,
      "step": 820
    },
    {
      "epoch": 0.2273608418720576,
      "grad_norm": 2.0051608085632324,
      "learning_rate": 0.0001777451230131737,
      "loss": 2.9223,
      "step": 821
    },
    {
      "epoch": 0.22763777346995293,
      "grad_norm": 4.079291820526123,
      "learning_rate": 0.0001776896060079773,
      "loss": 2.6139,
      "step": 822
    },
    {
      "epoch": 0.22791470506784825,
      "grad_norm": 2.510545015335083,
      "learning_rate": 0.0001776340285357438,
      "loss": 2.4995,
      "step": 823
    },
    {
      "epoch": 0.22819163666574357,
      "grad_norm": 4.3523406982421875,
      "learning_rate": 0.00017757839063972997,
      "loss": 3.0332,
      "step": 824
    },
    {
      "epoch": 0.22846856826363887,
      "grad_norm": 4.616940975189209,
      "learning_rate": 0.00017752269236323966,
      "loss": 3.6388,
      "step": 825
    },
    {
      "epoch": 0.2287454998615342,
      "grad_norm": 2.210371494293213,
      "learning_rate": 0.00017746693374962372,
      "loss": 2.7925,
      "step": 826
    },
    {
      "epoch": 0.22902243145942952,
      "grad_norm": 2.476367950439453,
      "learning_rate": 0.00017741111484227996,
      "loss": 2.9856,
      "step": 827
    },
    {
      "epoch": 0.22929936305732485,
      "grad_norm": 2.4525411128997803,
      "learning_rate": 0.00017735523568465305,
      "loss": 2.6746,
      "step": 828
    },
    {
      "epoch": 0.22957629465522017,
      "grad_norm": 2.463634490966797,
      "learning_rate": 0.00017729929632023472,
      "loss": 2.7279,
      "step": 829
    },
    {
      "epoch": 0.22985322625311547,
      "grad_norm": 2.8449554443359375,
      "learning_rate": 0.00017724329679256337,
      "loss": 2.9282,
      "step": 830
    },
    {
      "epoch": 0.2301301578510108,
      "grad_norm": 2.233956813812256,
      "learning_rate": 0.00017718723714522434,
      "loss": 2.7956,
      "step": 831
    },
    {
      "epoch": 0.23040708944890612,
      "grad_norm": 2.3343842029571533,
      "learning_rate": 0.00017713111742184975,
      "loss": 2.9016,
      "step": 832
    },
    {
      "epoch": 0.23068402104680144,
      "grad_norm": 1.9796818494796753,
      "learning_rate": 0.0001770749376661184,
      "loss": 2.8013,
      "step": 833
    },
    {
      "epoch": 0.23096095264469677,
      "grad_norm": 1.3682773113250732,
      "learning_rate": 0.00017701869792175593,
      "loss": 2.8476,
      "step": 834
    },
    {
      "epoch": 0.2312378842425921,
      "grad_norm": 1.4512349367141724,
      "learning_rate": 0.00017696239823253459,
      "loss": 2.7705,
      "step": 835
    },
    {
      "epoch": 0.2315148158404874,
      "grad_norm": 1.6896343231201172,
      "learning_rate": 0.0001769060386422733,
      "loss": 2.7434,
      "step": 836
    },
    {
      "epoch": 0.2317917474383827,
      "grad_norm": 2.7096025943756104,
      "learning_rate": 0.00017684961919483763,
      "loss": 2.9079,
      "step": 837
    },
    {
      "epoch": 0.23206867903627804,
      "grad_norm": 1.879219651222229,
      "learning_rate": 0.0001767931399341397,
      "loss": 2.9284,
      "step": 838
    },
    {
      "epoch": 0.23234561063417336,
      "grad_norm": 3.0728893280029297,
      "learning_rate": 0.00017673660090413823,
      "loss": 2.9426,
      "step": 839
    },
    {
      "epoch": 0.23262254223206869,
      "grad_norm": 3.179382562637329,
      "learning_rate": 0.00017668000214883845,
      "loss": 3.0977,
      "step": 840
    },
    {
      "epoch": 0.232899473829964,
      "grad_norm": 2.8578603267669678,
      "learning_rate": 0.000176623343712292,
      "loss": 2.7335,
      "step": 841
    },
    {
      "epoch": 0.2331764054278593,
      "grad_norm": 2.0931589603424072,
      "learning_rate": 0.00017656662563859702,
      "loss": 2.8404,
      "step": 842
    },
    {
      "epoch": 0.23345333702575463,
      "grad_norm": 4.9795074462890625,
      "learning_rate": 0.00017650984797189814,
      "loss": 3.2182,
      "step": 843
    },
    {
      "epoch": 0.23373026862364996,
      "grad_norm": 2.5955379009246826,
      "learning_rate": 0.00017645301075638634,
      "loss": 2.7909,
      "step": 844
    },
    {
      "epoch": 0.23400720022154528,
      "grad_norm": 2.148406744003296,
      "learning_rate": 0.0001763961140362988,
      "loss": 2.4648,
      "step": 845
    },
    {
      "epoch": 0.2342841318194406,
      "grad_norm": 2.5193843841552734,
      "learning_rate": 0.00017633915785591917,
      "loss": 2.9867,
      "step": 846
    },
    {
      "epoch": 0.23456106341733593,
      "grad_norm": 2.263282537460327,
      "learning_rate": 0.00017628214225957737,
      "loss": 2.7114,
      "step": 847
    },
    {
      "epoch": 0.23483799501523123,
      "grad_norm": 3.8058855533599854,
      "learning_rate": 0.00017622506729164948,
      "loss": 3.3003,
      "step": 848
    },
    {
      "epoch": 0.23511492661312655,
      "grad_norm": 3.6773746013641357,
      "learning_rate": 0.00017616793299655794,
      "loss": 3.128,
      "step": 849
    },
    {
      "epoch": 0.23539185821102188,
      "grad_norm": 5.275571346282959,
      "learning_rate": 0.00017611073941877113,
      "loss": 4.1249,
      "step": 850
    },
    {
      "epoch": 0.2356687898089172,
      "grad_norm": 1.4605493545532227,
      "learning_rate": 0.00017605348660280383,
      "loss": 2.6483,
      "step": 851
    },
    {
      "epoch": 0.23594572140681253,
      "grad_norm": 2.3449642658233643,
      "learning_rate": 0.00017599617459321674,
      "loss": 2.9185,
      "step": 852
    },
    {
      "epoch": 0.23622265300470785,
      "grad_norm": 1.8987805843353271,
      "learning_rate": 0.00017593880343461668,
      "loss": 2.5391,
      "step": 853
    },
    {
      "epoch": 0.23649958460260315,
      "grad_norm": 1.6489249467849731,
      "learning_rate": 0.00017588137317165657,
      "loss": 2.575,
      "step": 854
    },
    {
      "epoch": 0.23677651620049847,
      "grad_norm": 1.4591307640075684,
      "learning_rate": 0.00017582388384903522,
      "loss": 2.9114,
      "step": 855
    },
    {
      "epoch": 0.2370534477983938,
      "grad_norm": 1.4331414699554443,
      "learning_rate": 0.00017576633551149752,
      "loss": 2.7333,
      "step": 856
    },
    {
      "epoch": 0.23733037939628912,
      "grad_norm": 1.4502758979797363,
      "learning_rate": 0.0001757087282038343,
      "loss": 2.5141,
      "step": 857
    },
    {
      "epoch": 0.23760731099418445,
      "grad_norm": 2.1150856018066406,
      "learning_rate": 0.00017565106197088207,
      "loss": 2.9157,
      "step": 858
    },
    {
      "epoch": 0.23788424259207977,
      "grad_norm": 3.100188732147217,
      "learning_rate": 0.0001755933368575235,
      "loss": 2.8012,
      "step": 859
    },
    {
      "epoch": 0.23816117418997507,
      "grad_norm": 1.8641769886016846,
      "learning_rate": 0.00017553555290868687,
      "loss": 2.5764,
      "step": 860
    },
    {
      "epoch": 0.2384381057878704,
      "grad_norm": 1.5892128944396973,
      "learning_rate": 0.00017547771016934643,
      "loss": 2.947,
      "step": 861
    },
    {
      "epoch": 0.23871503738576572,
      "grad_norm": 1.791627287864685,
      "learning_rate": 0.00017541980868452197,
      "loss": 2.9522,
      "step": 862
    },
    {
      "epoch": 0.23899196898366104,
      "grad_norm": 2.1791608333587646,
      "learning_rate": 0.0001753618484992792,
      "loss": 2.9263,
      "step": 863
    },
    {
      "epoch": 0.23926890058155637,
      "grad_norm": 2.1276986598968506,
      "learning_rate": 0.0001753038296587294,
      "loss": 3.0579,
      "step": 864
    },
    {
      "epoch": 0.23954583217945166,
      "grad_norm": 1.7338223457336426,
      "learning_rate": 0.00017524575220802962,
      "loss": 2.9095,
      "step": 865
    },
    {
      "epoch": 0.239822763777347,
      "grad_norm": 1.9278991222381592,
      "learning_rate": 0.00017518761619238234,
      "loss": 2.6981,
      "step": 866
    },
    {
      "epoch": 0.2400996953752423,
      "grad_norm": 3.1734070777893066,
      "learning_rate": 0.00017512942165703578,
      "loss": 3.0285,
      "step": 867
    },
    {
      "epoch": 0.24037662697313764,
      "grad_norm": 1.7613427639007568,
      "learning_rate": 0.00017507116864728367,
      "loss": 2.678,
      "step": 868
    },
    {
      "epoch": 0.24065355857103296,
      "grad_norm": 1.979413390159607,
      "learning_rate": 0.00017501285720846523,
      "loss": 2.7302,
      "step": 869
    },
    {
      "epoch": 0.24093049016892829,
      "grad_norm": 2.640336513519287,
      "learning_rate": 0.00017495448738596517,
      "loss": 2.6632,
      "step": 870
    },
    {
      "epoch": 0.24120742176682358,
      "grad_norm": 1.7245389223098755,
      "learning_rate": 0.00017489605922521361,
      "loss": 2.5472,
      "step": 871
    },
    {
      "epoch": 0.2414843533647189,
      "grad_norm": 3.25132155418396,
      "learning_rate": 0.00017483757277168618,
      "loss": 2.6675,
      "step": 872
    },
    {
      "epoch": 0.24176128496261423,
      "grad_norm": 2.5426268577575684,
      "learning_rate": 0.0001747790280709037,
      "loss": 2.8879,
      "step": 873
    },
    {
      "epoch": 0.24203821656050956,
      "grad_norm": 2.1168370246887207,
      "learning_rate": 0.0001747204251684325,
      "loss": 3.1219,
      "step": 874
    },
    {
      "epoch": 0.24231514815840488,
      "grad_norm": 4.783705711364746,
      "learning_rate": 0.0001746617641098841,
      "loss": 3.5755,
      "step": 875
    },
    {
      "epoch": 0.2425920797563002,
      "grad_norm": 3.421607494354248,
      "learning_rate": 0.00017460304494091534,
      "loss": 3.0924,
      "step": 876
    },
    {
      "epoch": 0.2428690113541955,
      "grad_norm": 1.700899600982666,
      "learning_rate": 0.00017454426770722824,
      "loss": 2.8364,
      "step": 877
    },
    {
      "epoch": 0.24314594295209083,
      "grad_norm": 1.693171501159668,
      "learning_rate": 0.00017448543245457009,
      "loss": 2.8619,
      "step": 878
    },
    {
      "epoch": 0.24342287454998615,
      "grad_norm": 2.3933541774749756,
      "learning_rate": 0.00017442653922873327,
      "loss": 2.9582,
      "step": 879
    },
    {
      "epoch": 0.24369980614788148,
      "grad_norm": 1.2178289890289307,
      "learning_rate": 0.00017436758807555527,
      "loss": 2.6739,
      "step": 880
    },
    {
      "epoch": 0.2439767377457768,
      "grad_norm": 2.796721935272217,
      "learning_rate": 0.00017430857904091873,
      "loss": 2.6545,
      "step": 881
    },
    {
      "epoch": 0.24425366934367213,
      "grad_norm": 3.1098473072052,
      "learning_rate": 0.0001742495121707513,
      "loss": 2.8777,
      "step": 882
    },
    {
      "epoch": 0.24453060094156742,
      "grad_norm": 2.6938257217407227,
      "learning_rate": 0.00017419038751102564,
      "loss": 2.8174,
      "step": 883
    },
    {
      "epoch": 0.24480753253946275,
      "grad_norm": 2.1703615188598633,
      "learning_rate": 0.0001741312051077594,
      "loss": 2.6357,
      "step": 884
    },
    {
      "epoch": 0.24508446413735807,
      "grad_norm": 1.8589909076690674,
      "learning_rate": 0.00017407196500701517,
      "loss": 2.5496,
      "step": 885
    },
    {
      "epoch": 0.2453613957352534,
      "grad_norm": 2.470015525817871,
      "learning_rate": 0.00017401266725490048,
      "loss": 3.1467,
      "step": 886
    },
    {
      "epoch": 0.24563832733314872,
      "grad_norm": 1.9733291864395142,
      "learning_rate": 0.00017395331189756763,
      "loss": 2.5779,
      "step": 887
    },
    {
      "epoch": 0.24591525893104405,
      "grad_norm": 5.104317665100098,
      "learning_rate": 0.00017389389898121388,
      "loss": 2.9302,
      "step": 888
    },
    {
      "epoch": 0.24619219052893934,
      "grad_norm": 1.8060482740402222,
      "learning_rate": 0.00017383442855208124,
      "loss": 2.8656,
      "step": 889
    },
    {
      "epoch": 0.24646912212683467,
      "grad_norm": 3.628758430480957,
      "learning_rate": 0.00017377490065645642,
      "loss": 3.1442,
      "step": 890
    },
    {
      "epoch": 0.24674605372473,
      "grad_norm": 2.087409019470215,
      "learning_rate": 0.000173715315340671,
      "loss": 2.8172,
      "step": 891
    },
    {
      "epoch": 0.24702298532262532,
      "grad_norm": 2.6525936126708984,
      "learning_rate": 0.00017365567265110107,
      "loss": 2.9967,
      "step": 892
    },
    {
      "epoch": 0.24729991692052064,
      "grad_norm": 1.745984673500061,
      "learning_rate": 0.0001735959726341675,
      "loss": 2.9043,
      "step": 893
    },
    {
      "epoch": 0.24757684851841597,
      "grad_norm": 2.4623262882232666,
      "learning_rate": 0.00017353621533633583,
      "loss": 2.6212,
      "step": 894
    },
    {
      "epoch": 0.24785378011631126,
      "grad_norm": 1.3718273639678955,
      "learning_rate": 0.000173476400804116,
      "loss": 2.7948,
      "step": 895
    },
    {
      "epoch": 0.2481307117142066,
      "grad_norm": 1.9852874279022217,
      "learning_rate": 0.0001734165290840626,
      "loss": 2.7263,
      "step": 896
    },
    {
      "epoch": 0.2484076433121019,
      "grad_norm": 2.4187746047973633,
      "learning_rate": 0.0001733566002227748,
      "loss": 2.8173,
      "step": 897
    },
    {
      "epoch": 0.24868457490999724,
      "grad_norm": 2.9589130878448486,
      "learning_rate": 0.00017329661426689609,
      "loss": 3.0224,
      "step": 898
    },
    {
      "epoch": 0.24896150650789256,
      "grad_norm": 1.7410316467285156,
      "learning_rate": 0.00017323657126311454,
      "loss": 2.9229,
      "step": 899
    },
    {
      "epoch": 0.24923843810578786,
      "grad_norm": 3.568751811981201,
      "learning_rate": 0.00017317647125816255,
      "loss": 3.2714,
      "step": 900
    },
    {
      "epoch": 0.24951536970368318,
      "grad_norm": 7.780263423919678,
      "learning_rate": 0.00017311631429881686,
      "loss": 2.8428,
      "step": 901
    },
    {
      "epoch": 0.2497923013015785,
      "grad_norm": 1.9568160772323608,
      "learning_rate": 0.0001730561004318986,
      "loss": 2.6005,
      "step": 902
    },
    {
      "epoch": 0.2500692328994738,
      "grad_norm": 1.8098704814910889,
      "learning_rate": 0.00017299582970427317,
      "loss": 2.7199,
      "step": 903
    },
    {
      "epoch": 0.2500692328994738,
      "eval_loss": 0.36362597346305847,
      "eval_runtime": 808.5036,
      "eval_samples_per_second": 7.523,
      "eval_steps_per_second": 1.881,
      "step": 903
    },
    {
      "epoch": 0.25034616449736913,
      "grad_norm": 2.204132556915283,
      "learning_rate": 0.0001729355021628502,
      "loss": 2.7864,
      "step": 904
    },
    {
      "epoch": 0.25062309609526445,
      "grad_norm": 2.6328420639038086,
      "learning_rate": 0.00017287511785458358,
      "loss": 2.8572,
      "step": 905
    },
    {
      "epoch": 0.2509000276931598,
      "grad_norm": 1.911133885383606,
      "learning_rate": 0.0001728146768264714,
      "loss": 2.695,
      "step": 906
    },
    {
      "epoch": 0.2511769592910551,
      "grad_norm": 1.584742546081543,
      "learning_rate": 0.00017275417912555576,
      "loss": 2.8547,
      "step": 907
    },
    {
      "epoch": 0.2514538908889504,
      "grad_norm": 3.0724382400512695,
      "learning_rate": 0.00017269362479892304,
      "loss": 3.1775,
      "step": 908
    },
    {
      "epoch": 0.25173082248684575,
      "grad_norm": 2.9462833404541016,
      "learning_rate": 0.00017263301389370362,
      "loss": 2.9909,
      "step": 909
    },
    {
      "epoch": 0.2520077540847411,
      "grad_norm": 1.692214012145996,
      "learning_rate": 0.00017257234645707187,
      "loss": 2.5681,
      "step": 910
    },
    {
      "epoch": 0.2522846856826364,
      "grad_norm": 1.7105425596237183,
      "learning_rate": 0.00017251162253624624,
      "loss": 2.5708,
      "step": 911
    },
    {
      "epoch": 0.2525616172805317,
      "grad_norm": 1.436359167098999,
      "learning_rate": 0.00017245084217848913,
      "loss": 2.7043,
      "step": 912
    },
    {
      "epoch": 0.25283854887842705,
      "grad_norm": 1.3870446681976318,
      "learning_rate": 0.0001723900054311068,
      "loss": 2.8711,
      "step": 913
    },
    {
      "epoch": 0.2531154804763224,
      "grad_norm": 2.342883348464966,
      "learning_rate": 0.0001723291123414495,
      "loss": 2.8609,
      "step": 914
    },
    {
      "epoch": 0.25339241207421764,
      "grad_norm": 1.8659863471984863,
      "learning_rate": 0.00017226816295691123,
      "loss": 2.7894,
      "step": 915
    },
    {
      "epoch": 0.25366934367211297,
      "grad_norm": 2.27763032913208,
      "learning_rate": 0.0001722071573249298,
      "loss": 2.8256,
      "step": 916
    },
    {
      "epoch": 0.2539462752700083,
      "grad_norm": 2.1597366333007812,
      "learning_rate": 0.00017214609549298697,
      "loss": 2.4851,
      "step": 917
    },
    {
      "epoch": 0.2542232068679036,
      "grad_norm": 1.9769222736358643,
      "learning_rate": 0.00017208497750860805,
      "loss": 2.4697,
      "step": 918
    },
    {
      "epoch": 0.25450013846579894,
      "grad_norm": 2.303415536880493,
      "learning_rate": 0.00017202380341936212,
      "loss": 2.5831,
      "step": 919
    },
    {
      "epoch": 0.25477707006369427,
      "grad_norm": 2.2522268295288086,
      "learning_rate": 0.00017196257327286194,
      "loss": 2.8313,
      "step": 920
    },
    {
      "epoch": 0.2550540016615896,
      "grad_norm": 2.688744306564331,
      "learning_rate": 0.00017190128711676392,
      "loss": 2.8566,
      "step": 921
    },
    {
      "epoch": 0.2553309332594849,
      "grad_norm": 2.4292383193969727,
      "learning_rate": 0.00017183994499876798,
      "loss": 2.29,
      "step": 922
    },
    {
      "epoch": 0.25560786485738024,
      "grad_norm": 6.555003643035889,
      "learning_rate": 0.00017177854696661773,
      "loss": 3.0297,
      "step": 923
    },
    {
      "epoch": 0.25588479645527556,
      "grad_norm": 3.1714563369750977,
      "learning_rate": 0.00017171709306810012,
      "loss": 3.1388,
      "step": 924
    },
    {
      "epoch": 0.2561617280531709,
      "grad_norm": 6.414551734924316,
      "learning_rate": 0.00017165558335104577,
      "loss": 4.24,
      "step": 925
    },
    {
      "epoch": 0.2564386596510662,
      "grad_norm": 2.3539018630981445,
      "learning_rate": 0.00017159401786332864,
      "loss": 2.6739,
      "step": 926
    },
    {
      "epoch": 0.2567155912489615,
      "grad_norm": 1.710160255432129,
      "learning_rate": 0.00017153239665286605,
      "loss": 2.6524,
      "step": 927
    },
    {
      "epoch": 0.2569925228468568,
      "grad_norm": 2.419032573699951,
      "learning_rate": 0.0001714707197676188,
      "loss": 2.8205,
      "step": 928
    },
    {
      "epoch": 0.25726945444475213,
      "grad_norm": 2.155780076980591,
      "learning_rate": 0.000171408987255591,
      "loss": 2.5581,
      "step": 929
    },
    {
      "epoch": 0.25754638604264746,
      "grad_norm": 2.413773775100708,
      "learning_rate": 0.00017134719916482997,
      "loss": 2.9444,
      "step": 930
    },
    {
      "epoch": 0.2578233176405428,
      "grad_norm": 2.2128782272338867,
      "learning_rate": 0.00017128535554342643,
      "loss": 3.1043,
      "step": 931
    },
    {
      "epoch": 0.2581002492384381,
      "grad_norm": 3.0503475666046143,
      "learning_rate": 0.00017122345643951417,
      "loss": 2.7248,
      "step": 932
    },
    {
      "epoch": 0.25837718083633343,
      "grad_norm": 2.457817554473877,
      "learning_rate": 0.00017116150190127024,
      "loss": 2.4952,
      "step": 933
    },
    {
      "epoch": 0.25865411243422876,
      "grad_norm": 2.2783281803131104,
      "learning_rate": 0.00017109949197691485,
      "loss": 2.8066,
      "step": 934
    },
    {
      "epoch": 0.2589310440321241,
      "grad_norm": 3.6404027938842773,
      "learning_rate": 0.0001710374267147113,
      "loss": 3.5088,
      "step": 935
    },
    {
      "epoch": 0.2592079756300194,
      "grad_norm": 13.817218780517578,
      "learning_rate": 0.00017097530616296593,
      "loss": 2.9703,
      "step": 936
    },
    {
      "epoch": 0.25948490722791473,
      "grad_norm": 4.384236812591553,
      "learning_rate": 0.00017091313037002816,
      "loss": 2.7491,
      "step": 937
    },
    {
      "epoch": 0.25976183882581,
      "grad_norm": 36.29664993286133,
      "learning_rate": 0.00017085089938429038,
      "loss": 3.0183,
      "step": 938
    },
    {
      "epoch": 0.2600387704237053,
      "grad_norm": 2.643810987472534,
      "learning_rate": 0.00017078861325418797,
      "loss": 2.9312,
      "step": 939
    },
    {
      "epoch": 0.26031570202160065,
      "grad_norm": 2.287806272506714,
      "learning_rate": 0.0001707262720281991,
      "loss": 3.0836,
      "step": 940
    },
    {
      "epoch": 0.260592633619496,
      "grad_norm": 58.10905456542969,
      "learning_rate": 0.00017066387575484501,
      "loss": 3.0783,
      "step": 941
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 228.61538696289062,
      "learning_rate": 0.00017060142448268968,
      "loss": 3.4171,
      "step": 942
    },
    {
      "epoch": 0.2611464968152866,
      "grad_norm": 14.011260032653809,
      "learning_rate": 0.0001705389182603399,
      "loss": 2.582,
      "step": 943
    },
    {
      "epoch": 0.26142342841318195,
      "grad_norm": 3.490990400314331,
      "learning_rate": 0.00017047635713644528,
      "loss": 2.8077,
      "step": 944
    },
    {
      "epoch": 0.26170036001107727,
      "grad_norm": 1.8019381761550903,
      "learning_rate": 0.00017041374115969805,
      "loss": 2.9052,
      "step": 945
    },
    {
      "epoch": 0.2619772916089726,
      "grad_norm": 22.604679107666016,
      "learning_rate": 0.00017035107037883328,
      "loss": 3.0083,
      "step": 946
    },
    {
      "epoch": 0.2622542232068679,
      "grad_norm": 2.184748888015747,
      "learning_rate": 0.0001702883448426286,
      "loss": 3.1804,
      "step": 947
    },
    {
      "epoch": 0.26253115480476324,
      "grad_norm": 10.158811569213867,
      "learning_rate": 0.00017022556459990424,
      "loss": 2.8463,
      "step": 948
    },
    {
      "epoch": 0.26280808640265857,
      "grad_norm": 9.095647811889648,
      "learning_rate": 0.00017016272969952304,
      "loss": 3.5129,
      "step": 949
    },
    {
      "epoch": 0.26308501800055384,
      "grad_norm": 2.826272487640381,
      "learning_rate": 0.00017009984019039049,
      "loss": 3.419,
      "step": 950
    },
    {
      "epoch": 0.26336194959844916,
      "grad_norm": 3.0602076053619385,
      "learning_rate": 0.00017003689612145437,
      "loss": 2.8289,
      "step": 951
    },
    {
      "epoch": 0.2636388811963445,
      "grad_norm": 1.6230366230010986,
      "learning_rate": 0.00016997389754170507,
      "loss": 3.0666,
      "step": 952
    },
    {
      "epoch": 0.2639158127942398,
      "grad_norm": 1.6178385019302368,
      "learning_rate": 0.00016991084450017542,
      "loss": 2.7564,
      "step": 953
    },
    {
      "epoch": 0.26419274439213514,
      "grad_norm": 1.646022915840149,
      "learning_rate": 0.0001698477370459405,
      "loss": 2.8767,
      "step": 954
    },
    {
      "epoch": 0.26446967599003046,
      "grad_norm": 2.150141954421997,
      "learning_rate": 0.00016978457522811788,
      "loss": 2.8903,
      "step": 955
    },
    {
      "epoch": 0.2647466075879258,
      "grad_norm": 1.6503727436065674,
      "learning_rate": 0.00016972135909586742,
      "loss": 2.8843,
      "step": 956
    },
    {
      "epoch": 0.2650235391858211,
      "grad_norm": 2.434426784515381,
      "learning_rate": 0.00016965808869839116,
      "loss": 2.7896,
      "step": 957
    },
    {
      "epoch": 0.26530047078371644,
      "grad_norm": 2.7235426902770996,
      "learning_rate": 0.00016959476408493347,
      "loss": 2.73,
      "step": 958
    },
    {
      "epoch": 0.26557740238161176,
      "grad_norm": 2.0859873294830322,
      "learning_rate": 0.00016953138530478092,
      "loss": 3.0053,
      "step": 959
    },
    {
      "epoch": 0.2658543339795071,
      "grad_norm": 2.04044771194458,
      "learning_rate": 0.00016946795240726216,
      "loss": 2.813,
      "step": 960
    },
    {
      "epoch": 0.26613126557740235,
      "grad_norm": 1.8185943365097046,
      "learning_rate": 0.00016940446544174803,
      "loss": 2.8833,
      "step": 961
    },
    {
      "epoch": 0.2664081971752977,
      "grad_norm": 2.5354840755462646,
      "learning_rate": 0.00016934092445765144,
      "loss": 3.0756,
      "step": 962
    },
    {
      "epoch": 0.266685128773193,
      "grad_norm": 1.4968249797821045,
      "learning_rate": 0.00016927732950442728,
      "loss": 2.9121,
      "step": 963
    },
    {
      "epoch": 0.26696206037108833,
      "grad_norm": 15.088715553283691,
      "learning_rate": 0.0001692136806315726,
      "loss": 2.7988,
      "step": 964
    },
    {
      "epoch": 0.26723899196898365,
      "grad_norm": 1.238550066947937,
      "learning_rate": 0.00016914997788862622,
      "loss": 2.8324,
      "step": 965
    },
    {
      "epoch": 0.267515923566879,
      "grad_norm": 1.432619571685791,
      "learning_rate": 0.00016908622132516903,
      "loss": 2.5692,
      "step": 966
    },
    {
      "epoch": 0.2677928551647743,
      "grad_norm": 2.6180357933044434,
      "learning_rate": 0.00016902241099082374,
      "loss": 3.2998,
      "step": 967
    },
    {
      "epoch": 0.2680697867626696,
      "grad_norm": 1.5661367177963257,
      "learning_rate": 0.00016895854693525496,
      "loss": 2.6682,
      "step": 968
    },
    {
      "epoch": 0.26834671836056495,
      "grad_norm": 1.7412769794464111,
      "learning_rate": 0.00016889462920816902,
      "loss": 2.8722,
      "step": 969
    },
    {
      "epoch": 0.2686236499584603,
      "grad_norm": 1.6578108072280884,
      "learning_rate": 0.00016883065785931414,
      "loss": 2.6731,
      "step": 970
    },
    {
      "epoch": 0.2689005815563556,
      "grad_norm": 1.3283472061157227,
      "learning_rate": 0.00016876663293848026,
      "loss": 2.7675,
      "step": 971
    },
    {
      "epoch": 0.2691775131542509,
      "grad_norm": 1.5704797506332397,
      "learning_rate": 0.0001687025544954989,
      "loss": 2.7657,
      "step": 972
    },
    {
      "epoch": 0.2694544447521462,
      "grad_norm": 2.236462116241455,
      "learning_rate": 0.00016863842258024335,
      "loss": 3.0339,
      "step": 973
    },
    {
      "epoch": 0.2697313763500415,
      "grad_norm": 2.717695713043213,
      "learning_rate": 0.00016857423724262849,
      "loss": 3.0133,
      "step": 974
    },
    {
      "epoch": 0.27000830794793684,
      "grad_norm": 8.298233985900879,
      "learning_rate": 0.00016850999853261075,
      "loss": 3.9595,
      "step": 975
    },
    {
      "epoch": 0.27028523954583217,
      "grad_norm": 2.5315914154052734,
      "learning_rate": 0.0001684457065001882,
      "loss": 2.8995,
      "step": 976
    },
    {
      "epoch": 0.2705621711437275,
      "grad_norm": 2.19551944732666,
      "learning_rate": 0.00016838136119540022,
      "loss": 2.7672,
      "step": 977
    },
    {
      "epoch": 0.2708391027416228,
      "grad_norm": 1.4843145608901978,
      "learning_rate": 0.00016831696266832787,
      "loss": 2.7836,
      "step": 978
    },
    {
      "epoch": 0.27111603433951814,
      "grad_norm": 1.3755983114242554,
      "learning_rate": 0.00016825251096909343,
      "loss": 2.9166,
      "step": 979
    },
    {
      "epoch": 0.27139296593741347,
      "grad_norm": 2.4390645027160645,
      "learning_rate": 0.00016818800614786075,
      "loss": 2.4564,
      "step": 980
    },
    {
      "epoch": 0.2716698975353088,
      "grad_norm": 1.8554071187973022,
      "learning_rate": 0.00016812344825483492,
      "loss": 2.607,
      "step": 981
    },
    {
      "epoch": 0.2719468291332041,
      "grad_norm": 1.954175353050232,
      "learning_rate": 0.00016805883734026238,
      "loss": 3.053,
      "step": 982
    },
    {
      "epoch": 0.27222376073109944,
      "grad_norm": 1.6565766334533691,
      "learning_rate": 0.00016799417345443078,
      "loss": 2.6651,
      "step": 983
    },
    {
      "epoch": 0.27250069232899476,
      "grad_norm": 1.8020745515823364,
      "learning_rate": 0.00016792945664766907,
      "loss": 2.6281,
      "step": 984
    },
    {
      "epoch": 0.27277762392689003,
      "grad_norm": 2.614210367202759,
      "learning_rate": 0.0001678646869703473,
      "loss": 3.0437,
      "step": 985
    },
    {
      "epoch": 0.27305455552478536,
      "grad_norm": 1.832960844039917,
      "learning_rate": 0.00016779986447287677,
      "loss": 2.7291,
      "step": 986
    },
    {
      "epoch": 0.2733314871226807,
      "grad_norm": 2.2536306381225586,
      "learning_rate": 0.00016773498920570983,
      "loss": 3.2322,
      "step": 987
    },
    {
      "epoch": 0.273608418720576,
      "grad_norm": 2.4348583221435547,
      "learning_rate": 0.0001676700612193399,
      "loss": 2.9202,
      "step": 988
    },
    {
      "epoch": 0.27388535031847133,
      "grad_norm": 196.9659881591797,
      "learning_rate": 0.00016760508056430152,
      "loss": 3.5759,
      "step": 989
    },
    {
      "epoch": 0.27416228191636666,
      "grad_norm": 10.67332935333252,
      "learning_rate": 0.00016754004729117004,
      "loss": 2.7365,
      "step": 990
    },
    {
      "epoch": 0.274439213514262,
      "grad_norm": 1.9112107753753662,
      "learning_rate": 0.00016747496145056196,
      "loss": 2.8596,
      "step": 991
    },
    {
      "epoch": 0.2747161451121573,
      "grad_norm": 2.2627761363983154,
      "learning_rate": 0.0001674098230931346,
      "loss": 2.7513,
      "step": 992
    },
    {
      "epoch": 0.27499307671005263,
      "grad_norm": 1.7090967893600464,
      "learning_rate": 0.00016734463226958615,
      "loss": 2.8605,
      "step": 993
    },
    {
      "epoch": 0.27527000830794796,
      "grad_norm": 1.4351768493652344,
      "learning_rate": 0.0001672793890306556,
      "loss": 2.5909,
      "step": 994
    },
    {
      "epoch": 0.2755469399058433,
      "grad_norm": 1.6425434350967407,
      "learning_rate": 0.0001672140934271229,
      "loss": 2.6916,
      "step": 995
    },
    {
      "epoch": 0.27582387150373855,
      "grad_norm": 5.847080230712891,
      "learning_rate": 0.00016714874550980853,
      "loss": 3.4974,
      "step": 996
    },
    {
      "epoch": 0.2761008031016339,
      "grad_norm": 19.677980422973633,
      "learning_rate": 0.00016708334532957383,
      "loss": 4.3623,
      "step": 997
    },
    {
      "epoch": 0.2763777346995292,
      "grad_norm": 6.76720666885376,
      "learning_rate": 0.0001670178929373208,
      "loss": 4.0329,
      "step": 998
    },
    {
      "epoch": 0.2766546662974245,
      "grad_norm": 9.839909553527832,
      "learning_rate": 0.00016695238838399206,
      "loss": 3.7058,
      "step": 999
    },
    {
      "epoch": 0.27693159789531985,
      "grad_norm": 4.762269496917725,
      "learning_rate": 0.00016688683172057087,
      "loss": 3.7894,
      "step": 1000
    },
    {
      "epoch": 0.27720852949321517,
      "grad_norm": 1.6144564151763916,
      "learning_rate": 0.0001668212229980809,
      "loss": 2.6048,
      "step": 1001
    },
    {
      "epoch": 0.2774854610911105,
      "grad_norm": 2.5514297485351562,
      "learning_rate": 0.00016675556226758656,
      "loss": 2.943,
      "step": 1002
    },
    {
      "epoch": 0.2777623926890058,
      "grad_norm": 3.8206777572631836,
      "learning_rate": 0.00016668984958019255,
      "loss": 2.5958,
      "step": 1003
    },
    {
      "epoch": 0.27803932428690115,
      "grad_norm": 1.4613410234451294,
      "learning_rate": 0.0001666240849870441,
      "loss": 2.74,
      "step": 1004
    },
    {
      "epoch": 0.27831625588479647,
      "grad_norm": 1.8141924142837524,
      "learning_rate": 0.00016655826853932684,
      "loss": 2.9371,
      "step": 1005
    },
    {
      "epoch": 0.2785931874826918,
      "grad_norm": 1.7682191133499146,
      "learning_rate": 0.00016649240028826674,
      "loss": 2.6332,
      "step": 1006
    },
    {
      "epoch": 0.2788701190805871,
      "grad_norm": 2.4179537296295166,
      "learning_rate": 0.0001664264802851301,
      "loss": 3.1183,
      "step": 1007
    },
    {
      "epoch": 0.2791470506784824,
      "grad_norm": 2.072624444961548,
      "learning_rate": 0.00016636050858122344,
      "loss": 2.9504,
      "step": 1008
    },
    {
      "epoch": 0.2794239822763777,
      "grad_norm": 2.388211250305176,
      "learning_rate": 0.0001662944852278936,
      "loss": 2.5977,
      "step": 1009
    },
    {
      "epoch": 0.27970091387427304,
      "grad_norm": 2.1789684295654297,
      "learning_rate": 0.00016622841027652766,
      "loss": 2.941,
      "step": 1010
    },
    {
      "epoch": 0.27997784547216836,
      "grad_norm": 1.181632161140442,
      "learning_rate": 0.00016616228377855268,
      "loss": 2.882,
      "step": 1011
    },
    {
      "epoch": 0.2802547770700637,
      "grad_norm": 1.4988845586776733,
      "learning_rate": 0.00016609610578543598,
      "loss": 3.073,
      "step": 1012
    },
    {
      "epoch": 0.280531708667959,
      "grad_norm": 2.1757216453552246,
      "learning_rate": 0.000166029876348685,
      "loss": 2.6687,
      "step": 1013
    },
    {
      "epoch": 0.28080864026585434,
      "grad_norm": 2.580033779144287,
      "learning_rate": 0.00016596359551984704,
      "loss": 3.1823,
      "step": 1014
    },
    {
      "epoch": 0.28108557186374966,
      "grad_norm": 2.1309080123901367,
      "learning_rate": 0.00016589726335050956,
      "loss": 2.8604,
      "step": 1015
    },
    {
      "epoch": 0.281362503461645,
      "grad_norm": 2.224088191986084,
      "learning_rate": 0.00016583087989229997,
      "loss": 3.053,
      "step": 1016
    },
    {
      "epoch": 0.2816394350595403,
      "grad_norm": 1.8244150876998901,
      "learning_rate": 0.00016576444519688548,
      "loss": 3.1972,
      "step": 1017
    },
    {
      "epoch": 0.28191636665743564,
      "grad_norm": 2.5288257598876953,
      "learning_rate": 0.00016569795931597328,
      "loss": 2.8303,
      "step": 1018
    },
    {
      "epoch": 0.28219329825533096,
      "grad_norm": 1.8015689849853516,
      "learning_rate": 0.0001656314223013104,
      "loss": 2.6104,
      "step": 1019
    },
    {
      "epoch": 0.28247022985322623,
      "grad_norm": 8.759150505065918,
      "learning_rate": 0.0001655648342046836,
      "loss": 3.0135,
      "step": 1020
    },
    {
      "epoch": 0.28274716145112155,
      "grad_norm": 2.5134949684143066,
      "learning_rate": 0.00016549819507791943,
      "loss": 2.6073,
      "step": 1021
    },
    {
      "epoch": 0.2830240930490169,
      "grad_norm": 1.9912227392196655,
      "learning_rate": 0.00016543150497288424,
      "loss": 2.949,
      "step": 1022
    },
    {
      "epoch": 0.2833010246469122,
      "grad_norm": 2.0433785915374756,
      "learning_rate": 0.0001653647639414839,
      "loss": 2.6917,
      "step": 1023
    },
    {
      "epoch": 0.2835779562448075,
      "grad_norm": 1.5686930418014526,
      "learning_rate": 0.00016529797203566405,
      "loss": 3.1209,
      "step": 1024
    },
    {
      "epoch": 0.28385488784270285,
      "grad_norm": 4.718352317810059,
      "learning_rate": 0.00016523112930740986,
      "loss": 4.037,
      "step": 1025
    },
    {
      "epoch": 0.2841318194405982,
      "grad_norm": 2.3235116004943848,
      "learning_rate": 0.00016516423580874607,
      "loss": 2.7712,
      "step": 1026
    },
    {
      "epoch": 0.2844087510384935,
      "grad_norm": 1.234822392463684,
      "learning_rate": 0.00016509729159173696,
      "loss": 2.6417,
      "step": 1027
    },
    {
      "epoch": 0.2846856826363888,
      "grad_norm": 3.115663766860962,
      "learning_rate": 0.00016503029670848621,
      "loss": 3.3312,
      "step": 1028
    },
    {
      "epoch": 0.28496261423428415,
      "grad_norm": 1.5264135599136353,
      "learning_rate": 0.00016496325121113706,
      "loss": 2.9245,
      "step": 1029
    },
    {
      "epoch": 0.2852395458321795,
      "grad_norm": 1.799498438835144,
      "learning_rate": 0.000164896155151872,
      "loss": 2.9856,
      "step": 1030
    },
    {
      "epoch": 0.28551647743007474,
      "grad_norm": 1.908801794052124,
      "learning_rate": 0.000164829008582913,
      "loss": 2.6698,
      "step": 1031
    },
    {
      "epoch": 0.28579340902797007,
      "grad_norm": 1.5490243434906006,
      "learning_rate": 0.00016476181155652126,
      "loss": 2.6932,
      "step": 1032
    },
    {
      "epoch": 0.2860703406258654,
      "grad_norm": 2.0292749404907227,
      "learning_rate": 0.00016469456412499724,
      "loss": 2.6477,
      "step": 1033
    },
    {
      "epoch": 0.2863472722237607,
      "grad_norm": 1.593946099281311,
      "learning_rate": 0.00016462726634068075,
      "loss": 2.7718,
      "step": 1034
    },
    {
      "epoch": 0.28662420382165604,
      "grad_norm": 2.0860631465911865,
      "learning_rate": 0.00016455991825595066,
      "loss": 2.6695,
      "step": 1035
    },
    {
      "epoch": 0.28690113541955137,
      "grad_norm": 2.2195959091186523,
      "learning_rate": 0.00016449251992322505,
      "loss": 2.5107,
      "step": 1036
    },
    {
      "epoch": 0.2871780670174467,
      "grad_norm": 1.1555812358856201,
      "learning_rate": 0.0001644250713949611,
      "loss": 2.7165,
      "step": 1037
    },
    {
      "epoch": 0.287454998615342,
      "grad_norm": 1.8912162780761719,
      "learning_rate": 0.00016435757272365505,
      "loss": 2.9361,
      "step": 1038
    },
    {
      "epoch": 0.28773193021323734,
      "grad_norm": 1.9437050819396973,
      "learning_rate": 0.00016429002396184215,
      "loss": 2.8861,
      "step": 1039
    },
    {
      "epoch": 0.28800886181113267,
      "grad_norm": 1.3384404182434082,
      "learning_rate": 0.00016422242516209672,
      "loss": 2.7149,
      "step": 1040
    },
    {
      "epoch": 0.288285793409028,
      "grad_norm": 1.8033853769302368,
      "learning_rate": 0.0001641547763770319,
      "loss": 2.9564,
      "step": 1041
    },
    {
      "epoch": 0.2885627250069233,
      "grad_norm": 2.3526084423065186,
      "learning_rate": 0.00016408707765929985,
      "loss": 2.9201,
      "step": 1042
    },
    {
      "epoch": 0.2888396566048186,
      "grad_norm": 1.8857325315475464,
      "learning_rate": 0.0001640193290615915,
      "loss": 3.0058,
      "step": 1043
    },
    {
      "epoch": 0.2891165882027139,
      "grad_norm": 2.3668930530548096,
      "learning_rate": 0.00016395153063663667,
      "loss": 2.5814,
      "step": 1044
    },
    {
      "epoch": 0.28939351980060923,
      "grad_norm": 1.8008521795272827,
      "learning_rate": 0.00016388368243720392,
      "loss": 2.4955,
      "step": 1045
    },
    {
      "epoch": 0.28967045139850456,
      "grad_norm": 1.3481143712997437,
      "learning_rate": 0.00016381578451610062,
      "loss": 2.5505,
      "step": 1046
    },
    {
      "epoch": 0.2899473829963999,
      "grad_norm": 2.66656494140625,
      "learning_rate": 0.0001637478369261727,
      "loss": 2.4911,
      "step": 1047
    },
    {
      "epoch": 0.2902243145942952,
      "grad_norm": 1.821694016456604,
      "learning_rate": 0.0001636798397203049,
      "loss": 2.4427,
      "step": 1048
    },
    {
      "epoch": 0.29050124619219053,
      "grad_norm": 4.15843391418457,
      "learning_rate": 0.00016361179295142046,
      "loss": 3.3711,
      "step": 1049
    },
    {
      "epoch": 0.29077817779008586,
      "grad_norm": 4.490833282470703,
      "learning_rate": 0.00016354369667248128,
      "loss": 4.0469,
      "step": 1050
    },
    {
      "epoch": 0.2910551093879812,
      "grad_norm": 2.0667965412139893,
      "learning_rate": 0.00016347555093648776,
      "loss": 2.6667,
      "step": 1051
    },
    {
      "epoch": 0.2913320409858765,
      "grad_norm": 2.4750208854675293,
      "learning_rate": 0.0001634073557964788,
      "loss": 3.2475,
      "step": 1052
    },
    {
      "epoch": 0.29160897258377183,
      "grad_norm": 15.866964340209961,
      "learning_rate": 0.00016333911130553172,
      "loss": 2.8132,
      "step": 1053
    },
    {
      "epoch": 0.29188590418166716,
      "grad_norm": 2.471449375152588,
      "learning_rate": 0.00016327081751676227,
      "loss": 2.9777,
      "step": 1054
    },
    {
      "epoch": 0.2921628357795624,
      "grad_norm": 1.6980383396148682,
      "learning_rate": 0.0001632024744833246,
      "loss": 2.9916,
      "step": 1055
    },
    {
      "epoch": 0.29243976737745775,
      "grad_norm": 1.6648608446121216,
      "learning_rate": 0.00016313408225841118,
      "loss": 2.6989,
      "step": 1056
    },
    {
      "epoch": 0.2927166989753531,
      "grad_norm": 2.11301589012146,
      "learning_rate": 0.00016306564089525272,
      "loss": 2.9521,
      "step": 1057
    },
    {
      "epoch": 0.2929936305732484,
      "grad_norm": 1.940384030342102,
      "learning_rate": 0.0001629971504471182,
      "loss": 2.6839,
      "step": 1058
    },
    {
      "epoch": 0.2932705621711437,
      "grad_norm": 1.652581810951233,
      "learning_rate": 0.0001629286109673148,
      "loss": 2.9663,
      "step": 1059
    },
    {
      "epoch": 0.29354749376903905,
      "grad_norm": 1.4100010395050049,
      "learning_rate": 0.00016286002250918792,
      "loss": 2.8476,
      "step": 1060
    },
    {
      "epoch": 0.29382442536693437,
      "grad_norm": 2.3216185569763184,
      "learning_rate": 0.000162791385126121,
      "loss": 2.8777,
      "step": 1061
    },
    {
      "epoch": 0.2941013569648297,
      "grad_norm": 2.377795934677124,
      "learning_rate": 0.00016272269887153559,
      "loss": 2.8286,
      "step": 1062
    },
    {
      "epoch": 0.294378288562725,
      "grad_norm": 1.5239886045455933,
      "learning_rate": 0.00016265396379889127,
      "loss": 3.0177,
      "step": 1063
    },
    {
      "epoch": 0.29465522016062035,
      "grad_norm": 1.657238245010376,
      "learning_rate": 0.00016258517996168564,
      "loss": 2.6564,
      "step": 1064
    },
    {
      "epoch": 0.29493215175851567,
      "grad_norm": 3.806039571762085,
      "learning_rate": 0.00016251634741345423,
      "loss": 2.6983,
      "step": 1065
    },
    {
      "epoch": 0.29520908335641094,
      "grad_norm": 3.4286303520202637,
      "learning_rate": 0.0001624474662077705,
      "loss": 2.9889,
      "step": 1066
    },
    {
      "epoch": 0.29548601495430626,
      "grad_norm": 1.889386534690857,
      "learning_rate": 0.00016237853639824575,
      "loss": 2.8944,
      "step": 1067
    },
    {
      "epoch": 0.2957629465522016,
      "grad_norm": 2.8036301136016846,
      "learning_rate": 0.00016230955803852912,
      "loss": 3.0058,
      "step": 1068
    },
    {
      "epoch": 0.2960398781500969,
      "grad_norm": 2.555205821990967,
      "learning_rate": 0.0001622405311823076,
      "loss": 2.6748,
      "step": 1069
    },
    {
      "epoch": 0.29631680974799224,
      "grad_norm": 2.4662954807281494,
      "learning_rate": 0.00016217145588330584,
      "loss": 2.7295,
      "step": 1070
    },
    {
      "epoch": 0.29659374134588756,
      "grad_norm": 3.151988983154297,
      "learning_rate": 0.00016210233219528624,
      "loss": 2.9012,
      "step": 1071
    },
    {
      "epoch": 0.2968706729437829,
      "grad_norm": 1.1206333637237549,
      "learning_rate": 0.0001620331601720488,
      "loss": 2.6216,
      "step": 1072
    },
    {
      "epoch": 0.2971476045416782,
      "grad_norm": 2.9131011962890625,
      "learning_rate": 0.0001619639398674313,
      "loss": 2.7355,
      "step": 1073
    },
    {
      "epoch": 0.29742453613957354,
      "grad_norm": 3.1310834884643555,
      "learning_rate": 0.00016189467133530884,
      "loss": 3.048,
      "step": 1074
    },
    {
      "epoch": 0.29770146773746886,
      "grad_norm": 5.254419326782227,
      "learning_rate": 0.00016182535462959434,
      "loss": 4.0407,
      "step": 1075
    },
    {
      "epoch": 0.2979783993353642,
      "grad_norm": 1.8420734405517578,
      "learning_rate": 0.00016175598980423797,
      "loss": 3.0963,
      "step": 1076
    },
    {
      "epoch": 0.2982553309332595,
      "grad_norm": 1.8723719120025635,
      "learning_rate": 0.00016168657691322755,
      "loss": 2.793,
      "step": 1077
    },
    {
      "epoch": 0.2985322625311548,
      "grad_norm": 2.471306085586548,
      "learning_rate": 0.00016161711601058815,
      "loss": 2.654,
      "step": 1078
    },
    {
      "epoch": 0.2988091941290501,
      "grad_norm": 2.581939220428467,
      "learning_rate": 0.0001615476071503823,
      "loss": 3.0836,
      "step": 1079
    },
    {
      "epoch": 0.29908612572694543,
      "grad_norm": 2.18522572517395,
      "learning_rate": 0.00016147805038670986,
      "loss": 3.0257,
      "step": 1080
    },
    {
      "epoch": 0.29936305732484075,
      "grad_norm": 2.2486062049865723,
      "learning_rate": 0.00016140844577370791,
      "loss": 2.814,
      "step": 1081
    },
    {
      "epoch": 0.2996399889227361,
      "grad_norm": 2.070225238800049,
      "learning_rate": 0.00016133879336555085,
      "loss": 2.7629,
      "step": 1082
    },
    {
      "epoch": 0.2999169205206314,
      "grad_norm": 1.965770959854126,
      "learning_rate": 0.00016126909321645023,
      "loss": 2.7618,
      "step": 1083
    },
    {
      "epoch": 0.3001938521185267,
      "grad_norm": 3.0630273818969727,
      "learning_rate": 0.0001611993453806547,
      "loss": 2.7491,
      "step": 1084
    },
    {
      "epoch": 0.30047078371642205,
      "grad_norm": 1.745874047279358,
      "learning_rate": 0.00016112954991245022,
      "loss": 2.6349,
      "step": 1085
    },
    {
      "epoch": 0.3007477153143174,
      "grad_norm": 2.672760248184204,
      "learning_rate": 0.0001610597068661596,
      "loss": 2.659,
      "step": 1086
    },
    {
      "epoch": 0.3010246469122127,
      "grad_norm": 1.7880834341049194,
      "learning_rate": 0.00016098981629614277,
      "loss": 2.5338,
      "step": 1087
    },
    {
      "epoch": 0.301301578510108,
      "grad_norm": 1.5602076053619385,
      "learning_rate": 0.00016091987825679672,
      "loss": 2.9071,
      "step": 1088
    },
    {
      "epoch": 0.30157851010800335,
      "grad_norm": 2.032357931137085,
      "learning_rate": 0.0001608498928025553,
      "loss": 2.8791,
      "step": 1089
    },
    {
      "epoch": 0.3018554417058986,
      "grad_norm": 1.5926483869552612,
      "learning_rate": 0.00016077985998788925,
      "loss": 2.867,
      "step": 1090
    },
    {
      "epoch": 0.30213237330379394,
      "grad_norm": 3.0426523685455322,
      "learning_rate": 0.00016070977986730624,
      "loss": 3.0692,
      "step": 1091
    },
    {
      "epoch": 0.30240930490168927,
      "grad_norm": 1.8183352947235107,
      "learning_rate": 0.00016063965249535074,
      "loss": 2.8024,
      "step": 1092
    },
    {
      "epoch": 0.3026862364995846,
      "grad_norm": 1.9122745990753174,
      "learning_rate": 0.00016056947792660392,
      "loss": 2.7971,
      "step": 1093
    },
    {
      "epoch": 0.3029631680974799,
      "grad_norm": 1.4231312274932861,
      "learning_rate": 0.00016049925621568382,
      "loss": 2.4285,
      "step": 1094
    },
    {
      "epoch": 0.30324009969537524,
      "grad_norm": 1.6702405214309692,
      "learning_rate": 0.00016042898741724503,
      "loss": 2.5288,
      "step": 1095
    },
    {
      "epoch": 0.30351703129327057,
      "grad_norm": 2.072124719619751,
      "learning_rate": 0.0001603586715859789,
      "loss": 2.9134,
      "step": 1096
    },
    {
      "epoch": 0.3037939628911659,
      "grad_norm": 2.3705649375915527,
      "learning_rate": 0.00016028830877661335,
      "loss": 2.8175,
      "step": 1097
    },
    {
      "epoch": 0.3040708944890612,
      "grad_norm": 3.557770013809204,
      "learning_rate": 0.00016021789904391282,
      "loss": 3.0662,
      "step": 1098
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 2.803607940673828,
      "learning_rate": 0.00016014744244267833,
      "loss": 3.1365,
      "step": 1099
    },
    {
      "epoch": 0.30462475768485187,
      "grad_norm": 2.111748695373535,
      "learning_rate": 0.00016007693902774737,
      "loss": 3.2262,
      "step": 1100
    },
    {
      "epoch": 0.30490168928274713,
      "grad_norm": 2.4110352993011475,
      "learning_rate": 0.0001600063888539938,
      "loss": 2.7931,
      "step": 1101
    },
    {
      "epoch": 0.30517862088064246,
      "grad_norm": 1.7806950807571411,
      "learning_rate": 0.00015993579197632796,
      "loss": 2.8071,
      "step": 1102
    },
    {
      "epoch": 0.3054555524785378,
      "grad_norm": 1.2968131303787231,
      "learning_rate": 0.00015986514844969652,
      "loss": 2.594,
      "step": 1103
    },
    {
      "epoch": 0.3057324840764331,
      "grad_norm": 2.446486473083496,
      "learning_rate": 0.00015979445832908242,
      "loss": 2.6563,
      "step": 1104
    },
    {
      "epoch": 0.30600941567432843,
      "grad_norm": 1.2931456565856934,
      "learning_rate": 0.0001597237216695049,
      "loss": 2.8281,
      "step": 1105
    },
    {
      "epoch": 0.30628634727222376,
      "grad_norm": 2.0900042057037354,
      "learning_rate": 0.00015965293852601944,
      "loss": 2.7806,
      "step": 1106
    },
    {
      "epoch": 0.3065632788701191,
      "grad_norm": 2.632905960083008,
      "learning_rate": 0.0001595821089537176,
      "loss": 2.9521,
      "step": 1107
    },
    {
      "epoch": 0.3068402104680144,
      "grad_norm": 2.728095769882202,
      "learning_rate": 0.00015951123300772717,
      "loss": 2.8188,
      "step": 1108
    },
    {
      "epoch": 0.30711714206590973,
      "grad_norm": 1.6814217567443848,
      "learning_rate": 0.00015944031074321204,
      "loss": 2.4426,
      "step": 1109
    },
    {
      "epoch": 0.30739407366380506,
      "grad_norm": 1.18573796749115,
      "learning_rate": 0.0001593693422153721,
      "loss": 2.734,
      "step": 1110
    },
    {
      "epoch": 0.3076710052617004,
      "grad_norm": 2.4105420112609863,
      "learning_rate": 0.00015929832747944324,
      "loss": 3.0473,
      "step": 1111
    },
    {
      "epoch": 0.3079479368595957,
      "grad_norm": 2.6489510536193848,
      "learning_rate": 0.0001592272665906974,
      "loss": 3.3682,
      "step": 1112
    },
    {
      "epoch": 0.308224868457491,
      "grad_norm": 1.4610475301742554,
      "learning_rate": 0.00015915615960444233,
      "loss": 2.6737,
      "step": 1113
    },
    {
      "epoch": 0.3085018000553863,
      "grad_norm": 1.8061065673828125,
      "learning_rate": 0.00015908500657602174,
      "loss": 2.8074,
      "step": 1114
    },
    {
      "epoch": 0.3087787316532816,
      "grad_norm": 1.55250883102417,
      "learning_rate": 0.00015901380756081515,
      "loss": 2.9482,
      "step": 1115
    },
    {
      "epoch": 0.30905566325117695,
      "grad_norm": 1.9959843158721924,
      "learning_rate": 0.0001589425626142378,
      "loss": 3.1434,
      "step": 1116
    },
    {
      "epoch": 0.3093325948490723,
      "grad_norm": 1.5615752935409546,
      "learning_rate": 0.00015887127179174085,
      "loss": 2.9626,
      "step": 1117
    },
    {
      "epoch": 0.3096095264469676,
      "grad_norm": 1.5040863752365112,
      "learning_rate": 0.00015879993514881103,
      "loss": 3.0734,
      "step": 1118
    },
    {
      "epoch": 0.3098864580448629,
      "grad_norm": 2.326559066772461,
      "learning_rate": 0.0001587285527409707,
      "loss": 2.7417,
      "step": 1119
    },
    {
      "epoch": 0.31016338964275825,
      "grad_norm": 1.2279635667800903,
      "learning_rate": 0.00015865712462377802,
      "loss": 2.6415,
      "step": 1120
    },
    {
      "epoch": 0.31044032124065357,
      "grad_norm": 1.839401364326477,
      "learning_rate": 0.0001585856508528265,
      "loss": 2.6821,
      "step": 1121
    },
    {
      "epoch": 0.3107172528385489,
      "grad_norm": 1.4215173721313477,
      "learning_rate": 0.00015851413148374537,
      "loss": 2.674,
      "step": 1122
    },
    {
      "epoch": 0.3109941844364442,
      "grad_norm": 1.4573885202407837,
      "learning_rate": 0.00015844256657219927,
      "loss": 2.8351,
      "step": 1123
    },
    {
      "epoch": 0.31127111603433955,
      "grad_norm": 2.0855629444122314,
      "learning_rate": 0.00015837095617388827,
      "loss": 2.8772,
      "step": 1124
    },
    {
      "epoch": 0.3115480476322348,
      "grad_norm": 3.5365757942199707,
      "learning_rate": 0.00015829930034454783,
      "loss": 3.5358,
      "step": 1125
    },
    {
      "epoch": 0.31182497923013014,
      "grad_norm": 1.9074093103408813,
      "learning_rate": 0.00015822759913994888,
      "loss": 3.1212,
      "step": 1126
    },
    {
      "epoch": 0.31210191082802546,
      "grad_norm": 1.4721168279647827,
      "learning_rate": 0.00015815585261589753,
      "loss": 2.6182,
      "step": 1127
    },
    {
      "epoch": 0.3123788424259208,
      "grad_norm": 1.8700261116027832,
      "learning_rate": 0.00015808406082823528,
      "loss": 2.8233,
      "step": 1128
    },
    {
      "epoch": 0.3126557740238161,
      "grad_norm": 1.7332760095596313,
      "learning_rate": 0.0001580122238328387,
      "loss": 2.9136,
      "step": 1129
    },
    {
      "epoch": 0.31293270562171144,
      "grad_norm": 1.2909797430038452,
      "learning_rate": 0.00015794034168561982,
      "loss": 3.0554,
      "step": 1130
    },
    {
      "epoch": 0.31320963721960676,
      "grad_norm": 1.7902755737304688,
      "learning_rate": 0.00015786841444252546,
      "loss": 2.7218,
      "step": 1131
    },
    {
      "epoch": 0.3134865688175021,
      "grad_norm": 1.855206847190857,
      "learning_rate": 0.00015779644215953783,
      "loss": 3.0041,
      "step": 1132
    },
    {
      "epoch": 0.3137635004153974,
      "grad_norm": 1.2451910972595215,
      "learning_rate": 0.00015772442489267406,
      "loss": 2.6642,
      "step": 1133
    },
    {
      "epoch": 0.31404043201329274,
      "grad_norm": 1.616047739982605,
      "learning_rate": 0.00015765236269798627,
      "loss": 2.9032,
      "step": 1134
    },
    {
      "epoch": 0.31431736361118806,
      "grad_norm": 1.1861200332641602,
      "learning_rate": 0.00015758025563156167,
      "loss": 2.6461,
      "step": 1135
    },
    {
      "epoch": 0.31459429520908333,
      "grad_norm": 1.9919323921203613,
      "learning_rate": 0.00015750810374952226,
      "loss": 3.0355,
      "step": 1136
    },
    {
      "epoch": 0.31487122680697865,
      "grad_norm": 1.5378572940826416,
      "learning_rate": 0.00015743590710802505,
      "loss": 2.8951,
      "step": 1137
    },
    {
      "epoch": 0.315148158404874,
      "grad_norm": 1.3112939596176147,
      "learning_rate": 0.00015736366576326173,
      "loss": 2.727,
      "step": 1138
    },
    {
      "epoch": 0.3154250900027693,
      "grad_norm": 1.2822314500808716,
      "learning_rate": 0.00015729137977145893,
      "loss": 2.7409,
      "step": 1139
    },
    {
      "epoch": 0.31570202160066463,
      "grad_norm": 1.6411978006362915,
      "learning_rate": 0.00015721904918887796,
      "loss": 2.9995,
      "step": 1140
    },
    {
      "epoch": 0.31597895319855995,
      "grad_norm": 2.176082134246826,
      "learning_rate": 0.00015714667407181482,
      "loss": 2.84,
      "step": 1141
    },
    {
      "epoch": 0.3162558847964553,
      "grad_norm": 2.9109854698181152,
      "learning_rate": 0.00015707425447660021,
      "loss": 2.7467,
      "step": 1142
    },
    {
      "epoch": 0.3165328163943506,
      "grad_norm": 1.649181842803955,
      "learning_rate": 0.00015700179045959942,
      "loss": 2.6447,
      "step": 1143
    },
    {
      "epoch": 0.3168097479922459,
      "grad_norm": 2.1657958030700684,
      "learning_rate": 0.0001569292820772124,
      "loss": 2.8329,
      "step": 1144
    },
    {
      "epoch": 0.31708667959014125,
      "grad_norm": 2.1416025161743164,
      "learning_rate": 0.00015685672938587345,
      "loss": 2.6883,
      "step": 1145
    },
    {
      "epoch": 0.3173636111880366,
      "grad_norm": 2.3641982078552246,
      "learning_rate": 0.00015678413244205156,
      "loss": 2.6026,
      "step": 1146
    },
    {
      "epoch": 0.3176405427859319,
      "grad_norm": 2.626178741455078,
      "learning_rate": 0.00015671149130225005,
      "loss": 3.1148,
      "step": 1147
    },
    {
      "epoch": 0.31791747438382717,
      "grad_norm": 2.206259250640869,
      "learning_rate": 0.00015663880602300657,
      "loss": 2.8485,
      "step": 1148
    },
    {
      "epoch": 0.3181944059817225,
      "grad_norm": 2.893170118331909,
      "learning_rate": 0.00015656607666089334,
      "loss": 3.3303,
      "step": 1149
    },
    {
      "epoch": 0.3184713375796178,
      "grad_norm": 5.792006492614746,
      "learning_rate": 0.0001564933032725167,
      "loss": 4.157,
      "step": 1150
    },
    {
      "epoch": 0.31874826917751314,
      "grad_norm": 1.754286527633667,
      "learning_rate": 0.0001564204859145173,
      "loss": 2.7435,
      "step": 1151
    },
    {
      "epoch": 0.31902520077540847,
      "grad_norm": 1.5062384605407715,
      "learning_rate": 0.00015634762464357004,
      "loss": 2.7834,
      "step": 1152
    },
    {
      "epoch": 0.3193021323733038,
      "grad_norm": 1.6033989191055298,
      "learning_rate": 0.00015627471951638403,
      "loss": 2.8711,
      "step": 1153
    },
    {
      "epoch": 0.3195790639711991,
      "grad_norm": 3.486161231994629,
      "learning_rate": 0.0001562017705897024,
      "loss": 2.5715,
      "step": 1154
    },
    {
      "epoch": 0.31985599556909444,
      "grad_norm": 1.360491156578064,
      "learning_rate": 0.00015612877792030248,
      "loss": 2.769,
      "step": 1155
    },
    {
      "epoch": 0.32013292716698977,
      "grad_norm": 1.607000470161438,
      "learning_rate": 0.00015605574156499568,
      "loss": 2.6347,
      "step": 1156
    },
    {
      "epoch": 0.3204098587648851,
      "grad_norm": 1.483081579208374,
      "learning_rate": 0.0001559826615806272,
      "loss": 2.8894,
      "step": 1157
    },
    {
      "epoch": 0.3206867903627804,
      "grad_norm": 2.1092448234558105,
      "learning_rate": 0.00015590953802407644,
      "loss": 2.8086,
      "step": 1158
    },
    {
      "epoch": 0.32096372196067574,
      "grad_norm": 1.5998456478118896,
      "learning_rate": 0.00015583637095225656,
      "loss": 2.918,
      "step": 1159
    },
    {
      "epoch": 0.321240653558571,
      "grad_norm": 1.6553558111190796,
      "learning_rate": 0.00015576316042211467,
      "loss": 2.963,
      "step": 1160
    },
    {
      "epoch": 0.32151758515646633,
      "grad_norm": 2.043443441390991,
      "learning_rate": 0.00015568990649063164,
      "loss": 2.7848,
      "step": 1161
    },
    {
      "epoch": 0.32179451675436166,
      "grad_norm": 1.8811475038528442,
      "learning_rate": 0.00015561660921482217,
      "loss": 2.87,
      "step": 1162
    },
    {
      "epoch": 0.322071448352257,
      "grad_norm": 1.8709558248519897,
      "learning_rate": 0.00015554326865173469,
      "loss": 2.8862,
      "step": 1163
    },
    {
      "epoch": 0.3223483799501523,
      "grad_norm": 1.5902208089828491,
      "learning_rate": 0.00015546988485845125,
      "loss": 3.2132,
      "step": 1164
    },
    {
      "epoch": 0.32262531154804763,
      "grad_norm": 2.2096242904663086,
      "learning_rate": 0.00015539645789208772,
      "loss": 2.8945,
      "step": 1165
    },
    {
      "epoch": 0.32290224314594296,
      "grad_norm": 1.1872581243515015,
      "learning_rate": 0.00015532298780979336,
      "loss": 2.6953,
      "step": 1166
    },
    {
      "epoch": 0.3231791747438383,
      "grad_norm": 2.396256685256958,
      "learning_rate": 0.00015524947466875107,
      "loss": 2.9257,
      "step": 1167
    },
    {
      "epoch": 0.3234561063417336,
      "grad_norm": 1.961571455001831,
      "learning_rate": 0.00015517591852617736,
      "loss": 2.4849,
      "step": 1168
    },
    {
      "epoch": 0.32373303793962893,
      "grad_norm": 2.79026460647583,
      "learning_rate": 0.0001551023194393221,
      "loss": 2.7722,
      "step": 1169
    },
    {
      "epoch": 0.32400996953752426,
      "grad_norm": 1.9868056774139404,
      "learning_rate": 0.00015502867746546858,
      "loss": 2.6865,
      "step": 1170
    },
    {
      "epoch": 0.3242869011354195,
      "grad_norm": 1.6515250205993652,
      "learning_rate": 0.00015495499266193354,
      "loss": 2.9801,
      "step": 1171
    },
    {
      "epoch": 0.32456383273331485,
      "grad_norm": 1.5028882026672363,
      "learning_rate": 0.00015488126508606703,
      "loss": 2.7559,
      "step": 1172
    },
    {
      "epoch": 0.3248407643312102,
      "grad_norm": 2.3353559970855713,
      "learning_rate": 0.0001548074947952523,
      "loss": 2.845,
      "step": 1173
    },
    {
      "epoch": 0.3251176959291055,
      "grad_norm": 1.6773449182510376,
      "learning_rate": 0.00015473368184690597,
      "loss": 2.9368,
      "step": 1174
    },
    {
      "epoch": 0.3253946275270008,
      "grad_norm": 4.547130107879639,
      "learning_rate": 0.00015465982629847783,
      "loss": 4.1124,
      "step": 1175
    },
    {
      "epoch": 0.32567155912489615,
      "grad_norm": 1.3638783693313599,
      "learning_rate": 0.00015458592820745079,
      "loss": 2.9854,
      "step": 1176
    },
    {
      "epoch": 0.3259484907227915,
      "grad_norm": 1.6351149082183838,
      "learning_rate": 0.0001545119876313409,
      "loss": 2.6985,
      "step": 1177
    },
    {
      "epoch": 0.3262254223206868,
      "grad_norm": 2.6430916786193848,
      "learning_rate": 0.00015443800462769727,
      "loss": 3.2984,
      "step": 1178
    },
    {
      "epoch": 0.3265023539185821,
      "grad_norm": 1.3868046998977661,
      "learning_rate": 0.00015436397925410201,
      "loss": 2.6771,
      "step": 1179
    },
    {
      "epoch": 0.32677928551647745,
      "grad_norm": 2.804959535598755,
      "learning_rate": 0.00015428991156817029,
      "loss": 3.083,
      "step": 1180
    },
    {
      "epoch": 0.32705621711437277,
      "grad_norm": 1.8770880699157715,
      "learning_rate": 0.00015421580162755004,
      "loss": 2.8515,
      "step": 1181
    },
    {
      "epoch": 0.3273331487122681,
      "grad_norm": 1.4491260051727295,
      "learning_rate": 0.00015414164948992227,
      "loss": 2.7167,
      "step": 1182
    },
    {
      "epoch": 0.32761008031016337,
      "grad_norm": 1.5363740921020508,
      "learning_rate": 0.00015406745521300073,
      "loss": 2.74,
      "step": 1183
    },
    {
      "epoch": 0.3278870119080587,
      "grad_norm": 1.8011248111724854,
      "learning_rate": 0.00015399321885453202,
      "loss": 2.7164,
      "step": 1184
    },
    {
      "epoch": 0.328163943505954,
      "grad_norm": 2.1668336391448975,
      "learning_rate": 0.00015391894047229538,
      "loss": 2.9057,
      "step": 1185
    },
    {
      "epoch": 0.32844087510384934,
      "grad_norm": 1.4623689651489258,
      "learning_rate": 0.00015384462012410293,
      "loss": 2.7671,
      "step": 1186
    },
    {
      "epoch": 0.32871780670174466,
      "grad_norm": 1.0454082489013672,
      "learning_rate": 0.00015377025786779928,
      "loss": 2.982,
      "step": 1187
    },
    {
      "epoch": 0.32899473829964,
      "grad_norm": 1.496302604675293,
      "learning_rate": 0.0001536958537612618,
      "loss": 2.561,
      "step": 1188
    },
    {
      "epoch": 0.3292716698975353,
      "grad_norm": 2.425139904022217,
      "learning_rate": 0.00015362140786240035,
      "loss": 2.9257,
      "step": 1189
    },
    {
      "epoch": 0.32954860149543064,
      "grad_norm": 1.459738850593567,
      "learning_rate": 0.00015354692022915732,
      "loss": 3.065,
      "step": 1190
    },
    {
      "epoch": 0.32982553309332596,
      "grad_norm": 3.5508320331573486,
      "learning_rate": 0.00015347239091950763,
      "loss": 3.0286,
      "step": 1191
    },
    {
      "epoch": 0.3301024646912213,
      "grad_norm": 1.521572232246399,
      "learning_rate": 0.0001533978199914586,
      "loss": 2.7615,
      "step": 1192
    },
    {
      "epoch": 0.3303793962891166,
      "grad_norm": 2.401167392730713,
      "learning_rate": 0.00015332320750304992,
      "loss": 3.0155,
      "step": 1193
    },
    {
      "epoch": 0.3306563278870119,
      "grad_norm": 1.7472994327545166,
      "learning_rate": 0.00015324855351235372,
      "loss": 2.5185,
      "step": 1194
    },
    {
      "epoch": 0.3309332594849072,
      "grad_norm": 1.205310344696045,
      "learning_rate": 0.00015317385807747433,
      "loss": 2.5906,
      "step": 1195
    },
    {
      "epoch": 0.33121019108280253,
      "grad_norm": 1.574655294418335,
      "learning_rate": 0.0001530991212565484,
      "loss": 2.8073,
      "step": 1196
    },
    {
      "epoch": 0.33148712268069785,
      "grad_norm": 2.1971731185913086,
      "learning_rate": 0.00015302434310774472,
      "loss": 2.9192,
      "step": 1197
    },
    {
      "epoch": 0.3317640542785932,
      "grad_norm": 1.4974596500396729,
      "learning_rate": 0.00015294952368926433,
      "loss": 2.7797,
      "step": 1198
    },
    {
      "epoch": 0.3320409858764885,
      "grad_norm": 1.7164533138275146,
      "learning_rate": 0.00015287466305934037,
      "loss": 2.8779,
      "step": 1199
    },
    {
      "epoch": 0.33231791747438383,
      "grad_norm": 4.789784908294678,
      "learning_rate": 0.000152799761276238,
      "loss": 4.2732,
      "step": 1200
    },
    {
      "epoch": 0.33259484907227915,
      "grad_norm": 1.7506219148635864,
      "learning_rate": 0.00015272481839825452,
      "loss": 2.788,
      "step": 1201
    },
    {
      "epoch": 0.3328717806701745,
      "grad_norm": 2.5078749656677246,
      "learning_rate": 0.00015264983448371906,
      "loss": 3.158,
      "step": 1202
    },
    {
      "epoch": 0.3331487122680698,
      "grad_norm": 1.8120169639587402,
      "learning_rate": 0.00015257480959099282,
      "loss": 2.5553,
      "step": 1203
    },
    {
      "epoch": 0.3334256438659651,
      "grad_norm": 1.7837135791778564,
      "learning_rate": 0.0001524997437784689,
      "loss": 2.6184,
      "step": 1204
    },
    {
      "epoch": 0.33370257546386045,
      "grad_norm": 1.3695642948150635,
      "learning_rate": 0.00015242463710457208,
      "loss": 2.673,
      "step": 1205
    },
    {
      "epoch": 0.3339795070617557,
      "grad_norm": 1.6549372673034668,
      "learning_rate": 0.0001523494896277591,
      "loss": 2.6288,
      "step": 1206
    },
    {
      "epoch": 0.33425643865965105,
      "grad_norm": 2.5411343574523926,
      "learning_rate": 0.0001522743014065185,
      "loss": 2.8383,
      "step": 1207
    },
    {
      "epoch": 0.33453337025754637,
      "grad_norm": 2.1475119590759277,
      "learning_rate": 0.00015219907249937036,
      "loss": 2.9761,
      "step": 1208
    },
    {
      "epoch": 0.3348103018554417,
      "grad_norm": 2.9487884044647217,
      "learning_rate": 0.00015212380296486652,
      "loss": 2.8682,
      "step": 1209
    },
    {
      "epoch": 0.335087233453337,
      "grad_norm": 2.9902172088623047,
      "learning_rate": 0.00015204849286159052,
      "loss": 3.1509,
      "step": 1210
    },
    {
      "epoch": 0.33536416505123234,
      "grad_norm": 1.7279140949249268,
      "learning_rate": 0.0001519731422481573,
      "loss": 2.7741,
      "step": 1211
    },
    {
      "epoch": 0.33564109664912767,
      "grad_norm": 2.067688465118408,
      "learning_rate": 0.00015189775118321346,
      "loss": 2.9027,
      "step": 1212
    },
    {
      "epoch": 0.335918028247023,
      "grad_norm": 1.4905089139938354,
      "learning_rate": 0.00015182231972543707,
      "loss": 3.028,
      "step": 1213
    },
    {
      "epoch": 0.3361949598449183,
      "grad_norm": 1.398120403289795,
      "learning_rate": 0.0001517468479335376,
      "loss": 3.0139,
      "step": 1214
    },
    {
      "epoch": 0.33647189144281364,
      "grad_norm": 2.3304412364959717,
      "learning_rate": 0.00015167133586625587,
      "loss": 3.2774,
      "step": 1215
    },
    {
      "epoch": 0.33674882304070897,
      "grad_norm": 1.7044647932052612,
      "learning_rate": 0.00015159578358236422,
      "loss": 2.5921,
      "step": 1216
    },
    {
      "epoch": 0.3370257546386043,
      "grad_norm": 1.8110103607177734,
      "learning_rate": 0.00015152019114066608,
      "loss": 2.9911,
      "step": 1217
    },
    {
      "epoch": 0.33730268623649956,
      "grad_norm": 1.3686960935592651,
      "learning_rate": 0.0001514445585999962,
      "loss": 2.7397,
      "step": 1218
    },
    {
      "epoch": 0.3375796178343949,
      "grad_norm": 2.4159533977508545,
      "learning_rate": 0.00015136888601922072,
      "loss": 2.9673,
      "step": 1219
    },
    {
      "epoch": 0.3378565494322902,
      "grad_norm": 1.7092558145523071,
      "learning_rate": 0.00015129317345723665,
      "loss": 2.6965,
      "step": 1220
    },
    {
      "epoch": 0.33813348103018553,
      "grad_norm": 1.7533438205718994,
      "learning_rate": 0.00015121742097297237,
      "loss": 2.7309,
      "step": 1221
    },
    {
      "epoch": 0.33841041262808086,
      "grad_norm": 1.2231271266937256,
      "learning_rate": 0.00015114162862538713,
      "loss": 2.9981,
      "step": 1222
    },
    {
      "epoch": 0.3386873442259762,
      "grad_norm": 1.5578258037567139,
      "learning_rate": 0.00015106579647347136,
      "loss": 2.9053,
      "step": 1223
    },
    {
      "epoch": 0.3389642758238715,
      "grad_norm": 1.9765286445617676,
      "learning_rate": 0.0001509899245762464,
      "loss": 3.091,
      "step": 1224
    },
    {
      "epoch": 0.33924120742176683,
      "grad_norm": 4.126635551452637,
      "learning_rate": 0.00015091401299276456,
      "loss": 3.932,
      "step": 1225
    },
    {
      "epoch": 0.33951813901966216,
      "grad_norm": 2.4818685054779053,
      "learning_rate": 0.00015083806178210895,
      "loss": 2.709,
      "step": 1226
    },
    {
      "epoch": 0.3397950706175575,
      "grad_norm": 1.4024157524108887,
      "learning_rate": 0.00015076207100339363,
      "loss": 2.7262,
      "step": 1227
    },
    {
      "epoch": 0.3400720022154528,
      "grad_norm": 1.5914798974990845,
      "learning_rate": 0.00015068604071576344,
      "loss": 2.6496,
      "step": 1228
    },
    {
      "epoch": 0.3403489338133481,
      "grad_norm": 2.053748607635498,
      "learning_rate": 0.00015060997097839386,
      "loss": 2.8565,
      "step": 1229
    },
    {
      "epoch": 0.3406258654112434,
      "grad_norm": 1.3927969932556152,
      "learning_rate": 0.00015053386185049123,
      "loss": 2.8148,
      "step": 1230
    },
    {
      "epoch": 0.3409027970091387,
      "grad_norm": 1.5601234436035156,
      "learning_rate": 0.00015045771339129246,
      "loss": 2.6251,
      "step": 1231
    },
    {
      "epoch": 0.34117972860703405,
      "grad_norm": 2.194117546081543,
      "learning_rate": 0.00015038152566006509,
      "loss": 2.6468,
      "step": 1232
    },
    {
      "epoch": 0.3414566602049294,
      "grad_norm": 1.6312081813812256,
      "learning_rate": 0.00015030529871610718,
      "loss": 2.8919,
      "step": 1233
    },
    {
      "epoch": 0.3417335918028247,
      "grad_norm": 1.736798882484436,
      "learning_rate": 0.00015022903261874748,
      "loss": 2.6619,
      "step": 1234
    },
    {
      "epoch": 0.34201052340072,
      "grad_norm": 2.73242449760437,
      "learning_rate": 0.00015015272742734492,
      "loss": 3.0658,
      "step": 1235
    },
    {
      "epoch": 0.34228745499861535,
      "grad_norm": 2.544443130493164,
      "learning_rate": 0.00015007638320128913,
      "loss": 3.2522,
      "step": 1236
    },
    {
      "epoch": 0.3425643865965107,
      "grad_norm": 2.2239553928375244,
      "learning_rate": 0.00015000000000000001,
      "loss": 2.7484,
      "step": 1237
    },
    {
      "epoch": 0.342841318194406,
      "grad_norm": 1.7828017473220825,
      "learning_rate": 0.00014992357788292776,
      "loss": 2.6737,
      "step": 1238
    },
    {
      "epoch": 0.3431182497923013,
      "grad_norm": 1.5759692192077637,
      "learning_rate": 0.00014984711690955297,
      "loss": 2.7174,
      "step": 1239
    },
    {
      "epoch": 0.34339518139019665,
      "grad_norm": 1.4020365476608276,
      "learning_rate": 0.00014977061713938636,
      "loss": 2.8425,
      "step": 1240
    },
    {
      "epoch": 0.3436721129880919,
      "grad_norm": 1.759036660194397,
      "learning_rate": 0.0001496940786319689,
      "loss": 2.9138,
      "step": 1241
    },
    {
      "epoch": 0.34394904458598724,
      "grad_norm": 1.7949292659759521,
      "learning_rate": 0.00014961750144687174,
      "loss": 2.7142,
      "step": 1242
    },
    {
      "epoch": 0.34422597618388256,
      "grad_norm": 1.4500319957733154,
      "learning_rate": 0.0001495408856436961,
      "loss": 2.7205,
      "step": 1243
    },
    {
      "epoch": 0.3445029077817779,
      "grad_norm": 2.0495407581329346,
      "learning_rate": 0.00014946423128207322,
      "loss": 2.7994,
      "step": 1244
    },
    {
      "epoch": 0.3447798393796732,
      "grad_norm": 2.7560412883758545,
      "learning_rate": 0.00014938753842166444,
      "loss": 2.76,
      "step": 1245
    },
    {
      "epoch": 0.34505677097756854,
      "grad_norm": 2.827164888381958,
      "learning_rate": 0.00014931080712216104,
      "loss": 2.9787,
      "step": 1246
    },
    {
      "epoch": 0.34533370257546386,
      "grad_norm": 1.410880446434021,
      "learning_rate": 0.00014923403744328408,
      "loss": 3.0478,
      "step": 1247
    },
    {
      "epoch": 0.3456106341733592,
      "grad_norm": 1.8312749862670898,
      "learning_rate": 0.0001491572294447847,
      "loss": 2.7135,
      "step": 1248
    },
    {
      "epoch": 0.3458875657712545,
      "grad_norm": 3.0767037868499756,
      "learning_rate": 0.00014908038318644373,
      "loss": 3.2266,
      "step": 1249
    },
    {
      "epoch": 0.34616449736914984,
      "grad_norm": 5.677850723266602,
      "learning_rate": 0.00014900349872807182,
      "loss": 4.2348,
      "step": 1250
    },
    {
      "epoch": 0.34644142896704516,
      "grad_norm": 2.243927240371704,
      "learning_rate": 0.00014892657612950935,
      "loss": 2.9267,
      "step": 1251
    },
    {
      "epoch": 0.3467183605649405,
      "grad_norm": 2.399611234664917,
      "learning_rate": 0.0001488496154506264,
      "loss": 2.7608,
      "step": 1252
    },
    {
      "epoch": 0.34699529216283576,
      "grad_norm": 2.9935808181762695,
      "learning_rate": 0.00014877261675132265,
      "loss": 3.0759,
      "step": 1253
    },
    {
      "epoch": 0.3472722237607311,
      "grad_norm": 2.5465614795684814,
      "learning_rate": 0.0001486955800915274,
      "loss": 2.7247,
      "step": 1254
    },
    {
      "epoch": 0.3475491553586264,
      "grad_norm": 2.167867422103882,
      "learning_rate": 0.0001486185055311995,
      "loss": 2.6247,
      "step": 1255
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 2.093701124191284,
      "learning_rate": 0.00014854139313032726,
      "loss": 2.7427,
      "step": 1256
    },
    {
      "epoch": 0.34810301855441705,
      "grad_norm": 2.3673460483551025,
      "learning_rate": 0.00014846424294892848,
      "loss": 2.7806,
      "step": 1257
    },
    {
      "epoch": 0.3483799501523124,
      "grad_norm": 2.3842132091522217,
      "learning_rate": 0.0001483870550470504,
      "loss": 2.9437,
      "step": 1258
    },
    {
      "epoch": 0.3486568817502077,
      "grad_norm": 2.410362482070923,
      "learning_rate": 0.0001483098294847695,
      "loss": 2.848,
      "step": 1259
    },
    {
      "epoch": 0.34893381334810303,
      "grad_norm": 2.2598326206207275,
      "learning_rate": 0.00014823256632219168,
      "loss": 2.9541,
      "step": 1260
    },
    {
      "epoch": 0.34921074494599835,
      "grad_norm": 2.2390294075012207,
      "learning_rate": 0.0001481552656194521,
      "loss": 2.9345,
      "step": 1261
    },
    {
      "epoch": 0.3494876765438937,
      "grad_norm": 1.682488203048706,
      "learning_rate": 0.00014807792743671506,
      "loss": 2.7376,
      "step": 1262
    },
    {
      "epoch": 0.349764608141789,
      "grad_norm": 2.382782220840454,
      "learning_rate": 0.00014800055183417408,
      "loss": 2.9122,
      "step": 1263
    },
    {
      "epoch": 0.35004153973968427,
      "grad_norm": 2.7201778888702393,
      "learning_rate": 0.00014792313887205182,
      "loss": 3.1305,
      "step": 1264
    },
    {
      "epoch": 0.3503184713375796,
      "grad_norm": 2.838747501373291,
      "learning_rate": 0.0001478456886106,
      "loss": 2.876,
      "step": 1265
    },
    {
      "epoch": 0.3505954029354749,
      "grad_norm": 2.1791601181030273,
      "learning_rate": 0.00014776820111009936,
      "loss": 2.9689,
      "step": 1266
    },
    {
      "epoch": 0.35087233453337024,
      "grad_norm": 2.376283645629883,
      "learning_rate": 0.0001476906764308597,
      "loss": 2.8344,
      "step": 1267
    },
    {
      "epoch": 0.35114926613126557,
      "grad_norm": 1.7677280902862549,
      "learning_rate": 0.00014761311463321958,
      "loss": 2.8116,
      "step": 1268
    },
    {
      "epoch": 0.3514261977291609,
      "grad_norm": 2.8490333557128906,
      "learning_rate": 0.00014753551577754664,
      "loss": 2.9138,
      "step": 1269
    },
    {
      "epoch": 0.3517031293270562,
      "grad_norm": 1.9829556941986084,
      "learning_rate": 0.0001474578799242373,
      "loss": 2.7738,
      "step": 1270
    },
    {
      "epoch": 0.35198006092495154,
      "grad_norm": 2.358916997909546,
      "learning_rate": 0.00014738020713371666,
      "loss": 2.9176,
      "step": 1271
    },
    {
      "epoch": 0.35225699252284687,
      "grad_norm": 2.4208502769470215,
      "learning_rate": 0.0001473024974664388,
      "loss": 2.9986,
      "step": 1272
    },
    {
      "epoch": 0.3525339241207422,
      "grad_norm": 1.8187532424926758,
      "learning_rate": 0.00014722475098288625,
      "loss": 2.9038,
      "step": 1273
    },
    {
      "epoch": 0.3528108557186375,
      "grad_norm": 2.0144996643066406,
      "learning_rate": 0.0001471469677435704,
      "loss": 2.8533,
      "step": 1274
    },
    {
      "epoch": 0.35308778731653284,
      "grad_norm": 2.9319944381713867,
      "learning_rate": 0.00014706914780903113,
      "loss": 3.6354,
      "step": 1275
    },
    {
      "epoch": 0.3533647189144281,
      "grad_norm": 1.9010802507400513,
      "learning_rate": 0.0001469912912398369,
      "loss": 2.853,
      "step": 1276
    },
    {
      "epoch": 0.35364165051232344,
      "grad_norm": 2.424229621887207,
      "learning_rate": 0.00014691339809658473,
      "loss": 2.8508,
      "step": 1277
    },
    {
      "epoch": 0.35391858211021876,
      "grad_norm": 2.659932851791382,
      "learning_rate": 0.00014683546843990007,
      "loss": 2.8557,
      "step": 1278
    },
    {
      "epoch": 0.3541955137081141,
      "grad_norm": 1.6898329257965088,
      "learning_rate": 0.00014675750233043679,
      "loss": 2.9636,
      "step": 1279
    },
    {
      "epoch": 0.3544724453060094,
      "grad_norm": 1.7070603370666504,
      "learning_rate": 0.0001466794998288771,
      "loss": 2.6075,
      "step": 1280
    },
    {
      "epoch": 0.35474937690390473,
      "grad_norm": 2.196258068084717,
      "learning_rate": 0.0001466014609959316,
      "loss": 2.8587,
      "step": 1281
    },
    {
      "epoch": 0.35502630850180006,
      "grad_norm": 2.935746669769287,
      "learning_rate": 0.00014652338589233913,
      "loss": 2.6526,
      "step": 1282
    },
    {
      "epoch": 0.3553032400996954,
      "grad_norm": 2.032379627227783,
      "learning_rate": 0.0001464452745788668,
      "loss": 2.5506,
      "step": 1283
    },
    {
      "epoch": 0.3555801716975907,
      "grad_norm": 4.194707870483398,
      "learning_rate": 0.00014636712711630978,
      "loss": 2.7623,
      "step": 1284
    },
    {
      "epoch": 0.35585710329548603,
      "grad_norm": 2.7065067291259766,
      "learning_rate": 0.00014628894356549156,
      "loss": 2.569,
      "step": 1285
    },
    {
      "epoch": 0.35613403489338136,
      "grad_norm": 1.7280241250991821,
      "learning_rate": 0.00014621072398726356,
      "loss": 2.9045,
      "step": 1286
    },
    {
      "epoch": 0.3564109664912767,
      "grad_norm": 2.0148704051971436,
      "learning_rate": 0.0001461324684425053,
      "loss": 3.1877,
      "step": 1287
    },
    {
      "epoch": 0.35668789808917195,
      "grad_norm": 9.018928527832031,
      "learning_rate": 0.0001460541769921244,
      "loss": 3.0865,
      "step": 1288
    },
    {
      "epoch": 0.3569648296870673,
      "grad_norm": 4.8206562995910645,
      "learning_rate": 0.00014597584969705616,
      "loss": 2.631,
      "step": 1289
    },
    {
      "epoch": 0.3572417612849626,
      "grad_norm": 1.6684445142745972,
      "learning_rate": 0.00014589748661826402,
      "loss": 2.6218,
      "step": 1290
    },
    {
      "epoch": 0.3575186928828579,
      "grad_norm": 1.579044222831726,
      "learning_rate": 0.00014581908781673928,
      "loss": 2.9571,
      "step": 1291
    },
    {
      "epoch": 0.35779562448075325,
      "grad_norm": 1.2074787616729736,
      "learning_rate": 0.0001457406533535008,
      "loss": 2.7196,
      "step": 1292
    },
    {
      "epoch": 0.3580725560786486,
      "grad_norm": 2.3285152912139893,
      "learning_rate": 0.00014566218328959543,
      "loss": 2.6068,
      "step": 1293
    },
    {
      "epoch": 0.3583494876765439,
      "grad_norm": 1.3609609603881836,
      "learning_rate": 0.00014558367768609766,
      "loss": 2.5368,
      "step": 1294
    },
    {
      "epoch": 0.3586264192744392,
      "grad_norm": 1.4789910316467285,
      "learning_rate": 0.0001455051366041096,
      "loss": 2.7763,
      "step": 1295
    },
    {
      "epoch": 0.35890335087233455,
      "grad_norm": 1.3611867427825928,
      "learning_rate": 0.000145426560104761,
      "loss": 2.6266,
      "step": 1296
    },
    {
      "epoch": 0.3591802824702299,
      "grad_norm": 1.7433539628982544,
      "learning_rate": 0.00014534794824920923,
      "loss": 2.9559,
      "step": 1297
    },
    {
      "epoch": 0.3594572140681252,
      "grad_norm": 1.8061453104019165,
      "learning_rate": 0.00014526930109863906,
      "loss": 2.7681,
      "step": 1298
    },
    {
      "epoch": 0.35973414566602047,
      "grad_norm": 2.418627977371216,
      "learning_rate": 0.00014519061871426286,
      "loss": 3.0256,
      "step": 1299
    },
    {
      "epoch": 0.3600110772639158,
      "grad_norm": 2.375866174697876,
      "learning_rate": 0.00014511190115732033,
      "loss": 3.3018,
      "step": 1300
    },
    {
      "epoch": 0.3602880088618111,
      "grad_norm": 1.1713021993637085,
      "learning_rate": 0.00014503314848907857,
      "loss": 2.8696,
      "step": 1301
    },
    {
      "epoch": 0.36056494045970644,
      "grad_norm": 2.7128472328186035,
      "learning_rate": 0.00014495436077083203,
      "loss": 2.8759,
      "step": 1302
    },
    {
      "epoch": 0.36084187205760176,
      "grad_norm": 1.0512962341308594,
      "learning_rate": 0.00014487553806390245,
      "loss": 2.5979,
      "step": 1303
    },
    {
      "epoch": 0.3611188036554971,
      "grad_norm": 4.019785404205322,
      "learning_rate": 0.0001447966804296387,
      "loss": 2.7857,
      "step": 1304
    },
    {
      "epoch": 0.3613957352533924,
      "grad_norm": 2.5276665687561035,
      "learning_rate": 0.000144717787929417,
      "loss": 3.0544,
      "step": 1305
    },
    {
      "epoch": 0.36167266685128774,
      "grad_norm": 2.3366951942443848,
      "learning_rate": 0.00014463886062464056,
      "loss": 2.9433,
      "step": 1306
    },
    {
      "epoch": 0.36194959844918306,
      "grad_norm": 1.6152623891830444,
      "learning_rate": 0.00014455989857673976,
      "loss": 2.6153,
      "step": 1307
    },
    {
      "epoch": 0.3622265300470784,
      "grad_norm": 2.1880078315734863,
      "learning_rate": 0.000144480901847172,
      "loss": 2.5427,
      "step": 1308
    },
    {
      "epoch": 0.3625034616449737,
      "grad_norm": 1.9220170974731445,
      "learning_rate": 0.00014440187049742165,
      "loss": 2.9394,
      "step": 1309
    },
    {
      "epoch": 0.36278039324286904,
      "grad_norm": 1.6536104679107666,
      "learning_rate": 0.00014432280458900009,
      "loss": 2.6208,
      "step": 1310
    },
    {
      "epoch": 0.3630573248407643,
      "grad_norm": 2.5638813972473145,
      "learning_rate": 0.00014424370418344553,
      "loss": 2.5677,
      "step": 1311
    },
    {
      "epoch": 0.36333425643865963,
      "grad_norm": 1.7210147380828857,
      "learning_rate": 0.00014416456934232308,
      "loss": 2.701,
      "step": 1312
    },
    {
      "epoch": 0.36361118803655496,
      "grad_norm": 1.981189250946045,
      "learning_rate": 0.00014408540012722455,
      "loss": 2.8845,
      "step": 1313
    },
    {
      "epoch": 0.3638881196344503,
      "grad_norm": 2.588027238845825,
      "learning_rate": 0.00014400619659976863,
      "loss": 2.8426,
      "step": 1314
    },
    {
      "epoch": 0.3641650512323456,
      "grad_norm": 1.580614447593689,
      "learning_rate": 0.00014392695882160073,
      "loss": 2.8997,
      "step": 1315
    },
    {
      "epoch": 0.36444198283024093,
      "grad_norm": 1.2011154890060425,
      "learning_rate": 0.00014384768685439273,
      "loss": 2.6991,
      "step": 1316
    },
    {
      "epoch": 0.36471891442813625,
      "grad_norm": 3.265779733657837,
      "learning_rate": 0.00014376838075984334,
      "loss": 3.0898,
      "step": 1317
    },
    {
      "epoch": 0.3649958460260316,
      "grad_norm": 2.203280448913574,
      "learning_rate": 0.00014368904059967769,
      "loss": 2.6934,
      "step": 1318
    },
    {
      "epoch": 0.3652727776239269,
      "grad_norm": 2.0738298892974854,
      "learning_rate": 0.00014360966643564747,
      "loss": 2.8216,
      "step": 1319
    },
    {
      "epoch": 0.3655497092218222,
      "grad_norm": 2.641373634338379,
      "learning_rate": 0.0001435302583295308,
      "loss": 2.854,
      "step": 1320
    },
    {
      "epoch": 0.36582664081971755,
      "grad_norm": 1.9381299018859863,
      "learning_rate": 0.00014345081634313238,
      "loss": 2.7057,
      "step": 1321
    },
    {
      "epoch": 0.3661035724176129,
      "grad_norm": 1.7109977006912231,
      "learning_rate": 0.00014337134053828306,
      "loss": 2.9296,
      "step": 1322
    },
    {
      "epoch": 0.36638050401550815,
      "grad_norm": 1.5071414709091187,
      "learning_rate": 0.00014329183097684008,
      "loss": 2.6535,
      "step": 1323
    },
    {
      "epoch": 0.36665743561340347,
      "grad_norm": 1.8007851839065552,
      "learning_rate": 0.00014321228772068702,
      "loss": 2.8942,
      "step": 1324
    },
    {
      "epoch": 0.3669343672112988,
      "grad_norm": 3.320955753326416,
      "learning_rate": 0.00014313271083173364,
      "loss": 3.8501,
      "step": 1325
    },
    {
      "epoch": 0.3672112988091941,
      "grad_norm": 1.723261833190918,
      "learning_rate": 0.00014305310037191583,
      "loss": 2.4756,
      "step": 1326
    },
    {
      "epoch": 0.36748823040708944,
      "grad_norm": 1.984232783317566,
      "learning_rate": 0.00014297345640319574,
      "loss": 2.8204,
      "step": 1327
    },
    {
      "epoch": 0.36776516200498477,
      "grad_norm": 2.312904119491577,
      "learning_rate": 0.0001428937789875615,
      "loss": 2.8424,
      "step": 1328
    },
    {
      "epoch": 0.3680420936028801,
      "grad_norm": 1.8733426332473755,
      "learning_rate": 0.0001428140681870272,
      "loss": 2.7003,
      "step": 1329
    },
    {
      "epoch": 0.3683190252007754,
      "grad_norm": 2.389039993286133,
      "learning_rate": 0.00014273432406363313,
      "loss": 2.5219,
      "step": 1330
    },
    {
      "epoch": 0.36859595679867074,
      "grad_norm": 2.193798780441284,
      "learning_rate": 0.0001426545466794453,
      "loss": 3.1102,
      "step": 1331
    },
    {
      "epoch": 0.36887288839656607,
      "grad_norm": 1.861578345298767,
      "learning_rate": 0.00014257473609655575,
      "loss": 2.7627,
      "step": 1332
    },
    {
      "epoch": 0.3691498199944614,
      "grad_norm": 2.2872440814971924,
      "learning_rate": 0.00014249489237708226,
      "loss": 2.7749,
      "step": 1333
    },
    {
      "epoch": 0.36942675159235666,
      "grad_norm": 2.083334445953369,
      "learning_rate": 0.0001424150155831685,
      "loss": 2.8019,
      "step": 1334
    },
    {
      "epoch": 0.369703683190252,
      "grad_norm": 2.032106876373291,
      "learning_rate": 0.00014233510577698377,
      "loss": 2.8242,
      "step": 1335
    },
    {
      "epoch": 0.3699806147881473,
      "grad_norm": 1.8062784671783447,
      "learning_rate": 0.0001422551630207232,
      "loss": 2.7238,
      "step": 1336
    },
    {
      "epoch": 0.37025754638604264,
      "grad_norm": 2.386137008666992,
      "learning_rate": 0.00014217518737660742,
      "loss": 2.7767,
      "step": 1337
    },
    {
      "epoch": 0.37053447798393796,
      "grad_norm": 2.164215087890625,
      "learning_rate": 0.00014209517890688279,
      "loss": 3.1791,
      "step": 1338
    },
    {
      "epoch": 0.3708114095818333,
      "grad_norm": 2.2616584300994873,
      "learning_rate": 0.00014201513767382108,
      "loss": 2.9295,
      "step": 1339
    },
    {
      "epoch": 0.3710883411797286,
      "grad_norm": 2.6709892749786377,
      "learning_rate": 0.00014193506373971968,
      "loss": 2.8127,
      "step": 1340
    },
    {
      "epoch": 0.37136527277762393,
      "grad_norm": 2.624154567718506,
      "learning_rate": 0.0001418549571669014,
      "loss": 2.9689,
      "step": 1341
    },
    {
      "epoch": 0.37164220437551926,
      "grad_norm": 3.4025349617004395,
      "learning_rate": 0.00014177481801771439,
      "loss": 3.2772,
      "step": 1342
    },
    {
      "epoch": 0.3719191359734146,
      "grad_norm": 2.731710433959961,
      "learning_rate": 0.00014169464635453222,
      "loss": 2.5917,
      "step": 1343
    },
    {
      "epoch": 0.3721960675713099,
      "grad_norm": 2.406323194503784,
      "learning_rate": 0.00014161444223975383,
      "loss": 3.0855,
      "step": 1344
    },
    {
      "epoch": 0.37247299916920523,
      "grad_norm": 2.2487471103668213,
      "learning_rate": 0.00014153420573580322,
      "loss": 2.7417,
      "step": 1345
    },
    {
      "epoch": 0.3727499307671005,
      "grad_norm": 1.1252191066741943,
      "learning_rate": 0.0001414539369051298,
      "loss": 2.7274,
      "step": 1346
    },
    {
      "epoch": 0.3730268623649958,
      "grad_norm": 0.8789071440696716,
      "learning_rate": 0.00014137363581020803,
      "loss": 2.8867,
      "step": 1347
    },
    {
      "epoch": 0.37330379396289115,
      "grad_norm": 2.283622980117798,
      "learning_rate": 0.0001412933025135375,
      "loss": 2.8808,
      "step": 1348
    },
    {
      "epoch": 0.3735807255607865,
      "grad_norm": 2.586961030960083,
      "learning_rate": 0.0001412129370776429,
      "loss": 2.8784,
      "step": 1349
    },
    {
      "epoch": 0.3738576571586818,
      "grad_norm": 6.691732406616211,
      "learning_rate": 0.0001411325395650739,
      "loss": 3.9918,
      "step": 1350
    },
    {
      "epoch": 0.3741345887565771,
      "grad_norm": 3.5954439640045166,
      "learning_rate": 0.0001410521100384051,
      "loss": 2.8741,
      "step": 1351
    },
    {
      "epoch": 0.37441152035447245,
      "grad_norm": 3.1264543533325195,
      "learning_rate": 0.0001409716485602361,
      "loss": 3.2264,
      "step": 1352
    },
    {
      "epoch": 0.3746884519523678,
      "grad_norm": 2.292165994644165,
      "learning_rate": 0.00014089115519319134,
      "loss": 2.7182,
      "step": 1353
    },
    {
      "epoch": 0.3749653835502631,
      "grad_norm": 1.8987075090408325,
      "learning_rate": 0.00014081062999992005,
      "loss": 2.7812,
      "step": 1354
    },
    {
      "epoch": 0.3752423151481584,
      "grad_norm": 1.3458808660507202,
      "learning_rate": 0.00014073007304309625,
      "loss": 2.8054,
      "step": 1355
    },
    {
      "epoch": 0.37551924674605375,
      "grad_norm": 1.2150719165802002,
      "learning_rate": 0.0001406494843854187,
      "loss": 2.8242,
      "step": 1356
    },
    {
      "epoch": 0.37579617834394907,
      "grad_norm": 1.3368840217590332,
      "learning_rate": 0.00014056886408961078,
      "loss": 2.6038,
      "step": 1357
    },
    {
      "epoch": 0.37607310994184434,
      "grad_norm": 1.6765084266662598,
      "learning_rate": 0.00014048821221842052,
      "loss": 2.4216,
      "step": 1358
    },
    {
      "epoch": 0.37635004153973967,
      "grad_norm": 1.8513046503067017,
      "learning_rate": 0.0001404075288346206,
      "loss": 2.6081,
      "step": 1359
    },
    {
      "epoch": 0.376626973137635,
      "grad_norm": 2.7676756381988525,
      "learning_rate": 0.00014032681400100812,
      "loss": 2.8518,
      "step": 1360
    },
    {
      "epoch": 0.3769039047355303,
      "grad_norm": 2.3777315616607666,
      "learning_rate": 0.00014024606778040467,
      "loss": 2.5985,
      "step": 1361
    },
    {
      "epoch": 0.37718083633342564,
      "grad_norm": 1.4810510873794556,
      "learning_rate": 0.00014016529023565632,
      "loss": 2.9121,
      "step": 1362
    },
    {
      "epoch": 0.37745776793132096,
      "grad_norm": 2.7724931240081787,
      "learning_rate": 0.00014008448142963355,
      "loss": 2.6499,
      "step": 1363
    },
    {
      "epoch": 0.3777346995292163,
      "grad_norm": 1.3956025838851929,
      "learning_rate": 0.00014000364142523103,
      "loss": 2.8069,
      "step": 1364
    },
    {
      "epoch": 0.3780116311271116,
      "grad_norm": 2.4681835174560547,
      "learning_rate": 0.0001399227702853679,
      "loss": 2.8497,
      "step": 1365
    },
    {
      "epoch": 0.37828856272500694,
      "grad_norm": 1.7103861570358276,
      "learning_rate": 0.00013984186807298736,
      "loss": 2.6658,
      "step": 1366
    },
    {
      "epoch": 0.37856549432290226,
      "grad_norm": 2.5172946453094482,
      "learning_rate": 0.0001397609348510569,
      "loss": 2.9519,
      "step": 1367
    },
    {
      "epoch": 0.3788424259207976,
      "grad_norm": 1.6615102291107178,
      "learning_rate": 0.0001396799706825681,
      "loss": 2.6468,
      "step": 1368
    },
    {
      "epoch": 0.37911935751869286,
      "grad_norm": 1.980188012123108,
      "learning_rate": 0.00013959897563053662,
      "loss": 2.7114,
      "step": 1369
    },
    {
      "epoch": 0.3793962891165882,
      "grad_norm": 1.8296494483947754,
      "learning_rate": 0.00013951794975800222,
      "loss": 2.406,
      "step": 1370
    },
    {
      "epoch": 0.3796732207144835,
      "grad_norm": 1.7942696809768677,
      "learning_rate": 0.00013943689312802864,
      "loss": 2.5578,
      "step": 1371
    },
    {
      "epoch": 0.37995015231237883,
      "grad_norm": 2.6955957412719727,
      "learning_rate": 0.0001393558058037034,
      "loss": 2.8052,
      "step": 1372
    },
    {
      "epoch": 0.38022708391027416,
      "grad_norm": 2.0833001136779785,
      "learning_rate": 0.00013927468784813814,
      "loss": 3.0135,
      "step": 1373
    },
    {
      "epoch": 0.3805040155081695,
      "grad_norm": 1.5874693393707275,
      "learning_rate": 0.00013919353932446822,
      "loss": 2.8376,
      "step": 1374
    },
    {
      "epoch": 0.3807809471060648,
      "grad_norm": 6.984336853027344,
      "learning_rate": 0.00013911236029585274,
      "loss": 4.5221,
      "step": 1375
    },
    {
      "epoch": 0.38105787870396013,
      "grad_norm": 52.98382568359375,
      "learning_rate": 0.0001390311508254747,
      "loss": 3.6362,
      "step": 1376
    },
    {
      "epoch": 0.38133481030185545,
      "grad_norm": 1.5272222757339478,
      "learning_rate": 0.00013894991097654068,
      "loss": 2.8065,
      "step": 1377
    },
    {
      "epoch": 0.3816117418997508,
      "grad_norm": 1.4378056526184082,
      "learning_rate": 0.00013886864081228087,
      "loss": 2.6339,
      "step": 1378
    },
    {
      "epoch": 0.3818886734976461,
      "grad_norm": 2.1655452251434326,
      "learning_rate": 0.0001387873403959492,
      "loss": 2.9441,
      "step": 1379
    },
    {
      "epoch": 0.3821656050955414,
      "grad_norm": 2.181152105331421,
      "learning_rate": 0.000138706009790823,
      "loss": 2.7229,
      "step": 1380
    },
    {
      "epoch": 0.3824425366934367,
      "grad_norm": 1.6623657941818237,
      "learning_rate": 0.0001386246490602031,
      "loss": 2.8498,
      "step": 1381
    },
    {
      "epoch": 0.382719468291332,
      "grad_norm": 2.1751708984375,
      "learning_rate": 0.00013854325826741394,
      "loss": 2.6604,
      "step": 1382
    },
    {
      "epoch": 0.38299639988922735,
      "grad_norm": 1.7822431325912476,
      "learning_rate": 0.00013846183747580318,
      "loss": 2.7339,
      "step": 1383
    },
    {
      "epoch": 0.38327333148712267,
      "grad_norm": 2.052023410797119,
      "learning_rate": 0.00013838038674874193,
      "loss": 2.8987,
      "step": 1384
    },
    {
      "epoch": 0.383550263085018,
      "grad_norm": 1.9398444890975952,
      "learning_rate": 0.00013829890614962457,
      "loss": 2.8354,
      "step": 1385
    },
    {
      "epoch": 0.3838271946829133,
      "grad_norm": 1.7574689388275146,
      "learning_rate": 0.0001382173957418687,
      "loss": 2.9091,
      "step": 1386
    },
    {
      "epoch": 0.38410412628080864,
      "grad_norm": 1.563295602798462,
      "learning_rate": 0.00013813585558891518,
      "loss": 2.5467,
      "step": 1387
    },
    {
      "epoch": 0.38438105787870397,
      "grad_norm": 1.682442307472229,
      "learning_rate": 0.00013805428575422793,
      "loss": 2.6808,
      "step": 1388
    },
    {
      "epoch": 0.3846579894765993,
      "grad_norm": 2.0732030868530273,
      "learning_rate": 0.00013797268630129413,
      "loss": 2.7206,
      "step": 1389
    },
    {
      "epoch": 0.3849349210744946,
      "grad_norm": 1.4792330265045166,
      "learning_rate": 0.0001378910572936238,
      "loss": 2.8664,
      "step": 1390
    },
    {
      "epoch": 0.38521185267238994,
      "grad_norm": 2.685608148574829,
      "learning_rate": 0.00013780939879475012,
      "loss": 3.2621,
      "step": 1391
    },
    {
      "epoch": 0.3854887842702852,
      "grad_norm": 1.362554669380188,
      "learning_rate": 0.0001377277108682292,
      "loss": 2.8747,
      "step": 1392
    },
    {
      "epoch": 0.38576571586818054,
      "grad_norm": 1.9244787693023682,
      "learning_rate": 0.00013764599357764002,
      "loss": 2.6234,
      "step": 1393
    },
    {
      "epoch": 0.38604264746607586,
      "grad_norm": 1.9200278520584106,
      "learning_rate": 0.0001375642469865844,
      "loss": 2.565,
      "step": 1394
    },
    {
      "epoch": 0.3863195790639712,
      "grad_norm": 1.5326173305511475,
      "learning_rate": 0.00013748247115868704,
      "loss": 2.785,
      "step": 1395
    },
    {
      "epoch": 0.3865965106618665,
      "grad_norm": 1.5612823963165283,
      "learning_rate": 0.0001374006661575953,
      "loss": 2.6662,
      "step": 1396
    },
    {
      "epoch": 0.38687344225976183,
      "grad_norm": 0.9551043510437012,
      "learning_rate": 0.00013731883204697932,
      "loss": 2.6571,
      "step": 1397
    },
    {
      "epoch": 0.38715037385765716,
      "grad_norm": 1.4987437725067139,
      "learning_rate": 0.0001372369688905319,
      "loss": 2.7669,
      "step": 1398
    },
    {
      "epoch": 0.3874273054555525,
      "grad_norm": 3.515704870223999,
      "learning_rate": 0.00013715507675196836,
      "loss": 3.3099,
      "step": 1399
    },
    {
      "epoch": 0.3877042370534478,
      "grad_norm": 6.313563346862793,
      "learning_rate": 0.00013707315569502665,
      "loss": 4.2518,
      "step": 1400
    },
    {
      "epoch": 0.38798116865134313,
      "grad_norm": 5.393907070159912,
      "learning_rate": 0.00013699120578346727,
      "loss": 3.1773,
      "step": 1401
    },
    {
      "epoch": 0.38825810024923846,
      "grad_norm": 3.510133743286133,
      "learning_rate": 0.00013690922708107302,
      "loss": 2.8114,
      "step": 1402
    },
    {
      "epoch": 0.3885350318471338,
      "grad_norm": 3.648080825805664,
      "learning_rate": 0.00013682721965164926,
      "loss": 2.8152,
      "step": 1403
    },
    {
      "epoch": 0.38881196344502905,
      "grad_norm": 1.3213528394699097,
      "learning_rate": 0.0001367451835590237,
      "loss": 2.5302,
      "step": 1404
    },
    {
      "epoch": 0.3890888950429244,
      "grad_norm": 1.6521259546279907,
      "learning_rate": 0.0001366631188670462,
      "loss": 2.5972,
      "step": 1405
    },
    {
      "epoch": 0.3893658266408197,
      "grad_norm": 2.0183587074279785,
      "learning_rate": 0.0001365810256395891,
      "loss": 3.1236,
      "step": 1406
    },
    {
      "epoch": 0.389642758238715,
      "grad_norm": 17.155466079711914,
      "learning_rate": 0.0001364989039405468,
      "loss": 3.1734,
      "step": 1407
    },
    {
      "epoch": 0.38991968983661035,
      "grad_norm": 1.8948163986206055,
      "learning_rate": 0.0001364167538338359,
      "loss": 2.656,
      "step": 1408
    },
    {
      "epoch": 0.3901966214345057,
      "grad_norm": 1.8051021099090576,
      "learning_rate": 0.00013633457538339514,
      "loss": 3.0139,
      "step": 1409
    },
    {
      "epoch": 0.390473553032401,
      "grad_norm": 2.2459874153137207,
      "learning_rate": 0.00013625236865318533,
      "loss": 2.9797,
      "step": 1410
    },
    {
      "epoch": 0.3907504846302963,
      "grad_norm": 2.143657684326172,
      "learning_rate": 0.00013617013370718914,
      "loss": 2.7228,
      "step": 1411
    },
    {
      "epoch": 0.39102741622819165,
      "grad_norm": 2.0867865085601807,
      "learning_rate": 0.00013608787060941143,
      "loss": 2.9089,
      "step": 1412
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 1.8627058267593384,
      "learning_rate": 0.0001360055794238788,
      "loss": 2.9256,
      "step": 1413
    },
    {
      "epoch": 0.3915812794239823,
      "grad_norm": 1.8171625137329102,
      "learning_rate": 0.00013592326021463977,
      "loss": 3.2421,
      "step": 1414
    },
    {
      "epoch": 0.3918582110218776,
      "grad_norm": 2.448659896850586,
      "learning_rate": 0.00013584091304576467,
      "loss": 2.6718,
      "step": 1415
    },
    {
      "epoch": 0.3921351426197729,
      "grad_norm": 1.933889627456665,
      "learning_rate": 0.00013575853798134563,
      "loss": 2.7015,
      "step": 1416
    },
    {
      "epoch": 0.3924120742176682,
      "grad_norm": 2.313974618911743,
      "learning_rate": 0.0001356761350854964,
      "loss": 2.6308,
      "step": 1417
    },
    {
      "epoch": 0.39268900581556354,
      "grad_norm": 1.7314043045043945,
      "learning_rate": 0.0001355937044223525,
      "loss": 2.721,
      "step": 1418
    },
    {
      "epoch": 0.39296593741345887,
      "grad_norm": 2.7365071773529053,
      "learning_rate": 0.00013551124605607097,
      "loss": 2.579,
      "step": 1419
    },
    {
      "epoch": 0.3932428690113542,
      "grad_norm": 1.9637887477874756,
      "learning_rate": 0.00013542876005083045,
      "loss": 2.9192,
      "step": 1420
    },
    {
      "epoch": 0.3935198006092495,
      "grad_norm": 1.607652187347412,
      "learning_rate": 0.0001353462464708311,
      "loss": 2.6256,
      "step": 1421
    },
    {
      "epoch": 0.39379673220714484,
      "grad_norm": 2.0072946548461914,
      "learning_rate": 0.0001352637053802945,
      "loss": 2.7192,
      "step": 1422
    },
    {
      "epoch": 0.39407366380504016,
      "grad_norm": 4.196632385253906,
      "learning_rate": 0.00013518113684346373,
      "loss": 3.2017,
      "step": 1423
    },
    {
      "epoch": 0.3943505954029355,
      "grad_norm": 1.5665082931518555,
      "learning_rate": 0.00013509854092460312,
      "loss": 2.7653,
      "step": 1424
    },
    {
      "epoch": 0.3946275270008308,
      "grad_norm": 6.20534086227417,
      "learning_rate": 0.00013501591768799842,
      "loss": 4.5799,
      "step": 1425
    },
    {
      "epoch": 0.39490445859872614,
      "grad_norm": 1.3128141164779663,
      "learning_rate": 0.00013493326719795653,
      "loss": 2.6134,
      "step": 1426
    },
    {
      "epoch": 0.3951813901966214,
      "grad_norm": 1.984251618385315,
      "learning_rate": 0.00013485058951880568,
      "loss": 2.8511,
      "step": 1427
    },
    {
      "epoch": 0.39545832179451673,
      "grad_norm": 2.113044500350952,
      "learning_rate": 0.00013476788471489517,
      "loss": 2.8686,
      "step": 1428
    },
    {
      "epoch": 0.39573525339241206,
      "grad_norm": 2.3141541481018066,
      "learning_rate": 0.0001346851528505954,
      "loss": 2.9546,
      "step": 1429
    },
    {
      "epoch": 0.3960121849903074,
      "grad_norm": 2.277184009552002,
      "learning_rate": 0.00013460239399029796,
      "loss": 2.688,
      "step": 1430
    },
    {
      "epoch": 0.3962891165882027,
      "grad_norm": 2.426490068435669,
      "learning_rate": 0.00013451960819841537,
      "loss": 2.7827,
      "step": 1431
    },
    {
      "epoch": 0.39656604818609803,
      "grad_norm": 2.0649800300598145,
      "learning_rate": 0.00013443679553938102,
      "loss": 2.8816,
      "step": 1432
    },
    {
      "epoch": 0.39684297978399335,
      "grad_norm": 1.6541078090667725,
      "learning_rate": 0.00013435395607764935,
      "loss": 2.7599,
      "step": 1433
    },
    {
      "epoch": 0.3971199113818887,
      "grad_norm": 1.7938566207885742,
      "learning_rate": 0.00013427108987769566,
      "loss": 2.5367,
      "step": 1434
    },
    {
      "epoch": 0.397396842979784,
      "grad_norm": 2.0776562690734863,
      "learning_rate": 0.0001341881970040159,
      "loss": 3.1443,
      "step": 1435
    },
    {
      "epoch": 0.39767377457767933,
      "grad_norm": 1.541983962059021,
      "learning_rate": 0.000134105277521127,
      "loss": 2.7567,
      "step": 1436
    },
    {
      "epoch": 0.39795070617557465,
      "grad_norm": 1.157357096672058,
      "learning_rate": 0.00013402233149356643,
      "loss": 2.6953,
      "step": 1437
    },
    {
      "epoch": 0.39822763777347,
      "grad_norm": 2.230661153793335,
      "learning_rate": 0.00013393935898589238,
      "loss": 2.8742,
      "step": 1438
    },
    {
      "epoch": 0.39850456937136525,
      "grad_norm": 2.2509617805480957,
      "learning_rate": 0.00013385636006268368,
      "loss": 2.9729,
      "step": 1439
    },
    {
      "epoch": 0.39878150096926057,
      "grad_norm": 1.4444016218185425,
      "learning_rate": 0.00013377333478853967,
      "loss": 2.7014,
      "step": 1440
    },
    {
      "epoch": 0.3990584325671559,
      "grad_norm": 1.4741573333740234,
      "learning_rate": 0.00013369028322808023,
      "loss": 2.9002,
      "step": 1441
    },
    {
      "epoch": 0.3993353641650512,
      "grad_norm": 1.3836803436279297,
      "learning_rate": 0.0001336072054459457,
      "loss": 2.6069,
      "step": 1442
    },
    {
      "epoch": 0.39961229576294655,
      "grad_norm": 1.7925223112106323,
      "learning_rate": 0.00013352410150679683,
      "loss": 2.9793,
      "step": 1443
    },
    {
      "epoch": 0.39988922736084187,
      "grad_norm": 1.3602858781814575,
      "learning_rate": 0.00013344097147531469,
      "loss": 2.7197,
      "step": 1444
    },
    {
      "epoch": 0.4001661589587372,
      "grad_norm": 2.3307301998138428,
      "learning_rate": 0.0001333578154162007,
      "loss": 2.7677,
      "step": 1445
    },
    {
      "epoch": 0.4004430905566325,
      "grad_norm": 2.375763416290283,
      "learning_rate": 0.00013327463339417653,
      "loss": 2.7946,
      "step": 1446
    },
    {
      "epoch": 0.40072002215452784,
      "grad_norm": 1.6489684581756592,
      "learning_rate": 0.000133191425473984,
      "loss": 2.7883,
      "step": 1447
    },
    {
      "epoch": 0.40099695375242317,
      "grad_norm": 1.7327191829681396,
      "learning_rate": 0.0001331081917203852,
      "loss": 2.8872,
      "step": 1448
    },
    {
      "epoch": 0.4012738853503185,
      "grad_norm": 2.1036574840545654,
      "learning_rate": 0.00013302493219816223,
      "loss": 3.1464,
      "step": 1449
    },
    {
      "epoch": 0.4015508169482138,
      "grad_norm": 4.300667762756348,
      "learning_rate": 0.00013294164697211732,
      "loss": 4.0743,
      "step": 1450
    },
    {
      "epoch": 0.4018277485461091,
      "grad_norm": 3.243358612060547,
      "learning_rate": 0.0001328583361070726,
      "loss": 3.1086,
      "step": 1451
    },
    {
      "epoch": 0.4021046801440044,
      "grad_norm": 1.1075637340545654,
      "learning_rate": 0.0001327749996678703,
      "loss": 2.6491,
      "step": 1452
    },
    {
      "epoch": 0.40238161174189974,
      "grad_norm": 2.0181403160095215,
      "learning_rate": 0.0001326916377193724,
      "loss": 2.6164,
      "step": 1453
    },
    {
      "epoch": 0.40265854333979506,
      "grad_norm": 1.3411420583724976,
      "learning_rate": 0.00013260825032646083,
      "loss": 2.5476,
      "step": 1454
    },
    {
      "epoch": 0.4029354749376904,
      "grad_norm": 1.5511304140090942,
      "learning_rate": 0.00013252483755403735,
      "loss": 2.9308,
      "step": 1455
    },
    {
      "epoch": 0.4032124065355857,
      "grad_norm": 1.9710975885391235,
      "learning_rate": 0.00013244139946702336,
      "loss": 2.8186,
      "step": 1456
    },
    {
      "epoch": 0.40348933813348103,
      "grad_norm": 1.543416976928711,
      "learning_rate": 0.0001323579361303601,
      "loss": 2.5958,
      "step": 1457
    },
    {
      "epoch": 0.40376626973137636,
      "grad_norm": 1.2788368463516235,
      "learning_rate": 0.00013227444760900836,
      "loss": 2.7018,
      "step": 1458
    },
    {
      "epoch": 0.4040432013292717,
      "grad_norm": 2.103099822998047,
      "learning_rate": 0.00013219093396794852,
      "loss": 2.6884,
      "step": 1459
    },
    {
      "epoch": 0.404320132927167,
      "grad_norm": 1.5627678632736206,
      "learning_rate": 0.00013210739527218064,
      "loss": 2.9022,
      "step": 1460
    },
    {
      "epoch": 0.40459706452506233,
      "grad_norm": 1.1994236707687378,
      "learning_rate": 0.00013202383158672412,
      "loss": 2.9379,
      "step": 1461
    },
    {
      "epoch": 0.4048739961229576,
      "grad_norm": 1.8488538265228271,
      "learning_rate": 0.00013194024297661793,
      "loss": 2.8452,
      "step": 1462
    },
    {
      "epoch": 0.4051509277208529,
      "grad_norm": 1.3858424425125122,
      "learning_rate": 0.00013185662950692035,
      "loss": 2.5327,
      "step": 1463
    },
    {
      "epoch": 0.40542785931874825,
      "grad_norm": 1.5252209901809692,
      "learning_rate": 0.00013177299124270911,
      "loss": 2.8002,
      "step": 1464
    },
    {
      "epoch": 0.4057047909166436,
      "grad_norm": 2.618840456008911,
      "learning_rate": 0.00013168932824908115,
      "loss": 2.7631,
      "step": 1465
    },
    {
      "epoch": 0.4059817225145389,
      "grad_norm": 1.9298254251480103,
      "learning_rate": 0.0001316056405911527,
      "loss": 3.064,
      "step": 1466
    },
    {
      "epoch": 0.4062586541124342,
      "grad_norm": 2.1758623123168945,
      "learning_rate": 0.00013152192833405918,
      "loss": 2.7877,
      "step": 1467
    },
    {
      "epoch": 0.40653558571032955,
      "grad_norm": 1.194083333015442,
      "learning_rate": 0.0001314381915429551,
      "loss": 2.8273,
      "step": 1468
    },
    {
      "epoch": 0.4068125173082249,
      "grad_norm": 2.225982666015625,
      "learning_rate": 0.0001313544302830142,
      "loss": 2.6538,
      "step": 1469
    },
    {
      "epoch": 0.4070894489061202,
      "grad_norm": 2.0385043621063232,
      "learning_rate": 0.00013127064461942913,
      "loss": 2.8584,
      "step": 1470
    },
    {
      "epoch": 0.4073663805040155,
      "grad_norm": 1.949174165725708,
      "learning_rate": 0.00013118683461741155,
      "loss": 2.7196,
      "step": 1471
    },
    {
      "epoch": 0.40764331210191085,
      "grad_norm": 0.9881244897842407,
      "learning_rate": 0.00013110300034219216,
      "loss": 2.606,
      "step": 1472
    },
    {
      "epoch": 0.4079202436998062,
      "grad_norm": 3.028315544128418,
      "learning_rate": 0.00013101914185902045,
      "loss": 2.9581,
      "step": 1473
    },
    {
      "epoch": 0.40819717529770144,
      "grad_norm": 2.262873888015747,
      "learning_rate": 0.00013093525923316482,
      "loss": 3.0206,
      "step": 1474
    },
    {
      "epoch": 0.40847410689559677,
      "grad_norm": 3.912444829940796,
      "learning_rate": 0.00013085135252991238,
      "loss": 4.0442,
      "step": 1475
    },
    {
      "epoch": 0.4087510384934921,
      "grad_norm": 1.941103219985962,
      "learning_rate": 0.00013076742181456914,
      "loss": 2.9139,
      "step": 1476
    },
    {
      "epoch": 0.4090279700913874,
      "grad_norm": 1.6083106994628906,
      "learning_rate": 0.00013068346715245957,
      "loss": 2.5113,
      "step": 1477
    },
    {
      "epoch": 0.40930490168928274,
      "grad_norm": 1.6494433879852295,
      "learning_rate": 0.00013059948860892696,
      "loss": 2.6644,
      "step": 1478
    },
    {
      "epoch": 0.40958183328717807,
      "grad_norm": 1.2976964712142944,
      "learning_rate": 0.00013051548624933314,
      "loss": 2.4618,
      "step": 1479
    },
    {
      "epoch": 0.4098587648850734,
      "grad_norm": 1.8355804681777954,
      "learning_rate": 0.0001304314601390584,
      "loss": 2.9373,
      "step": 1480
    },
    {
      "epoch": 0.4101356964829687,
      "grad_norm": 1.3326092958450317,
      "learning_rate": 0.00013034741034350161,
      "loss": 2.6393,
      "step": 1481
    },
    {
      "epoch": 0.41041262808086404,
      "grad_norm": 1.550118088722229,
      "learning_rate": 0.00013026333692808006,
      "loss": 2.7761,
      "step": 1482
    },
    {
      "epoch": 0.41068955967875936,
      "grad_norm": 2.0922064781188965,
      "learning_rate": 0.00013017923995822938,
      "loss": 2.9785,
      "step": 1483
    },
    {
      "epoch": 0.4109664912766547,
      "grad_norm": 1.86045241355896,
      "learning_rate": 0.00013009511949940358,
      "loss": 2.5891,
      "step": 1484
    },
    {
      "epoch": 0.41124342287455,
      "grad_norm": 1.8999186754226685,
      "learning_rate": 0.00013001097561707493,
      "loss": 2.4681,
      "step": 1485
    },
    {
      "epoch": 0.4115203544724453,
      "grad_norm": 0.8147444725036621,
      "learning_rate": 0.00012992680837673392,
      "loss": 2.7766,
      "step": 1486
    },
    {
      "epoch": 0.4117972860703406,
      "grad_norm": 1.3010146617889404,
      "learning_rate": 0.00012984261784388922,
      "loss": 2.5141,
      "step": 1487
    },
    {
      "epoch": 0.41207421766823593,
      "grad_norm": 1.1852285861968994,
      "learning_rate": 0.00012975840408406767,
      "loss": 2.6221,
      "step": 1488
    },
    {
      "epoch": 0.41235114926613126,
      "grad_norm": 33.655174255371094,
      "learning_rate": 0.00012967416716281414,
      "loss": 2.9547,
      "step": 1489
    },
    {
      "epoch": 0.4126280808640266,
      "grad_norm": 1.393959641456604,
      "learning_rate": 0.00012958990714569154,
      "loss": 2.6042,
      "step": 1490
    },
    {
      "epoch": 0.4129050124619219,
      "grad_norm": 2.544478416442871,
      "learning_rate": 0.0001295056240982808,
      "loss": 2.6323,
      "step": 1491
    },
    {
      "epoch": 0.41318194405981723,
      "grad_norm": 1.5582716464996338,
      "learning_rate": 0.00012942131808618067,
      "loss": 2.9899,
      "step": 1492
    },
    {
      "epoch": 0.41345887565771255,
      "grad_norm": 1.7537245750427246,
      "learning_rate": 0.00012933698917500788,
      "loss": 2.6883,
      "step": 1493
    },
    {
      "epoch": 0.4137358072556079,
      "grad_norm": 1.275985836982727,
      "learning_rate": 0.00012925263743039693,
      "loss": 2.7156,
      "step": 1494
    },
    {
      "epoch": 0.4140127388535032,
      "grad_norm": 1.2964093685150146,
      "learning_rate": 0.0001291682629180001,
      "loss": 2.7783,
      "step": 1495
    },
    {
      "epoch": 0.41428967045139853,
      "grad_norm": 1.6701529026031494,
      "learning_rate": 0.0001290838657034874,
      "loss": 2.7645,
      "step": 1496
    },
    {
      "epoch": 0.4145666020492938,
      "grad_norm": 1.611763834953308,
      "learning_rate": 0.00012899944585254656,
      "loss": 2.8794,
      "step": 1497
    },
    {
      "epoch": 0.4148435336471891,
      "grad_norm": 1.6163182258605957,
      "learning_rate": 0.00012891500343088278,
      "loss": 2.9532,
      "step": 1498
    },
    {
      "epoch": 0.41512046524508445,
      "grad_norm": 2.047060489654541,
      "learning_rate": 0.00012883053850421897,
      "loss": 3.1522,
      "step": 1499
    },
    {
      "epoch": 0.41539739684297977,
      "grad_norm": 3.8433516025543213,
      "learning_rate": 0.00012874605113829553,
      "loss": 4.1045,
      "step": 1500
    },
    {
      "epoch": 0.4156743284408751,
      "grad_norm": 2.038623094558716,
      "learning_rate": 0.0001286615413988702,
      "loss": 2.7263,
      "step": 1501
    },
    {
      "epoch": 0.4159512600387704,
      "grad_norm": 1.697388768196106,
      "learning_rate": 0.00012857700935171835,
      "loss": 2.7766,
      "step": 1502
    },
    {
      "epoch": 0.41622819163666575,
      "grad_norm": 1.923000693321228,
      "learning_rate": 0.00012849245506263256,
      "loss": 2.8755,
      "step": 1503
    },
    {
      "epoch": 0.41650512323456107,
      "grad_norm": 1.566849708557129,
      "learning_rate": 0.00012840787859742266,
      "loss": 2.9213,
      "step": 1504
    },
    {
      "epoch": 0.4167820548324564,
      "grad_norm": 1.4963854551315308,
      "learning_rate": 0.00012832328002191599,
      "loss": 2.6829,
      "step": 1505
    },
    {
      "epoch": 0.4170589864303517,
      "grad_norm": 1.7939183712005615,
      "learning_rate": 0.00012823865940195684,
      "loss": 2.5174,
      "step": 1506
    },
    {
      "epoch": 0.41733591802824704,
      "grad_norm": 1.2367196083068848,
      "learning_rate": 0.00012815401680340682,
      "loss": 2.9059,
      "step": 1507
    },
    {
      "epoch": 0.41761284962614237,
      "grad_norm": 1.3630605936050415,
      "learning_rate": 0.00012806935229214456,
      "loss": 2.7343,
      "step": 1508
    },
    {
      "epoch": 0.41788978122403764,
      "grad_norm": 10.827095985412598,
      "learning_rate": 0.00012798466593406583,
      "loss": 2.8514,
      "step": 1509
    },
    {
      "epoch": 0.41816671282193296,
      "grad_norm": 1.4653106927871704,
      "learning_rate": 0.00012789995779508327,
      "loss": 2.8998,
      "step": 1510
    },
    {
      "epoch": 0.4184436444198283,
      "grad_norm": 1.8175551891326904,
      "learning_rate": 0.00012781522794112657,
      "loss": 2.8933,
      "step": 1511
    },
    {
      "epoch": 0.4187205760177236,
      "grad_norm": 1.614170789718628,
      "learning_rate": 0.00012773047643814235,
      "loss": 2.7967,
      "step": 1512
    },
    {
      "epoch": 0.41899750761561894,
      "grad_norm": 1.9364715814590454,
      "learning_rate": 0.000127645703352094,
      "loss": 2.7361,
      "step": 1513
    },
    {
      "epoch": 0.41927443921351426,
      "grad_norm": 2.4313740730285645,
      "learning_rate": 0.00012756090874896172,
      "loss": 3.164,
      "step": 1514
    },
    {
      "epoch": 0.4195513708114096,
      "grad_norm": 1.7980979681015015,
      "learning_rate": 0.0001274760926947425,
      "loss": 3.0288,
      "step": 1515
    },
    {
      "epoch": 0.4198283024093049,
      "grad_norm": 1.5087634325027466,
      "learning_rate": 0.00012739125525545002,
      "loss": 2.9844,
      "step": 1516
    },
    {
      "epoch": 0.42010523400720023,
      "grad_norm": 1.5665369033813477,
      "learning_rate": 0.00012730639649711454,
      "loss": 2.6869,
      "step": 1517
    },
    {
      "epoch": 0.42038216560509556,
      "grad_norm": 1.50995934009552,
      "learning_rate": 0.00012722151648578302,
      "loss": 2.8659,
      "step": 1518
    },
    {
      "epoch": 0.4206590972029909,
      "grad_norm": 1.8009982109069824,
      "learning_rate": 0.00012713661528751888,
      "loss": 2.5628,
      "step": 1519
    },
    {
      "epoch": 0.4209360288008862,
      "grad_norm": 1.7115381956100464,
      "learning_rate": 0.00012705169296840202,
      "loss": 2.5867,
      "step": 1520
    },
    {
      "epoch": 0.4212129603987815,
      "grad_norm": 1.8676518201828003,
      "learning_rate": 0.00012696674959452887,
      "loss": 2.7795,
      "step": 1521
    },
    {
      "epoch": 0.4214898919966768,
      "grad_norm": 1.6781612634658813,
      "learning_rate": 0.00012688178523201214,
      "loss": 2.7978,
      "step": 1522
    },
    {
      "epoch": 0.4217668235945721,
      "grad_norm": 2.0392162799835205,
      "learning_rate": 0.0001267967999469809,
      "loss": 2.8138,
      "step": 1523
    },
    {
      "epoch": 0.42204375519246745,
      "grad_norm": 2.3748555183410645,
      "learning_rate": 0.00012671179380558062,
      "loss": 3.1233,
      "step": 1524
    },
    {
      "epoch": 0.4223206867903628,
      "grad_norm": 5.822400093078613,
      "learning_rate": 0.00012662676687397284,
      "loss": 4.4801,
      "step": 1525
    },
    {
      "epoch": 0.4225976183882581,
      "grad_norm": 1.2015067338943481,
      "learning_rate": 0.00012654171921833534,
      "loss": 2.6184,
      "step": 1526
    },
    {
      "epoch": 0.4228745499861534,
      "grad_norm": 1.7967123985290527,
      "learning_rate": 0.00012645665090486212,
      "loss": 2.6241,
      "step": 1527
    },
    {
      "epoch": 0.42315148158404875,
      "grad_norm": 1.2164345979690552,
      "learning_rate": 0.0001263715619997631,
      "loss": 2.4707,
      "step": 1528
    },
    {
      "epoch": 0.4234284131819441,
      "grad_norm": 1.5270050764083862,
      "learning_rate": 0.00012628645256926438,
      "loss": 2.5566,
      "step": 1529
    },
    {
      "epoch": 0.4237053447798394,
      "grad_norm": 1.3630717992782593,
      "learning_rate": 0.00012620132267960788,
      "loss": 2.8082,
      "step": 1530
    },
    {
      "epoch": 0.4239822763777347,
      "grad_norm": 1.9305917024612427,
      "learning_rate": 0.00012611617239705161,
      "loss": 2.859,
      "step": 1531
    },
    {
      "epoch": 0.42425920797563,
      "grad_norm": 1.6518409252166748,
      "learning_rate": 0.00012603100178786928,
      "loss": 2.688,
      "step": 1532
    },
    {
      "epoch": 0.4245361395735253,
      "grad_norm": 2.4489521980285645,
      "learning_rate": 0.00012594581091835062,
      "loss": 2.9602,
      "step": 1533
    },
    {
      "epoch": 0.42481307117142064,
      "grad_norm": 3.1953954696655273,
      "learning_rate": 0.0001258605998548009,
      "loss": 2.9738,
      "step": 1534
    },
    {
      "epoch": 0.42509000276931597,
      "grad_norm": 1.9063570499420166,
      "learning_rate": 0.00012577536866354135,
      "loss": 2.8507,
      "step": 1535
    },
    {
      "epoch": 0.4253669343672113,
      "grad_norm": 1.2348942756652832,
      "learning_rate": 0.00012569011741090863,
      "loss": 2.2727,
      "step": 1536
    },
    {
      "epoch": 0.4256438659651066,
      "grad_norm": 1.3155874013900757,
      "learning_rate": 0.00012560484616325513,
      "loss": 2.7515,
      "step": 1537
    },
    {
      "epoch": 0.42592079756300194,
      "grad_norm": 1.3448861837387085,
      "learning_rate": 0.0001255195549869489,
      "loss": 2.7228,
      "step": 1538
    },
    {
      "epoch": 0.42619772916089727,
      "grad_norm": 1.4402925968170166,
      "learning_rate": 0.0001254342439483733,
      "loss": 2.7855,
      "step": 1539
    },
    {
      "epoch": 0.4264746607587926,
      "grad_norm": 1.7898362874984741,
      "learning_rate": 0.0001253489131139273,
      "loss": 2.8563,
      "step": 1540
    },
    {
      "epoch": 0.4267515923566879,
      "grad_norm": 1.267905592918396,
      "learning_rate": 0.0001252635625500252,
      "loss": 2.9075,
      "step": 1541
    },
    {
      "epoch": 0.42702852395458324,
      "grad_norm": 1.5784574747085571,
      "learning_rate": 0.0001251781923230967,
      "loss": 2.6532,
      "step": 1542
    },
    {
      "epoch": 0.42730545555247856,
      "grad_norm": 2.0321223735809326,
      "learning_rate": 0.00012509280249958674,
      "loss": 2.7816,
      "step": 1543
    },
    {
      "epoch": 0.42758238715037383,
      "grad_norm": 1.8454991579055786,
      "learning_rate": 0.00012500739314595563,
      "loss": 2.7301,
      "step": 1544
    },
    {
      "epoch": 0.42785931874826916,
      "grad_norm": 1.7442742586135864,
      "learning_rate": 0.00012492196432867874,
      "loss": 2.719,
      "step": 1545
    },
    {
      "epoch": 0.4281362503461645,
      "grad_norm": 2.122340202331543,
      "learning_rate": 0.00012483651611424666,
      "loss": 2.9033,
      "step": 1546
    },
    {
      "epoch": 0.4284131819440598,
      "grad_norm": 1.8782052993774414,
      "learning_rate": 0.00012475104856916511,
      "loss": 2.6263,
      "step": 1547
    },
    {
      "epoch": 0.42869011354195513,
      "grad_norm": 1.387713074684143,
      "learning_rate": 0.0001246655617599548,
      "loss": 2.5894,
      "step": 1548
    },
    {
      "epoch": 0.42896704513985046,
      "grad_norm": 1.0579252243041992,
      "learning_rate": 0.00012458005575315147,
      "loss": 3.0989,
      "step": 1549
    },
    {
      "epoch": 0.4292439767377458,
      "grad_norm": 4.583604335784912,
      "learning_rate": 0.0001244945306153058,
      "loss": 4.2209,
      "step": 1550
    },
    {
      "epoch": 0.4295209083356411,
      "grad_norm": 1.498977541923523,
      "learning_rate": 0.00012440898641298328,
      "loss": 2.6145,
      "step": 1551
    },
    {
      "epoch": 0.42979783993353643,
      "grad_norm": 1.6750907897949219,
      "learning_rate": 0.00012432342321276436,
      "loss": 2.6988,
      "step": 1552
    },
    {
      "epoch": 0.43007477153143175,
      "grad_norm": 2.225656270980835,
      "learning_rate": 0.0001242378410812442,
      "loss": 2.9361,
      "step": 1553
    },
    {
      "epoch": 0.4303517031293271,
      "grad_norm": 2.292717933654785,
      "learning_rate": 0.0001241522400850327,
      "loss": 2.9439,
      "step": 1554
    },
    {
      "epoch": 0.4306286347272224,
      "grad_norm": 1.349919319152832,
      "learning_rate": 0.00012406662029075446,
      "loss": 2.628,
      "step": 1555
    },
    {
      "epoch": 0.4309055663251177,
      "grad_norm": 2.0709357261657715,
      "learning_rate": 0.00012398098176504872,
      "loss": 2.6342,
      "step": 1556
    },
    {
      "epoch": 0.431182497923013,
      "grad_norm": 1.249013900756836,
      "learning_rate": 0.0001238953245745693,
      "loss": 2.7055,
      "step": 1557
    },
    {
      "epoch": 0.4314594295209083,
      "grad_norm": 1.735985517501831,
      "learning_rate": 0.0001238096487859845,
      "loss": 2.658,
      "step": 1558
    },
    {
      "epoch": 0.43173636111880365,
      "grad_norm": 1.6668094396591187,
      "learning_rate": 0.0001237239544659771,
      "loss": 2.8608,
      "step": 1559
    },
    {
      "epoch": 0.43201329271669897,
      "grad_norm": 1.760405421257019,
      "learning_rate": 0.0001236382416812444,
      "loss": 2.6073,
      "step": 1560
    },
    {
      "epoch": 0.4322902243145943,
      "grad_norm": 1.0794938802719116,
      "learning_rate": 0.000123552510498498,
      "loss": 2.6189,
      "step": 1561
    },
    {
      "epoch": 0.4325671559124896,
      "grad_norm": 2.733625650405884,
      "learning_rate": 0.00012346676098446378,
      "loss": 3.4352,
      "step": 1562
    },
    {
      "epoch": 0.43284408751038495,
      "grad_norm": 2.011298418045044,
      "learning_rate": 0.00012338099320588193,
      "loss": 2.7116,
      "step": 1563
    },
    {
      "epoch": 0.43312101910828027,
      "grad_norm": 1.5651980638504028,
      "learning_rate": 0.0001232952072295069,
      "loss": 2.9265,
      "step": 1564
    },
    {
      "epoch": 0.4333979507061756,
      "grad_norm": 2.379638671875,
      "learning_rate": 0.00012320940312210725,
      "loss": 2.8316,
      "step": 1565
    },
    {
      "epoch": 0.4336748823040709,
      "grad_norm": 0.860095202922821,
      "learning_rate": 0.00012312358095046562,
      "loss": 2.7123,
      "step": 1566
    },
    {
      "epoch": 0.4339518139019662,
      "grad_norm": 1.6845221519470215,
      "learning_rate": 0.00012303774078137872,
      "loss": 2.7341,
      "step": 1567
    },
    {
      "epoch": 0.4342287454998615,
      "grad_norm": 1.8354023694992065,
      "learning_rate": 0.00012295188268165742,
      "loss": 2.9535,
      "step": 1568
    },
    {
      "epoch": 0.43450567709775684,
      "grad_norm": 1.8640836477279663,
      "learning_rate": 0.0001228660067181263,
      "loss": 2.5892,
      "step": 1569
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 1.708167552947998,
      "learning_rate": 0.00012278011295762406,
      "loss": 2.7308,
      "step": 1570
    },
    {
      "epoch": 0.4350595402935475,
      "grad_norm": 2.214993476867676,
      "learning_rate": 0.00012269420146700312,
      "loss": 2.6486,
      "step": 1571
    },
    {
      "epoch": 0.4353364718914428,
      "grad_norm": 1.652694582939148,
      "learning_rate": 0.00012260827231312974,
      "loss": 2.951,
      "step": 1572
    },
    {
      "epoch": 0.43561340348933814,
      "grad_norm": 1.8127185106277466,
      "learning_rate": 0.00012252232556288393,
      "loss": 2.5346,
      "step": 1573
    },
    {
      "epoch": 0.43589033508723346,
      "grad_norm": 1.5010112524032593,
      "learning_rate": 0.00012243636128315939,
      "loss": 3.0699,
      "step": 1574
    },
    {
      "epoch": 0.4361672666851288,
      "grad_norm": 3.595019578933716,
      "learning_rate": 0.00012235037954086344,
      "loss": 3.8706,
      "step": 1575
    },
    {
      "epoch": 0.4364441982830241,
      "grad_norm": 2.001720428466797,
      "learning_rate": 0.00012226438040291703,
      "loss": 2.8349,
      "step": 1576
    },
    {
      "epoch": 0.43672112988091943,
      "grad_norm": 1.3290728330612183,
      "learning_rate": 0.0001221783639362547,
      "loss": 2.7987,
      "step": 1577
    },
    {
      "epoch": 0.43699806147881476,
      "grad_norm": 2.3499648571014404,
      "learning_rate": 0.0001220923302078243,
      "loss": 2.571,
      "step": 1578
    },
    {
      "epoch": 0.43727499307671003,
      "grad_norm": 1.8382792472839355,
      "learning_rate": 0.0001220062792845873,
      "loss": 2.715,
      "step": 1579
    },
    {
      "epoch": 0.43755192467460535,
      "grad_norm": 2.051205635070801,
      "learning_rate": 0.00012192021123351847,
      "loss": 2.7925,
      "step": 1580
    },
    {
      "epoch": 0.4378288562725007,
      "grad_norm": 2.2316954135894775,
      "learning_rate": 0.00012183412612160592,
      "loss": 2.7037,
      "step": 1581
    },
    {
      "epoch": 0.438105787870396,
      "grad_norm": 1.6185784339904785,
      "learning_rate": 0.00012174802401585106,
      "loss": 2.8406,
      "step": 1582
    },
    {
      "epoch": 0.4383827194682913,
      "grad_norm": 1.404736876487732,
      "learning_rate": 0.00012166190498326849,
      "loss": 2.6622,
      "step": 1583
    },
    {
      "epoch": 0.43865965106618665,
      "grad_norm": 1.912427544593811,
      "learning_rate": 0.00012157576909088599,
      "loss": 2.8172,
      "step": 1584
    },
    {
      "epoch": 0.438936582664082,
      "grad_norm": 2.464752674102783,
      "learning_rate": 0.00012148961640574448,
      "loss": 3.0364,
      "step": 1585
    },
    {
      "epoch": 0.4392135142619773,
      "grad_norm": 1.2723188400268555,
      "learning_rate": 0.00012140344699489797,
      "loss": 2.9328,
      "step": 1586
    },
    {
      "epoch": 0.4394904458598726,
      "grad_norm": 1.8504717350006104,
      "learning_rate": 0.0001213172609254134,
      "loss": 2.9263,
      "step": 1587
    },
    {
      "epoch": 0.43976737745776795,
      "grad_norm": 2.0745506286621094,
      "learning_rate": 0.0001212310582643708,
      "loss": 2.8592,
      "step": 1588
    },
    {
      "epoch": 0.4400443090556633,
      "grad_norm": 1.8228636980056763,
      "learning_rate": 0.00012114483907886308,
      "loss": 2.6974,
      "step": 1589
    },
    {
      "epoch": 0.4403212406535586,
      "grad_norm": 2.4309403896331787,
      "learning_rate": 0.00012105860343599587,
      "loss": 3.0777,
      "step": 1590
    },
    {
      "epoch": 0.44059817225145387,
      "grad_norm": 1.2440286874771118,
      "learning_rate": 0.00012097235140288778,
      "loss": 2.6633,
      "step": 1591
    },
    {
      "epoch": 0.4408751038493492,
      "grad_norm": 1.2311338186264038,
      "learning_rate": 0.00012088608304667013,
      "loss": 2.9672,
      "step": 1592
    },
    {
      "epoch": 0.4411520354472445,
      "grad_norm": 1.4380064010620117,
      "learning_rate": 0.00012079979843448695,
      "loss": 2.7795,
      "step": 1593
    },
    {
      "epoch": 0.44142896704513984,
      "grad_norm": 1.5805429220199585,
      "learning_rate": 0.00012071349763349484,
      "loss": 2.8181,
      "step": 1594
    },
    {
      "epoch": 0.44170589864303517,
      "grad_norm": 1.2909247875213623,
      "learning_rate": 0.00012062718071086317,
      "loss": 2.7215,
      "step": 1595
    },
    {
      "epoch": 0.4419828302409305,
      "grad_norm": 1.5421581268310547,
      "learning_rate": 0.00012054084773377364,
      "loss": 2.9609,
      "step": 1596
    },
    {
      "epoch": 0.4422597618388258,
      "grad_norm": 1.854814887046814,
      "learning_rate": 0.0001204544987694206,
      "loss": 2.6241,
      "step": 1597
    },
    {
      "epoch": 0.44253669343672114,
      "grad_norm": 3.275557279586792,
      "learning_rate": 0.00012036813388501085,
      "loss": 3.2255,
      "step": 1598
    },
    {
      "epoch": 0.44281362503461646,
      "grad_norm": 1.4313879013061523,
      "learning_rate": 0.00012028175314776344,
      "loss": 2.7054,
      "step": 1599
    },
    {
      "epoch": 0.4430905566325118,
      "grad_norm": 3.1973049640655518,
      "learning_rate": 0.00012019535662490992,
      "loss": 3.1645,
      "step": 1600
    },
    {
      "epoch": 0.4433674882304071,
      "grad_norm": 1.480504035949707,
      "learning_rate": 0.00012010894438369404,
      "loss": 2.9397,
      "step": 1601
    },
    {
      "epoch": 0.4436444198283024,
      "grad_norm": 2.36940598487854,
      "learning_rate": 0.00012002251649137179,
      "loss": 3.103,
      "step": 1602
    },
    {
      "epoch": 0.4439213514261977,
      "grad_norm": 1.381364345550537,
      "learning_rate": 0.00011993607301521137,
      "loss": 2.5824,
      "step": 1603
    },
    {
      "epoch": 0.44419828302409303,
      "grad_norm": 1.7203359603881836,
      "learning_rate": 0.00011984961402249311,
      "loss": 2.6152,
      "step": 1604
    },
    {
      "epoch": 0.44447521462198836,
      "grad_norm": 1.387699007987976,
      "learning_rate": 0.00011976313958050933,
      "loss": 2.6626,
      "step": 1605
    },
    {
      "epoch": 0.4447521462198837,
      "grad_norm": 1.2170424461364746,
      "learning_rate": 0.0001196766497565645,
      "loss": 2.5803,
      "step": 1606
    },
    {
      "epoch": 0.445029077817779,
      "grad_norm": 1.3350768089294434,
      "learning_rate": 0.00011959014461797497,
      "loss": 3.0191,
      "step": 1607
    },
    {
      "epoch": 0.44530600941567433,
      "grad_norm": 1.831148624420166,
      "learning_rate": 0.00011950362423206907,
      "loss": 2.6024,
      "step": 1608
    },
    {
      "epoch": 0.44558294101356966,
      "grad_norm": 1.5087878704071045,
      "learning_rate": 0.00011941708866618697,
      "loss": 2.7182,
      "step": 1609
    },
    {
      "epoch": 0.445859872611465,
      "grad_norm": 1.560234546661377,
      "learning_rate": 0.00011933053798768066,
      "loss": 2.8327,
      "step": 1610
    },
    {
      "epoch": 0.4461368042093603,
      "grad_norm": 1.6062860488891602,
      "learning_rate": 0.00011924397226391384,
      "loss": 2.7957,
      "step": 1611
    },
    {
      "epoch": 0.44641373580725563,
      "grad_norm": 1.3197059631347656,
      "learning_rate": 0.00011915739156226202,
      "loss": 2.9193,
      "step": 1612
    },
    {
      "epoch": 0.44669066740515095,
      "grad_norm": 1.0454370975494385,
      "learning_rate": 0.00011907079595011231,
      "loss": 2.9694,
      "step": 1613
    },
    {
      "epoch": 0.4469675990030462,
      "grad_norm": 1.8847899436950684,
      "learning_rate": 0.0001189841854948634,
      "loss": 3.1058,
      "step": 1614
    },
    {
      "epoch": 0.44724453060094155,
      "grad_norm": 1.7060329914093018,
      "learning_rate": 0.00011889756026392562,
      "loss": 3.068,
      "step": 1615
    },
    {
      "epoch": 0.4475214621988369,
      "grad_norm": 1.8782567977905273,
      "learning_rate": 0.00011881092032472073,
      "loss": 2.6358,
      "step": 1616
    },
    {
      "epoch": 0.4477983937967322,
      "grad_norm": 1.6177935600280762,
      "learning_rate": 0.00011872426574468187,
      "loss": 2.5237,
      "step": 1617
    },
    {
      "epoch": 0.4480753253946275,
      "grad_norm": 1.101818323135376,
      "learning_rate": 0.00011863759659125376,
      "loss": 2.6257,
      "step": 1618
    },
    {
      "epoch": 0.44835225699252285,
      "grad_norm": 1.8147088289260864,
      "learning_rate": 0.00011855091293189234,
      "loss": 2.5104,
      "step": 1619
    },
    {
      "epoch": 0.44862918859041817,
      "grad_norm": 1.6569304466247559,
      "learning_rate": 0.0001184642148340648,
      "loss": 2.9456,
      "step": 1620
    },
    {
      "epoch": 0.4489061201883135,
      "grad_norm": 1.4350814819335938,
      "learning_rate": 0.00011837750236524967,
      "loss": 2.6975,
      "step": 1621
    },
    {
      "epoch": 0.4491830517862088,
      "grad_norm": 1.4598429203033447,
      "learning_rate": 0.00011829077559293664,
      "loss": 2.7264,
      "step": 1622
    },
    {
      "epoch": 0.44945998338410414,
      "grad_norm": 1.9063446521759033,
      "learning_rate": 0.00011820403458462647,
      "loss": 3.161,
      "step": 1623
    },
    {
      "epoch": 0.44973691498199947,
      "grad_norm": 1.749877691268921,
      "learning_rate": 0.00011811727940783108,
      "loss": 3.0542,
      "step": 1624
    },
    {
      "epoch": 0.45001384657989474,
      "grad_norm": 3.7590386867523193,
      "learning_rate": 0.00011803051013007336,
      "loss": 3.8809,
      "step": 1625
    },
    {
      "epoch": 0.45029077817779006,
      "grad_norm": 1.150923728942871,
      "learning_rate": 0.00011794372681888721,
      "loss": 2.7087,
      "step": 1626
    },
    {
      "epoch": 0.4505677097756854,
      "grad_norm": 1.7766743898391724,
      "learning_rate": 0.00011785692954181746,
      "loss": 3.1327,
      "step": 1627
    },
    {
      "epoch": 0.4508446413735807,
      "grad_norm": 1.7171368598937988,
      "learning_rate": 0.00011777011836641978,
      "loss": 3.0311,
      "step": 1628
    },
    {
      "epoch": 0.45112157297147604,
      "grad_norm": 1.6420807838439941,
      "learning_rate": 0.00011768329336026062,
      "loss": 2.8227,
      "step": 1629
    },
    {
      "epoch": 0.45139850456937136,
      "grad_norm": 1.121830940246582,
      "learning_rate": 0.00011759645459091731,
      "loss": 2.5743,
      "step": 1630
    },
    {
      "epoch": 0.4516754361672667,
      "grad_norm": 1.3861570358276367,
      "learning_rate": 0.0001175096021259778,
      "loss": 2.6474,
      "step": 1631
    },
    {
      "epoch": 0.451952367765162,
      "grad_norm": 1.0731266736984253,
      "learning_rate": 0.0001174227360330407,
      "loss": 2.8231,
      "step": 1632
    },
    {
      "epoch": 0.45222929936305734,
      "grad_norm": 1.2852985858917236,
      "learning_rate": 0.00011733585637971527,
      "loss": 2.6447,
      "step": 1633
    },
    {
      "epoch": 0.45250623096095266,
      "grad_norm": 1.646578311920166,
      "learning_rate": 0.0001172489632336213,
      "loss": 2.6627,
      "step": 1634
    },
    {
      "epoch": 0.452783162558848,
      "grad_norm": 1.2219154834747314,
      "learning_rate": 0.00011716205666238908,
      "loss": 2.4261,
      "step": 1635
    },
    {
      "epoch": 0.4530600941567433,
      "grad_norm": 2.7012853622436523,
      "learning_rate": 0.0001170751367336594,
      "loss": 2.7471,
      "step": 1636
    },
    {
      "epoch": 0.4533370257546386,
      "grad_norm": 1.415909767150879,
      "learning_rate": 0.00011698820351508335,
      "loss": 2.6987,
      "step": 1637
    },
    {
      "epoch": 0.4536139573525339,
      "grad_norm": 1.3519766330718994,
      "learning_rate": 0.00011690125707432244,
      "loss": 2.9456,
      "step": 1638
    },
    {
      "epoch": 0.4538908889504292,
      "grad_norm": 1.2360529899597168,
      "learning_rate": 0.00011681429747904842,
      "loss": 2.5995,
      "step": 1639
    },
    {
      "epoch": 0.45416782054832455,
      "grad_norm": 2.9068734645843506,
      "learning_rate": 0.00011672732479694338,
      "loss": 3.1497,
      "step": 1640
    },
    {
      "epoch": 0.4544447521462199,
      "grad_norm": 1.7364581823349,
      "learning_rate": 0.0001166403390956994,
      "loss": 2.7091,
      "step": 1641
    },
    {
      "epoch": 0.4547216837441152,
      "grad_norm": 1.1984012126922607,
      "learning_rate": 0.00011655334044301892,
      "loss": 2.6776,
      "step": 1642
    },
    {
      "epoch": 0.4549986153420105,
      "grad_norm": 2.4119842052459717,
      "learning_rate": 0.00011646632890661431,
      "loss": 2.5245,
      "step": 1643
    },
    {
      "epoch": 0.45527554693990585,
      "grad_norm": 1.5090466737747192,
      "learning_rate": 0.00011637930455420798,
      "loss": 2.5598,
      "step": 1644
    },
    {
      "epoch": 0.4555524785378012,
      "grad_norm": 1.5950679779052734,
      "learning_rate": 0.00011629226745353242,
      "loss": 2.5069,
      "step": 1645
    },
    {
      "epoch": 0.4558294101356965,
      "grad_norm": 2.2791571617126465,
      "learning_rate": 0.00011620521767232988,
      "loss": 2.6385,
      "step": 1646
    },
    {
      "epoch": 0.4561063417335918,
      "grad_norm": 1.426064372062683,
      "learning_rate": 0.00011611815527835266,
      "loss": 2.8296,
      "step": 1647
    },
    {
      "epoch": 0.45638327333148715,
      "grad_norm": 1.7675575017929077,
      "learning_rate": 0.00011603108033936271,
      "loss": 2.5534,
      "step": 1648
    },
    {
      "epoch": 0.4566602049293824,
      "grad_norm": 1.77476966381073,
      "learning_rate": 0.00011594399292313192,
      "loss": 3.0106,
      "step": 1649
    },
    {
      "epoch": 0.45693713652727774,
      "grad_norm": 4.373526573181152,
      "learning_rate": 0.00011585689309744166,
      "loss": 4.0497,
      "step": 1650
    },
    {
      "epoch": 0.45721406812517307,
      "grad_norm": 1.977187156677246,
      "learning_rate": 0.00011576978093008317,
      "loss": 2.6723,
      "step": 1651
    },
    {
      "epoch": 0.4574909997230684,
      "grad_norm": 2.139688014984131,
      "learning_rate": 0.00011568265648885721,
      "loss": 2.4705,
      "step": 1652
    },
    {
      "epoch": 0.4577679313209637,
      "grad_norm": 1.7218035459518433,
      "learning_rate": 0.00011559551984157404,
      "loss": 2.761,
      "step": 1653
    },
    {
      "epoch": 0.45804486291885904,
      "grad_norm": 1.5717421770095825,
      "learning_rate": 0.00011550837105605354,
      "loss": 2.5358,
      "step": 1654
    },
    {
      "epoch": 0.45832179451675437,
      "grad_norm": 2.267744541168213,
      "learning_rate": 0.00011542121020012497,
      "loss": 2.9883,
      "step": 1655
    },
    {
      "epoch": 0.4585987261146497,
      "grad_norm": 1.3252437114715576,
      "learning_rate": 0.00011533403734162696,
      "loss": 2.6564,
      "step": 1656
    },
    {
      "epoch": 0.458875657712545,
      "grad_norm": 1.4101369380950928,
      "learning_rate": 0.00011524685254840752,
      "loss": 2.7786,
      "step": 1657
    },
    {
      "epoch": 0.45915258931044034,
      "grad_norm": 1.5715200901031494,
      "learning_rate": 0.00011515965588832393,
      "loss": 2.735,
      "step": 1658
    },
    {
      "epoch": 0.45942952090833566,
      "grad_norm": 1.70425283908844,
      "learning_rate": 0.00011507244742924274,
      "loss": 2.7144,
      "step": 1659
    },
    {
      "epoch": 0.45970645250623093,
      "grad_norm": 2.2731897830963135,
      "learning_rate": 0.00011498522723903966,
      "loss": 2.8134,
      "step": 1660
    },
    {
      "epoch": 0.45998338410412626,
      "grad_norm": 2.732503652572632,
      "learning_rate": 0.00011489799538559953,
      "loss": 2.8805,
      "step": 1661
    },
    {
      "epoch": 0.4602603157020216,
      "grad_norm": 1.5472544431686401,
      "learning_rate": 0.00011481075193681625,
      "loss": 2.8424,
      "step": 1662
    },
    {
      "epoch": 0.4605372472999169,
      "grad_norm": 1.6334117650985718,
      "learning_rate": 0.00011472349696059275,
      "loss": 2.7526,
      "step": 1663
    },
    {
      "epoch": 0.46081417889781223,
      "grad_norm": 1.8457192182540894,
      "learning_rate": 0.000114636230524841,
      "loss": 3.1816,
      "step": 1664
    },
    {
      "epoch": 0.46109111049570756,
      "grad_norm": 1.5061466693878174,
      "learning_rate": 0.00011454895269748176,
      "loss": 2.8831,
      "step": 1665
    },
    {
      "epoch": 0.4613680420936029,
      "grad_norm": 2.3925230503082275,
      "learning_rate": 0.00011446166354644479,
      "loss": 2.8143,
      "step": 1666
    },
    {
      "epoch": 0.4616449736914982,
      "grad_norm": 1.9008785486221313,
      "learning_rate": 0.00011437436313966856,
      "loss": 2.8392,
      "step": 1667
    },
    {
      "epoch": 0.46192190528939353,
      "grad_norm": 0.9940817952156067,
      "learning_rate": 0.0001142870515451004,
      "loss": 2.6159,
      "step": 1668
    },
    {
      "epoch": 0.46219883688728886,
      "grad_norm": 2.2903690338134766,
      "learning_rate": 0.00011419972883069623,
      "loss": 3.1102,
      "step": 1669
    },
    {
      "epoch": 0.4624757684851842,
      "grad_norm": 1.6850765943527222,
      "learning_rate": 0.00011411239506442073,
      "loss": 2.7338,
      "step": 1670
    },
    {
      "epoch": 0.4627527000830795,
      "grad_norm": 2.8360297679901123,
      "learning_rate": 0.0001140250503142471,
      "loss": 2.7281,
      "step": 1671
    },
    {
      "epoch": 0.4630296316809748,
      "grad_norm": 1.7955050468444824,
      "learning_rate": 0.00011393769464815718,
      "loss": 2.5796,
      "step": 1672
    },
    {
      "epoch": 0.4633065632788701,
      "grad_norm": 2.1702804565429688,
      "learning_rate": 0.0001138503281341412,
      "loss": 3.0308,
      "step": 1673
    },
    {
      "epoch": 0.4635834948767654,
      "grad_norm": 1.1264212131500244,
      "learning_rate": 0.00011376295084019792,
      "loss": 2.9888,
      "step": 1674
    },
    {
      "epoch": 0.46386042647466075,
      "grad_norm": 5.740417957305908,
      "learning_rate": 0.00011367556283433445,
      "loss": 4.6839,
      "step": 1675
    },
    {
      "epoch": 0.46413735807255607,
      "grad_norm": 1.6483699083328247,
      "learning_rate": 0.00011358816418456624,
      "loss": 2.6919,
      "step": 1676
    },
    {
      "epoch": 0.4644142896704514,
      "grad_norm": 1.5567094087600708,
      "learning_rate": 0.000113500754958917,
      "loss": 2.6648,
      "step": 1677
    },
    {
      "epoch": 0.4646912212683467,
      "grad_norm": 1.4033349752426147,
      "learning_rate": 0.00011341333522541873,
      "loss": 2.5723,
      "step": 1678
    },
    {
      "epoch": 0.46496815286624205,
      "grad_norm": 1.4402943849563599,
      "learning_rate": 0.00011332590505211159,
      "loss": 2.7539,
      "step": 1679
    },
    {
      "epoch": 0.46524508446413737,
      "grad_norm": 1.549176573753357,
      "learning_rate": 0.00011323846450704381,
      "loss": 2.756,
      "step": 1680
    },
    {
      "epoch": 0.4655220160620327,
      "grad_norm": 2.345027446746826,
      "learning_rate": 0.00011315101365827177,
      "loss": 2.9999,
      "step": 1681
    },
    {
      "epoch": 0.465798947659928,
      "grad_norm": 1.1298154592514038,
      "learning_rate": 0.00011306355257385985,
      "loss": 2.6519,
      "step": 1682
    },
    {
      "epoch": 0.46607587925782334,
      "grad_norm": 2.0884218215942383,
      "learning_rate": 0.00011297608132188035,
      "loss": 2.8539,
      "step": 1683
    },
    {
      "epoch": 0.4663528108557186,
      "grad_norm": 1.6356163024902344,
      "learning_rate": 0.00011288859997041353,
      "loss": 2.7576,
      "step": 1684
    },
    {
      "epoch": 0.46662974245361394,
      "grad_norm": 2.01662540435791,
      "learning_rate": 0.00011280110858754749,
      "loss": 2.8964,
      "step": 1685
    },
    {
      "epoch": 0.46690667405150926,
      "grad_norm": 1.871184229850769,
      "learning_rate": 0.00011271360724137812,
      "loss": 2.7625,
      "step": 1686
    },
    {
      "epoch": 0.4671836056494046,
      "grad_norm": 0.8342902064323425,
      "learning_rate": 0.00011262609600000913,
      "loss": 2.7472,
      "step": 1687
    },
    {
      "epoch": 0.4674605372472999,
      "grad_norm": 2.2938618659973145,
      "learning_rate": 0.00011253857493155188,
      "loss": 2.9402,
      "step": 1688
    },
    {
      "epoch": 0.46773746884519524,
      "grad_norm": 1.5873247385025024,
      "learning_rate": 0.00011245104410412537,
      "loss": 2.9376,
      "step": 1689
    },
    {
      "epoch": 0.46801440044309056,
      "grad_norm": 1.8236016035079956,
      "learning_rate": 0.00011236350358585618,
      "loss": 2.7425,
      "step": 1690
    },
    {
      "epoch": 0.4682913320409859,
      "grad_norm": 2.1926429271698,
      "learning_rate": 0.0001122759534448786,
      "loss": 2.9251,
      "step": 1691
    },
    {
      "epoch": 0.4685682636388812,
      "grad_norm": 2.688744068145752,
      "learning_rate": 0.00011218839374933413,
      "loss": 3.0736,
      "step": 1692
    },
    {
      "epoch": 0.46884519523677654,
      "grad_norm": 2.1417758464813232,
      "learning_rate": 0.00011210082456737192,
      "loss": 2.7558,
      "step": 1693
    },
    {
      "epoch": 0.46912212683467186,
      "grad_norm": 1.4778650999069214,
      "learning_rate": 0.00011201324596714844,
      "loss": 2.7607,
      "step": 1694
    },
    {
      "epoch": 0.46939905843256713,
      "grad_norm": 1.9460140466690063,
      "learning_rate": 0.00011192565801682744,
      "loss": 2.9039,
      "step": 1695
    },
    {
      "epoch": 0.46967599003046245,
      "grad_norm": 1.354663610458374,
      "learning_rate": 0.00011183806078458003,
      "loss": 2.8078,
      "step": 1696
    },
    {
      "epoch": 0.4699529216283578,
      "grad_norm": 1.3584784269332886,
      "learning_rate": 0.00011175045433858456,
      "loss": 2.8078,
      "step": 1697
    },
    {
      "epoch": 0.4702298532262531,
      "grad_norm": 2.069406032562256,
      "learning_rate": 0.00011166283874702638,
      "loss": 2.8907,
      "step": 1698
    },
    {
      "epoch": 0.4705067848241484,
      "grad_norm": 1.4888218641281128,
      "learning_rate": 0.00011157521407809815,
      "loss": 2.8859,
      "step": 1699
    },
    {
      "epoch": 0.47078371642204375,
      "grad_norm": 3.220172643661499,
      "learning_rate": 0.00011148758039999952,
      "loss": 3.7449,
      "step": 1700
    },
    {
      "epoch": 0.4710606480199391,
      "grad_norm": 1.4528577327728271,
      "learning_rate": 0.00011139993778093711,
      "loss": 2.6247,
      "step": 1701
    },
    {
      "epoch": 0.4713375796178344,
      "grad_norm": 0.8638042211532593,
      "learning_rate": 0.00011131228628912462,
      "loss": 2.6005,
      "step": 1702
    },
    {
      "epoch": 0.4716145112157297,
      "grad_norm": 2.3900153636932373,
      "learning_rate": 0.00011122462599278254,
      "loss": 3.0792,
      "step": 1703
    },
    {
      "epoch": 0.47189144281362505,
      "grad_norm": 1.8319112062454224,
      "learning_rate": 0.00011113695696013824,
      "loss": 2.6175,
      "step": 1704
    },
    {
      "epoch": 0.4721683744115204,
      "grad_norm": 1.171964168548584,
      "learning_rate": 0.0001110492792594259,
      "loss": 2.6471,
      "step": 1705
    },
    {
      "epoch": 0.4724453060094157,
      "grad_norm": 1.1825437545776367,
      "learning_rate": 0.00011096159295888646,
      "loss": 2.7537,
      "step": 1706
    },
    {
      "epoch": 0.47272223760731097,
      "grad_norm": 1.0841095447540283,
      "learning_rate": 0.00011087389812676754,
      "loss": 2.7579,
      "step": 1707
    },
    {
      "epoch": 0.4729991692052063,
      "grad_norm": 2.2057273387908936,
      "learning_rate": 0.0001107861948313234,
      "loss": 2.9958,
      "step": 1708
    },
    {
      "epoch": 0.4732761008031016,
      "grad_norm": 2.306844711303711,
      "learning_rate": 0.0001106984831408149,
      "loss": 3.1338,
      "step": 1709
    },
    {
      "epoch": 0.47355303240099694,
      "grad_norm": 1.9412044286727905,
      "learning_rate": 0.0001106107631235094,
      "loss": 2.8777,
      "step": 1710
    },
    {
      "epoch": 0.47382996399889227,
      "grad_norm": 2.0891644954681396,
      "learning_rate": 0.00011052303484768076,
      "loss": 2.8106,
      "step": 1711
    },
    {
      "epoch": 0.4741068955967876,
      "grad_norm": 1.4442548751831055,
      "learning_rate": 0.0001104352983816093,
      "loss": 2.8836,
      "step": 1712
    },
    {
      "epoch": 0.4743838271946829,
      "grad_norm": 1.5611284971237183,
      "learning_rate": 0.00011034755379358166,
      "loss": 2.72,
      "step": 1713
    },
    {
      "epoch": 0.47466075879257824,
      "grad_norm": 1.841544508934021,
      "learning_rate": 0.00011025980115189086,
      "loss": 2.7075,
      "step": 1714
    },
    {
      "epoch": 0.47493769039047357,
      "grad_norm": 1.9898557662963867,
      "learning_rate": 0.00011017204052483613,
      "loss": 2.8212,
      "step": 1715
    },
    {
      "epoch": 0.4752146219883689,
      "grad_norm": 1.2505582571029663,
      "learning_rate": 0.00011008427198072294,
      "loss": 2.8529,
      "step": 1716
    },
    {
      "epoch": 0.4754915535862642,
      "grad_norm": 1.7798211574554443,
      "learning_rate": 0.00010999649558786291,
      "loss": 2.7741,
      "step": 1717
    },
    {
      "epoch": 0.47576848518415954,
      "grad_norm": 1.4395679235458374,
      "learning_rate": 0.00010990871141457382,
      "loss": 2.9254,
      "step": 1718
    },
    {
      "epoch": 0.4760454167820548,
      "grad_norm": 1.0878676176071167,
      "learning_rate": 0.00010982091952917943,
      "loss": 2.8418,
      "step": 1719
    },
    {
      "epoch": 0.47632234837995013,
      "grad_norm": 1.452322244644165,
      "learning_rate": 0.00010973312000000956,
      "loss": 2.5341,
      "step": 1720
    },
    {
      "epoch": 0.47659927997784546,
      "grad_norm": 0.7994594573974609,
      "learning_rate": 0.00010964531289539996,
      "loss": 2.7033,
      "step": 1721
    },
    {
      "epoch": 0.4768762115757408,
      "grad_norm": 2.953162908554077,
      "learning_rate": 0.00010955749828369224,
      "loss": 3.1346,
      "step": 1722
    },
    {
      "epoch": 0.4771531431736361,
      "grad_norm": 1.1681833267211914,
      "learning_rate": 0.00010946967623323394,
      "loss": 2.8322,
      "step": 1723
    },
    {
      "epoch": 0.47743007477153143,
      "grad_norm": 2.5147387981414795,
      "learning_rate": 0.00010938184681237833,
      "loss": 2.9224,
      "step": 1724
    },
    {
      "epoch": 0.47770700636942676,
      "grad_norm": 3.0960545539855957,
      "learning_rate": 0.00010929401008948437,
      "loss": 3.5451,
      "step": 1725
    },
    {
      "epoch": 0.4779839379673221,
      "grad_norm": 1.759600043296814,
      "learning_rate": 0.00010920616613291682,
      "loss": 3.0218,
      "step": 1726
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 2.5177507400512695,
      "learning_rate": 0.00010911831501104597,
      "loss": 2.9697,
      "step": 1727
    },
    {
      "epoch": 0.47853780116311273,
      "grad_norm": 1.8307939767837524,
      "learning_rate": 0.00010903045679224772,
      "loss": 2.6927,
      "step": 1728
    },
    {
      "epoch": 0.47881473276100806,
      "grad_norm": 1.502057671546936,
      "learning_rate": 0.00010894259154490354,
      "loss": 2.6628,
      "step": 1729
    },
    {
      "epoch": 0.4790916643589033,
      "grad_norm": 1.1559317111968994,
      "learning_rate": 0.0001088547193374003,
      "loss": 2.8771,
      "step": 1730
    },
    {
      "epoch": 0.47936859595679865,
      "grad_norm": 1.6867003440856934,
      "learning_rate": 0.00010876684023813029,
      "loss": 3.0452,
      "step": 1731
    },
    {
      "epoch": 0.479645527554694,
      "grad_norm": 1.1228195428848267,
      "learning_rate": 0.00010867895431549122,
      "loss": 2.617,
      "step": 1732
    },
    {
      "epoch": 0.4799224591525893,
      "grad_norm": 2.0624401569366455,
      "learning_rate": 0.00010859106163788608,
      "loss": 2.8471,
      "step": 1733
    },
    {
      "epoch": 0.4801993907504846,
      "grad_norm": 2.141110420227051,
      "learning_rate": 0.00010850316227372312,
      "loss": 2.9351,
      "step": 1734
    },
    {
      "epoch": 0.48047632234837995,
      "grad_norm": 1.0106289386749268,
      "learning_rate": 0.0001084152562914158,
      "loss": 2.757,
      "step": 1735
    },
    {
      "epoch": 0.48075325394627527,
      "grad_norm": 1.6883660554885864,
      "learning_rate": 0.00010832734375938269,
      "loss": 2.8802,
      "step": 1736
    },
    {
      "epoch": 0.4810301855441706,
      "grad_norm": 1.8972655534744263,
      "learning_rate": 0.00010823942474604752,
      "loss": 2.8388,
      "step": 1737
    },
    {
      "epoch": 0.4813071171420659,
      "grad_norm": 2.012760639190674,
      "learning_rate": 0.00010815149931983902,
      "loss": 2.9814,
      "step": 1738
    },
    {
      "epoch": 0.48158404873996125,
      "grad_norm": 1.6839152574539185,
      "learning_rate": 0.00010806356754919091,
      "loss": 2.8423,
      "step": 1739
    },
    {
      "epoch": 0.48186098033785657,
      "grad_norm": 1.588602900505066,
      "learning_rate": 0.0001079756295025419,
      "loss": 2.4987,
      "step": 1740
    },
    {
      "epoch": 0.4821379119357519,
      "grad_norm": 2.094052791595459,
      "learning_rate": 0.00010788768524833555,
      "loss": 2.9114,
      "step": 1741
    },
    {
      "epoch": 0.48241484353364716,
      "grad_norm": 2.5224688053131104,
      "learning_rate": 0.0001077997348550202,
      "loss": 3.3598,
      "step": 1742
    },
    {
      "epoch": 0.4826917751315425,
      "grad_norm": 2.207646608352661,
      "learning_rate": 0.00010771177839104907,
      "loss": 2.9569,
      "step": 1743
    },
    {
      "epoch": 0.4829687067294378,
      "grad_norm": 1.835942268371582,
      "learning_rate": 0.00010762381592488002,
      "loss": 2.6772,
      "step": 1744
    },
    {
      "epoch": 0.48324563832733314,
      "grad_norm": 1.3604185581207275,
      "learning_rate": 0.00010753584752497565,
      "loss": 2.6844,
      "step": 1745
    },
    {
      "epoch": 0.48352256992522846,
      "grad_norm": 1.3155308961868286,
      "learning_rate": 0.0001074478732598031,
      "loss": 2.7029,
      "step": 1746
    },
    {
      "epoch": 0.4837995015231238,
      "grad_norm": 2.120432138442993,
      "learning_rate": 0.00010735989319783418,
      "loss": 2.6089,
      "step": 1747
    },
    {
      "epoch": 0.4840764331210191,
      "grad_norm": 1.0304890871047974,
      "learning_rate": 0.0001072719074075451,
      "loss": 2.577,
      "step": 1748
    },
    {
      "epoch": 0.48435336471891444,
      "grad_norm": 1.767606258392334,
      "learning_rate": 0.00010718391595741657,
      "loss": 2.8686,
      "step": 1749
    },
    {
      "epoch": 0.48463029631680976,
      "grad_norm": 4.137774467468262,
      "learning_rate": 0.00010709591891593376,
      "loss": 4.0072,
      "step": 1750
    },
    {
      "epoch": 0.4849072279147051,
      "grad_norm": 1.735060691833496,
      "learning_rate": 0.00010700791635158611,
      "loss": 2.7465,
      "step": 1751
    },
    {
      "epoch": 0.4851841595126004,
      "grad_norm": 1.67653489112854,
      "learning_rate": 0.0001069199083328674,
      "loss": 2.8397,
      "step": 1752
    },
    {
      "epoch": 0.48546109111049573,
      "grad_norm": 1.4346954822540283,
      "learning_rate": 0.00010683189492827567,
      "loss": 2.7064,
      "step": 1753
    },
    {
      "epoch": 0.485738022708391,
      "grad_norm": 1.6197758913040161,
      "learning_rate": 0.00010674387620631308,
      "loss": 2.8717,
      "step": 1754
    },
    {
      "epoch": 0.48601495430628633,
      "grad_norm": 1.7235110998153687,
      "learning_rate": 0.00010665585223548604,
      "loss": 2.5654,
      "step": 1755
    },
    {
      "epoch": 0.48629188590418165,
      "grad_norm": 1.6998274326324463,
      "learning_rate": 0.00010656782308430498,
      "loss": 2.8146,
      "step": 1756
    },
    {
      "epoch": 0.486568817502077,
      "grad_norm": 1.2824156284332275,
      "learning_rate": 0.00010647978882128431,
      "loss": 2.7132,
      "step": 1757
    },
    {
      "epoch": 0.4868457490999723,
      "grad_norm": 1.7642462253570557,
      "learning_rate": 0.00010639174951494253,
      "loss": 2.9732,
      "step": 1758
    },
    {
      "epoch": 0.4871226806978676,
      "grad_norm": 1.8423540592193604,
      "learning_rate": 0.00010630370523380202,
      "loss": 2.9273,
      "step": 1759
    },
    {
      "epoch": 0.48739961229576295,
      "grad_norm": 3.762970209121704,
      "learning_rate": 0.00010621565604638897,
      "loss": 2.8981,
      "step": 1760
    },
    {
      "epoch": 0.4876765438936583,
      "grad_norm": 2.405668258666992,
      "learning_rate": 0.00010612760202123346,
      "loss": 2.6629,
      "step": 1761
    },
    {
      "epoch": 0.4879534754915536,
      "grad_norm": 1.2584848403930664,
      "learning_rate": 0.00010603954322686935,
      "loss": 2.6843,
      "step": 1762
    },
    {
      "epoch": 0.4882304070894489,
      "grad_norm": 1.6214781999588013,
      "learning_rate": 0.00010595147973183415,
      "loss": 2.8037,
      "step": 1763
    },
    {
      "epoch": 0.48850733868734425,
      "grad_norm": 1.881946325302124,
      "learning_rate": 0.00010586341160466904,
      "loss": 3.0972,
      "step": 1764
    },
    {
      "epoch": 0.4887842702852395,
      "grad_norm": 1.7505426406860352,
      "learning_rate": 0.00010577533891391884,
      "loss": 2.8725,
      "step": 1765
    },
    {
      "epoch": 0.48906120188313484,
      "grad_norm": 2.402600049972534,
      "learning_rate": 0.00010568726172813193,
      "loss": 2.8975,
      "step": 1766
    },
    {
      "epoch": 0.48933813348103017,
      "grad_norm": 2.2055821418762207,
      "learning_rate": 0.00010559918011586014,
      "loss": 2.9041,
      "step": 1767
    },
    {
      "epoch": 0.4896150650789255,
      "grad_norm": 2.2101831436157227,
      "learning_rate": 0.00010551109414565878,
      "loss": 2.9287,
      "step": 1768
    },
    {
      "epoch": 0.4898919966768208,
      "grad_norm": 2.1074306964874268,
      "learning_rate": 0.00010542300388608652,
      "loss": 2.8658,
      "step": 1769
    },
    {
      "epoch": 0.49016892827471614,
      "grad_norm": 2.2460777759552,
      "learning_rate": 0.00010533490940570541,
      "loss": 2.598,
      "step": 1770
    },
    {
      "epoch": 0.49044585987261147,
      "grad_norm": 1.70467209815979,
      "learning_rate": 0.00010524681077308077,
      "loss": 2.8554,
      "step": 1771
    },
    {
      "epoch": 0.4907227914705068,
      "grad_norm": 1.0729258060455322,
      "learning_rate": 0.00010515870805678109,
      "loss": 2.7869,
      "step": 1772
    },
    {
      "epoch": 0.4909997230684021,
      "grad_norm": 1.8093794584274292,
      "learning_rate": 0.00010507060132537813,
      "loss": 3.181,
      "step": 1773
    },
    {
      "epoch": 0.49127665466629744,
      "grad_norm": 1.5102819204330444,
      "learning_rate": 0.00010498249064744679,
      "loss": 2.9491,
      "step": 1774
    },
    {
      "epoch": 0.49155358626419277,
      "grad_norm": 3.7557647228240967,
      "learning_rate": 0.00010489437609156492,
      "loss": 3.8743,
      "step": 1775
    },
    {
      "epoch": 0.4918305178620881,
      "grad_norm": 3.593411445617676,
      "learning_rate": 0.00010480625772631347,
      "loss": 2.7411,
      "step": 1776
    },
    {
      "epoch": 0.49210744945998336,
      "grad_norm": 2.1997084617614746,
      "learning_rate": 0.00010471813562027634,
      "loss": 2.8769,
      "step": 1777
    },
    {
      "epoch": 0.4923843810578787,
      "grad_norm": 1.8847730159759521,
      "learning_rate": 0.00010463000984204038,
      "loss": 2.8433,
      "step": 1778
    },
    {
      "epoch": 0.492661312655774,
      "grad_norm": 1.8461335897445679,
      "learning_rate": 0.00010454188046019524,
      "loss": 2.6241,
      "step": 1779
    },
    {
      "epoch": 0.49293824425366933,
      "grad_norm": 1.412909984588623,
      "learning_rate": 0.00010445374754333344,
      "loss": 2.8483,
      "step": 1780
    },
    {
      "epoch": 0.49321517585156466,
      "grad_norm": 2.3904054164886475,
      "learning_rate": 0.00010436561116005012,
      "loss": 3.296,
      "step": 1781
    },
    {
      "epoch": 0.49349210744946,
      "grad_norm": 1.9982287883758545,
      "learning_rate": 0.00010427747137894328,
      "loss": 2.91,
      "step": 1782
    },
    {
      "epoch": 0.4937690390473553,
      "grad_norm": 2.10746169090271,
      "learning_rate": 0.00010418932826861349,
      "loss": 2.8191,
      "step": 1783
    },
    {
      "epoch": 0.49404597064525063,
      "grad_norm": 1.9534249305725098,
      "learning_rate": 0.00010410118189766387,
      "loss": 2.6736,
      "step": 1784
    },
    {
      "epoch": 0.49432290224314596,
      "grad_norm": 1.4590779542922974,
      "learning_rate": 0.00010401303233470014,
      "loss": 2.5901,
      "step": 1785
    },
    {
      "epoch": 0.4945998338410413,
      "grad_norm": 1.5505417585372925,
      "learning_rate": 0.00010392487964833051,
      "loss": 2.7745,
      "step": 1786
    },
    {
      "epoch": 0.4948767654389366,
      "grad_norm": 1.32901930809021,
      "learning_rate": 0.00010383672390716556,
      "loss": 2.7089,
      "step": 1787
    },
    {
      "epoch": 0.49515369703683193,
      "grad_norm": 1.4571396112442017,
      "learning_rate": 0.00010374856517981832,
      "loss": 3.0791,
      "step": 1788
    },
    {
      "epoch": 0.4954306286347272,
      "grad_norm": 1.337172508239746,
      "learning_rate": 0.0001036604035349041,
      "loss": 2.7896,
      "step": 1789
    },
    {
      "epoch": 0.4957075602326225,
      "grad_norm": 2.150773286819458,
      "learning_rate": 0.00010357223904104045,
      "loss": 2.8772,
      "step": 1790
    },
    {
      "epoch": 0.49598449183051785,
      "grad_norm": 2.0338151454925537,
      "learning_rate": 0.00010348407176684723,
      "loss": 2.8992,
      "step": 1791
    },
    {
      "epoch": 0.4962614234284132,
      "grad_norm": 1.7818877696990967,
      "learning_rate": 0.00010339590178094638,
      "loss": 2.8217,
      "step": 1792
    },
    {
      "epoch": 0.4965383550263085,
      "grad_norm": 1.5991873741149902,
      "learning_rate": 0.00010330772915196199,
      "loss": 2.8831,
      "step": 1793
    },
    {
      "epoch": 0.4968152866242038,
      "grad_norm": 4.544161796569824,
      "learning_rate": 0.00010321955394852018,
      "loss": 3.1074,
      "step": 1794
    },
    {
      "epoch": 0.49709221822209915,
      "grad_norm": 1.8289427757263184,
      "learning_rate": 0.00010313137623924913,
      "loss": 2.7945,
      "step": 1795
    },
    {
      "epoch": 0.49736914981999447,
      "grad_norm": 1.4768778085708618,
      "learning_rate": 0.00010304319609277888,
      "loss": 2.8313,
      "step": 1796
    },
    {
      "epoch": 0.4976460814178898,
      "grad_norm": 2.174567222595215,
      "learning_rate": 0.00010295501357774146,
      "loss": 2.9793,
      "step": 1797
    },
    {
      "epoch": 0.4979230130157851,
      "grad_norm": 3.003870725631714,
      "learning_rate": 0.00010286682876277069,
      "loss": 2.7108,
      "step": 1798
    },
    {
      "epoch": 0.49819994461368045,
      "grad_norm": 1.7297585010528564,
      "learning_rate": 0.0001027786417165022,
      "loss": 2.899,
      "step": 1799
    },
    {
      "epoch": 0.4984768762115757,
      "grad_norm": 2.468169689178467,
      "learning_rate": 0.00010269045250757332,
      "loss": 3.4341,
      "step": 1800
    },
    {
      "epoch": 0.49875380780947104,
      "grad_norm": 1.8460444211959839,
      "learning_rate": 0.00010260226120462315,
      "loss": 2.7304,
      "step": 1801
    },
    {
      "epoch": 0.49903073940736636,
      "grad_norm": 1.7109407186508179,
      "learning_rate": 0.00010251406787629232,
      "loss": 2.7065,
      "step": 1802
    },
    {
      "epoch": 0.4993076710052617,
      "grad_norm": 2.210998773574829,
      "learning_rate": 0.00010242587259122307,
      "loss": 2.7809,
      "step": 1803
    },
    {
      "epoch": 0.499584602603157,
      "grad_norm": 1.3211313486099243,
      "learning_rate": 0.0001023376754180592,
      "loss": 2.6021,
      "step": 1804
    },
    {
      "epoch": 0.49986153420105234,
      "grad_norm": 2.163815975189209,
      "learning_rate": 0.00010224947642544593,
      "loss": 2.6656,
      "step": 1805
    },
    {
      "epoch": 0.5001384657989476,
      "grad_norm": 1.8360047340393066,
      "learning_rate": 0.00010216127568202994,
      "loss": 3.0708,
      "step": 1806
    },
    {
      "epoch": 0.5001384657989476,
      "eval_loss": 0.3538380563259125,
      "eval_runtime": 801.5582,
      "eval_samples_per_second": 7.588,
      "eval_steps_per_second": 1.898,
      "step": 1806
    },
    {
      "epoch": 0.5004153973968429,
      "grad_norm": 1.0672329664230347,
      "learning_rate": 0.00010207307325645923,
      "loss": 2.6721,
      "step": 1807
    },
    {
      "epoch": 0.5006923289947383,
      "grad_norm": 1.2056281566619873,
      "learning_rate": 0.00010198486921738313,
      "loss": 2.6439,
      "step": 1808
    },
    {
      "epoch": 0.5009692605926336,
      "grad_norm": 2.3176612854003906,
      "learning_rate": 0.00010189666363345223,
      "loss": 2.9849,
      "step": 1809
    },
    {
      "epoch": 0.5012461921905289,
      "grad_norm": 2.192941427230835,
      "learning_rate": 0.00010180845657331832,
      "loss": 3.0648,
      "step": 1810
    },
    {
      "epoch": 0.5015231237884242,
      "grad_norm": 1.4033616781234741,
      "learning_rate": 0.00010172024810563434,
      "loss": 2.6953,
      "step": 1811
    },
    {
      "epoch": 0.5018000553863196,
      "grad_norm": 1.9293441772460938,
      "learning_rate": 0.0001016320382990543,
      "loss": 2.7721,
      "step": 1812
    },
    {
      "epoch": 0.5020769869842149,
      "grad_norm": 1.4902727603912354,
      "learning_rate": 0.00010154382722223333,
      "loss": 2.6199,
      "step": 1813
    },
    {
      "epoch": 0.5023539185821102,
      "grad_norm": 1.641219973564148,
      "learning_rate": 0.00010145561494382742,
      "loss": 2.8022,
      "step": 1814
    },
    {
      "epoch": 0.5026308501800055,
      "grad_norm": 2.4387571811676025,
      "learning_rate": 0.0001013674015324936,
      "loss": 2.9015,
      "step": 1815
    },
    {
      "epoch": 0.5029077817779009,
      "grad_norm": 2.294370651245117,
      "learning_rate": 0.00010127918705688978,
      "loss": 2.7876,
      "step": 1816
    },
    {
      "epoch": 0.5031847133757962,
      "grad_norm": 1.5861742496490479,
      "learning_rate": 0.00010119097158567461,
      "loss": 2.8373,
      "step": 1817
    },
    {
      "epoch": 0.5034616449736915,
      "grad_norm": 2.003089666366577,
      "learning_rate": 0.0001011027551875076,
      "loss": 2.5265,
      "step": 1818
    },
    {
      "epoch": 0.5037385765715868,
      "grad_norm": 1.5193865299224854,
      "learning_rate": 0.00010101453793104898,
      "loss": 2.617,
      "step": 1819
    },
    {
      "epoch": 0.5040155081694822,
      "grad_norm": 2.096163511276245,
      "learning_rate": 0.00010092631988495957,
      "loss": 2.9376,
      "step": 1820
    },
    {
      "epoch": 0.5042924397673775,
      "grad_norm": 1.358824610710144,
      "learning_rate": 0.0001008381011179009,
      "loss": 2.6227,
      "step": 1821
    },
    {
      "epoch": 0.5045693713652728,
      "grad_norm": 1.8758240938186646,
      "learning_rate": 0.00010074988169853503,
      "loss": 2.7398,
      "step": 1822
    },
    {
      "epoch": 0.5048463029631681,
      "grad_norm": 1.6822946071624756,
      "learning_rate": 0.00010066166169552444,
      "loss": 3.0444,
      "step": 1823
    },
    {
      "epoch": 0.5051232345610634,
      "grad_norm": 2.4263274669647217,
      "learning_rate": 0.00010057344117753222,
      "loss": 2.781,
      "step": 1824
    },
    {
      "epoch": 0.5054001661589588,
      "grad_norm": 3.9306766986846924,
      "learning_rate": 0.00010048522021322176,
      "loss": 3.7365,
      "step": 1825
    },
    {
      "epoch": 0.5056770977568541,
      "grad_norm": 1.615681529045105,
      "learning_rate": 0.00010039699887125678,
      "loss": 2.6159,
      "step": 1826
    },
    {
      "epoch": 0.5059540293547494,
      "grad_norm": 2.1996114253997803,
      "learning_rate": 0.00010030877722030136,
      "loss": 3.1963,
      "step": 1827
    },
    {
      "epoch": 0.5062309609526447,
      "grad_norm": 1.7249773740768433,
      "learning_rate": 0.00010022055532901984,
      "loss": 2.8976,
      "step": 1828
    },
    {
      "epoch": 0.50650789255054,
      "grad_norm": 1.4680061340332031,
      "learning_rate": 0.00010013233326607661,
      "loss": 2.9497,
      "step": 1829
    },
    {
      "epoch": 0.5067848241484353,
      "grad_norm": 1.1771354675292969,
      "learning_rate": 0.00010004411110013634,
      "loss": 2.7234,
      "step": 1830
    },
    {
      "epoch": 0.5070617557463306,
      "grad_norm": 2.575040340423584,
      "learning_rate": 9.99558888998637e-05,
      "loss": 3.0858,
      "step": 1831
    },
    {
      "epoch": 0.5073386873442259,
      "grad_norm": 2.529264450073242,
      "learning_rate": 9.986766673392344e-05,
      "loss": 2.6105,
      "step": 1832
    },
    {
      "epoch": 0.5076156189421213,
      "grad_norm": 1.6454603672027588,
      "learning_rate": 9.977944467098018e-05,
      "loss": 2.6702,
      "step": 1833
    },
    {
      "epoch": 0.5078925505400166,
      "grad_norm": 2.0674197673797607,
      "learning_rate": 9.969122277969865e-05,
      "loss": 2.8699,
      "step": 1834
    },
    {
      "epoch": 0.5081694821379119,
      "grad_norm": 3.0945839881896973,
      "learning_rate": 9.960300112874326e-05,
      "loss": 3.1807,
      "step": 1835
    },
    {
      "epoch": 0.5084464137358072,
      "grad_norm": 1.3725661039352417,
      "learning_rate": 9.951477978677826e-05,
      "loss": 2.8623,
      "step": 1836
    },
    {
      "epoch": 0.5087233453337026,
      "grad_norm": 1.615574836730957,
      "learning_rate": 9.942655882246781e-05,
      "loss": 3.1129,
      "step": 1837
    },
    {
      "epoch": 0.5090002769315979,
      "grad_norm": 2.0210142135620117,
      "learning_rate": 9.93383383044756e-05,
      "loss": 2.7283,
      "step": 1838
    },
    {
      "epoch": 0.5092772085294932,
      "grad_norm": 2.330843925476074,
      "learning_rate": 9.9250118301465e-05,
      "loss": 2.8052,
      "step": 1839
    },
    {
      "epoch": 0.5095541401273885,
      "grad_norm": 1.1833444833755493,
      "learning_rate": 9.916189888209912e-05,
      "loss": 2.967,
      "step": 1840
    },
    {
      "epoch": 0.5098310717252839,
      "grad_norm": 1.0780837535858154,
      "learning_rate": 9.907368011504045e-05,
      "loss": 2.8319,
      "step": 1841
    },
    {
      "epoch": 0.5101080033231792,
      "grad_norm": 1.9871891736984253,
      "learning_rate": 9.898546206895106e-05,
      "loss": 2.8931,
      "step": 1842
    },
    {
      "epoch": 0.5103849349210745,
      "grad_norm": 1.5821150541305542,
      "learning_rate": 9.88972448124924e-05,
      "loss": 2.9264,
      "step": 1843
    },
    {
      "epoch": 0.5106618665189698,
      "grad_norm": 1.4421741962432861,
      "learning_rate": 9.880902841432544e-05,
      "loss": 2.5686,
      "step": 1844
    },
    {
      "epoch": 0.5109387981168652,
      "grad_norm": 1.768431544303894,
      "learning_rate": 9.872081294311024e-05,
      "loss": 2.6354,
      "step": 1845
    },
    {
      "epoch": 0.5112157297147605,
      "grad_norm": 1.4881033897399902,
      "learning_rate": 9.863259846750641e-05,
      "loss": 2.4879,
      "step": 1846
    },
    {
      "epoch": 0.5114926613126558,
      "grad_norm": 1.8734623193740845,
      "learning_rate": 9.854438505617263e-05,
      "loss": 2.5476,
      "step": 1847
    },
    {
      "epoch": 0.5117695929105511,
      "grad_norm": 2.301177501678467,
      "learning_rate": 9.84561727777667e-05,
      "loss": 3.0524,
      "step": 1848
    },
    {
      "epoch": 0.5120465245084465,
      "grad_norm": 1.8294378519058228,
      "learning_rate": 9.836796170094571e-05,
      "loss": 2.6988,
      "step": 1849
    },
    {
      "epoch": 0.5123234561063418,
      "grad_norm": 3.631112813949585,
      "learning_rate": 9.827975189436571e-05,
      "loss": 3.7903,
      "step": 1850
    },
    {
      "epoch": 0.5126003877042371,
      "grad_norm": 1.5857528448104858,
      "learning_rate": 9.81915434266817e-05,
      "loss": 2.7797,
      "step": 1851
    },
    {
      "epoch": 0.5128773193021324,
      "grad_norm": 1.5357552766799927,
      "learning_rate": 9.81033363665478e-05,
      "loss": 2.6727,
      "step": 1852
    },
    {
      "epoch": 0.5131542509000276,
      "grad_norm": 2.5034070014953613,
      "learning_rate": 9.801513078261692e-05,
      "loss": 2.9577,
      "step": 1853
    },
    {
      "epoch": 0.513431182497923,
      "grad_norm": 2.242323875427246,
      "learning_rate": 9.792692674354079e-05,
      "loss": 2.8901,
      "step": 1854
    },
    {
      "epoch": 0.5137081140958183,
      "grad_norm": 2.157565116882324,
      "learning_rate": 9.783872431797008e-05,
      "loss": 2.9966,
      "step": 1855
    },
    {
      "epoch": 0.5139850456937136,
      "grad_norm": 1.0566247701644897,
      "learning_rate": 9.77505235745541e-05,
      "loss": 2.6681,
      "step": 1856
    },
    {
      "epoch": 0.5142619772916089,
      "grad_norm": 1.577938437461853,
      "learning_rate": 9.766232458194082e-05,
      "loss": 2.8405,
      "step": 1857
    },
    {
      "epoch": 0.5145389088895043,
      "grad_norm": 0.847844660282135,
      "learning_rate": 9.757412740877697e-05,
      "loss": 2.7065,
      "step": 1858
    },
    {
      "epoch": 0.5148158404873996,
      "grad_norm": 1.4090830087661743,
      "learning_rate": 9.748593212370773e-05,
      "loss": 2.5943,
      "step": 1859
    },
    {
      "epoch": 0.5150927720852949,
      "grad_norm": 1.3234925270080566,
      "learning_rate": 9.739773879537687e-05,
      "loss": 2.6805,
      "step": 1860
    },
    {
      "epoch": 0.5153697036831902,
      "grad_norm": 1.1672470569610596,
      "learning_rate": 9.73095474924267e-05,
      "loss": 2.8095,
      "step": 1861
    },
    {
      "epoch": 0.5156466352810856,
      "grad_norm": 1.7627800703048706,
      "learning_rate": 9.722135828349784e-05,
      "loss": 2.6748,
      "step": 1862
    },
    {
      "epoch": 0.5159235668789809,
      "grad_norm": 1.1423671245574951,
      "learning_rate": 9.713317123722933e-05,
      "loss": 2.7358,
      "step": 1863
    },
    {
      "epoch": 0.5162004984768762,
      "grad_norm": 3.5104258060455322,
      "learning_rate": 9.704498642225856e-05,
      "loss": 3.2547,
      "step": 1864
    },
    {
      "epoch": 0.5164774300747715,
      "grad_norm": 2.085096836090088,
      "learning_rate": 9.695680390722116e-05,
      "loss": 2.9059,
      "step": 1865
    },
    {
      "epoch": 0.5167543616726669,
      "grad_norm": 1.6688218116760254,
      "learning_rate": 9.68686237607509e-05,
      "loss": 2.8208,
      "step": 1866
    },
    {
      "epoch": 0.5170312932705622,
      "grad_norm": 1.3693195581436157,
      "learning_rate": 9.678044605147984e-05,
      "loss": 2.737,
      "step": 1867
    },
    {
      "epoch": 0.5173082248684575,
      "grad_norm": 1.1143624782562256,
      "learning_rate": 9.669227084803806e-05,
      "loss": 2.8413,
      "step": 1868
    },
    {
      "epoch": 0.5175851564663528,
      "grad_norm": 1.4536685943603516,
      "learning_rate": 9.660409821905363e-05,
      "loss": 2.6045,
      "step": 1869
    },
    {
      "epoch": 0.5178620880642482,
      "grad_norm": 1.3811722993850708,
      "learning_rate": 9.65159282331528e-05,
      "loss": 2.4863,
      "step": 1870
    },
    {
      "epoch": 0.5181390196621435,
      "grad_norm": 2.0206351280212402,
      "learning_rate": 9.642776095895959e-05,
      "loss": 2.6733,
      "step": 1871
    },
    {
      "epoch": 0.5184159512600388,
      "grad_norm": 1.589301586151123,
      "learning_rate": 9.633959646509592e-05,
      "loss": 2.573,
      "step": 1872
    },
    {
      "epoch": 0.5186928828579341,
      "grad_norm": 1.077090859413147,
      "learning_rate": 9.62514348201817e-05,
      "loss": 2.8819,
      "step": 1873
    },
    {
      "epoch": 0.5189698144558295,
      "grad_norm": 1.8220001459121704,
      "learning_rate": 9.616327609283445e-05,
      "loss": 2.8533,
      "step": 1874
    },
    {
      "epoch": 0.5192467460537248,
      "grad_norm": 4.240963459014893,
      "learning_rate": 9.607512035166951e-05,
      "loss": 4.215,
      "step": 1875
    },
    {
      "epoch": 0.51952367765162,
      "grad_norm": 1.659738540649414,
      "learning_rate": 9.598696766529987e-05,
      "loss": 2.5913,
      "step": 1876
    },
    {
      "epoch": 0.5198006092495153,
      "grad_norm": 2.6843607425689697,
      "learning_rate": 9.589881810233617e-05,
      "loss": 3.0707,
      "step": 1877
    },
    {
      "epoch": 0.5200775408474106,
      "grad_norm": 1.4031602144241333,
      "learning_rate": 9.581067173138653e-05,
      "loss": 2.9531,
      "step": 1878
    },
    {
      "epoch": 0.520354472445306,
      "grad_norm": 1.6640127897262573,
      "learning_rate": 9.572252862105673e-05,
      "loss": 2.8823,
      "step": 1879
    },
    {
      "epoch": 0.5206314040432013,
      "grad_norm": 1.616991639137268,
      "learning_rate": 9.563438883994992e-05,
      "loss": 2.8805,
      "step": 1880
    },
    {
      "epoch": 0.5209083356410966,
      "grad_norm": 1.947402000427246,
      "learning_rate": 9.554625245666658e-05,
      "loss": 2.8934,
      "step": 1881
    },
    {
      "epoch": 0.521185267238992,
      "grad_norm": 1.7645297050476074,
      "learning_rate": 9.545811953980477e-05,
      "loss": 2.6654,
      "step": 1882
    },
    {
      "epoch": 0.5214621988368873,
      "grad_norm": 1.5257445573806763,
      "learning_rate": 9.536999015795963e-05,
      "loss": 2.9309,
      "step": 1883
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 1.9620789289474487,
      "learning_rate": 9.528186437972368e-05,
      "loss": 2.8087,
      "step": 1884
    },
    {
      "epoch": 0.5220160620326779,
      "grad_norm": 1.9745149612426758,
      "learning_rate": 9.519374227368656e-05,
      "loss": 2.5654,
      "step": 1885
    },
    {
      "epoch": 0.5222929936305732,
      "grad_norm": 1.953885793685913,
      "learning_rate": 9.510562390843513e-05,
      "loss": 2.6027,
      "step": 1886
    },
    {
      "epoch": 0.5225699252284686,
      "grad_norm": 1.6615899801254272,
      "learning_rate": 9.501750935255322e-05,
      "loss": 2.7731,
      "step": 1887
    },
    {
      "epoch": 0.5228468568263639,
      "grad_norm": 1.7982279062271118,
      "learning_rate": 9.492939867462188e-05,
      "loss": 2.7539,
      "step": 1888
    },
    {
      "epoch": 0.5231237884242592,
      "grad_norm": 0.8459172248840332,
      "learning_rate": 9.484129194321896e-05,
      "loss": 2.8204,
      "step": 1889
    },
    {
      "epoch": 0.5234007200221545,
      "grad_norm": 1.4594982862472534,
      "learning_rate": 9.475318922691926e-05,
      "loss": 2.9914,
      "step": 1890
    },
    {
      "epoch": 0.5236776516200499,
      "grad_norm": 1.65652596950531,
      "learning_rate": 9.466509059429461e-05,
      "loss": 2.8122,
      "step": 1891
    },
    {
      "epoch": 0.5239545832179452,
      "grad_norm": 1.3848004341125488,
      "learning_rate": 9.457699611391352e-05,
      "loss": 2.8318,
      "step": 1892
    },
    {
      "epoch": 0.5242315148158405,
      "grad_norm": 1.2636313438415527,
      "learning_rate": 9.448890585434124e-05,
      "loss": 2.8518,
      "step": 1893
    },
    {
      "epoch": 0.5245084464137358,
      "grad_norm": 1.5513272285461426,
      "learning_rate": 9.440081988413987e-05,
      "loss": 2.7005,
      "step": 1894
    },
    {
      "epoch": 0.5247853780116312,
      "grad_norm": 1.6557773351669312,
      "learning_rate": 9.431273827186808e-05,
      "loss": 2.6995,
      "step": 1895
    },
    {
      "epoch": 0.5250623096095265,
      "grad_norm": 1.735222578048706,
      "learning_rate": 9.422466108608117e-05,
      "loss": 2.676,
      "step": 1896
    },
    {
      "epoch": 0.5253392412074218,
      "grad_norm": 0.8899129629135132,
      "learning_rate": 9.413658839533099e-05,
      "loss": 2.614,
      "step": 1897
    },
    {
      "epoch": 0.5256161728053171,
      "grad_norm": 2.3472046852111816,
      "learning_rate": 9.40485202681659e-05,
      "loss": 2.9799,
      "step": 1898
    },
    {
      "epoch": 0.5258931044032124,
      "grad_norm": 2.162396192550659,
      "learning_rate": 9.396045677313067e-05,
      "loss": 3.197,
      "step": 1899
    },
    {
      "epoch": 0.5261700360011077,
      "grad_norm": 4.4616169929504395,
      "learning_rate": 9.387239797876657e-05,
      "loss": 3.7858,
      "step": 1900
    },
    {
      "epoch": 0.526446967599003,
      "grad_norm": 1.3809014558792114,
      "learning_rate": 9.378434395361108e-05,
      "loss": 2.8094,
      "step": 1901
    },
    {
      "epoch": 0.5267238991968983,
      "grad_norm": 1.2681607007980347,
      "learning_rate": 9.369629476619801e-05,
      "loss": 2.649,
      "step": 1902
    },
    {
      "epoch": 0.5270008307947937,
      "grad_norm": 1.460354208946228,
      "learning_rate": 9.360825048505749e-05,
      "loss": 2.6172,
      "step": 1903
    },
    {
      "epoch": 0.527277762392689,
      "grad_norm": 1.7847353219985962,
      "learning_rate": 9.352021117871574e-05,
      "loss": 2.7001,
      "step": 1904
    },
    {
      "epoch": 0.5275546939905843,
      "grad_norm": 1.2793521881103516,
      "learning_rate": 9.343217691569506e-05,
      "loss": 2.9796,
      "step": 1905
    },
    {
      "epoch": 0.5278316255884796,
      "grad_norm": 1.5768253803253174,
      "learning_rate": 9.3344147764514e-05,
      "loss": 2.7394,
      "step": 1906
    },
    {
      "epoch": 0.528108557186375,
      "grad_norm": 1.7183265686035156,
      "learning_rate": 9.325612379368695e-05,
      "loss": 2.5647,
      "step": 1907
    },
    {
      "epoch": 0.5283854887842703,
      "grad_norm": 2.033125638961792,
      "learning_rate": 9.316810507172435e-05,
      "loss": 2.7214,
      "step": 1908
    },
    {
      "epoch": 0.5286624203821656,
      "grad_norm": 1.1333674192428589,
      "learning_rate": 9.308009166713263e-05,
      "loss": 2.6175,
      "step": 1909
    },
    {
      "epoch": 0.5289393519800609,
      "grad_norm": 1.6264960765838623,
      "learning_rate": 9.299208364841394e-05,
      "loss": 2.7965,
      "step": 1910
    },
    {
      "epoch": 0.5292162835779562,
      "grad_norm": 1.4833307266235352,
      "learning_rate": 9.290408108406626e-05,
      "loss": 2.9848,
      "step": 1911
    },
    {
      "epoch": 0.5294932151758516,
      "grad_norm": 1.5251394510269165,
      "learning_rate": 9.281608404258347e-05,
      "loss": 2.8359,
      "step": 1912
    },
    {
      "epoch": 0.5297701467737469,
      "grad_norm": 1.435335636138916,
      "learning_rate": 9.272809259245497e-05,
      "loss": 2.7089,
      "step": 1913
    },
    {
      "epoch": 0.5300470783716422,
      "grad_norm": 1.2831498384475708,
      "learning_rate": 9.264010680216583e-05,
      "loss": 2.8127,
      "step": 1914
    },
    {
      "epoch": 0.5303240099695375,
      "grad_norm": 1.6412091255187988,
      "learning_rate": 9.255212674019691e-05,
      "loss": 2.8101,
      "step": 1915
    },
    {
      "epoch": 0.5306009415674329,
      "grad_norm": 1.5583209991455078,
      "learning_rate": 9.246415247502437e-05,
      "loss": 2.671,
      "step": 1916
    },
    {
      "epoch": 0.5308778731653282,
      "grad_norm": 0.9175875782966614,
      "learning_rate": 9.237618407512e-05,
      "loss": 2.8082,
      "step": 1917
    },
    {
      "epoch": 0.5311548047632235,
      "grad_norm": 1.8662338256835938,
      "learning_rate": 9.228822160895095e-05,
      "loss": 2.5191,
      "step": 1918
    },
    {
      "epoch": 0.5314317363611188,
      "grad_norm": 1.1972931623458862,
      "learning_rate": 9.220026514497983e-05,
      "loss": 2.4665,
      "step": 1919
    },
    {
      "epoch": 0.5317086679590142,
      "grad_norm": 1.1261391639709473,
      "learning_rate": 9.211231475166447e-05,
      "loss": 2.6903,
      "step": 1920
    },
    {
      "epoch": 0.5319855995569095,
      "grad_norm": 1.0068024396896362,
      "learning_rate": 9.202437049745812e-05,
      "loss": 2.5293,
      "step": 1921
    },
    {
      "epoch": 0.5322625311548047,
      "grad_norm": 1.2183352708816528,
      "learning_rate": 9.193643245080913e-05,
      "loss": 2.6087,
      "step": 1922
    },
    {
      "epoch": 0.5325394627527,
      "grad_norm": 0.8475114107131958,
      "learning_rate": 9.184850068016099e-05,
      "loss": 2.6615,
      "step": 1923
    },
    {
      "epoch": 0.5328163943505954,
      "grad_norm": 2.806544780731201,
      "learning_rate": 9.176057525395252e-05,
      "loss": 3.0558,
      "step": 1924
    },
    {
      "epoch": 0.5330933259484907,
      "grad_norm": 4.840540885925293,
      "learning_rate": 9.167265624061735e-05,
      "loss": 4.3507,
      "step": 1925
    },
    {
      "epoch": 0.533370257546386,
      "grad_norm": 2.502866744995117,
      "learning_rate": 9.158474370858421e-05,
      "loss": 2.9533,
      "step": 1926
    },
    {
      "epoch": 0.5336471891442813,
      "grad_norm": 1.1832548379898071,
      "learning_rate": 9.149683772627689e-05,
      "loss": 2.6061,
      "step": 1927
    },
    {
      "epoch": 0.5339241207421767,
      "grad_norm": 1.4272520542144775,
      "learning_rate": 9.140893836211393e-05,
      "loss": 2.6801,
      "step": 1928
    },
    {
      "epoch": 0.534201052340072,
      "grad_norm": 2.6324303150177,
      "learning_rate": 9.132104568450879e-05,
      "loss": 2.8803,
      "step": 1929
    },
    {
      "epoch": 0.5344779839379673,
      "grad_norm": 1.6218551397323608,
      "learning_rate": 9.123315976186972e-05,
      "loss": 2.8786,
      "step": 1930
    },
    {
      "epoch": 0.5347549155358626,
      "grad_norm": 2.4477269649505615,
      "learning_rate": 9.114528066259975e-05,
      "loss": 3.1828,
      "step": 1931
    },
    {
      "epoch": 0.535031847133758,
      "grad_norm": 2.003542423248291,
      "learning_rate": 9.105740845509647e-05,
      "loss": 3.0627,
      "step": 1932
    },
    {
      "epoch": 0.5353087787316533,
      "grad_norm": 1.9033154249191284,
      "learning_rate": 9.09695432077523e-05,
      "loss": 2.8415,
      "step": 1933
    },
    {
      "epoch": 0.5355857103295486,
      "grad_norm": 1.7393341064453125,
      "learning_rate": 9.088168498895408e-05,
      "loss": 2.8617,
      "step": 1934
    },
    {
      "epoch": 0.5358626419274439,
      "grad_norm": 2.2522337436676025,
      "learning_rate": 9.07938338670832e-05,
      "loss": 2.8665,
      "step": 1935
    },
    {
      "epoch": 0.5361395735253393,
      "grad_norm": 3.2727909088134766,
      "learning_rate": 9.070598991051565e-05,
      "loss": 3.087,
      "step": 1936
    },
    {
      "epoch": 0.5364165051232346,
      "grad_norm": 1.747339129447937,
      "learning_rate": 9.061815318762172e-05,
      "loss": 2.9148,
      "step": 1937
    },
    {
      "epoch": 0.5366934367211299,
      "grad_norm": 1.5096392631530762,
      "learning_rate": 9.053032376676607e-05,
      "loss": 2.3975,
      "step": 1938
    },
    {
      "epoch": 0.5369703683190252,
      "grad_norm": 1.3254399299621582,
      "learning_rate": 9.044250171630778e-05,
      "loss": 2.593,
      "step": 1939
    },
    {
      "epoch": 0.5372472999169206,
      "grad_norm": 1.1389713287353516,
      "learning_rate": 9.035468710460006e-05,
      "loss": 2.6224,
      "step": 1940
    },
    {
      "epoch": 0.5375242315148159,
      "grad_norm": 1.5871886014938354,
      "learning_rate": 9.026687999999045e-05,
      "loss": 2.8183,
      "step": 1941
    },
    {
      "epoch": 0.5378011631127112,
      "grad_norm": 1.3882852792739868,
      "learning_rate": 9.017908047082059e-05,
      "loss": 2.8155,
      "step": 1942
    },
    {
      "epoch": 0.5380780947106065,
      "grad_norm": 1.0509225130081177,
      "learning_rate": 9.009128858542621e-05,
      "loss": 2.7038,
      "step": 1943
    },
    {
      "epoch": 0.5383550263085018,
      "grad_norm": 1.0775872468948364,
      "learning_rate": 9.000350441213708e-05,
      "loss": 2.4207,
      "step": 1944
    },
    {
      "epoch": 0.5386319579063972,
      "grad_norm": 0.9771863222122192,
      "learning_rate": 8.991572801927709e-05,
      "loss": 2.5425,
      "step": 1945
    },
    {
      "epoch": 0.5389088895042924,
      "grad_norm": 0.9501717686653137,
      "learning_rate": 8.982795947516392e-05,
      "loss": 2.7157,
      "step": 1946
    },
    {
      "epoch": 0.5391858211021877,
      "grad_norm": 2.0657010078430176,
      "learning_rate": 8.974019884810915e-05,
      "loss": 2.4117,
      "step": 1947
    },
    {
      "epoch": 0.539462752700083,
      "grad_norm": 1.415224313735962,
      "learning_rate": 8.965244620641835e-05,
      "loss": 2.9617,
      "step": 1948
    },
    {
      "epoch": 0.5397396842979784,
      "grad_norm": 3.221825361251831,
      "learning_rate": 8.956470161839072e-05,
      "loss": 3.4721,
      "step": 1949
    },
    {
      "epoch": 0.5400166158958737,
      "grad_norm": 4.618506908416748,
      "learning_rate": 8.947696515231925e-05,
      "loss": 4.0782,
      "step": 1950
    },
    {
      "epoch": 0.540293547493769,
      "grad_norm": 2.003098249435425,
      "learning_rate": 8.938923687649063e-05,
      "loss": 2.8448,
      "step": 1951
    },
    {
      "epoch": 0.5405704790916643,
      "grad_norm": 1.5508081912994385,
      "learning_rate": 8.930151685918513e-05,
      "loss": 2.8655,
      "step": 1952
    },
    {
      "epoch": 0.5408474106895597,
      "grad_norm": 2.273536443710327,
      "learning_rate": 8.92138051686766e-05,
      "loss": 2.9843,
      "step": 1953
    },
    {
      "epoch": 0.541124342287455,
      "grad_norm": 2.2980072498321533,
      "learning_rate": 8.912610187323248e-05,
      "loss": 3.1316,
      "step": 1954
    },
    {
      "epoch": 0.5414012738853503,
      "grad_norm": 1.8171645402908325,
      "learning_rate": 8.903840704111357e-05,
      "loss": 2.7456,
      "step": 1955
    },
    {
      "epoch": 0.5416782054832456,
      "grad_norm": 1.8774298429489136,
      "learning_rate": 8.895072074057411e-05,
      "loss": 2.837,
      "step": 1956
    },
    {
      "epoch": 0.541955137081141,
      "grad_norm": 1.7538028955459595,
      "learning_rate": 8.886304303986179e-05,
      "loss": 2.8964,
      "step": 1957
    },
    {
      "epoch": 0.5422320686790363,
      "grad_norm": 1.317739486694336,
      "learning_rate": 8.87753740072175e-05,
      "loss": 2.5798,
      "step": 1958
    },
    {
      "epoch": 0.5425090002769316,
      "grad_norm": 1.340731143951416,
      "learning_rate": 8.868771371087539e-05,
      "loss": 2.7468,
      "step": 1959
    },
    {
      "epoch": 0.5427859318748269,
      "grad_norm": 1.9598355293273926,
      "learning_rate": 8.86000622190629e-05,
      "loss": 2.6228,
      "step": 1960
    },
    {
      "epoch": 0.5430628634727223,
      "grad_norm": 1.5818569660186768,
      "learning_rate": 8.851241960000052e-05,
      "loss": 2.8423,
      "step": 1961
    },
    {
      "epoch": 0.5433397950706176,
      "grad_norm": 1.6524901390075684,
      "learning_rate": 8.842478592190188e-05,
      "loss": 2.8836,
      "step": 1962
    },
    {
      "epoch": 0.5436167266685129,
      "grad_norm": 1.3244292736053467,
      "learning_rate": 8.833716125297364e-05,
      "loss": 2.7687,
      "step": 1963
    },
    {
      "epoch": 0.5438936582664082,
      "grad_norm": 1.4066139459609985,
      "learning_rate": 8.82495456614155e-05,
      "loss": 2.7047,
      "step": 1964
    },
    {
      "epoch": 0.5441705898643036,
      "grad_norm": 0.9185974597930908,
      "learning_rate": 8.816193921541995e-05,
      "loss": 2.8846,
      "step": 1965
    },
    {
      "epoch": 0.5444475214621989,
      "grad_norm": 1.8787202835083008,
      "learning_rate": 8.807434198317257e-05,
      "loss": 2.7159,
      "step": 1966
    },
    {
      "epoch": 0.5447244530600942,
      "grad_norm": 1.6922743320465088,
      "learning_rate": 8.79867540328516e-05,
      "loss": 2.5007,
      "step": 1967
    },
    {
      "epoch": 0.5450013846579895,
      "grad_norm": 1.678337812423706,
      "learning_rate": 8.78991754326281e-05,
      "loss": 2.9345,
      "step": 1968
    },
    {
      "epoch": 0.5452783162558847,
      "grad_norm": 1.3174742460250854,
      "learning_rate": 8.781160625066588e-05,
      "loss": 2.8186,
      "step": 1969
    },
    {
      "epoch": 0.5455552478537801,
      "grad_norm": 1.35786771774292,
      "learning_rate": 8.772404655512145e-05,
      "loss": 2.5213,
      "step": 1970
    },
    {
      "epoch": 0.5458321794516754,
      "grad_norm": 1.3507778644561768,
      "learning_rate": 8.76364964141438e-05,
      "loss": 2.7484,
      "step": 1971
    },
    {
      "epoch": 0.5461091110495707,
      "grad_norm": 1.12258780002594,
      "learning_rate": 8.754895589587466e-05,
      "loss": 2.6031,
      "step": 1972
    },
    {
      "epoch": 0.546386042647466,
      "grad_norm": 1.8437641859054565,
      "learning_rate": 8.746142506844815e-05,
      "loss": 2.9779,
      "step": 1973
    },
    {
      "epoch": 0.5466629742453614,
      "grad_norm": 1.7055060863494873,
      "learning_rate": 8.737390399999086e-05,
      "loss": 2.9552,
      "step": 1974
    },
    {
      "epoch": 0.5469399058432567,
      "grad_norm": 3.4427578449249268,
      "learning_rate": 8.72863927586219e-05,
      "loss": 3.9147,
      "step": 1975
    },
    {
      "epoch": 0.547216837441152,
      "grad_norm": 1.0980651378631592,
      "learning_rate": 8.719889141245256e-05,
      "loss": 2.902,
      "step": 1976
    },
    {
      "epoch": 0.5474937690390473,
      "grad_norm": 1.8571419715881348,
      "learning_rate": 8.711140002958648e-05,
      "loss": 2.6336,
      "step": 1977
    },
    {
      "epoch": 0.5477707006369427,
      "grad_norm": 2.336714506149292,
      "learning_rate": 8.702391867811967e-05,
      "loss": 2.7682,
      "step": 1978
    },
    {
      "epoch": 0.548047632234838,
      "grad_norm": 2.097658157348633,
      "learning_rate": 8.693644742614017e-05,
      "loss": 2.7868,
      "step": 1979
    },
    {
      "epoch": 0.5483245638327333,
      "grad_norm": 1.3417327404022217,
      "learning_rate": 8.684898634172823e-05,
      "loss": 2.6392,
      "step": 1980
    },
    {
      "epoch": 0.5486014954306286,
      "grad_norm": 1.5336050987243652,
      "learning_rate": 8.676153549295622e-05,
      "loss": 2.6745,
      "step": 1981
    },
    {
      "epoch": 0.548878427028524,
      "grad_norm": 1.5942492485046387,
      "learning_rate": 8.667409494788844e-05,
      "loss": 2.6761,
      "step": 1982
    },
    {
      "epoch": 0.5491553586264193,
      "grad_norm": 1.1920937299728394,
      "learning_rate": 8.658666477458128e-05,
      "loss": 2.9014,
      "step": 1983
    },
    {
      "epoch": 0.5494322902243146,
      "grad_norm": 2.115769386291504,
      "learning_rate": 8.649924504108302e-05,
      "loss": 2.8111,
      "step": 1984
    },
    {
      "epoch": 0.5497092218222099,
      "grad_norm": 1.9710432291030884,
      "learning_rate": 8.641183581543381e-05,
      "loss": 2.6635,
      "step": 1985
    },
    {
      "epoch": 0.5499861534201053,
      "grad_norm": 1.1491562128067017,
      "learning_rate": 8.632443716566556e-05,
      "loss": 2.8456,
      "step": 1986
    },
    {
      "epoch": 0.5502630850180006,
      "grad_norm": 1.9223709106445312,
      "learning_rate": 8.623704915980209e-05,
      "loss": 3.2156,
      "step": 1987
    },
    {
      "epoch": 0.5505400166158959,
      "grad_norm": 1.284507155418396,
      "learning_rate": 8.614967186585882e-05,
      "loss": 2.8755,
      "step": 1988
    },
    {
      "epoch": 0.5508169482137912,
      "grad_norm": 1.6829004287719727,
      "learning_rate": 8.606230535184283e-05,
      "loss": 3.087,
      "step": 1989
    },
    {
      "epoch": 0.5510938798116866,
      "grad_norm": 0.9009906053543091,
      "learning_rate": 8.59749496857529e-05,
      "loss": 2.7426,
      "step": 1990
    },
    {
      "epoch": 0.5513708114095819,
      "grad_norm": 1.6485047340393066,
      "learning_rate": 8.588760493557932e-05,
      "loss": 2.6128,
      "step": 1991
    },
    {
      "epoch": 0.5516477430074771,
      "grad_norm": 1.8770582675933838,
      "learning_rate": 8.580027116930378e-05,
      "loss": 3.0946,
      "step": 1992
    },
    {
      "epoch": 0.5519246746053724,
      "grad_norm": 1.648174524307251,
      "learning_rate": 8.571294845489963e-05,
      "loss": 2.6195,
      "step": 1993
    },
    {
      "epoch": 0.5522016062032677,
      "grad_norm": 1.851514458656311,
      "learning_rate": 8.562563686033145e-05,
      "loss": 2.7964,
      "step": 1994
    },
    {
      "epoch": 0.5524785378011631,
      "grad_norm": 1.098483681678772,
      "learning_rate": 8.553833645355524e-05,
      "loss": 2.727,
      "step": 1995
    },
    {
      "epoch": 0.5527554693990584,
      "grad_norm": 2.0957040786743164,
      "learning_rate": 8.545104730251824e-05,
      "loss": 2.4808,
      "step": 1996
    },
    {
      "epoch": 0.5530324009969537,
      "grad_norm": 1.257961392402649,
      "learning_rate": 8.536376947515905e-05,
      "loss": 2.4464,
      "step": 1997
    },
    {
      "epoch": 0.553309332594849,
      "grad_norm": 1.606569766998291,
      "learning_rate": 8.527650303940725e-05,
      "loss": 2.5888,
      "step": 1998
    },
    {
      "epoch": 0.5535862641927444,
      "grad_norm": 2.9079225063323975,
      "learning_rate": 8.518924806318378e-05,
      "loss": 3.1359,
      "step": 1999
    },
    {
      "epoch": 0.5538631957906397,
      "grad_norm": 3.6829776763916016,
      "learning_rate": 8.510200461440051e-05,
      "loss": 3.8721,
      "step": 2000
    },
    {
      "epoch": 0.554140127388535,
      "grad_norm": 1.0013054609298706,
      "learning_rate": 8.501477276096033e-05,
      "loss": 2.6452,
      "step": 2001
    },
    {
      "epoch": 0.5544170589864303,
      "grad_norm": 1.6629724502563477,
      "learning_rate": 8.492755257075728e-05,
      "loss": 2.8157,
      "step": 2002
    },
    {
      "epoch": 0.5546939905843257,
      "grad_norm": 2.2238128185272217,
      "learning_rate": 8.48403441116761e-05,
      "loss": 3.124,
      "step": 2003
    },
    {
      "epoch": 0.554970922182221,
      "grad_norm": 1.7178497314453125,
      "learning_rate": 8.47531474515925e-05,
      "loss": 2.6766,
      "step": 2004
    },
    {
      "epoch": 0.5552478537801163,
      "grad_norm": 2.264681816101074,
      "learning_rate": 8.466596265837305e-05,
      "loss": 2.7763,
      "step": 2005
    },
    {
      "epoch": 0.5555247853780116,
      "grad_norm": 1.8057752847671509,
      "learning_rate": 8.457878979987507e-05,
      "loss": 2.9713,
      "step": 2006
    },
    {
      "epoch": 0.555801716975907,
      "grad_norm": 1.1006498336791992,
      "learning_rate": 8.449162894394646e-05,
      "loss": 2.6146,
      "step": 2007
    },
    {
      "epoch": 0.5560786485738023,
      "grad_norm": 1.485538363456726,
      "learning_rate": 8.440448015842597e-05,
      "loss": 2.6359,
      "step": 2008
    },
    {
      "epoch": 0.5563555801716976,
      "grad_norm": 1.3723620176315308,
      "learning_rate": 8.431734351114284e-05,
      "loss": 2.6276,
      "step": 2009
    },
    {
      "epoch": 0.5566325117695929,
      "grad_norm": 1.5234686136245728,
      "learning_rate": 8.423021906991684e-05,
      "loss": 3.0212,
      "step": 2010
    },
    {
      "epoch": 0.5569094433674883,
      "grad_norm": 2.5109102725982666,
      "learning_rate": 8.414310690255835e-05,
      "loss": 3.2222,
      "step": 2011
    },
    {
      "epoch": 0.5571863749653836,
      "grad_norm": 1.1213618516921997,
      "learning_rate": 8.405600707686814e-05,
      "loss": 2.6941,
      "step": 2012
    },
    {
      "epoch": 0.5574633065632789,
      "grad_norm": 1.080403208732605,
      "learning_rate": 8.396891966063727e-05,
      "loss": 2.7642,
      "step": 2013
    },
    {
      "epoch": 0.5577402381611742,
      "grad_norm": 1.604928970336914,
      "learning_rate": 8.388184472164736e-05,
      "loss": 2.6035,
      "step": 2014
    },
    {
      "epoch": 0.5580171697590696,
      "grad_norm": 1.7442255020141602,
      "learning_rate": 8.379478232767013e-05,
      "loss": 2.7144,
      "step": 2015
    },
    {
      "epoch": 0.5582941013569648,
      "grad_norm": 1.8785520792007446,
      "learning_rate": 8.370773254646762e-05,
      "loss": 2.7974,
      "step": 2016
    },
    {
      "epoch": 0.5585710329548601,
      "grad_norm": 1.1109877824783325,
      "learning_rate": 8.362069544579205e-05,
      "loss": 2.5566,
      "step": 2017
    },
    {
      "epoch": 0.5588479645527554,
      "grad_norm": 1.8859199285507202,
      "learning_rate": 8.353367109338575e-05,
      "loss": 2.542,
      "step": 2018
    },
    {
      "epoch": 0.5591248961506508,
      "grad_norm": 1.1205683946609497,
      "learning_rate": 8.34466595569811e-05,
      "loss": 2.8586,
      "step": 2019
    },
    {
      "epoch": 0.5594018277485461,
      "grad_norm": 1.04397714138031,
      "learning_rate": 8.335966090430062e-05,
      "loss": 2.6522,
      "step": 2020
    },
    {
      "epoch": 0.5596787593464414,
      "grad_norm": 1.1475187540054321,
      "learning_rate": 8.327267520305669e-05,
      "loss": 2.3847,
      "step": 2021
    },
    {
      "epoch": 0.5599556909443367,
      "grad_norm": 1.7953752279281616,
      "learning_rate": 8.318570252095157e-05,
      "loss": 2.8365,
      "step": 2022
    },
    {
      "epoch": 0.560232622542232,
      "grad_norm": 1.6694520711898804,
      "learning_rate": 8.309874292567758e-05,
      "loss": 3.0579,
      "step": 2023
    },
    {
      "epoch": 0.5605095541401274,
      "grad_norm": 2.378661632537842,
      "learning_rate": 8.301179648491669e-05,
      "loss": 3.0315,
      "step": 2024
    },
    {
      "epoch": 0.5607864857380227,
      "grad_norm": 3.761888027191162,
      "learning_rate": 8.292486326634061e-05,
      "loss": 4.1084,
      "step": 2025
    },
    {
      "epoch": 0.561063417335918,
      "grad_norm": 1.8311100006103516,
      "learning_rate": 8.283794333761093e-05,
      "loss": 2.8907,
      "step": 2026
    },
    {
      "epoch": 0.5613403489338133,
      "grad_norm": 2.8452889919281006,
      "learning_rate": 8.275103676637871e-05,
      "loss": 3.1021,
      "step": 2027
    },
    {
      "epoch": 0.5616172805317087,
      "grad_norm": 1.8134194612503052,
      "learning_rate": 8.266414362028475e-05,
      "loss": 2.8781,
      "step": 2028
    },
    {
      "epoch": 0.561894212129604,
      "grad_norm": 1.291551947593689,
      "learning_rate": 8.257726396695933e-05,
      "loss": 2.9156,
      "step": 2029
    },
    {
      "epoch": 0.5621711437274993,
      "grad_norm": 1.6090056896209717,
      "learning_rate": 8.249039787402224e-05,
      "loss": 2.832,
      "step": 2030
    },
    {
      "epoch": 0.5624480753253946,
      "grad_norm": 1.1308784484863281,
      "learning_rate": 8.240354540908269e-05,
      "loss": 2.6733,
      "step": 2031
    },
    {
      "epoch": 0.56272500692329,
      "grad_norm": 1.1666631698608398,
      "learning_rate": 8.231670663973939e-05,
      "loss": 2.6943,
      "step": 2032
    },
    {
      "epoch": 0.5630019385211853,
      "grad_norm": 1.3602391481399536,
      "learning_rate": 8.222988163358027e-05,
      "loss": 2.7505,
      "step": 2033
    },
    {
      "epoch": 0.5632788701190806,
      "grad_norm": 1.7966910600662231,
      "learning_rate": 8.214307045818254e-05,
      "loss": 2.8815,
      "step": 2034
    },
    {
      "epoch": 0.563555801716976,
      "grad_norm": 2.1865363121032715,
      "learning_rate": 8.20562731811128e-05,
      "loss": 2.7694,
      "step": 2035
    },
    {
      "epoch": 0.5638327333148713,
      "grad_norm": 2.047175645828247,
      "learning_rate": 8.196948986992666e-05,
      "loss": 2.9757,
      "step": 2036
    },
    {
      "epoch": 0.5641096649127666,
      "grad_norm": 1.7177631855010986,
      "learning_rate": 8.188272059216894e-05,
      "loss": 3.0657,
      "step": 2037
    },
    {
      "epoch": 0.5643865965106619,
      "grad_norm": 2.061433792114258,
      "learning_rate": 8.179596541537355e-05,
      "loss": 3.3796,
      "step": 2038
    },
    {
      "epoch": 0.5646635281085571,
      "grad_norm": 1.3437374830245972,
      "learning_rate": 8.17092244070634e-05,
      "loss": 3.1446,
      "step": 2039
    },
    {
      "epoch": 0.5649404597064525,
      "grad_norm": 1.3183518648147583,
      "learning_rate": 8.162249763475034e-05,
      "loss": 2.9589,
      "step": 2040
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 1.7386242151260376,
      "learning_rate": 8.153578516593524e-05,
      "loss": 2.8237,
      "step": 2041
    },
    {
      "epoch": 0.5654943229022431,
      "grad_norm": 1.5994588136672974,
      "learning_rate": 8.144908706810773e-05,
      "loss": 2.8691,
      "step": 2042
    },
    {
      "epoch": 0.5657712545001384,
      "grad_norm": 1.195237636566162,
      "learning_rate": 8.136240340874624e-05,
      "loss": 2.7326,
      "step": 2043
    },
    {
      "epoch": 0.5660481860980338,
      "grad_norm": 2.3551249504089355,
      "learning_rate": 8.127573425531814e-05,
      "loss": 3.0542,
      "step": 2044
    },
    {
      "epoch": 0.5663251176959291,
      "grad_norm": 1.8700072765350342,
      "learning_rate": 8.118907967527934e-05,
      "loss": 2.5069,
      "step": 2045
    },
    {
      "epoch": 0.5666020492938244,
      "grad_norm": 1.3926947116851807,
      "learning_rate": 8.110243973607439e-05,
      "loss": 2.6471,
      "step": 2046
    },
    {
      "epoch": 0.5668789808917197,
      "grad_norm": 2.0446665287017822,
      "learning_rate": 8.10158145051366e-05,
      "loss": 3.096,
      "step": 2047
    },
    {
      "epoch": 0.567155912489615,
      "grad_norm": 2.3419084548950195,
      "learning_rate": 8.09292040498877e-05,
      "loss": 2.8887,
      "step": 2048
    },
    {
      "epoch": 0.5674328440875104,
      "grad_norm": 1.5965373516082764,
      "learning_rate": 8.084260843773799e-05,
      "loss": 2.7957,
      "step": 2049
    },
    {
      "epoch": 0.5677097756854057,
      "grad_norm": 2.7372281551361084,
      "learning_rate": 8.075602773608617e-05,
      "loss": 3.4179,
      "step": 2050
    },
    {
      "epoch": 0.567986707283301,
      "grad_norm": 1.4773602485656738,
      "learning_rate": 8.06694620123194e-05,
      "loss": 2.5247,
      "step": 2051
    },
    {
      "epoch": 0.5682636388811964,
      "grad_norm": 1.3511680364608765,
      "learning_rate": 8.058291133381304e-05,
      "loss": 2.929,
      "step": 2052
    },
    {
      "epoch": 0.5685405704790917,
      "grad_norm": 1.5803459882736206,
      "learning_rate": 8.049637576793094e-05,
      "loss": 2.8925,
      "step": 2053
    },
    {
      "epoch": 0.568817502076987,
      "grad_norm": 1.283042311668396,
      "learning_rate": 8.040985538202505e-05,
      "loss": 2.8373,
      "step": 2054
    },
    {
      "epoch": 0.5690944336748823,
      "grad_norm": 1.295939326286316,
      "learning_rate": 8.032335024343551e-05,
      "loss": 2.8411,
      "step": 2055
    },
    {
      "epoch": 0.5693713652727777,
      "grad_norm": 1.0056647062301636,
      "learning_rate": 8.02368604194907e-05,
      "loss": 2.8974,
      "step": 2056
    },
    {
      "epoch": 0.569648296870673,
      "grad_norm": 1.825939416885376,
      "learning_rate": 8.015038597750694e-05,
      "loss": 2.735,
      "step": 2057
    },
    {
      "epoch": 0.5699252284685683,
      "grad_norm": 1.263617992401123,
      "learning_rate": 8.006392698478862e-05,
      "loss": 2.811,
      "step": 2058
    },
    {
      "epoch": 0.5702021600664636,
      "grad_norm": 1.8211050033569336,
      "learning_rate": 7.997748350862822e-05,
      "loss": 3.0062,
      "step": 2059
    },
    {
      "epoch": 0.570479091664359,
      "grad_norm": 1.1094549894332886,
      "learning_rate": 7.989105561630598e-05,
      "loss": 2.9058,
      "step": 2060
    },
    {
      "epoch": 0.5707560232622543,
      "grad_norm": 1.377359390258789,
      "learning_rate": 7.980464337509006e-05,
      "loss": 2.6647,
      "step": 2061
    },
    {
      "epoch": 0.5710329548601495,
      "grad_norm": 1.1790040731430054,
      "learning_rate": 7.971824685223657e-05,
      "loss": 2.7497,
      "step": 2062
    },
    {
      "epoch": 0.5713098864580448,
      "grad_norm": 1.1479779481887817,
      "learning_rate": 7.96318661149892e-05,
      "loss": 2.726,
      "step": 2063
    },
    {
      "epoch": 0.5715868180559401,
      "grad_norm": 1.4475488662719727,
      "learning_rate": 7.954550123057939e-05,
      "loss": 2.7639,
      "step": 2064
    },
    {
      "epoch": 0.5718637496538355,
      "grad_norm": 1.5319483280181885,
      "learning_rate": 7.945915226622639e-05,
      "loss": 2.8743,
      "step": 2065
    },
    {
      "epoch": 0.5721406812517308,
      "grad_norm": 1.368962287902832,
      "learning_rate": 7.937281928913688e-05,
      "loss": 2.9782,
      "step": 2066
    },
    {
      "epoch": 0.5724176128496261,
      "grad_norm": 1.8736423254013062,
      "learning_rate": 7.928650236650514e-05,
      "loss": 2.857,
      "step": 2067
    },
    {
      "epoch": 0.5726945444475214,
      "grad_norm": 1.5408180952072144,
      "learning_rate": 7.920020156551307e-05,
      "loss": 2.5916,
      "step": 2068
    },
    {
      "epoch": 0.5729714760454168,
      "grad_norm": 1.3159997463226318,
      "learning_rate": 7.911391695332988e-05,
      "loss": 2.6835,
      "step": 2069
    },
    {
      "epoch": 0.5732484076433121,
      "grad_norm": 1.2888193130493164,
      "learning_rate": 7.902764859711223e-05,
      "loss": 2.6142,
      "step": 2070
    },
    {
      "epoch": 0.5735253392412074,
      "grad_norm": 1.3029569387435913,
      "learning_rate": 7.894139656400417e-05,
      "loss": 2.581,
      "step": 2071
    },
    {
      "epoch": 0.5738022708391027,
      "grad_norm": 1.5423462390899658,
      "learning_rate": 7.885516092113699e-05,
      "loss": 2.7403,
      "step": 2072
    },
    {
      "epoch": 0.5740792024369981,
      "grad_norm": 2.0652363300323486,
      "learning_rate": 7.876894173562919e-05,
      "loss": 3.1884,
      "step": 2073
    },
    {
      "epoch": 0.5743561340348934,
      "grad_norm": 2.0919711589813232,
      "learning_rate": 7.868273907458661e-05,
      "loss": 2.9806,
      "step": 2074
    },
    {
      "epoch": 0.5746330656327887,
      "grad_norm": 14.729402542114258,
      "learning_rate": 7.859655300510208e-05,
      "loss": 3.9938,
      "step": 2075
    },
    {
      "epoch": 0.574909997230684,
      "grad_norm": 2.284801721572876,
      "learning_rate": 7.851038359425553e-05,
      "loss": 3.2097,
      "step": 2076
    },
    {
      "epoch": 0.5751869288285794,
      "grad_norm": 1.1180181503295898,
      "learning_rate": 7.842423090911403e-05,
      "loss": 2.8244,
      "step": 2077
    },
    {
      "epoch": 0.5754638604264747,
      "grad_norm": 1.5196349620819092,
      "learning_rate": 7.833809501673155e-05,
      "loss": 2.8955,
      "step": 2078
    },
    {
      "epoch": 0.57574079202437,
      "grad_norm": 1.8050135374069214,
      "learning_rate": 7.825197598414895e-05,
      "loss": 2.8498,
      "step": 2079
    },
    {
      "epoch": 0.5760177236222653,
      "grad_norm": 1.4506398439407349,
      "learning_rate": 7.81658738783941e-05,
      "loss": 2.6669,
      "step": 2080
    },
    {
      "epoch": 0.5762946552201607,
      "grad_norm": 0.9477134943008423,
      "learning_rate": 7.807978876648154e-05,
      "loss": 2.6933,
      "step": 2081
    },
    {
      "epoch": 0.576571586818056,
      "grad_norm": 2.4606692790985107,
      "learning_rate": 7.799372071541272e-05,
      "loss": 2.8029,
      "step": 2082
    },
    {
      "epoch": 0.5768485184159513,
      "grad_norm": 1.804126262664795,
      "learning_rate": 7.790766979217571e-05,
      "loss": 2.7009,
      "step": 2083
    },
    {
      "epoch": 0.5771254500138466,
      "grad_norm": 2.0358893871307373,
      "learning_rate": 7.782163606374536e-05,
      "loss": 2.7267,
      "step": 2084
    },
    {
      "epoch": 0.577402381611742,
      "grad_norm": 1.8686617612838745,
      "learning_rate": 7.773561959708298e-05,
      "loss": 2.7292,
      "step": 2085
    },
    {
      "epoch": 0.5776793132096372,
      "grad_norm": 1.3002493381500244,
      "learning_rate": 7.76496204591366e-05,
      "loss": 3.0942,
      "step": 2086
    },
    {
      "epoch": 0.5779562448075325,
      "grad_norm": 1.4342671632766724,
      "learning_rate": 7.756363871684066e-05,
      "loss": 2.9262,
      "step": 2087
    },
    {
      "epoch": 0.5782331764054278,
      "grad_norm": 1.3472551107406616,
      "learning_rate": 7.747767443711609e-05,
      "loss": 2.9602,
      "step": 2088
    },
    {
      "epoch": 0.5785101080033231,
      "grad_norm": 1.9964295625686646,
      "learning_rate": 7.739172768687028e-05,
      "loss": 3.0501,
      "step": 2089
    },
    {
      "epoch": 0.5787870396012185,
      "grad_norm": 1.5070070028305054,
      "learning_rate": 7.730579853299691e-05,
      "loss": 2.8507,
      "step": 2090
    },
    {
      "epoch": 0.5790639711991138,
      "grad_norm": 1.8707159757614136,
      "learning_rate": 7.721988704237593e-05,
      "loss": 2.7029,
      "step": 2091
    },
    {
      "epoch": 0.5793409027970091,
      "grad_norm": 1.4172472953796387,
      "learning_rate": 7.713399328187371e-05,
      "loss": 3.0517,
      "step": 2092
    },
    {
      "epoch": 0.5796178343949044,
      "grad_norm": 1.8085438013076782,
      "learning_rate": 7.70481173183426e-05,
      "loss": 2.5827,
      "step": 2093
    },
    {
      "epoch": 0.5798947659927998,
      "grad_norm": 1.3533729314804077,
      "learning_rate": 7.696225921862126e-05,
      "loss": 2.5971,
      "step": 2094
    },
    {
      "epoch": 0.5801716975906951,
      "grad_norm": 1.7932556867599487,
      "learning_rate": 7.687641904953443e-05,
      "loss": 2.6052,
      "step": 2095
    },
    {
      "epoch": 0.5804486291885904,
      "grad_norm": 1.3073060512542725,
      "learning_rate": 7.67905968778928e-05,
      "loss": 2.7844,
      "step": 2096
    },
    {
      "epoch": 0.5807255607864857,
      "grad_norm": 1.4766454696655273,
      "learning_rate": 7.67047927704931e-05,
      "loss": 2.774,
      "step": 2097
    },
    {
      "epoch": 0.5810024923843811,
      "grad_norm": 1.1093850135803223,
      "learning_rate": 7.661900679411806e-05,
      "loss": 2.4939,
      "step": 2098
    },
    {
      "epoch": 0.5812794239822764,
      "grad_norm": 2.555997371673584,
      "learning_rate": 7.653323901553625e-05,
      "loss": 2.9841,
      "step": 2099
    },
    {
      "epoch": 0.5815563555801717,
      "grad_norm": 4.845314979553223,
      "learning_rate": 7.6447489501502e-05,
      "loss": 4.4183,
      "step": 2100
    },
    {
      "epoch": 0.581833287178067,
      "grad_norm": 3.8324882984161377,
      "learning_rate": 7.63617583187556e-05,
      "loss": 3.7171,
      "step": 2101
    },
    {
      "epoch": 0.5821102187759624,
      "grad_norm": 1.3293406963348389,
      "learning_rate": 7.627604553402291e-05,
      "loss": 2.6822,
      "step": 2102
    },
    {
      "epoch": 0.5823871503738577,
      "grad_norm": 1.512557029724121,
      "learning_rate": 7.619035121401554e-05,
      "loss": 2.8265,
      "step": 2103
    },
    {
      "epoch": 0.582664081971753,
      "grad_norm": 1.215285062789917,
      "learning_rate": 7.610467542543073e-05,
      "loss": 2.6078,
      "step": 2104
    },
    {
      "epoch": 0.5829410135696483,
      "grad_norm": 0.7829873561859131,
      "learning_rate": 7.60190182349513e-05,
      "loss": 2.6538,
      "step": 2105
    },
    {
      "epoch": 0.5832179451675437,
      "grad_norm": 2.242455005645752,
      "learning_rate": 7.593337970924554e-05,
      "loss": 2.7853,
      "step": 2106
    },
    {
      "epoch": 0.583494876765439,
      "grad_norm": 1.0244053602218628,
      "learning_rate": 7.584775991496732e-05,
      "loss": 2.7967,
      "step": 2107
    },
    {
      "epoch": 0.5837718083633343,
      "grad_norm": 2.7727086544036865,
      "learning_rate": 7.576215891875584e-05,
      "loss": 2.9253,
      "step": 2108
    },
    {
      "epoch": 0.5840487399612295,
      "grad_norm": 1.8041883707046509,
      "learning_rate": 7.567657678723565e-05,
      "loss": 2.6955,
      "step": 2109
    },
    {
      "epoch": 0.5843256715591248,
      "grad_norm": 2.026834726333618,
      "learning_rate": 7.559101358701673e-05,
      "loss": 2.5201,
      "step": 2110
    },
    {
      "epoch": 0.5846026031570202,
      "grad_norm": 1.2291382551193237,
      "learning_rate": 7.550546938469424e-05,
      "loss": 2.71,
      "step": 2111
    },
    {
      "epoch": 0.5848795347549155,
      "grad_norm": 1.3409702777862549,
      "learning_rate": 7.541994424684852e-05,
      "loss": 2.7428,
      "step": 2112
    },
    {
      "epoch": 0.5851564663528108,
      "grad_norm": 1.8270671367645264,
      "learning_rate": 7.53344382400452e-05,
      "loss": 2.8873,
      "step": 2113
    },
    {
      "epoch": 0.5854333979507061,
      "grad_norm": 2.524343729019165,
      "learning_rate": 7.52489514308349e-05,
      "loss": 3.2532,
      "step": 2114
    },
    {
      "epoch": 0.5857103295486015,
      "grad_norm": 1.66667902469635,
      "learning_rate": 7.516348388575336e-05,
      "loss": 2.6943,
      "step": 2115
    },
    {
      "epoch": 0.5859872611464968,
      "grad_norm": 2.1451516151428223,
      "learning_rate": 7.50780356713213e-05,
      "loss": 2.6747,
      "step": 2116
    },
    {
      "epoch": 0.5862641927443921,
      "grad_norm": 1.5665041208267212,
      "learning_rate": 7.499260685404442e-05,
      "loss": 2.6168,
      "step": 2117
    },
    {
      "epoch": 0.5865411243422874,
      "grad_norm": 1.6107885837554932,
      "learning_rate": 7.490719750041326e-05,
      "loss": 2.7984,
      "step": 2118
    },
    {
      "epoch": 0.5868180559401828,
      "grad_norm": 1.0318968296051025,
      "learning_rate": 7.482180767690334e-05,
      "loss": 2.6345,
      "step": 2119
    },
    {
      "epoch": 0.5870949875380781,
      "grad_norm": 1.556569218635559,
      "learning_rate": 7.473643744997483e-05,
      "loss": 2.885,
      "step": 2120
    },
    {
      "epoch": 0.5873719191359734,
      "grad_norm": 1.727760910987854,
      "learning_rate": 7.46510868860727e-05,
      "loss": 2.608,
      "step": 2121
    },
    {
      "epoch": 0.5876488507338687,
      "grad_norm": 1.6430820226669312,
      "learning_rate": 7.456575605162672e-05,
      "loss": 2.635,
      "step": 2122
    },
    {
      "epoch": 0.5879257823317641,
      "grad_norm": 1.2055941820144653,
      "learning_rate": 7.448044501305114e-05,
      "loss": 2.7858,
      "step": 2123
    },
    {
      "epoch": 0.5882027139296594,
      "grad_norm": 1.0556684732437134,
      "learning_rate": 7.439515383674485e-05,
      "loss": 2.8217,
      "step": 2124
    },
    {
      "epoch": 0.5884796455275547,
      "grad_norm": 3.524182081222534,
      "learning_rate": 7.43098825890914e-05,
      "loss": 3.5602,
      "step": 2125
    },
    {
      "epoch": 0.58875657712545,
      "grad_norm": 1.06099271774292,
      "learning_rate": 7.42246313364587e-05,
      "loss": 2.6188,
      "step": 2126
    },
    {
      "epoch": 0.5890335087233454,
      "grad_norm": 1.5477778911590576,
      "learning_rate": 7.413940014519907e-05,
      "loss": 2.6959,
      "step": 2127
    },
    {
      "epoch": 0.5893104403212407,
      "grad_norm": 1.5380752086639404,
      "learning_rate": 7.405418908164939e-05,
      "loss": 2.7056,
      "step": 2128
    },
    {
      "epoch": 0.589587371919136,
      "grad_norm": 1.5953619480133057,
      "learning_rate": 7.396899821213072e-05,
      "loss": 2.6368,
      "step": 2129
    },
    {
      "epoch": 0.5898643035170313,
      "grad_norm": 1.9024546146392822,
      "learning_rate": 7.38838276029484e-05,
      "loss": 2.8164,
      "step": 2130
    },
    {
      "epoch": 0.5901412351149267,
      "grad_norm": 1.7428923845291138,
      "learning_rate": 7.379867732039213e-05,
      "loss": 2.7139,
      "step": 2131
    },
    {
      "epoch": 0.5904181667128219,
      "grad_norm": 1.7151989936828613,
      "learning_rate": 7.371354743073566e-05,
      "loss": 2.6295,
      "step": 2132
    },
    {
      "epoch": 0.5906950983107172,
      "grad_norm": 1.5807915925979614,
      "learning_rate": 7.362843800023689e-05,
      "loss": 2.7809,
      "step": 2133
    },
    {
      "epoch": 0.5909720299086125,
      "grad_norm": 1.5456976890563965,
      "learning_rate": 7.354334909513791e-05,
      "loss": 2.6879,
      "step": 2134
    },
    {
      "epoch": 0.5912489615065079,
      "grad_norm": 1.4076478481292725,
      "learning_rate": 7.345828078166466e-05,
      "loss": 2.6555,
      "step": 2135
    },
    {
      "epoch": 0.5915258931044032,
      "grad_norm": 1.5420868396759033,
      "learning_rate": 7.337323312602718e-05,
      "loss": 2.6094,
      "step": 2136
    },
    {
      "epoch": 0.5918028247022985,
      "grad_norm": 1.3896031379699707,
      "learning_rate": 7.32882061944194e-05,
      "loss": 2.9571,
      "step": 2137
    },
    {
      "epoch": 0.5920797563001938,
      "grad_norm": 1.683767318725586,
      "learning_rate": 7.320320005301911e-05,
      "loss": 3.0786,
      "step": 2138
    },
    {
      "epoch": 0.5923566878980892,
      "grad_norm": 1.792648196220398,
      "learning_rate": 7.311821476798789e-05,
      "loss": 2.8892,
      "step": 2139
    },
    {
      "epoch": 0.5926336194959845,
      "grad_norm": 2.0883829593658447,
      "learning_rate": 7.303325040547116e-05,
      "loss": 3.319,
      "step": 2140
    },
    {
      "epoch": 0.5929105510938798,
      "grad_norm": 1.1606462001800537,
      "learning_rate": 7.2948307031598e-05,
      "loss": 2.8071,
      "step": 2141
    },
    {
      "epoch": 0.5931874826917751,
      "grad_norm": 1.9280019998550415,
      "learning_rate": 7.286338471248113e-05,
      "loss": 2.7506,
      "step": 2142
    },
    {
      "epoch": 0.5934644142896704,
      "grad_norm": 0.847400426864624,
      "learning_rate": 7.277848351421699e-05,
      "loss": 2.8121,
      "step": 2143
    },
    {
      "epoch": 0.5937413458875658,
      "grad_norm": 1.5167869329452515,
      "learning_rate": 7.269360350288547e-05,
      "loss": 2.6086,
      "step": 2144
    },
    {
      "epoch": 0.5940182774854611,
      "grad_norm": 1.0383349657058716,
      "learning_rate": 7.260874474455e-05,
      "loss": 2.715,
      "step": 2145
    },
    {
      "epoch": 0.5942952090833564,
      "grad_norm": 1.9108144044876099,
      "learning_rate": 7.25239073052575e-05,
      "loss": 2.9016,
      "step": 2146
    },
    {
      "epoch": 0.5945721406812517,
      "grad_norm": 1.9010608196258545,
      "learning_rate": 7.243909125103829e-05,
      "loss": 2.6386,
      "step": 2147
    },
    {
      "epoch": 0.5948490722791471,
      "grad_norm": 1.4881986379623413,
      "learning_rate": 7.235429664790603e-05,
      "loss": 2.7531,
      "step": 2148
    },
    {
      "epoch": 0.5951260038770424,
      "grad_norm": 1.6744920015335083,
      "learning_rate": 7.226952356185765e-05,
      "loss": 3.2777,
      "step": 2149
    },
    {
      "epoch": 0.5954029354749377,
      "grad_norm": 3.5773537158966064,
      "learning_rate": 7.218477205887344e-05,
      "loss": 3.7091,
      "step": 2150
    },
    {
      "epoch": 0.595679867072833,
      "grad_norm": 1.0836102962493896,
      "learning_rate": 7.210004220491673e-05,
      "loss": 2.7561,
      "step": 2151
    },
    {
      "epoch": 0.5959567986707284,
      "grad_norm": 1.4775099754333496,
      "learning_rate": 7.20153340659342e-05,
      "loss": 2.831,
      "step": 2152
    },
    {
      "epoch": 0.5962337302686237,
      "grad_norm": 1.6922560930252075,
      "learning_rate": 7.193064770785545e-05,
      "loss": 2.6291,
      "step": 2153
    },
    {
      "epoch": 0.596510661866519,
      "grad_norm": 1.7395472526550293,
      "learning_rate": 7.184598319659317e-05,
      "loss": 2.9307,
      "step": 2154
    },
    {
      "epoch": 0.5967875934644142,
      "grad_norm": 1.068514108657837,
      "learning_rate": 7.176134059804316e-05,
      "loss": 2.8556,
      "step": 2155
    },
    {
      "epoch": 0.5970645250623096,
      "grad_norm": 1.519996166229248,
      "learning_rate": 7.167671997808405e-05,
      "loss": 2.6497,
      "step": 2156
    },
    {
      "epoch": 0.5973414566602049,
      "grad_norm": 1.3461872339248657,
      "learning_rate": 7.159212140257734e-05,
      "loss": 2.6785,
      "step": 2157
    },
    {
      "epoch": 0.5976183882581002,
      "grad_norm": 1.668176293373108,
      "learning_rate": 7.150754493736749e-05,
      "loss": 2.9134,
      "step": 2158
    },
    {
      "epoch": 0.5978953198559955,
      "grad_norm": 1.2623629570007324,
      "learning_rate": 7.142299064828169e-05,
      "loss": 2.7,
      "step": 2159
    },
    {
      "epoch": 0.5981722514538909,
      "grad_norm": 1.7657973766326904,
      "learning_rate": 7.133845860112978e-05,
      "loss": 2.9739,
      "step": 2160
    },
    {
      "epoch": 0.5984491830517862,
      "grad_norm": 1.0858269929885864,
      "learning_rate": 7.12539488617045e-05,
      "loss": 2.5533,
      "step": 2161
    },
    {
      "epoch": 0.5987261146496815,
      "grad_norm": 1.3624961376190186,
      "learning_rate": 7.116946149578105e-05,
      "loss": 2.7661,
      "step": 2162
    },
    {
      "epoch": 0.5990030462475768,
      "grad_norm": 1.414823055267334,
      "learning_rate": 7.108499656911721e-05,
      "loss": 2.6578,
      "step": 2163
    },
    {
      "epoch": 0.5992799778454722,
      "grad_norm": 2.4731411933898926,
      "learning_rate": 7.100055414745346e-05,
      "loss": 3.1674,
      "step": 2164
    },
    {
      "epoch": 0.5995569094433675,
      "grad_norm": 1.1124169826507568,
      "learning_rate": 7.09161342965126e-05,
      "loss": 2.8148,
      "step": 2165
    },
    {
      "epoch": 0.5998338410412628,
      "grad_norm": 2.0510025024414062,
      "learning_rate": 7.08317370819999e-05,
      "loss": 3.0599,
      "step": 2166
    },
    {
      "epoch": 0.6001107726391581,
      "grad_norm": 1.8278013467788696,
      "learning_rate": 7.07473625696031e-05,
      "loss": 2.9529,
      "step": 2167
    },
    {
      "epoch": 0.6003877042370535,
      "grad_norm": 2.326231002807617,
      "learning_rate": 7.066301082499216e-05,
      "loss": 2.791,
      "step": 2168
    },
    {
      "epoch": 0.6006646358349488,
      "grad_norm": 1.308354139328003,
      "learning_rate": 7.057868191381936e-05,
      "loss": 2.6107,
      "step": 2169
    },
    {
      "epoch": 0.6009415674328441,
      "grad_norm": 1.7721244096755981,
      "learning_rate": 7.049437590171924e-05,
      "loss": 2.5734,
      "step": 2170
    },
    {
      "epoch": 0.6012184990307394,
      "grad_norm": 1.9452602863311768,
      "learning_rate": 7.041009285430848e-05,
      "loss": 2.8181,
      "step": 2171
    },
    {
      "epoch": 0.6014954306286348,
      "grad_norm": 1.4054052829742432,
      "learning_rate": 7.032583283718585e-05,
      "loss": 2.7896,
      "step": 2172
    },
    {
      "epoch": 0.6017723622265301,
      "grad_norm": 1.540339708328247,
      "learning_rate": 7.024159591593233e-05,
      "loss": 2.7709,
      "step": 2173
    },
    {
      "epoch": 0.6020492938244254,
      "grad_norm": 1.711843490600586,
      "learning_rate": 7.015738215611079e-05,
      "loss": 3.3467,
      "step": 2174
    },
    {
      "epoch": 0.6023262254223207,
      "grad_norm": 2.023811101913452,
      "learning_rate": 7.007319162326607e-05,
      "loss": 3.4878,
      "step": 2175
    },
    {
      "epoch": 0.602603157020216,
      "grad_norm": 0.9673023223876953,
      "learning_rate": 6.998902438292508e-05,
      "loss": 2.6065,
      "step": 2176
    },
    {
      "epoch": 0.6028800886181114,
      "grad_norm": 1.6274542808532715,
      "learning_rate": 6.990488050059644e-05,
      "loss": 2.6886,
      "step": 2177
    },
    {
      "epoch": 0.6031570202160067,
      "grad_norm": 2.1140644550323486,
      "learning_rate": 6.982076004177061e-05,
      "loss": 3.0166,
      "step": 2178
    },
    {
      "epoch": 0.6034339518139019,
      "grad_norm": 1.175340175628662,
      "learning_rate": 6.973666307191996e-05,
      "loss": 2.7278,
      "step": 2179
    },
    {
      "epoch": 0.6037108834117972,
      "grad_norm": 1.7920030355453491,
      "learning_rate": 6.965258965649841e-05,
      "loss": 2.8499,
      "step": 2180
    },
    {
      "epoch": 0.6039878150096926,
      "grad_norm": 1.4756419658660889,
      "learning_rate": 6.956853986094163e-05,
      "loss": 2.8933,
      "step": 2181
    },
    {
      "epoch": 0.6042647466075879,
      "grad_norm": 2.0288076400756836,
      "learning_rate": 6.948451375066689e-05,
      "loss": 2.9402,
      "step": 2182
    },
    {
      "epoch": 0.6045416782054832,
      "grad_norm": 1.8753596544265747,
      "learning_rate": 6.940051139107306e-05,
      "loss": 2.8504,
      "step": 2183
    },
    {
      "epoch": 0.6048186098033785,
      "grad_norm": 1.7512092590332031,
      "learning_rate": 6.931653284754042e-05,
      "loss": 2.8587,
      "step": 2184
    },
    {
      "epoch": 0.6050955414012739,
      "grad_norm": 0.7080219388008118,
      "learning_rate": 6.923257818543088e-05,
      "loss": 2.7284,
      "step": 2185
    },
    {
      "epoch": 0.6053724729991692,
      "grad_norm": 1.5267521142959595,
      "learning_rate": 6.914864747008762e-05,
      "loss": 3.099,
      "step": 2186
    },
    {
      "epoch": 0.6056494045970645,
      "grad_norm": 1.6850520372390747,
      "learning_rate": 6.906474076683519e-05,
      "loss": 2.7281,
      "step": 2187
    },
    {
      "epoch": 0.6059263361949598,
      "grad_norm": 1.2631269693374634,
      "learning_rate": 6.898085814097957e-05,
      "loss": 2.911,
      "step": 2188
    },
    {
      "epoch": 0.6062032677928552,
      "grad_norm": 1.7381616830825806,
      "learning_rate": 6.889699965780787e-05,
      "loss": 2.6787,
      "step": 2189
    },
    {
      "epoch": 0.6064801993907505,
      "grad_norm": 1.130190372467041,
      "learning_rate": 6.881316538258846e-05,
      "loss": 2.9246,
      "step": 2190
    },
    {
      "epoch": 0.6067571309886458,
      "grad_norm": 1.3303025960922241,
      "learning_rate": 6.87293553805709e-05,
      "loss": 3.1455,
      "step": 2191
    },
    {
      "epoch": 0.6070340625865411,
      "grad_norm": 1.4642069339752197,
      "learning_rate": 6.864556971698583e-05,
      "loss": 2.6106,
      "step": 2192
    },
    {
      "epoch": 0.6073109941844365,
      "grad_norm": 1.1034564971923828,
      "learning_rate": 6.856180845704488e-05,
      "loss": 2.7239,
      "step": 2193
    },
    {
      "epoch": 0.6075879257823318,
      "grad_norm": 1.9593942165374756,
      "learning_rate": 6.847807166594083e-05,
      "loss": 2.7019,
      "step": 2194
    },
    {
      "epoch": 0.6078648573802271,
      "grad_norm": 1.1505136489868164,
      "learning_rate": 6.839435940884731e-05,
      "loss": 2.6409,
      "step": 2195
    },
    {
      "epoch": 0.6081417889781224,
      "grad_norm": 1.1177148818969727,
      "learning_rate": 6.831067175091884e-05,
      "loss": 2.7674,
      "step": 2196
    },
    {
      "epoch": 0.6084187205760178,
      "grad_norm": 1.723169207572937,
      "learning_rate": 6.82270087572909e-05,
      "loss": 2.5711,
      "step": 2197
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 1.8692177534103394,
      "learning_rate": 6.814337049307966e-05,
      "loss": 2.8186,
      "step": 2198
    },
    {
      "epoch": 0.6089725837718084,
      "grad_norm": 1.2709190845489502,
      "learning_rate": 6.805975702338208e-05,
      "loss": 2.909,
      "step": 2199
    },
    {
      "epoch": 0.6092495153697037,
      "grad_norm": 3.416496515274048,
      "learning_rate": 6.797616841327591e-05,
      "loss": 3.9108,
      "step": 2200
    },
    {
      "epoch": 0.6095264469675991,
      "grad_norm": 1.9776725769042969,
      "learning_rate": 6.78926047278194e-05,
      "loss": 2.8953,
      "step": 2201
    },
    {
      "epoch": 0.6098033785654943,
      "grad_norm": 1.553315281867981,
      "learning_rate": 6.780906603205148e-05,
      "loss": 2.733,
      "step": 2202
    },
    {
      "epoch": 0.6100803101633896,
      "grad_norm": 1.8651763200759888,
      "learning_rate": 6.772555239099166e-05,
      "loss": 2.9544,
      "step": 2203
    },
    {
      "epoch": 0.6103572417612849,
      "grad_norm": 1.6313016414642334,
      "learning_rate": 6.764206386963991e-05,
      "loss": 2.6848,
      "step": 2204
    },
    {
      "epoch": 0.6106341733591802,
      "grad_norm": 1.7985180616378784,
      "learning_rate": 6.755860053297662e-05,
      "loss": 2.5212,
      "step": 2205
    },
    {
      "epoch": 0.6109111049570756,
      "grad_norm": 1.282670497894287,
      "learning_rate": 6.747516244596267e-05,
      "loss": 2.8279,
      "step": 2206
    },
    {
      "epoch": 0.6111880365549709,
      "grad_norm": 2.2696824073791504,
      "learning_rate": 6.73917496735392e-05,
      "loss": 2.9487,
      "step": 2207
    },
    {
      "epoch": 0.6114649681528662,
      "grad_norm": 1.2766075134277344,
      "learning_rate": 6.730836228062763e-05,
      "loss": 2.6696,
      "step": 2208
    },
    {
      "epoch": 0.6117418997507615,
      "grad_norm": 1.6926491260528564,
      "learning_rate": 6.722500033212974e-05,
      "loss": 2.8545,
      "step": 2209
    },
    {
      "epoch": 0.6120188313486569,
      "grad_norm": 1.9839706420898438,
      "learning_rate": 6.714166389292744e-05,
      "loss": 2.5835,
      "step": 2210
    },
    {
      "epoch": 0.6122957629465522,
      "grad_norm": 1.2539972066879272,
      "learning_rate": 6.70583530278827e-05,
      "loss": 3.0236,
      "step": 2211
    },
    {
      "epoch": 0.6125726945444475,
      "grad_norm": 1.5419610738754272,
      "learning_rate": 6.697506780183778e-05,
      "loss": 2.8228,
      "step": 2212
    },
    {
      "epoch": 0.6128496261423428,
      "grad_norm": 1.4121469259262085,
      "learning_rate": 6.689180827961481e-05,
      "loss": 3.0115,
      "step": 2213
    },
    {
      "epoch": 0.6131265577402382,
      "grad_norm": 1.744856834411621,
      "learning_rate": 6.680857452601598e-05,
      "loss": 2.7659,
      "step": 2214
    },
    {
      "epoch": 0.6134034893381335,
      "grad_norm": 2.4373672008514404,
      "learning_rate": 6.67253666058235e-05,
      "loss": 3.2158,
      "step": 2215
    },
    {
      "epoch": 0.6136804209360288,
      "grad_norm": 1.1923551559448242,
      "learning_rate": 6.664218458379933e-05,
      "loss": 2.6316,
      "step": 2216
    },
    {
      "epoch": 0.6139573525339241,
      "grad_norm": 1.2512011528015137,
      "learning_rate": 6.65590285246853e-05,
      "loss": 2.6847,
      "step": 2217
    },
    {
      "epoch": 0.6142342841318195,
      "grad_norm": 1.0812313556671143,
      "learning_rate": 6.647589849320318e-05,
      "loss": 2.5979,
      "step": 2218
    },
    {
      "epoch": 0.6145112157297148,
      "grad_norm": 1.2180825471878052,
      "learning_rate": 6.639279455405432e-05,
      "loss": 2.6674,
      "step": 2219
    },
    {
      "epoch": 0.6147881473276101,
      "grad_norm": 0.8857952356338501,
      "learning_rate": 6.630971677191978e-05,
      "loss": 2.6687,
      "step": 2220
    },
    {
      "epoch": 0.6150650789255054,
      "grad_norm": 1.5782952308654785,
      "learning_rate": 6.622666521146036e-05,
      "loss": 2.7259,
      "step": 2221
    },
    {
      "epoch": 0.6153420105234008,
      "grad_norm": 2.128387212753296,
      "learning_rate": 6.614363993731636e-05,
      "loss": 2.8938,
      "step": 2222
    },
    {
      "epoch": 0.6156189421212961,
      "grad_norm": 0.9730496406555176,
      "learning_rate": 6.606064101410765e-05,
      "loss": 2.5285,
      "step": 2223
    },
    {
      "epoch": 0.6158958737191914,
      "grad_norm": 1.520876407623291,
      "learning_rate": 6.597766850643361e-05,
      "loss": 3.1274,
      "step": 2224
    },
    {
      "epoch": 0.6161728053170866,
      "grad_norm": 2.834122657775879,
      "learning_rate": 6.589472247887305e-05,
      "loss": 3.5855,
      "step": 2225
    },
    {
      "epoch": 0.616449736914982,
      "grad_norm": 1.961204171180725,
      "learning_rate": 6.58118029959841e-05,
      "loss": 2.9324,
      "step": 2226
    },
    {
      "epoch": 0.6167266685128773,
      "grad_norm": 1.0947190523147583,
      "learning_rate": 6.572891012230438e-05,
      "loss": 2.5908,
      "step": 2227
    },
    {
      "epoch": 0.6170036001107726,
      "grad_norm": 1.51961350440979,
      "learning_rate": 6.564604392235066e-05,
      "loss": 2.8876,
      "step": 2228
    },
    {
      "epoch": 0.6172805317086679,
      "grad_norm": 2.206125020980835,
      "learning_rate": 6.556320446061902e-05,
      "loss": 2.8758,
      "step": 2229
    },
    {
      "epoch": 0.6175574633065632,
      "grad_norm": 1.4404202699661255,
      "learning_rate": 6.548039180158466e-05,
      "loss": 2.9083,
      "step": 2230
    },
    {
      "epoch": 0.6178343949044586,
      "grad_norm": 1.103936791419983,
      "learning_rate": 6.539760600970205e-05,
      "loss": 2.7771,
      "step": 2231
    },
    {
      "epoch": 0.6181113265023539,
      "grad_norm": 1.6521354913711548,
      "learning_rate": 6.531484714940461e-05,
      "loss": 2.7143,
      "step": 2232
    },
    {
      "epoch": 0.6183882581002492,
      "grad_norm": 1.1557974815368652,
      "learning_rate": 6.523211528510487e-05,
      "loss": 2.8761,
      "step": 2233
    },
    {
      "epoch": 0.6186651896981445,
      "grad_norm": 0.951933741569519,
      "learning_rate": 6.514941048119435e-05,
      "loss": 2.7942,
      "step": 2234
    },
    {
      "epoch": 0.6189421212960399,
      "grad_norm": 1.1569383144378662,
      "learning_rate": 6.50667328020435e-05,
      "loss": 2.7508,
      "step": 2235
    },
    {
      "epoch": 0.6192190528939352,
      "grad_norm": 1.6564579010009766,
      "learning_rate": 6.498408231200159e-05,
      "loss": 2.6084,
      "step": 2236
    },
    {
      "epoch": 0.6194959844918305,
      "grad_norm": 1.7123055458068848,
      "learning_rate": 6.490145907539689e-05,
      "loss": 2.4802,
      "step": 2237
    },
    {
      "epoch": 0.6197729160897258,
      "grad_norm": 1.9307233095169067,
      "learning_rate": 6.481886315653632e-05,
      "loss": 2.8049,
      "step": 2238
    },
    {
      "epoch": 0.6200498476876212,
      "grad_norm": 1.2563745975494385,
      "learning_rate": 6.47362946197055e-05,
      "loss": 2.9327,
      "step": 2239
    },
    {
      "epoch": 0.6203267792855165,
      "grad_norm": 1.546926498413086,
      "learning_rate": 6.465375352916893e-05,
      "loss": 2.7569,
      "step": 2240
    },
    {
      "epoch": 0.6206037108834118,
      "grad_norm": 1.681423306465149,
      "learning_rate": 6.45712399491696e-05,
      "loss": 3.1465,
      "step": 2241
    },
    {
      "epoch": 0.6208806424813071,
      "grad_norm": 1.571089267730713,
      "learning_rate": 6.448875394392905e-05,
      "loss": 2.8521,
      "step": 2242
    },
    {
      "epoch": 0.6211575740792025,
      "grad_norm": 1.9650115966796875,
      "learning_rate": 6.440629557764753e-05,
      "loss": 2.8507,
      "step": 2243
    },
    {
      "epoch": 0.6214345056770978,
      "grad_norm": 1.1810646057128906,
      "learning_rate": 6.432386491450361e-05,
      "loss": 2.5333,
      "step": 2244
    },
    {
      "epoch": 0.6217114372749931,
      "grad_norm": 1.0897647142410278,
      "learning_rate": 6.42414620186544e-05,
      "loss": 2.8562,
      "step": 2245
    },
    {
      "epoch": 0.6219883688728884,
      "grad_norm": 0.9666537046432495,
      "learning_rate": 6.415908695423534e-05,
      "loss": 2.8046,
      "step": 2246
    },
    {
      "epoch": 0.6222653004707838,
      "grad_norm": 1.246010422706604,
      "learning_rate": 6.407673978536029e-05,
      "loss": 2.749,
      "step": 2247
    },
    {
      "epoch": 0.6225422320686791,
      "grad_norm": 1.62389075756073,
      "learning_rate": 6.399442057612122e-05,
      "loss": 2.8788,
      "step": 2248
    },
    {
      "epoch": 0.6228191636665743,
      "grad_norm": 1.7508984804153442,
      "learning_rate": 6.391212939058861e-05,
      "loss": 3.1259,
      "step": 2249
    },
    {
      "epoch": 0.6230960952644696,
      "grad_norm": 3.678032159805298,
      "learning_rate": 6.38298662928109e-05,
      "loss": 3.8598,
      "step": 2250
    },
    {
      "epoch": 0.623373026862365,
      "grad_norm": 1.0360453128814697,
      "learning_rate": 6.374763134681472e-05,
      "loss": 2.6876,
      "step": 2251
    },
    {
      "epoch": 0.6236499584602603,
      "grad_norm": 1.6751775741577148,
      "learning_rate": 6.366542461660488e-05,
      "loss": 3.1079,
      "step": 2252
    },
    {
      "epoch": 0.6239268900581556,
      "grad_norm": 1.4977526664733887,
      "learning_rate": 6.358324616616412e-05,
      "loss": 2.695,
      "step": 2253
    },
    {
      "epoch": 0.6242038216560509,
      "grad_norm": 1.6655688285827637,
      "learning_rate": 6.350109605945323e-05,
      "loss": 3.0501,
      "step": 2254
    },
    {
      "epoch": 0.6244807532539463,
      "grad_norm": 1.2852628231048584,
      "learning_rate": 6.341897436041093e-05,
      "loss": 3.0004,
      "step": 2255
    },
    {
      "epoch": 0.6247576848518416,
      "grad_norm": 1.231870412826538,
      "learning_rate": 6.333688113295384e-05,
      "loss": 2.5332,
      "step": 2256
    },
    {
      "epoch": 0.6250346164497369,
      "grad_norm": 1.5554944276809692,
      "learning_rate": 6.325481644097635e-05,
      "loss": 2.5779,
      "step": 2257
    },
    {
      "epoch": 0.6253115480476322,
      "grad_norm": 1.6678074598312378,
      "learning_rate": 6.317278034835077e-05,
      "loss": 2.7953,
      "step": 2258
    },
    {
      "epoch": 0.6255884796455276,
      "grad_norm": 1.0044925212860107,
      "learning_rate": 6.309077291892702e-05,
      "loss": 2.6219,
      "step": 2259
    },
    {
      "epoch": 0.6258654112434229,
      "grad_norm": 1.7450201511383057,
      "learning_rate": 6.300879421653277e-05,
      "loss": 2.8333,
      "step": 2260
    },
    {
      "epoch": 0.6261423428413182,
      "grad_norm": 2.0873069763183594,
      "learning_rate": 6.292684430497335e-05,
      "loss": 2.6769,
      "step": 2261
    },
    {
      "epoch": 0.6264192744392135,
      "grad_norm": 1.1980091333389282,
      "learning_rate": 6.284492324803167e-05,
      "loss": 2.7962,
      "step": 2262
    },
    {
      "epoch": 0.6266962060371088,
      "grad_norm": 1.5165520906448364,
      "learning_rate": 6.27630311094681e-05,
      "loss": 2.92,
      "step": 2263
    },
    {
      "epoch": 0.6269731376350042,
      "grad_norm": 1.4691338539123535,
      "learning_rate": 6.268116795302068e-05,
      "loss": 3.0009,
      "step": 2264
    },
    {
      "epoch": 0.6272500692328995,
      "grad_norm": 1.3578917980194092,
      "learning_rate": 6.25993338424047e-05,
      "loss": 2.5752,
      "step": 2265
    },
    {
      "epoch": 0.6275270008307948,
      "grad_norm": 1.5438745021820068,
      "learning_rate": 6.251752884131298e-05,
      "loss": 2.6971,
      "step": 2266
    },
    {
      "epoch": 0.6278039324286901,
      "grad_norm": 1.9460550546646118,
      "learning_rate": 6.243575301341561e-05,
      "loss": 2.7949,
      "step": 2267
    },
    {
      "epoch": 0.6280808640265855,
      "grad_norm": 1.9470924139022827,
      "learning_rate": 6.235400642236002e-05,
      "loss": 2.6991,
      "step": 2268
    },
    {
      "epoch": 0.6283577956244808,
      "grad_norm": 1.1154978275299072,
      "learning_rate": 6.227228913177081e-05,
      "loss": 2.8878,
      "step": 2269
    },
    {
      "epoch": 0.6286347272223761,
      "grad_norm": 1.426668405532837,
      "learning_rate": 6.21906012052499e-05,
      "loss": 2.8709,
      "step": 2270
    },
    {
      "epoch": 0.6289116588202714,
      "grad_norm": 1.4238054752349854,
      "learning_rate": 6.210894270637625e-05,
      "loss": 2.6498,
      "step": 2271
    },
    {
      "epoch": 0.6291885904181667,
      "grad_norm": 1.6661927700042725,
      "learning_rate": 6.202731369870591e-05,
      "loss": 2.978,
      "step": 2272
    },
    {
      "epoch": 0.629465522016062,
      "grad_norm": 1.2813878059387207,
      "learning_rate": 6.19457142457721e-05,
      "loss": 2.8663,
      "step": 2273
    },
    {
      "epoch": 0.6297424536139573,
      "grad_norm": 1.130333662033081,
      "learning_rate": 6.186414441108487e-05,
      "loss": 2.8716,
      "step": 2274
    },
    {
      "epoch": 0.6300193852118526,
      "grad_norm": 3.8945274353027344,
      "learning_rate": 6.178260425813131e-05,
      "loss": 3.9808,
      "step": 2275
    },
    {
      "epoch": 0.630296316809748,
      "grad_norm": 1.5076262950897217,
      "learning_rate": 6.170109385037545e-05,
      "loss": 2.7333,
      "step": 2276
    },
    {
      "epoch": 0.6305732484076433,
      "grad_norm": 1.3850371837615967,
      "learning_rate": 6.161961325125809e-05,
      "loss": 2.9242,
      "step": 2277
    },
    {
      "epoch": 0.6308501800055386,
      "grad_norm": 1.3438189029693604,
      "learning_rate": 6.153816252419682e-05,
      "loss": 3.0425,
      "step": 2278
    },
    {
      "epoch": 0.6311271116034339,
      "grad_norm": 1.1912837028503418,
      "learning_rate": 6.14567417325861e-05,
      "loss": 2.773,
      "step": 2279
    },
    {
      "epoch": 0.6314040432013293,
      "grad_norm": 1.4491552114486694,
      "learning_rate": 6.137535093979694e-05,
      "loss": 2.6852,
      "step": 2280
    },
    {
      "epoch": 0.6316809747992246,
      "grad_norm": 1.8994818925857544,
      "learning_rate": 6.129399020917705e-05,
      "loss": 2.9605,
      "step": 2281
    },
    {
      "epoch": 0.6319579063971199,
      "grad_norm": 0.9408001899719238,
      "learning_rate": 6.121265960405085e-05,
      "loss": 2.9011,
      "step": 2282
    },
    {
      "epoch": 0.6322348379950152,
      "grad_norm": 1.6518540382385254,
      "learning_rate": 6.113135918771915e-05,
      "loss": 2.9159,
      "step": 2283
    },
    {
      "epoch": 0.6325117695929106,
      "grad_norm": 0.8682531118392944,
      "learning_rate": 6.105008902345935e-05,
      "loss": 2.584,
      "step": 2284
    },
    {
      "epoch": 0.6327887011908059,
      "grad_norm": 2.14064884185791,
      "learning_rate": 6.096884917452531e-05,
      "loss": 2.7146,
      "step": 2285
    },
    {
      "epoch": 0.6330656327887012,
      "grad_norm": 1.1299852132797241,
      "learning_rate": 6.088763970414726e-05,
      "loss": 3.0074,
      "step": 2286
    },
    {
      "epoch": 0.6333425643865965,
      "grad_norm": 1.2496439218521118,
      "learning_rate": 6.080646067553182e-05,
      "loss": 2.9153,
      "step": 2287
    },
    {
      "epoch": 0.6336194959844919,
      "grad_norm": 1.5232523679733276,
      "learning_rate": 6.0725312151861866e-05,
      "loss": 2.6882,
      "step": 2288
    },
    {
      "epoch": 0.6338964275823872,
      "grad_norm": 1.2904765605926514,
      "learning_rate": 6.064419419629662e-05,
      "loss": 3.0187,
      "step": 2289
    },
    {
      "epoch": 0.6341733591802825,
      "grad_norm": 1.1545737981796265,
      "learning_rate": 6.0563106871971385e-05,
      "loss": 2.7736,
      "step": 2290
    },
    {
      "epoch": 0.6344502907781778,
      "grad_norm": 1.1595368385314941,
      "learning_rate": 6.048205024199778e-05,
      "loss": 2.8019,
      "step": 2291
    },
    {
      "epoch": 0.6347272223760732,
      "grad_norm": 1.647827386856079,
      "learning_rate": 6.04010243694634e-05,
      "loss": 2.8922,
      "step": 2292
    },
    {
      "epoch": 0.6350041539739685,
      "grad_norm": 2.090649366378784,
      "learning_rate": 6.0320029317431924e-05,
      "loss": 2.5161,
      "step": 2293
    },
    {
      "epoch": 0.6352810855718638,
      "grad_norm": 1.726109266281128,
      "learning_rate": 6.023906514894313e-05,
      "loss": 2.709,
      "step": 2294
    },
    {
      "epoch": 0.635558017169759,
      "grad_norm": 1.3686494827270508,
      "learning_rate": 6.015813192701267e-05,
      "loss": 2.967,
      "step": 2295
    },
    {
      "epoch": 0.6358349487676543,
      "grad_norm": 21.516817092895508,
      "learning_rate": 6.007722971463211e-05,
      "loss": 2.955,
      "step": 2296
    },
    {
      "epoch": 0.6361118803655497,
      "grad_norm": 1.523563027381897,
      "learning_rate": 5.999635857476897e-05,
      "loss": 2.693,
      "step": 2297
    },
    {
      "epoch": 0.636388811963445,
      "grad_norm": 1.3220926523208618,
      "learning_rate": 5.991551857036648e-05,
      "loss": 3.0974,
      "step": 2298
    },
    {
      "epoch": 0.6366657435613403,
      "grad_norm": 1.6534843444824219,
      "learning_rate": 5.983470976434369e-05,
      "loss": 2.9154,
      "step": 2299
    },
    {
      "epoch": 0.6369426751592356,
      "grad_norm": 4.144333362579346,
      "learning_rate": 5.9753932219595356e-05,
      "loss": 3.9512,
      "step": 2300
    },
    {
      "epoch": 0.637219606757131,
      "grad_norm": 1.7014082670211792,
      "learning_rate": 5.967318599899194e-05,
      "loss": 2.7434,
      "step": 2301
    },
    {
      "epoch": 0.6374965383550263,
      "grad_norm": 1.1944373846054077,
      "learning_rate": 5.9592471165379425e-05,
      "loss": 2.746,
      "step": 2302
    },
    {
      "epoch": 0.6377734699529216,
      "grad_norm": 1.1372411251068115,
      "learning_rate": 5.95117877815795e-05,
      "loss": 2.7947,
      "step": 2303
    },
    {
      "epoch": 0.6380504015508169,
      "grad_norm": 1.6045831441879272,
      "learning_rate": 5.943113591038928e-05,
      "loss": 2.8579,
      "step": 2304
    },
    {
      "epoch": 0.6383273331487123,
      "grad_norm": 1.6579948663711548,
      "learning_rate": 5.935051561458134e-05,
      "loss": 2.7143,
      "step": 2305
    },
    {
      "epoch": 0.6386042647466076,
      "grad_norm": 0.9162551760673523,
      "learning_rate": 5.926992695690378e-05,
      "loss": 2.5414,
      "step": 2306
    },
    {
      "epoch": 0.6388811963445029,
      "grad_norm": 1.522512435913086,
      "learning_rate": 5.918937000007999e-05,
      "loss": 2.6506,
      "step": 2307
    },
    {
      "epoch": 0.6391581279423982,
      "grad_norm": 2.271165609359741,
      "learning_rate": 5.910884480680867e-05,
      "loss": 2.9811,
      "step": 2308
    },
    {
      "epoch": 0.6394350595402936,
      "grad_norm": 1.3790143728256226,
      "learning_rate": 5.902835143976393e-05,
      "loss": 2.6826,
      "step": 2309
    },
    {
      "epoch": 0.6397119911381889,
      "grad_norm": 1.4355623722076416,
      "learning_rate": 5.894788996159493e-05,
      "loss": 2.7139,
      "step": 2310
    },
    {
      "epoch": 0.6399889227360842,
      "grad_norm": 1.4164738655090332,
      "learning_rate": 5.8867460434926125e-05,
      "loss": 2.7957,
      "step": 2311
    },
    {
      "epoch": 0.6402658543339795,
      "grad_norm": 1.381259799003601,
      "learning_rate": 5.8787062922357115e-05,
      "loss": 2.5392,
      "step": 2312
    },
    {
      "epoch": 0.6405427859318749,
      "grad_norm": 1.610293984413147,
      "learning_rate": 5.870669748646254e-05,
      "loss": 2.7615,
      "step": 2313
    },
    {
      "epoch": 0.6408197175297702,
      "grad_norm": 1.2964593172073364,
      "learning_rate": 5.862636418979198e-05,
      "loss": 2.737,
      "step": 2314
    },
    {
      "epoch": 0.6410966491276655,
      "grad_norm": 1.647544264793396,
      "learning_rate": 5.854606309487023e-05,
      "loss": 2.6559,
      "step": 2315
    },
    {
      "epoch": 0.6413735807255608,
      "grad_norm": 1.4856895208358765,
      "learning_rate": 5.8465794264196815e-05,
      "loss": 2.8889,
      "step": 2316
    },
    {
      "epoch": 0.6416505123234562,
      "grad_norm": 1.3166649341583252,
      "learning_rate": 5.8385557760246203e-05,
      "loss": 2.6272,
      "step": 2317
    },
    {
      "epoch": 0.6419274439213515,
      "grad_norm": 6.107063293457031,
      "learning_rate": 5.830535364546779e-05,
      "loss": 3.0925,
      "step": 2318
    },
    {
      "epoch": 0.6422043755192467,
      "grad_norm": 1.8524757623672485,
      "learning_rate": 5.822518198228565e-05,
      "loss": 2.5587,
      "step": 2319
    },
    {
      "epoch": 0.642481307117142,
      "grad_norm": 1.1845844984054565,
      "learning_rate": 5.8145042833098654e-05,
      "loss": 2.5099,
      "step": 2320
    },
    {
      "epoch": 0.6427582387150373,
      "grad_norm": 1.5074481964111328,
      "learning_rate": 5.8064936260280335e-05,
      "loss": 2.7683,
      "step": 2321
    },
    {
      "epoch": 0.6430351703129327,
      "grad_norm": 0.9466500282287598,
      "learning_rate": 5.798486232617898e-05,
      "loss": 2.5384,
      "step": 2322
    },
    {
      "epoch": 0.643312101910828,
      "grad_norm": 1.6233863830566406,
      "learning_rate": 5.790482109311723e-05,
      "loss": 2.8482,
      "step": 2323
    },
    {
      "epoch": 0.6435890335087233,
      "grad_norm": 1.5645900964736938,
      "learning_rate": 5.782481262339261e-05,
      "loss": 3.1355,
      "step": 2324
    },
    {
      "epoch": 0.6438659651066186,
      "grad_norm": 5.980441093444824,
      "learning_rate": 5.774483697927684e-05,
      "loss": 4.6119,
      "step": 2325
    },
    {
      "epoch": 0.644142896704514,
      "grad_norm": 1.4056565761566162,
      "learning_rate": 5.766489422301624e-05,
      "loss": 2.7215,
      "step": 2326
    },
    {
      "epoch": 0.6444198283024093,
      "grad_norm": 1.7121729850769043,
      "learning_rate": 5.7584984416831535e-05,
      "loss": 2.7691,
      "step": 2327
    },
    {
      "epoch": 0.6446967599003046,
      "grad_norm": 1.0393214225769043,
      "learning_rate": 5.750510762291775e-05,
      "loss": 2.8974,
      "step": 2328
    },
    {
      "epoch": 0.6449736914981999,
      "grad_norm": 1.7014375925064087,
      "learning_rate": 5.742526390344427e-05,
      "loss": 2.8694,
      "step": 2329
    },
    {
      "epoch": 0.6452506230960953,
      "grad_norm": 1.6189837455749512,
      "learning_rate": 5.734545332055471e-05,
      "loss": 2.6544,
      "step": 2330
    },
    {
      "epoch": 0.6455275546939906,
      "grad_norm": 0.8944665193557739,
      "learning_rate": 5.7265675936366925e-05,
      "loss": 2.6017,
      "step": 2331
    },
    {
      "epoch": 0.6458044862918859,
      "grad_norm": 2.0103485584259033,
      "learning_rate": 5.718593181297278e-05,
      "loss": 3.1379,
      "step": 2332
    },
    {
      "epoch": 0.6460814178897812,
      "grad_norm": 1.183464527130127,
      "learning_rate": 5.710622101243857e-05,
      "loss": 2.8474,
      "step": 2333
    },
    {
      "epoch": 0.6463583494876766,
      "grad_norm": 1.8213006258010864,
      "learning_rate": 5.702654359680428e-05,
      "loss": 2.7316,
      "step": 2334
    },
    {
      "epoch": 0.6466352810855719,
      "grad_norm": 1.498777985572815,
      "learning_rate": 5.694689962808417e-05,
      "loss": 2.9369,
      "step": 2335
    },
    {
      "epoch": 0.6469122126834672,
      "grad_norm": 2.145932197570801,
      "learning_rate": 5.68672891682664e-05,
      "loss": 2.756,
      "step": 2336
    },
    {
      "epoch": 0.6471891442813625,
      "grad_norm": 1.2577202320098877,
      "learning_rate": 5.6787712279313e-05,
      "loss": 2.5537,
      "step": 2337
    },
    {
      "epoch": 0.6474660758792579,
      "grad_norm": 1.0808194875717163,
      "learning_rate": 5.670816902315994e-05,
      "loss": 2.6741,
      "step": 2338
    },
    {
      "epoch": 0.6477430074771532,
      "grad_norm": 1.7351064682006836,
      "learning_rate": 5.662865946171696e-05,
      "loss": 3.1183,
      "step": 2339
    },
    {
      "epoch": 0.6480199390750485,
      "grad_norm": 1.717112421989441,
      "learning_rate": 5.654918365686766e-05,
      "loss": 2.9241,
      "step": 2340
    },
    {
      "epoch": 0.6482968706729438,
      "grad_norm": 1.0554273128509521,
      "learning_rate": 5.6469741670469144e-05,
      "loss": 2.8183,
      "step": 2341
    },
    {
      "epoch": 0.648573802270839,
      "grad_norm": 12.077258110046387,
      "learning_rate": 5.639033356435257e-05,
      "loss": 2.8132,
      "step": 2342
    },
    {
      "epoch": 0.6488507338687344,
      "grad_norm": 1.9363653659820557,
      "learning_rate": 5.631095940032235e-05,
      "loss": 2.3197,
      "step": 2343
    },
    {
      "epoch": 0.6491276654666297,
      "grad_norm": 1.1245683431625366,
      "learning_rate": 5.6231619240156694e-05,
      "loss": 2.7323,
      "step": 2344
    },
    {
      "epoch": 0.649404597064525,
      "grad_norm": 1.9193580150604248,
      "learning_rate": 5.615231314560727e-05,
      "loss": 2.5539,
      "step": 2345
    },
    {
      "epoch": 0.6496815286624203,
      "grad_norm": 1.237788200378418,
      "learning_rate": 5.607304117839929e-05,
      "loss": 2.4474,
      "step": 2346
    },
    {
      "epoch": 0.6499584602603157,
      "grad_norm": 0.9905593395233154,
      "learning_rate": 5.599380340023135e-05,
      "loss": 2.7252,
      "step": 2347
    },
    {
      "epoch": 0.650235391858211,
      "grad_norm": 1.0714815855026245,
      "learning_rate": 5.591459987277545e-05,
      "loss": 2.7466,
      "step": 2348
    },
    {
      "epoch": 0.6505123234561063,
      "grad_norm": 1.6997700929641724,
      "learning_rate": 5.5835430657676976e-05,
      "loss": 2.9485,
      "step": 2349
    },
    {
      "epoch": 0.6507892550540016,
      "grad_norm": 3.4924404621124268,
      "learning_rate": 5.575629581655445e-05,
      "loss": 3.946,
      "step": 2350
    },
    {
      "epoch": 0.651066186651897,
      "grad_norm": 1.172761082649231,
      "learning_rate": 5.567719541099992e-05,
      "loss": 2.8241,
      "step": 2351
    },
    {
      "epoch": 0.6513431182497923,
      "grad_norm": 1.1470800638198853,
      "learning_rate": 5.5598129502578344e-05,
      "loss": 2.5238,
      "step": 2352
    },
    {
      "epoch": 0.6516200498476876,
      "grad_norm": 1.4182355403900146,
      "learning_rate": 5.551909815282801e-05,
      "loss": 2.7498,
      "step": 2353
    },
    {
      "epoch": 0.651896981445583,
      "grad_norm": 1.447364330291748,
      "learning_rate": 5.544010142326026e-05,
      "loss": 2.9252,
      "step": 2354
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 1.1676846742630005,
      "learning_rate": 5.5361139375359494e-05,
      "loss": 2.7525,
      "step": 2355
    },
    {
      "epoch": 0.6524508446413736,
      "grad_norm": 1.3512132167816162,
      "learning_rate": 5.5282212070583015e-05,
      "loss": 2.6566,
      "step": 2356
    },
    {
      "epoch": 0.6527277762392689,
      "grad_norm": 1.911013126373291,
      "learning_rate": 5.520331957036134e-05,
      "loss": 2.9136,
      "step": 2357
    },
    {
      "epoch": 0.6530047078371642,
      "grad_norm": 2.6728639602661133,
      "learning_rate": 5.5124461936097605e-05,
      "loss": 3.1852,
      "step": 2358
    },
    {
      "epoch": 0.6532816394350596,
      "grad_norm": 2.6511623859405518,
      "learning_rate": 5.504563922916799e-05,
      "loss": 3.2279,
      "step": 2359
    },
    {
      "epoch": 0.6535585710329549,
      "grad_norm": 1.7919533252716064,
      "learning_rate": 5.496685151092145e-05,
      "loss": 2.777,
      "step": 2360
    },
    {
      "epoch": 0.6538355026308502,
      "grad_norm": 1.2835693359375,
      "learning_rate": 5.4888098842679694e-05,
      "loss": 2.9625,
      "step": 2361
    },
    {
      "epoch": 0.6541124342287455,
      "grad_norm": 1.2808218002319336,
      "learning_rate": 5.480938128573716e-05,
      "loss": 2.789,
      "step": 2362
    },
    {
      "epoch": 0.6543893658266409,
      "grad_norm": 1.460275650024414,
      "learning_rate": 5.473069890136093e-05,
      "loss": 2.7713,
      "step": 2363
    },
    {
      "epoch": 0.6546662974245362,
      "grad_norm": 1.2018828392028809,
      "learning_rate": 5.4652051750790825e-05,
      "loss": 2.8828,
      "step": 2364
    },
    {
      "epoch": 0.6549432290224314,
      "grad_norm": 1.755499243736267,
      "learning_rate": 5.4573439895238976e-05,
      "loss": 2.449,
      "step": 2365
    },
    {
      "epoch": 0.6552201606203267,
      "grad_norm": 1.0556343793869019,
      "learning_rate": 5.449486339589043e-05,
      "loss": 2.7824,
      "step": 2366
    },
    {
      "epoch": 0.655497092218222,
      "grad_norm": 2.3796725273132324,
      "learning_rate": 5.441632231390237e-05,
      "loss": 3.0264,
      "step": 2367
    },
    {
      "epoch": 0.6557740238161174,
      "grad_norm": 1.304976224899292,
      "learning_rate": 5.4337816710404585e-05,
      "loss": 2.5941,
      "step": 2368
    },
    {
      "epoch": 0.6560509554140127,
      "grad_norm": 1.4646837711334229,
      "learning_rate": 5.425934664649921e-05,
      "loss": 2.7932,
      "step": 2369
    },
    {
      "epoch": 0.656327887011908,
      "grad_norm": 1.8304128646850586,
      "learning_rate": 5.4180912183260756e-05,
      "loss": 2.7483,
      "step": 2370
    },
    {
      "epoch": 0.6566048186098034,
      "grad_norm": 1.6258549690246582,
      "learning_rate": 5.410251338173595e-05,
      "loss": 2.8172,
      "step": 2371
    },
    {
      "epoch": 0.6568817502076987,
      "grad_norm": 1.4259107112884521,
      "learning_rate": 5.402415030294384e-05,
      "loss": 2.9041,
      "step": 2372
    },
    {
      "epoch": 0.657158681805594,
      "grad_norm": 1.149336576461792,
      "learning_rate": 5.3945823007875676e-05,
      "loss": 2.8761,
      "step": 2373
    },
    {
      "epoch": 0.6574356134034893,
      "grad_norm": 1.3595845699310303,
      "learning_rate": 5.3867531557494674e-05,
      "loss": 2.8705,
      "step": 2374
    },
    {
      "epoch": 0.6577125450013847,
      "grad_norm": 8.332207679748535,
      "learning_rate": 5.378927601273648e-05,
      "loss": 4.7642,
      "step": 2375
    },
    {
      "epoch": 0.65798947659928,
      "grad_norm": 1.699044942855835,
      "learning_rate": 5.371105643450849e-05,
      "loss": 2.8592,
      "step": 2376
    },
    {
      "epoch": 0.6582664081971753,
      "grad_norm": 1.3306677341461182,
      "learning_rate": 5.363287288369023e-05,
      "loss": 2.7023,
      "step": 2377
    },
    {
      "epoch": 0.6585433397950706,
      "grad_norm": 1.9283133745193481,
      "learning_rate": 5.355472542113325e-05,
      "loss": 2.8911,
      "step": 2378
    },
    {
      "epoch": 0.658820271392966,
      "grad_norm": 1.7714849710464478,
      "learning_rate": 5.347661410766087e-05,
      "loss": 2.8308,
      "step": 2379
    },
    {
      "epoch": 0.6590972029908613,
      "grad_norm": 1.4751148223876953,
      "learning_rate": 5.3398539004068415e-05,
      "loss": 2.7754,
      "step": 2380
    },
    {
      "epoch": 0.6593741345887566,
      "grad_norm": 1.3037173748016357,
      "learning_rate": 5.3320500171122914e-05,
      "loss": 2.9529,
      "step": 2381
    },
    {
      "epoch": 0.6596510661866519,
      "grad_norm": 1.1674219369888306,
      "learning_rate": 5.3242497669563264e-05,
      "loss": 3.0056,
      "step": 2382
    },
    {
      "epoch": 0.6599279977845472,
      "grad_norm": 0.8783468008041382,
      "learning_rate": 5.316453156009992e-05,
      "loss": 2.7169,
      "step": 2383
    },
    {
      "epoch": 0.6602049293824426,
      "grad_norm": 1.2833096981048584,
      "learning_rate": 5.308660190341528e-05,
      "loss": 2.6623,
      "step": 2384
    },
    {
      "epoch": 0.6604818609803379,
      "grad_norm": 1.4621068239212036,
      "learning_rate": 5.3008708760163106e-05,
      "loss": 2.7997,
      "step": 2385
    },
    {
      "epoch": 0.6607587925782332,
      "grad_norm": 2.31742525100708,
      "learning_rate": 5.293085219096889e-05,
      "loss": 2.8536,
      "step": 2386
    },
    {
      "epoch": 0.6610357241761285,
      "grad_norm": 0.8828467726707458,
      "learning_rate": 5.2853032256429614e-05,
      "loss": 2.4077,
      "step": 2387
    },
    {
      "epoch": 0.6613126557740238,
      "grad_norm": 1.3070199489593506,
      "learning_rate": 5.2775249017113796e-05,
      "loss": 2.9468,
      "step": 2388
    },
    {
      "epoch": 0.6615895873719191,
      "grad_norm": 0.896975040435791,
      "learning_rate": 5.2697502533561226e-05,
      "loss": 2.6763,
      "step": 2389
    },
    {
      "epoch": 0.6618665189698144,
      "grad_norm": 1.2044589519500732,
      "learning_rate": 5.2619792866283355e-05,
      "loss": 2.827,
      "step": 2390
    },
    {
      "epoch": 0.6621434505677097,
      "grad_norm": 1.2612204551696777,
      "learning_rate": 5.2542120075762755e-05,
      "loss": 2.7524,
      "step": 2391
    },
    {
      "epoch": 0.6624203821656051,
      "grad_norm": 0.9731190204620361,
      "learning_rate": 5.246448422245337e-05,
      "loss": 2.671,
      "step": 2392
    },
    {
      "epoch": 0.6626973137635004,
      "grad_norm": 1.6314771175384521,
      "learning_rate": 5.238688536678042e-05,
      "loss": 2.684,
      "step": 2393
    },
    {
      "epoch": 0.6629742453613957,
      "grad_norm": 1.4474800825119019,
      "learning_rate": 5.230932356914032e-05,
      "loss": 2.8404,
      "step": 2394
    },
    {
      "epoch": 0.663251176959291,
      "grad_norm": 1.5060440301895142,
      "learning_rate": 5.223179888990062e-05,
      "loss": 2.7726,
      "step": 2395
    },
    {
      "epoch": 0.6635281085571864,
      "grad_norm": 1.4345086812973022,
      "learning_rate": 5.215431138939999e-05,
      "loss": 3.0741,
      "step": 2396
    },
    {
      "epoch": 0.6638050401550817,
      "grad_norm": 1.7585291862487793,
      "learning_rate": 5.207686112794822e-05,
      "loss": 2.6842,
      "step": 2397
    },
    {
      "epoch": 0.664081971752977,
      "grad_norm": 1.4770677089691162,
      "learning_rate": 5.199944816582592e-05,
      "loss": 2.7869,
      "step": 2398
    },
    {
      "epoch": 0.6643589033508723,
      "grad_norm": 2.3265128135681152,
      "learning_rate": 5.1922072563284986e-05,
      "loss": 3.0738,
      "step": 2399
    },
    {
      "epoch": 0.6646358349487677,
      "grad_norm": 3.2085912227630615,
      "learning_rate": 5.184473438054793e-05,
      "loss": 3.7115,
      "step": 2400
    },
    {
      "epoch": 0.664912766546663,
      "grad_norm": 1.9501490592956543,
      "learning_rate": 5.176743367780833e-05,
      "loss": 2.9766,
      "step": 2401
    },
    {
      "epoch": 0.6651896981445583,
      "grad_norm": 1.0873230695724487,
      "learning_rate": 5.1690170515230504e-05,
      "loss": 2.7145,
      "step": 2402
    },
    {
      "epoch": 0.6654666297424536,
      "grad_norm": 1.1702440977096558,
      "learning_rate": 5.1612944952949615e-05,
      "loss": 2.7145,
      "step": 2403
    },
    {
      "epoch": 0.665743561340349,
      "grad_norm": 1.6476362943649292,
      "learning_rate": 5.153575705107152e-05,
      "loss": 2.803,
      "step": 2404
    },
    {
      "epoch": 0.6660204929382443,
      "grad_norm": 1.682246446609497,
      "learning_rate": 5.145860686967274e-05,
      "loss": 2.8331,
      "step": 2405
    },
    {
      "epoch": 0.6662974245361396,
      "grad_norm": 1.1325786113739014,
      "learning_rate": 5.138149446880054e-05,
      "loss": 2.6439,
      "step": 2406
    },
    {
      "epoch": 0.6665743561340349,
      "grad_norm": 1.6570894718170166,
      "learning_rate": 5.130441990847259e-05,
      "loss": 2.9284,
      "step": 2407
    },
    {
      "epoch": 0.6668512877319303,
      "grad_norm": 1.747573971748352,
      "learning_rate": 5.122738324867737e-05,
      "loss": 2.7495,
      "step": 2408
    },
    {
      "epoch": 0.6671282193298256,
      "grad_norm": 1.1928094625473022,
      "learning_rate": 5.115038454937362e-05,
      "loss": 2.8352,
      "step": 2409
    },
    {
      "epoch": 0.6674051509277209,
      "grad_norm": 1.263839840888977,
      "learning_rate": 5.1073423870490665e-05,
      "loss": 2.7625,
      "step": 2410
    },
    {
      "epoch": 0.6676820825256162,
      "grad_norm": 1.6746209859848022,
      "learning_rate": 5.09965012719282e-05,
      "loss": 2.7107,
      "step": 2411
    },
    {
      "epoch": 0.6679590141235114,
      "grad_norm": 1.419972538948059,
      "learning_rate": 5.0919616813556284e-05,
      "loss": 2.7692,
      "step": 2412
    },
    {
      "epoch": 0.6682359457214068,
      "grad_norm": 1.947323203086853,
      "learning_rate": 5.0842770555215316e-05,
      "loss": 2.9443,
      "step": 2413
    },
    {
      "epoch": 0.6685128773193021,
      "grad_norm": 2.007458209991455,
      "learning_rate": 5.076596255671592e-05,
      "loss": 2.7715,
      "step": 2414
    },
    {
      "epoch": 0.6687898089171974,
      "grad_norm": 1.3769513368606567,
      "learning_rate": 5.068919287783902e-05,
      "loss": 3.0913,
      "step": 2415
    },
    {
      "epoch": 0.6690667405150927,
      "grad_norm": 1.4708313941955566,
      "learning_rate": 5.0612461578335526e-05,
      "loss": 2.757,
      "step": 2416
    },
    {
      "epoch": 0.6693436721129881,
      "grad_norm": 1.393153190612793,
      "learning_rate": 5.053576871792678e-05,
      "loss": 2.5286,
      "step": 2417
    },
    {
      "epoch": 0.6696206037108834,
      "grad_norm": 1.5462419986724854,
      "learning_rate": 5.0459114356303926e-05,
      "loss": 2.5453,
      "step": 2418
    },
    {
      "epoch": 0.6698975353087787,
      "grad_norm": 1.400042176246643,
      "learning_rate": 5.0382498553128265e-05,
      "loss": 2.4766,
      "step": 2419
    },
    {
      "epoch": 0.670174466906674,
      "grad_norm": 2.019869089126587,
      "learning_rate": 5.030592136803112e-05,
      "loss": 3.0333,
      "step": 2420
    },
    {
      "epoch": 0.6704513985045694,
      "grad_norm": 1.8234835863113403,
      "learning_rate": 5.022938286061369e-05,
      "loss": 3.2183,
      "step": 2421
    },
    {
      "epoch": 0.6707283301024647,
      "grad_norm": 1.0895123481750488,
      "learning_rate": 5.0152883090447044e-05,
      "loss": 2.5655,
      "step": 2422
    },
    {
      "epoch": 0.67100526170036,
      "grad_norm": 1.3848090171813965,
      "learning_rate": 5.0076422117072266e-05,
      "loss": 2.7499,
      "step": 2423
    },
    {
      "epoch": 0.6712821932982553,
      "grad_norm": 1.6396257877349854,
      "learning_rate": 5.000000000000002e-05,
      "loss": 3.0039,
      "step": 2424
    },
    {
      "epoch": 0.6715591248961507,
      "grad_norm": 2.707336664199829,
      "learning_rate": 4.9923616798710884e-05,
      "loss": 3.4594,
      "step": 2425
    },
    {
      "epoch": 0.671836056494046,
      "grad_norm": 1.0716488361358643,
      "learning_rate": 4.984727257265509e-05,
      "loss": 2.7283,
      "step": 2426
    },
    {
      "epoch": 0.6721129880919413,
      "grad_norm": 1.407761812210083,
      "learning_rate": 4.9770967381252557e-05,
      "loss": 2.7461,
      "step": 2427
    },
    {
      "epoch": 0.6723899196898366,
      "grad_norm": 1.3955081701278687,
      "learning_rate": 4.9694701283892796e-05,
      "loss": 2.8607,
      "step": 2428
    },
    {
      "epoch": 0.672666851287732,
      "grad_norm": 1.2413393259048462,
      "learning_rate": 4.9618474339934916e-05,
      "loss": 2.7155,
      "step": 2429
    },
    {
      "epoch": 0.6729437828856273,
      "grad_norm": 2.0657734870910645,
      "learning_rate": 4.954228660870758e-05,
      "loss": 3.2198,
      "step": 2430
    },
    {
      "epoch": 0.6732207144835226,
      "grad_norm": 1.1644593477249146,
      "learning_rate": 4.946613814950877e-05,
      "loss": 2.873,
      "step": 2431
    },
    {
      "epoch": 0.6734976460814179,
      "grad_norm": 1.8703086376190186,
      "learning_rate": 4.939002902160616e-05,
      "loss": 2.7932,
      "step": 2432
    },
    {
      "epoch": 0.6737745776793133,
      "grad_norm": 0.9289453029632568,
      "learning_rate": 4.9313959284236614e-05,
      "loss": 2.774,
      "step": 2433
    },
    {
      "epoch": 0.6740515092772086,
      "grad_norm": 1.319032073020935,
      "learning_rate": 4.9237928996606384e-05,
      "loss": 2.6889,
      "step": 2434
    },
    {
      "epoch": 0.6743284408751038,
      "grad_norm": 1.0665626525878906,
      "learning_rate": 4.9161938217891065e-05,
      "loss": 2.7208,
      "step": 2435
    },
    {
      "epoch": 0.6746053724729991,
      "grad_norm": 1.254766821861267,
      "learning_rate": 4.9085987007235466e-05,
      "loss": 2.6566,
      "step": 2436
    },
    {
      "epoch": 0.6748823040708944,
      "grad_norm": 1.3757987022399902,
      "learning_rate": 4.90100754237536e-05,
      "loss": 2.8939,
      "step": 2437
    },
    {
      "epoch": 0.6751592356687898,
      "grad_norm": 1.6149036884307861,
      "learning_rate": 4.8934203526528634e-05,
      "loss": 2.7772,
      "step": 2438
    },
    {
      "epoch": 0.6754361672666851,
      "grad_norm": 1.2694714069366455,
      "learning_rate": 4.88583713746129e-05,
      "loss": 2.7618,
      "step": 2439
    },
    {
      "epoch": 0.6757130988645804,
      "grad_norm": 1.7957725524902344,
      "learning_rate": 4.878257902702764e-05,
      "loss": 2.6743,
      "step": 2440
    },
    {
      "epoch": 0.6759900304624757,
      "grad_norm": 1.4955267906188965,
      "learning_rate": 4.870682654276336e-05,
      "loss": 2.9218,
      "step": 2441
    },
    {
      "epoch": 0.6762669620603711,
      "grad_norm": 1.2260961532592773,
      "learning_rate": 4.863111398077932e-05,
      "loss": 2.5895,
      "step": 2442
    },
    {
      "epoch": 0.6765438936582664,
      "grad_norm": 1.1047837734222412,
      "learning_rate": 4.85554414000038e-05,
      "loss": 2.6782,
      "step": 2443
    },
    {
      "epoch": 0.6768208252561617,
      "grad_norm": 1.667965292930603,
      "learning_rate": 4.8479808859333964e-05,
      "loss": 2.9919,
      "step": 2444
    },
    {
      "epoch": 0.677097756854057,
      "grad_norm": 1.1275628805160522,
      "learning_rate": 4.840421641763582e-05,
      "loss": 2.4726,
      "step": 2445
    },
    {
      "epoch": 0.6773746884519524,
      "grad_norm": 1.742850661277771,
      "learning_rate": 4.832866413374413e-05,
      "loss": 2.7017,
      "step": 2446
    },
    {
      "epoch": 0.6776516200498477,
      "grad_norm": 1.8891174793243408,
      "learning_rate": 4.8253152066462416e-05,
      "loss": 2.9787,
      "step": 2447
    },
    {
      "epoch": 0.677928551647743,
      "grad_norm": 1.3550724983215332,
      "learning_rate": 4.817768027456296e-05,
      "loss": 2.8453,
      "step": 2448
    },
    {
      "epoch": 0.6782054832456383,
      "grad_norm": 1.646095633506775,
      "learning_rate": 4.810224881678652e-05,
      "loss": 2.8204,
      "step": 2449
    },
    {
      "epoch": 0.6784824148435337,
      "grad_norm": 3.411214590072632,
      "learning_rate": 4.802685775184272e-05,
      "loss": 3.8809,
      "step": 2450
    },
    {
      "epoch": 0.678759346441429,
      "grad_norm": 1.5782161951065063,
      "learning_rate": 4.7951507138409516e-05,
      "loss": 2.5259,
      "step": 2451
    },
    {
      "epoch": 0.6790362780393243,
      "grad_norm": 1.3184013366699219,
      "learning_rate": 4.787619703513348e-05,
      "loss": 2.7041,
      "step": 2452
    },
    {
      "epoch": 0.6793132096372196,
      "grad_norm": 1.0506107807159424,
      "learning_rate": 4.780092750062967e-05,
      "loss": 2.5995,
      "step": 2453
    },
    {
      "epoch": 0.679590141235115,
      "grad_norm": 1.323937177658081,
      "learning_rate": 4.772569859348156e-05,
      "loss": 2.7424,
      "step": 2454
    },
    {
      "epoch": 0.6798670728330103,
      "grad_norm": 1.750764012336731,
      "learning_rate": 4.7650510372240895e-05,
      "loss": 2.8218,
      "step": 2455
    },
    {
      "epoch": 0.6801440044309056,
      "grad_norm": 1.2859998941421509,
      "learning_rate": 4.757536289542798e-05,
      "loss": 2.8217,
      "step": 2456
    },
    {
      "epoch": 0.6804209360288009,
      "grad_norm": 1.068601131439209,
      "learning_rate": 4.750025622153117e-05,
      "loss": 2.8105,
      "step": 2457
    },
    {
      "epoch": 0.6806978676266962,
      "grad_norm": 1.5132895708084106,
      "learning_rate": 4.742519040900719e-05,
      "loss": 2.8082,
      "step": 2458
    },
    {
      "epoch": 0.6809747992245915,
      "grad_norm": 1.374457836151123,
      "learning_rate": 4.735016551628095e-05,
      "loss": 2.7911,
      "step": 2459
    },
    {
      "epoch": 0.6812517308224868,
      "grad_norm": 1.1537971496582031,
      "learning_rate": 4.72751816017455e-05,
      "loss": 2.7979,
      "step": 2460
    },
    {
      "epoch": 0.6815286624203821,
      "grad_norm": 1.4131252765655518,
      "learning_rate": 4.720023872376199e-05,
      "loss": 2.808,
      "step": 2461
    },
    {
      "epoch": 0.6818055940182774,
      "grad_norm": 1.4175875186920166,
      "learning_rate": 4.712533694065964e-05,
      "loss": 2.5207,
      "step": 2462
    },
    {
      "epoch": 0.6820825256161728,
      "grad_norm": 1.0395691394805908,
      "learning_rate": 4.7050476310735704e-05,
      "loss": 2.8166,
      "step": 2463
    },
    {
      "epoch": 0.6823594572140681,
      "grad_norm": 1.8296945095062256,
      "learning_rate": 4.697565689225528e-05,
      "loss": 3.0081,
      "step": 2464
    },
    {
      "epoch": 0.6826363888119634,
      "grad_norm": 1.247381329536438,
      "learning_rate": 4.690087874345165e-05,
      "loss": 2.9156,
      "step": 2465
    },
    {
      "epoch": 0.6829133204098587,
      "grad_norm": 1.3310344219207764,
      "learning_rate": 4.68261419225257e-05,
      "loss": 2.7953,
      "step": 2466
    },
    {
      "epoch": 0.6831902520077541,
      "grad_norm": 1.7048262357711792,
      "learning_rate": 4.67514464876463e-05,
      "loss": 2.9409,
      "step": 2467
    },
    {
      "epoch": 0.6834671836056494,
      "grad_norm": 1.1385921239852905,
      "learning_rate": 4.6676792496950095e-05,
      "loss": 2.6641,
      "step": 2468
    },
    {
      "epoch": 0.6837441152035447,
      "grad_norm": 1.2046301364898682,
      "learning_rate": 4.660218000854143e-05,
      "loss": 2.7068,
      "step": 2469
    },
    {
      "epoch": 0.68402104680144,
      "grad_norm": 1.1131768226623535,
      "learning_rate": 4.6527609080492385e-05,
      "loss": 2.6678,
      "step": 2470
    },
    {
      "epoch": 0.6842979783993354,
      "grad_norm": 1.2756351232528687,
      "learning_rate": 4.6453079770842686e-05,
      "loss": 2.6196,
      "step": 2471
    },
    {
      "epoch": 0.6845749099972307,
      "grad_norm": 2.282907009124756,
      "learning_rate": 4.637859213759971e-05,
      "loss": 2.6622,
      "step": 2472
    },
    {
      "epoch": 0.684851841595126,
      "grad_norm": 1.1799778938293457,
      "learning_rate": 4.6304146238738196e-05,
      "loss": 2.5564,
      "step": 2473
    },
    {
      "epoch": 0.6851287731930213,
      "grad_norm": 1.7189338207244873,
      "learning_rate": 4.6229742132200746e-05,
      "loss": 2.8177,
      "step": 2474
    },
    {
      "epoch": 0.6854057047909167,
      "grad_norm": 2.7451953887939453,
      "learning_rate": 4.615537987589711e-05,
      "loss": 3.7004,
      "step": 2475
    },
    {
      "epoch": 0.685682636388812,
      "grad_norm": 1.2103180885314941,
      "learning_rate": 4.608105952770464e-05,
      "loss": 2.5203,
      "step": 2476
    },
    {
      "epoch": 0.6859595679867073,
      "grad_norm": 1.9908820390701294,
      "learning_rate": 4.600678114546801e-05,
      "loss": 2.9552,
      "step": 2477
    },
    {
      "epoch": 0.6862364995846026,
      "grad_norm": 1.184588074684143,
      "learning_rate": 4.593254478699928e-05,
      "loss": 2.84,
      "step": 2478
    },
    {
      "epoch": 0.686513431182498,
      "grad_norm": 1.2031408548355103,
      "learning_rate": 4.585835051007774e-05,
      "loss": 2.8386,
      "step": 2479
    },
    {
      "epoch": 0.6867903627803933,
      "grad_norm": 0.9946233034133911,
      "learning_rate": 4.578419837244997e-05,
      "loss": 2.7443,
      "step": 2480
    },
    {
      "epoch": 0.6870672943782886,
      "grad_norm": 1.1824885606765747,
      "learning_rate": 4.571008843182978e-05,
      "loss": 2.6317,
      "step": 2481
    },
    {
      "epoch": 0.6873442259761838,
      "grad_norm": 0.902035653591156,
      "learning_rate": 4.5636020745897975e-05,
      "loss": 2.6174,
      "step": 2482
    },
    {
      "epoch": 0.6876211575740792,
      "grad_norm": 1.003700852394104,
      "learning_rate": 4.5561995372302756e-05,
      "loss": 2.916,
      "step": 2483
    },
    {
      "epoch": 0.6878980891719745,
      "grad_norm": 1.6476397514343262,
      "learning_rate": 4.548801236865912e-05,
      "loss": 2.5126,
      "step": 2484
    },
    {
      "epoch": 0.6881750207698698,
      "grad_norm": 1.7157683372497559,
      "learning_rate": 4.5414071792549217e-05,
      "loss": 2.9392,
      "step": 2485
    },
    {
      "epoch": 0.6884519523677651,
      "grad_norm": 1.2440541982650757,
      "learning_rate": 4.534017370152218e-05,
      "loss": 2.9517,
      "step": 2486
    },
    {
      "epoch": 0.6887288839656605,
      "grad_norm": 1.5872433185577393,
      "learning_rate": 4.5266318153094076e-05,
      "loss": 2.7537,
      "step": 2487
    },
    {
      "epoch": 0.6890058155635558,
      "grad_norm": 1.8673579692840576,
      "learning_rate": 4.5192505204747716e-05,
      "loss": 2.785,
      "step": 2488
    },
    {
      "epoch": 0.6892827471614511,
      "grad_norm": 2.1917476654052734,
      "learning_rate": 4.511873491393304e-05,
      "loss": 2.5704,
      "step": 2489
    },
    {
      "epoch": 0.6895596787593464,
      "grad_norm": 1.8062355518341064,
      "learning_rate": 4.504500733806648e-05,
      "loss": 2.8255,
      "step": 2490
    },
    {
      "epoch": 0.6898366103572418,
      "grad_norm": 1.379191279411316,
      "learning_rate": 4.497132253453143e-05,
      "loss": 3.0313,
      "step": 2491
    },
    {
      "epoch": 0.6901135419551371,
      "grad_norm": 1.0113121271133423,
      "learning_rate": 4.489768056067792e-05,
      "loss": 2.6995,
      "step": 2492
    },
    {
      "epoch": 0.6903904735530324,
      "grad_norm": 1.1049953699111938,
      "learning_rate": 4.4824081473822646e-05,
      "loss": 2.8133,
      "step": 2493
    },
    {
      "epoch": 0.6906674051509277,
      "grad_norm": 1.151623010635376,
      "learning_rate": 4.475052533124893e-05,
      "loss": 2.6096,
      "step": 2494
    },
    {
      "epoch": 0.690944336748823,
      "grad_norm": 1.7494882345199585,
      "learning_rate": 4.467701219020667e-05,
      "loss": 2.5964,
      "step": 2495
    },
    {
      "epoch": 0.6912212683467184,
      "grad_norm": 1.7969138622283936,
      "learning_rate": 4.460354210791233e-05,
      "loss": 2.4509,
      "step": 2496
    },
    {
      "epoch": 0.6914981999446137,
      "grad_norm": 0.80321204662323,
      "learning_rate": 4.453011514154872e-05,
      "loss": 2.7233,
      "step": 2497
    },
    {
      "epoch": 0.691775131542509,
      "grad_norm": 1.1336066722869873,
      "learning_rate": 4.4456731348265344e-05,
      "loss": 2.6503,
      "step": 2498
    },
    {
      "epoch": 0.6920520631404043,
      "grad_norm": 2.478153944015503,
      "learning_rate": 4.438339078517785e-05,
      "loss": 3.31,
      "step": 2499
    },
    {
      "epoch": 0.6923289947382997,
      "grad_norm": 3.321981191635132,
      "learning_rate": 4.431009350936838e-05,
      "loss": 3.7714,
      "step": 2500
    },
    {
      "epoch": 0.692605926336195,
      "grad_norm": 1.3748323917388916,
      "learning_rate": 4.4236839577885345e-05,
      "loss": 2.969,
      "step": 2501
    },
    {
      "epoch": 0.6928828579340903,
      "grad_norm": 1.1666796207427979,
      "learning_rate": 4.416362904774345e-05,
      "loss": 2.4245,
      "step": 2502
    },
    {
      "epoch": 0.6931597895319856,
      "grad_norm": 1.0594323873519897,
      "learning_rate": 4.4090461975923566e-05,
      "loss": 2.6162,
      "step": 2503
    },
    {
      "epoch": 0.693436721129881,
      "grad_norm": 1.1525086164474487,
      "learning_rate": 4.401733841937279e-05,
      "loss": 2.4683,
      "step": 2504
    },
    {
      "epoch": 0.6937136527277762,
      "grad_norm": 2.1008377075195312,
      "learning_rate": 4.394425843500437e-05,
      "loss": 3.0293,
      "step": 2505
    },
    {
      "epoch": 0.6939905843256715,
      "grad_norm": 1.5007545948028564,
      "learning_rate": 4.387122207969748e-05,
      "loss": 2.9255,
      "step": 2506
    },
    {
      "epoch": 0.6942675159235668,
      "grad_norm": 1.5370593070983887,
      "learning_rate": 4.379822941029762e-05,
      "loss": 2.9777,
      "step": 2507
    },
    {
      "epoch": 0.6945444475214622,
      "grad_norm": 2.107943058013916,
      "learning_rate": 4.372528048361601e-05,
      "loss": 2.8037,
      "step": 2508
    },
    {
      "epoch": 0.6948213791193575,
      "grad_norm": 1.1811412572860718,
      "learning_rate": 4.3652375356429974e-05,
      "loss": 2.716,
      "step": 2509
    },
    {
      "epoch": 0.6950983107172528,
      "grad_norm": 1.3295549154281616,
      "learning_rate": 4.357951408548272e-05,
      "loss": 2.7155,
      "step": 2510
    },
    {
      "epoch": 0.6953752423151481,
      "grad_norm": 1.7516978979110718,
      "learning_rate": 4.350669672748332e-05,
      "loss": 2.7683,
      "step": 2511
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 0.7216670513153076,
      "learning_rate": 4.3433923339106666e-05,
      "loss": 2.6673,
      "step": 2512
    },
    {
      "epoch": 0.6959291055109388,
      "grad_norm": 1.3093624114990234,
      "learning_rate": 4.336119397699341e-05,
      "loss": 2.8426,
      "step": 2513
    },
    {
      "epoch": 0.6962060371088341,
      "grad_norm": 1.3083853721618652,
      "learning_rate": 4.328850869775001e-05,
      "loss": 2.4453,
      "step": 2514
    },
    {
      "epoch": 0.6964829687067294,
      "grad_norm": 1.5862066745758057,
      "learning_rate": 4.321586755794843e-05,
      "loss": 2.7283,
      "step": 2515
    },
    {
      "epoch": 0.6967599003046248,
      "grad_norm": 0.8611969351768494,
      "learning_rate": 4.314327061412656e-05,
      "loss": 2.6224,
      "step": 2516
    },
    {
      "epoch": 0.6970368319025201,
      "grad_norm": 1.286521553993225,
      "learning_rate": 4.307071792278764e-05,
      "loss": 2.7307,
      "step": 2517
    },
    {
      "epoch": 0.6973137635004154,
      "grad_norm": 1.4638444185256958,
      "learning_rate": 4.299820954040058e-05,
      "loss": 2.827,
      "step": 2518
    },
    {
      "epoch": 0.6975906950983107,
      "grad_norm": 1.2252277135849,
      "learning_rate": 4.292574552339981e-05,
      "loss": 2.5911,
      "step": 2519
    },
    {
      "epoch": 0.6978676266962061,
      "grad_norm": 1.9520236253738403,
      "learning_rate": 4.2853325928185236e-05,
      "loss": 2.8218,
      "step": 2520
    },
    {
      "epoch": 0.6981445582941014,
      "grad_norm": 1.7045632600784302,
      "learning_rate": 4.2780950811122056e-05,
      "loss": 2.809,
      "step": 2521
    },
    {
      "epoch": 0.6984214898919967,
      "grad_norm": 1.407806158065796,
      "learning_rate": 4.2708620228541105e-05,
      "loss": 3.1178,
      "step": 2522
    },
    {
      "epoch": 0.698698421489892,
      "grad_norm": 1.407745361328125,
      "learning_rate": 4.26363342367383e-05,
      "loss": 3.02,
      "step": 2523
    },
    {
      "epoch": 0.6989753530877874,
      "grad_norm": 1.6395858526229858,
      "learning_rate": 4.256409289197495e-05,
      "loss": 3.1189,
      "step": 2524
    },
    {
      "epoch": 0.6992522846856827,
      "grad_norm": 2.629124879837036,
      "learning_rate": 4.2491896250477734e-05,
      "loss": 3.7059,
      "step": 2525
    },
    {
      "epoch": 0.699529216283578,
      "grad_norm": 1.2712671756744385,
      "learning_rate": 4.241974436843834e-05,
      "loss": 2.6529,
      "step": 2526
    },
    {
      "epoch": 0.6998061478814733,
      "grad_norm": 1.899190902709961,
      "learning_rate": 4.234763730201373e-05,
      "loss": 2.6366,
      "step": 2527
    },
    {
      "epoch": 0.7000830794793685,
      "grad_norm": 1.3905671834945679,
      "learning_rate": 4.227557510732596e-05,
      "loss": 2.9294,
      "step": 2528
    },
    {
      "epoch": 0.7003600110772639,
      "grad_norm": 1.6453317403793335,
      "learning_rate": 4.2203557840462214e-05,
      "loss": 2.8451,
      "step": 2529
    },
    {
      "epoch": 0.7006369426751592,
      "grad_norm": 2.208395004272461,
      "learning_rate": 4.213158555747454e-05,
      "loss": 3.1687,
      "step": 2530
    },
    {
      "epoch": 0.7009138742730545,
      "grad_norm": 1.2020130157470703,
      "learning_rate": 4.2059658314380234e-05,
      "loss": 2.8466,
      "step": 2531
    },
    {
      "epoch": 0.7011908058709498,
      "grad_norm": 1.0927914381027222,
      "learning_rate": 4.1987776167161294e-05,
      "loss": 2.6193,
      "step": 2532
    },
    {
      "epoch": 0.7014677374688452,
      "grad_norm": 2.179379940032959,
      "learning_rate": 4.191593917176475e-05,
      "loss": 2.5691,
      "step": 2533
    },
    {
      "epoch": 0.7017446690667405,
      "grad_norm": 1.823103666305542,
      "learning_rate": 4.184414738410248e-05,
      "loss": 2.8192,
      "step": 2534
    },
    {
      "epoch": 0.7020216006646358,
      "grad_norm": 1.3351765871047974,
      "learning_rate": 4.177240086005113e-05,
      "loss": 2.732,
      "step": 2535
    },
    {
      "epoch": 0.7022985322625311,
      "grad_norm": 1.3181465864181519,
      "learning_rate": 4.170069965545217e-05,
      "loss": 2.7464,
      "step": 2536
    },
    {
      "epoch": 0.7025754638604265,
      "grad_norm": 1.428551197052002,
      "learning_rate": 4.162904382611174e-05,
      "loss": 2.5484,
      "step": 2537
    },
    {
      "epoch": 0.7028523954583218,
      "grad_norm": 1.301132321357727,
      "learning_rate": 4.155743342780077e-05,
      "loss": 2.7892,
      "step": 2538
    },
    {
      "epoch": 0.7031293270562171,
      "grad_norm": 2.054124593734741,
      "learning_rate": 4.148586851625461e-05,
      "loss": 2.8681,
      "step": 2539
    },
    {
      "epoch": 0.7034062586541124,
      "grad_norm": 1.7845115661621094,
      "learning_rate": 4.1414349147173514e-05,
      "loss": 2.9608,
      "step": 2540
    },
    {
      "epoch": 0.7036831902520078,
      "grad_norm": 2.0754706859588623,
      "learning_rate": 4.1342875376222015e-05,
      "loss": 3.1372,
      "step": 2541
    },
    {
      "epoch": 0.7039601218499031,
      "grad_norm": 1.381613850593567,
      "learning_rate": 4.12714472590293e-05,
      "loss": 2.8474,
      "step": 2542
    },
    {
      "epoch": 0.7042370534477984,
      "grad_norm": 1.4121874570846558,
      "learning_rate": 4.1200064851189e-05,
      "loss": 2.6797,
      "step": 2543
    },
    {
      "epoch": 0.7045139850456937,
      "grad_norm": 2.215932607650757,
      "learning_rate": 4.112872820825915e-05,
      "loss": 2.6889,
      "step": 2544
    },
    {
      "epoch": 0.7047909166435891,
      "grad_norm": 1.5485789775848389,
      "learning_rate": 4.105743738576219e-05,
      "loss": 2.772,
      "step": 2545
    },
    {
      "epoch": 0.7050678482414844,
      "grad_norm": 0.7619416117668152,
      "learning_rate": 4.0986192439184864e-05,
      "loss": 2.6577,
      "step": 2546
    },
    {
      "epoch": 0.7053447798393797,
      "grad_norm": 1.0292936563491821,
      "learning_rate": 4.09149934239783e-05,
      "loss": 2.9299,
      "step": 2547
    },
    {
      "epoch": 0.705621711437275,
      "grad_norm": 1.4957592487335205,
      "learning_rate": 4.0843840395557664e-05,
      "loss": 2.7241,
      "step": 2548
    },
    {
      "epoch": 0.7058986430351704,
      "grad_norm": 0.8321310877799988,
      "learning_rate": 4.077273340930263e-05,
      "loss": 2.9524,
      "step": 2549
    },
    {
      "epoch": 0.7061755746330657,
      "grad_norm": 2.9555342197418213,
      "learning_rate": 4.070167252055676e-05,
      "loss": 3.4832,
      "step": 2550
    },
    {
      "epoch": 0.7064525062309609,
      "grad_norm": 3.6347153186798096,
      "learning_rate": 4.063065778462792e-05,
      "loss": 3.2612,
      "step": 2551
    },
    {
      "epoch": 0.7067294378288562,
      "grad_norm": 1.2320882081985474,
      "learning_rate": 4.055968925678797e-05,
      "loss": 2.6501,
      "step": 2552
    },
    {
      "epoch": 0.7070063694267515,
      "grad_norm": 1.7431206703186035,
      "learning_rate": 4.0488766992272865e-05,
      "loss": 3.2253,
      "step": 2553
    },
    {
      "epoch": 0.7072833010246469,
      "grad_norm": 1.1208981275558472,
      "learning_rate": 4.041789104628241e-05,
      "loss": 2.9444,
      "step": 2554
    },
    {
      "epoch": 0.7075602326225422,
      "grad_norm": 0.957042396068573,
      "learning_rate": 4.034706147398061e-05,
      "loss": 2.7775,
      "step": 2555
    },
    {
      "epoch": 0.7078371642204375,
      "grad_norm": 1.6962882280349731,
      "learning_rate": 4.027627833049511e-05,
      "loss": 2.7378,
      "step": 2556
    },
    {
      "epoch": 0.7081140958183328,
      "grad_norm": 1.5339020490646362,
      "learning_rate": 4.020554167091756e-05,
      "loss": 2.6384,
      "step": 2557
    },
    {
      "epoch": 0.7083910274162282,
      "grad_norm": 1.851544737815857,
      "learning_rate": 4.0134851550303496e-05,
      "loss": 2.8828,
      "step": 2558
    },
    {
      "epoch": 0.7086679590141235,
      "grad_norm": 0.9925166964530945,
      "learning_rate": 4.006420802367205e-05,
      "loss": 2.5696,
      "step": 2559
    },
    {
      "epoch": 0.7089448906120188,
      "grad_norm": 1.440911889076233,
      "learning_rate": 3.999361114600621e-05,
      "loss": 2.6636,
      "step": 2560
    },
    {
      "epoch": 0.7092218222099141,
      "grad_norm": 1.3923132419586182,
      "learning_rate": 3.9923060972252655e-05,
      "loss": 2.7548,
      "step": 2561
    },
    {
      "epoch": 0.7094987538078095,
      "grad_norm": 1.1634244918823242,
      "learning_rate": 3.9852557557321714e-05,
      "loss": 3.009,
      "step": 2562
    },
    {
      "epoch": 0.7097756854057048,
      "grad_norm": 1.1615941524505615,
      "learning_rate": 3.978210095608718e-05,
      "loss": 2.8798,
      "step": 2563
    },
    {
      "epoch": 0.7100526170036001,
      "grad_norm": 2.0248026847839355,
      "learning_rate": 3.971169122338668e-05,
      "loss": 2.9412,
      "step": 2564
    },
    {
      "epoch": 0.7103295486014954,
      "grad_norm": 1.8589743375778198,
      "learning_rate": 3.964132841402112e-05,
      "loss": 2.8227,
      "step": 2565
    },
    {
      "epoch": 0.7106064801993908,
      "grad_norm": 1.61398184299469,
      "learning_rate": 3.9571012582754985e-05,
      "loss": 2.7969,
      "step": 2566
    },
    {
      "epoch": 0.7108834117972861,
      "grad_norm": 1.543515682220459,
      "learning_rate": 3.9500743784316206e-05,
      "loss": 3.1315,
      "step": 2567
    },
    {
      "epoch": 0.7111603433951814,
      "grad_norm": 1.421244502067566,
      "learning_rate": 3.9430522073396084e-05,
      "loss": 2.7719,
      "step": 2568
    },
    {
      "epoch": 0.7114372749930767,
      "grad_norm": 1.5355627536773682,
      "learning_rate": 3.936034750464927e-05,
      "loss": 2.613,
      "step": 2569
    },
    {
      "epoch": 0.7117142065909721,
      "grad_norm": 1.0952552556991577,
      "learning_rate": 3.9290220132693756e-05,
      "loss": 2.6977,
      "step": 2570
    },
    {
      "epoch": 0.7119911381888674,
      "grad_norm": 1.621482253074646,
      "learning_rate": 3.922014001211077e-05,
      "loss": 2.7291,
      "step": 2571
    },
    {
      "epoch": 0.7122680697867627,
      "grad_norm": 2.1047937870025635,
      "learning_rate": 3.91501071974447e-05,
      "loss": 2.8873,
      "step": 2572
    },
    {
      "epoch": 0.712545001384658,
      "grad_norm": 1.1262298822402954,
      "learning_rate": 3.908012174320329e-05,
      "loss": 2.9252,
      "step": 2573
    },
    {
      "epoch": 0.7128219329825534,
      "grad_norm": 1.6226671934127808,
      "learning_rate": 3.901018370385724e-05,
      "loss": 2.7067,
      "step": 2574
    },
    {
      "epoch": 0.7130988645804486,
      "grad_norm": 2.3573596477508545,
      "learning_rate": 3.894029313384043e-05,
      "loss": 3.3536,
      "step": 2575
    },
    {
      "epoch": 0.7133757961783439,
      "grad_norm": 2.8473141193389893,
      "learning_rate": 3.88704500875498e-05,
      "loss": 3.2805,
      "step": 2576
    },
    {
      "epoch": 0.7136527277762392,
      "grad_norm": 1.263624668121338,
      "learning_rate": 3.880065461934529e-05,
      "loss": 2.7721,
      "step": 2577
    },
    {
      "epoch": 0.7139296593741346,
      "grad_norm": 1.3694475889205933,
      "learning_rate": 3.8730906783549795e-05,
      "loss": 2.8154,
      "step": 2578
    },
    {
      "epoch": 0.7142065909720299,
      "grad_norm": 0.9628450870513916,
      "learning_rate": 3.866120663444914e-05,
      "loss": 2.8262,
      "step": 2579
    },
    {
      "epoch": 0.7144835225699252,
      "grad_norm": 1.5288922786712646,
      "learning_rate": 3.85915542262921e-05,
      "loss": 2.9611,
      "step": 2580
    },
    {
      "epoch": 0.7147604541678205,
      "grad_norm": 1.7090123891830444,
      "learning_rate": 3.852194961329013e-05,
      "loss": 2.7685,
      "step": 2581
    },
    {
      "epoch": 0.7150373857657158,
      "grad_norm": 1.9330387115478516,
      "learning_rate": 3.8452392849617715e-05,
      "loss": 2.9505,
      "step": 2582
    },
    {
      "epoch": 0.7153143173636112,
      "grad_norm": 1.3753412961959839,
      "learning_rate": 3.838288398941187e-05,
      "loss": 2.9888,
      "step": 2583
    },
    {
      "epoch": 0.7155912489615065,
      "grad_norm": 1.3932002782821655,
      "learning_rate": 3.831342308677247e-05,
      "loss": 2.7885,
      "step": 2584
    },
    {
      "epoch": 0.7158681805594018,
      "grad_norm": 1.121376395225525,
      "learning_rate": 3.824401019576203e-05,
      "loss": 2.8153,
      "step": 2585
    },
    {
      "epoch": 0.7161451121572971,
      "grad_norm": 1.4889800548553467,
      "learning_rate": 3.817464537040571e-05,
      "loss": 2.7574,
      "step": 2586
    },
    {
      "epoch": 0.7164220437551925,
      "grad_norm": 0.9660783410072327,
      "learning_rate": 3.810532866469114e-05,
      "loss": 2.5652,
      "step": 2587
    },
    {
      "epoch": 0.7166989753530878,
      "grad_norm": 0.9699069261550903,
      "learning_rate": 3.803606013256871e-05,
      "loss": 2.8056,
      "step": 2588
    },
    {
      "epoch": 0.7169759069509831,
      "grad_norm": 0.8376581072807312,
      "learning_rate": 3.7966839827951196e-05,
      "loss": 2.7144,
      "step": 2589
    },
    {
      "epoch": 0.7172528385488784,
      "grad_norm": 1.4266955852508545,
      "learning_rate": 3.7897667804713756e-05,
      "loss": 2.9554,
      "step": 2590
    },
    {
      "epoch": 0.7175297701467738,
      "grad_norm": 1.1173572540283203,
      "learning_rate": 3.7828544116694176e-05,
      "loss": 2.5609,
      "step": 2591
    },
    {
      "epoch": 0.7178067017446691,
      "grad_norm": 1.240865707397461,
      "learning_rate": 3.775946881769241e-05,
      "loss": 2.7688,
      "step": 2592
    },
    {
      "epoch": 0.7180836333425644,
      "grad_norm": 1.3590009212493896,
      "learning_rate": 3.769044196147088e-05,
      "loss": 2.9172,
      "step": 2593
    },
    {
      "epoch": 0.7183605649404597,
      "grad_norm": 1.9907031059265137,
      "learning_rate": 3.762146360175427e-05,
      "loss": 2.5582,
      "step": 2594
    },
    {
      "epoch": 0.7186374965383551,
      "grad_norm": 1.6450612545013428,
      "learning_rate": 3.755253379222955e-05,
      "loss": 2.8313,
      "step": 2595
    },
    {
      "epoch": 0.7189144281362504,
      "grad_norm": 1.57441246509552,
      "learning_rate": 3.748365258654577e-05,
      "loss": 2.6225,
      "step": 2596
    },
    {
      "epoch": 0.7191913597341457,
      "grad_norm": 0.8331193327903748,
      "learning_rate": 3.741482003831439e-05,
      "loss": 2.6939,
      "step": 2597
    },
    {
      "epoch": 0.7194682913320409,
      "grad_norm": 0.8779714703559875,
      "learning_rate": 3.734603620110876e-05,
      "loss": 2.6911,
      "step": 2598
    },
    {
      "epoch": 0.7197452229299363,
      "grad_norm": 1.293872356414795,
      "learning_rate": 3.727730112846444e-05,
      "loss": 2.7169,
      "step": 2599
    },
    {
      "epoch": 0.7200221545278316,
      "grad_norm": 2.3049585819244385,
      "learning_rate": 3.720861487387901e-05,
      "loss": 3.6294,
      "step": 2600
    },
    {
      "epoch": 0.7202990861257269,
      "grad_norm": 1.5383538007736206,
      "learning_rate": 3.7139977490812087e-05,
      "loss": 2.8187,
      "step": 2601
    },
    {
      "epoch": 0.7205760177236222,
      "grad_norm": 1.1297353506088257,
      "learning_rate": 3.707138903268519e-05,
      "loss": 2.7199,
      "step": 2602
    },
    {
      "epoch": 0.7208529493215176,
      "grad_norm": 1.559241771697998,
      "learning_rate": 3.7002849552881815e-05,
      "loss": 2.7379,
      "step": 2603
    },
    {
      "epoch": 0.7211298809194129,
      "grad_norm": 1.2448389530181885,
      "learning_rate": 3.693435910474732e-05,
      "loss": 2.7702,
      "step": 2604
    },
    {
      "epoch": 0.7214068125173082,
      "grad_norm": 1.3755321502685547,
      "learning_rate": 3.686591774158882e-05,
      "loss": 2.8438,
      "step": 2605
    },
    {
      "epoch": 0.7216837441152035,
      "grad_norm": 1.7060540914535522,
      "learning_rate": 3.679752551667541e-05,
      "loss": 3.1769,
      "step": 2606
    },
    {
      "epoch": 0.7219606757130989,
      "grad_norm": 1.2973065376281738,
      "learning_rate": 3.672918248323774e-05,
      "loss": 2.9313,
      "step": 2607
    },
    {
      "epoch": 0.7222376073109942,
      "grad_norm": 0.9247957468032837,
      "learning_rate": 3.666088869446831e-05,
      "loss": 2.7732,
      "step": 2608
    },
    {
      "epoch": 0.7225145389088895,
      "grad_norm": 0.8578128814697266,
      "learning_rate": 3.659264420352122e-05,
      "loss": 2.7906,
      "step": 2609
    },
    {
      "epoch": 0.7227914705067848,
      "grad_norm": 1.6201186180114746,
      "learning_rate": 3.652444906351223e-05,
      "loss": 2.8034,
      "step": 2610
    },
    {
      "epoch": 0.7230684021046802,
      "grad_norm": 1.3790613412857056,
      "learning_rate": 3.645630332751871e-05,
      "loss": 2.8627,
      "step": 2611
    },
    {
      "epoch": 0.7233453337025755,
      "grad_norm": 1.420621395111084,
      "learning_rate": 3.638820704857954e-05,
      "loss": 2.9343,
      "step": 2612
    },
    {
      "epoch": 0.7236222653004708,
      "grad_norm": 1.4299603700637817,
      "learning_rate": 3.632016027969514e-05,
      "loss": 2.7154,
      "step": 2613
    },
    {
      "epoch": 0.7238991968983661,
      "grad_norm": 1.3731162548065186,
      "learning_rate": 3.6252163073827294e-05,
      "loss": 2.8569,
      "step": 2614
    },
    {
      "epoch": 0.7241761284962615,
      "grad_norm": 0.7518453598022461,
      "learning_rate": 3.618421548389942e-05,
      "loss": 2.7739,
      "step": 2615
    },
    {
      "epoch": 0.7244530600941568,
      "grad_norm": 0.7167025208473206,
      "learning_rate": 3.6116317562796085e-05,
      "loss": 2.7477,
      "step": 2616
    },
    {
      "epoch": 0.7247299916920521,
      "grad_norm": 1.2134454250335693,
      "learning_rate": 3.6048469363363345e-05,
      "loss": 2.4328,
      "step": 2617
    },
    {
      "epoch": 0.7250069232899474,
      "grad_norm": 10.561895370483398,
      "learning_rate": 3.5980670938408514e-05,
      "loss": 2.8107,
      "step": 2618
    },
    {
      "epoch": 0.7252838548878427,
      "grad_norm": 1.1714071035385132,
      "learning_rate": 3.5912922340700206e-05,
      "loss": 2.6758,
      "step": 2619
    },
    {
      "epoch": 0.7255607864857381,
      "grad_norm": 1.4013359546661377,
      "learning_rate": 3.5845223622968104e-05,
      "loss": 2.8461,
      "step": 2620
    },
    {
      "epoch": 0.7258377180836333,
      "grad_norm": 1.1809953451156616,
      "learning_rate": 3.577757483790329e-05,
      "loss": 2.8659,
      "step": 2621
    },
    {
      "epoch": 0.7261146496815286,
      "grad_norm": 1.393582820892334,
      "learning_rate": 3.5709976038157875e-05,
      "loss": 2.6336,
      "step": 2622
    },
    {
      "epoch": 0.7263915812794239,
      "grad_norm": 1.4034349918365479,
      "learning_rate": 3.564242727634499e-05,
      "loss": 2.73,
      "step": 2623
    },
    {
      "epoch": 0.7266685128773193,
      "grad_norm": 1.565304160118103,
      "learning_rate": 3.557492860503893e-05,
      "loss": 3.0669,
      "step": 2624
    },
    {
      "epoch": 0.7269454444752146,
      "grad_norm": 2.7715139389038086,
      "learning_rate": 3.550748007677497e-05,
      "loss": 3.332,
      "step": 2625
    },
    {
      "epoch": 0.7272223760731099,
      "grad_norm": 1.4593929052352905,
      "learning_rate": 3.544008174404938e-05,
      "loss": 2.7901,
      "step": 2626
    },
    {
      "epoch": 0.7274993076710052,
      "grad_norm": 1.7429101467132568,
      "learning_rate": 3.5372733659319256e-05,
      "loss": 2.5106,
      "step": 2627
    },
    {
      "epoch": 0.7277762392689006,
      "grad_norm": 1.2671338319778442,
      "learning_rate": 3.530543587500278e-05,
      "loss": 2.5782,
      "step": 2628
    },
    {
      "epoch": 0.7280531708667959,
      "grad_norm": 2.224750280380249,
      "learning_rate": 3.5238188443478795e-05,
      "loss": 2.864,
      "step": 2629
    },
    {
      "epoch": 0.7283301024646912,
      "grad_norm": 1.1255524158477783,
      "learning_rate": 3.517099141708703e-05,
      "loss": 2.4131,
      "step": 2630
    },
    {
      "epoch": 0.7286070340625865,
      "grad_norm": 1.5960115194320679,
      "learning_rate": 3.510384484812802e-05,
      "loss": 2.8515,
      "step": 2631
    },
    {
      "epoch": 0.7288839656604819,
      "grad_norm": 2.0236940383911133,
      "learning_rate": 3.5036748788862974e-05,
      "loss": 2.8617,
      "step": 2632
    },
    {
      "epoch": 0.7291608972583772,
      "grad_norm": 1.7315971851348877,
      "learning_rate": 3.49697032915138e-05,
      "loss": 2.6889,
      "step": 2633
    },
    {
      "epoch": 0.7294378288562725,
      "grad_norm": 1.8834228515625,
      "learning_rate": 3.4902708408263066e-05,
      "loss": 2.6978,
      "step": 2634
    },
    {
      "epoch": 0.7297147604541678,
      "grad_norm": 0.9408883452415466,
      "learning_rate": 3.483576419125397e-05,
      "loss": 2.5433,
      "step": 2635
    },
    {
      "epoch": 0.7299916920520632,
      "grad_norm": 1.3076893091201782,
      "learning_rate": 3.4768870692590147e-05,
      "loss": 3.1385,
      "step": 2636
    },
    {
      "epoch": 0.7302686236499585,
      "grad_norm": 1.5100501775741577,
      "learning_rate": 3.470202796433598e-05,
      "loss": 2.6895,
      "step": 2637
    },
    {
      "epoch": 0.7305455552478538,
      "grad_norm": 1.9968551397323608,
      "learning_rate": 3.463523605851613e-05,
      "loss": 2.5876,
      "step": 2638
    },
    {
      "epoch": 0.7308224868457491,
      "grad_norm": 1.1709613800048828,
      "learning_rate": 3.45684950271158e-05,
      "loss": 2.5664,
      "step": 2639
    },
    {
      "epoch": 0.7310994184436445,
      "grad_norm": 0.9514749050140381,
      "learning_rate": 3.450180492208058e-05,
      "loss": 2.6495,
      "step": 2640
    },
    {
      "epoch": 0.7313763500415398,
      "grad_norm": 1.7250149250030518,
      "learning_rate": 3.443516579531644e-05,
      "loss": 2.9695,
      "step": 2641
    },
    {
      "epoch": 0.7316532816394351,
      "grad_norm": 2.7542314529418945,
      "learning_rate": 3.436857769868963e-05,
      "loss": 2.8169,
      "step": 2642
    },
    {
      "epoch": 0.7319302132373304,
      "grad_norm": 1.302791953086853,
      "learning_rate": 3.4302040684026736e-05,
      "loss": 3.0475,
      "step": 2643
    },
    {
      "epoch": 0.7322071448352258,
      "grad_norm": 1.1405366659164429,
      "learning_rate": 3.423555480311457e-05,
      "loss": 2.685,
      "step": 2644
    },
    {
      "epoch": 0.732484076433121,
      "grad_norm": 1.650712251663208,
      "learning_rate": 3.416912010770005e-05,
      "loss": 2.65,
      "step": 2645
    },
    {
      "epoch": 0.7327610080310163,
      "grad_norm": 1.5690234899520874,
      "learning_rate": 3.410273664949045e-05,
      "loss": 2.4833,
      "step": 2646
    },
    {
      "epoch": 0.7330379396289116,
      "grad_norm": 0.912711501121521,
      "learning_rate": 3.403640448015298e-05,
      "loss": 2.6781,
      "step": 2647
    },
    {
      "epoch": 0.7333148712268069,
      "grad_norm": 1.6189790964126587,
      "learning_rate": 3.3970123651315046e-05,
      "loss": 2.6668,
      "step": 2648
    },
    {
      "epoch": 0.7335918028247023,
      "grad_norm": 1.7619707584381104,
      "learning_rate": 3.3903894214564026e-05,
      "loss": 2.7105,
      "step": 2649
    },
    {
      "epoch": 0.7338687344225976,
      "grad_norm": 3.2235915660858154,
      "learning_rate": 3.383771622144738e-05,
      "loss": 3.5835,
      "step": 2650
    },
    {
      "epoch": 0.7341456660204929,
      "grad_norm": 0.9661833047866821,
      "learning_rate": 3.3771589723472365e-05,
      "loss": 2.617,
      "step": 2651
    },
    {
      "epoch": 0.7344225976183882,
      "grad_norm": 1.3732293844223022,
      "learning_rate": 3.3705514772106415e-05,
      "loss": 2.8641,
      "step": 2652
    },
    {
      "epoch": 0.7346995292162836,
      "grad_norm": 0.6539769768714905,
      "learning_rate": 3.363949141877659e-05,
      "loss": 2.7455,
      "step": 2653
    },
    {
      "epoch": 0.7349764608141789,
      "grad_norm": 2.1681718826293945,
      "learning_rate": 3.3573519714869914e-05,
      "loss": 2.5595,
      "step": 2654
    },
    {
      "epoch": 0.7352533924120742,
      "grad_norm": 1.89735746383667,
      "learning_rate": 3.3507599711733286e-05,
      "loss": 2.5669,
      "step": 2655
    },
    {
      "epoch": 0.7355303240099695,
      "grad_norm": 1.5284838676452637,
      "learning_rate": 3.344173146067317e-05,
      "loss": 2.6923,
      "step": 2656
    },
    {
      "epoch": 0.7358072556078649,
      "grad_norm": 1.8220961093902588,
      "learning_rate": 3.337591501295592e-05,
      "loss": 2.6526,
      "step": 2657
    },
    {
      "epoch": 0.7360841872057602,
      "grad_norm": 1.5740773677825928,
      "learning_rate": 3.3310150419807474e-05,
      "loss": 2.8735,
      "step": 2658
    },
    {
      "epoch": 0.7363611188036555,
      "grad_norm": 1.1814055442810059,
      "learning_rate": 3.324443773241349e-05,
      "loss": 2.6313,
      "step": 2659
    },
    {
      "epoch": 0.7366380504015508,
      "grad_norm": 1.3792455196380615,
      "learning_rate": 3.317877700191909e-05,
      "loss": 2.7861,
      "step": 2660
    },
    {
      "epoch": 0.7369149819994462,
      "grad_norm": 2.363563299179077,
      "learning_rate": 3.311316827942918e-05,
      "loss": 2.2229,
      "step": 2661
    },
    {
      "epoch": 0.7371919135973415,
      "grad_norm": 1.0998777151107788,
      "learning_rate": 3.3047611616007945e-05,
      "loss": 2.8149,
      "step": 2662
    },
    {
      "epoch": 0.7374688451952368,
      "grad_norm": 1.206363320350647,
      "learning_rate": 3.2982107062679215e-05,
      "loss": 2.762,
      "step": 2663
    },
    {
      "epoch": 0.7377457767931321,
      "grad_norm": 0.9043403267860413,
      "learning_rate": 3.291665467042618e-05,
      "loss": 2.6419,
      "step": 2664
    },
    {
      "epoch": 0.7380227083910275,
      "grad_norm": 2.09918212890625,
      "learning_rate": 3.2851254490191495e-05,
      "loss": 2.7888,
      "step": 2665
    },
    {
      "epoch": 0.7382996399889228,
      "grad_norm": 1.7726424932479858,
      "learning_rate": 3.278590657287713e-05,
      "loss": 3.0603,
      "step": 2666
    },
    {
      "epoch": 0.7385765715868181,
      "grad_norm": 1.3921743631362915,
      "learning_rate": 3.272061096934439e-05,
      "loss": 2.6677,
      "step": 2667
    },
    {
      "epoch": 0.7388535031847133,
      "grad_norm": 1.326472520828247,
      "learning_rate": 3.26553677304139e-05,
      "loss": 2.7886,
      "step": 2668
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 1.3895070552825928,
      "learning_rate": 3.25901769068654e-05,
      "loss": 2.5025,
      "step": 2669
    },
    {
      "epoch": 0.739407366380504,
      "grad_norm": 1.382885217666626,
      "learning_rate": 3.252503854943806e-05,
      "loss": 2.7819,
      "step": 2670
    },
    {
      "epoch": 0.7396842979783993,
      "grad_norm": 1.269974946975708,
      "learning_rate": 3.2459952708829974e-05,
      "loss": 2.7521,
      "step": 2671
    },
    {
      "epoch": 0.7399612295762946,
      "grad_norm": 1.6350727081298828,
      "learning_rate": 3.2394919435698524e-05,
      "loss": 2.5446,
      "step": 2672
    },
    {
      "epoch": 0.74023816117419,
      "grad_norm": 1.1815497875213623,
      "learning_rate": 3.232993878066011e-05,
      "loss": 2.8673,
      "step": 2673
    },
    {
      "epoch": 0.7405150927720853,
      "grad_norm": 1.97178053855896,
      "learning_rate": 3.2265010794290195e-05,
      "loss": 2.8683,
      "step": 2674
    },
    {
      "epoch": 0.7407920243699806,
      "grad_norm": 4.066892147064209,
      "learning_rate": 3.220013552712325e-05,
      "loss": 4.1637,
      "step": 2675
    },
    {
      "epoch": 0.7410689559678759,
      "grad_norm": 1.6811068058013916,
      "learning_rate": 3.213531302965271e-05,
      "loss": 2.9421,
      "step": 2676
    },
    {
      "epoch": 0.7413458875657712,
      "grad_norm": 1.3643163442611694,
      "learning_rate": 3.2070543352330986e-05,
      "loss": 2.6982,
      "step": 2677
    },
    {
      "epoch": 0.7416228191636666,
      "grad_norm": 0.915283739566803,
      "learning_rate": 3.2005826545569214e-05,
      "loss": 2.9205,
      "step": 2678
    },
    {
      "epoch": 0.7418997507615619,
      "grad_norm": 1.3957204818725586,
      "learning_rate": 3.1941162659737647e-05,
      "loss": 2.5736,
      "step": 2679
    },
    {
      "epoch": 0.7421766823594572,
      "grad_norm": 1.4447449445724487,
      "learning_rate": 3.1876551745165084e-05,
      "loss": 2.4436,
      "step": 2680
    },
    {
      "epoch": 0.7424536139573525,
      "grad_norm": 1.5328930616378784,
      "learning_rate": 3.1811993852139255e-05,
      "loss": 2.7714,
      "step": 2681
    },
    {
      "epoch": 0.7427305455552479,
      "grad_norm": 1.225975751876831,
      "learning_rate": 3.174748903090658e-05,
      "loss": 2.7398,
      "step": 2682
    },
    {
      "epoch": 0.7430074771531432,
      "grad_norm": 2.1260430812835693,
      "learning_rate": 3.1683037331672206e-05,
      "loss": 3.0236,
      "step": 2683
    },
    {
      "epoch": 0.7432844087510385,
      "grad_norm": 1.9688694477081299,
      "learning_rate": 3.16186388045998e-05,
      "loss": 2.9523,
      "step": 2684
    },
    {
      "epoch": 0.7435613403489338,
      "grad_norm": 2.151796340942383,
      "learning_rate": 3.1554293499811826e-05,
      "loss": 2.4532,
      "step": 2685
    },
    {
      "epoch": 0.7438382719468292,
      "grad_norm": 1.460901141166687,
      "learning_rate": 3.149000146738927e-05,
      "loss": 2.7071,
      "step": 2686
    },
    {
      "epoch": 0.7441152035447245,
      "grad_norm": 1.6640523672103882,
      "learning_rate": 3.142576275737151e-05,
      "loss": 3.0941,
      "step": 2687
    },
    {
      "epoch": 0.7443921351426198,
      "grad_norm": 1.4269442558288574,
      "learning_rate": 3.136157741975668e-05,
      "loss": 2.5218,
      "step": 2688
    },
    {
      "epoch": 0.7446690667405151,
      "grad_norm": 2.3087620735168457,
      "learning_rate": 3.129744550450113e-05,
      "loss": 2.9008,
      "step": 2689
    },
    {
      "epoch": 0.7449459983384105,
      "grad_norm": 1.1492661237716675,
      "learning_rate": 3.123336706151978e-05,
      "loss": 2.8982,
      "step": 2690
    },
    {
      "epoch": 0.7452229299363057,
      "grad_norm": 2.044564723968506,
      "learning_rate": 3.1169342140685865e-05,
      "loss": 3.0544,
      "step": 2691
    },
    {
      "epoch": 0.745499861534201,
      "grad_norm": 2.40684175491333,
      "learning_rate": 3.110537079183104e-05,
      "loss": 2.7255,
      "step": 2692
    },
    {
      "epoch": 0.7457767931320963,
      "grad_norm": 1.2087938785552979,
      "learning_rate": 3.104145306474507e-05,
      "loss": 2.635,
      "step": 2693
    },
    {
      "epoch": 0.7460537247299917,
      "grad_norm": 1.0230350494384766,
      "learning_rate": 3.09775890091763e-05,
      "loss": 2.5559,
      "step": 2694
    },
    {
      "epoch": 0.746330656327887,
      "grad_norm": 1.2716301679611206,
      "learning_rate": 3.091377867483101e-05,
      "loss": 2.6127,
      "step": 2695
    },
    {
      "epoch": 0.7466075879257823,
      "grad_norm": 1.2278804779052734,
      "learning_rate": 3.08500221113738e-05,
      "loss": 2.8742,
      "step": 2696
    },
    {
      "epoch": 0.7468845195236776,
      "grad_norm": 1.224388599395752,
      "learning_rate": 3.078631936842742e-05,
      "loss": 2.9862,
      "step": 2697
    },
    {
      "epoch": 0.747161451121573,
      "grad_norm": 0.8031061291694641,
      "learning_rate": 3.0722670495572714e-05,
      "loss": 2.8578,
      "step": 2698
    },
    {
      "epoch": 0.7474383827194683,
      "grad_norm": 1.8247153759002686,
      "learning_rate": 3.065907554234858e-05,
      "loss": 2.6747,
      "step": 2699
    },
    {
      "epoch": 0.7477153143173636,
      "grad_norm": 3.208592414855957,
      "learning_rate": 3.059553455825198e-05,
      "loss": 3.7767,
      "step": 2700
    },
    {
      "epoch": 0.7479922459152589,
      "grad_norm": 1.3665844202041626,
      "learning_rate": 3.0532047592737875e-05,
      "loss": 2.7423,
      "step": 2701
    },
    {
      "epoch": 0.7482691775131542,
      "grad_norm": 1.562633991241455,
      "learning_rate": 3.046861469521909e-05,
      "loss": 2.8209,
      "step": 2702
    },
    {
      "epoch": 0.7485461091110496,
      "grad_norm": 0.8740465044975281,
      "learning_rate": 3.0405235915066553e-05,
      "loss": 2.8583,
      "step": 2703
    },
    {
      "epoch": 0.7488230407089449,
      "grad_norm": 1.4273957014083862,
      "learning_rate": 3.034191130160887e-05,
      "loss": 2.9819,
      "step": 2704
    },
    {
      "epoch": 0.7490999723068402,
      "grad_norm": 1.8278026580810547,
      "learning_rate": 3.0278640904132625e-05,
      "loss": 2.9547,
      "step": 2705
    },
    {
      "epoch": 0.7493769039047355,
      "grad_norm": 1.4489752054214478,
      "learning_rate": 3.021542477188213e-05,
      "loss": 2.5978,
      "step": 2706
    },
    {
      "epoch": 0.7496538355026309,
      "grad_norm": 1.9576112031936646,
      "learning_rate": 3.015226295405952e-05,
      "loss": 2.6865,
      "step": 2707
    },
    {
      "epoch": 0.7499307671005262,
      "grad_norm": 0.8638213276863098,
      "learning_rate": 3.0089155499824605e-05,
      "loss": 2.599,
      "step": 2708
    },
    {
      "epoch": 0.7502076986984215,
      "grad_norm": 1.1353836059570312,
      "learning_rate": 3.0026102458294924e-05,
      "loss": 2.9723,
      "step": 2709
    },
    {
      "epoch": 0.7502076986984215,
      "eval_loss": 0.3528745770454407,
      "eval_runtime": 804.2518,
      "eval_samples_per_second": 7.562,
      "eval_steps_per_second": 1.891,
      "step": 2709
    },
    {
      "epoch": 0.7504846302963168,
      "grad_norm": 1.2876255512237549,
      "learning_rate": 2.9963103878545662e-05,
      "loss": 2.6112,
      "step": 2710
    },
    {
      "epoch": 0.7507615618942122,
      "grad_norm": 1.2612682580947876,
      "learning_rate": 2.9900159809609517e-05,
      "loss": 2.7906,
      "step": 2711
    },
    {
      "epoch": 0.7510384934921075,
      "grad_norm": 1.1340705156326294,
      "learning_rate": 2.983727030047696e-05,
      "loss": 2.8441,
      "step": 2712
    },
    {
      "epoch": 0.7513154250900028,
      "grad_norm": 1.7764126062393188,
      "learning_rate": 2.9774435400095802e-05,
      "loss": 3.2773,
      "step": 2713
    },
    {
      "epoch": 0.7515923566878981,
      "grad_norm": 1.3928422927856445,
      "learning_rate": 2.9711655157371443e-05,
      "loss": 2.603,
      "step": 2714
    },
    {
      "epoch": 0.7518692882857934,
      "grad_norm": 1.1370984315872192,
      "learning_rate": 2.9648929621166742e-05,
      "loss": 2.827,
      "step": 2715
    },
    {
      "epoch": 0.7521462198836887,
      "grad_norm": 1.6648856401443481,
      "learning_rate": 2.9586258840301983e-05,
      "loss": 2.8053,
      "step": 2716
    },
    {
      "epoch": 0.752423151481584,
      "grad_norm": 1.1349536180496216,
      "learning_rate": 2.9523642863554746e-05,
      "loss": 3.0446,
      "step": 2717
    },
    {
      "epoch": 0.7527000830794793,
      "grad_norm": 1.5090161561965942,
      "learning_rate": 2.9461081739660103e-05,
      "loss": 2.8436,
      "step": 2718
    },
    {
      "epoch": 0.7529770146773747,
      "grad_norm": 1.2340972423553467,
      "learning_rate": 2.9398575517310355e-05,
      "loss": 3.0514,
      "step": 2719
    },
    {
      "epoch": 0.75325394627527,
      "grad_norm": 0.8456771969795227,
      "learning_rate": 2.9336124245154995e-05,
      "loss": 2.5057,
      "step": 2720
    },
    {
      "epoch": 0.7535308778731653,
      "grad_norm": 1.490369439125061,
      "learning_rate": 2.9273727971800935e-05,
      "loss": 2.7719,
      "step": 2721
    },
    {
      "epoch": 0.7538078094710606,
      "grad_norm": 1.4499380588531494,
      "learning_rate": 2.9211386745812084e-05,
      "loss": 2.5641,
      "step": 2722
    },
    {
      "epoch": 0.754084741068956,
      "grad_norm": 1.4747735261917114,
      "learning_rate": 2.9149100615709634e-05,
      "loss": 3.0209,
      "step": 2723
    },
    {
      "epoch": 0.7543616726668513,
      "grad_norm": 1.836558222770691,
      "learning_rate": 2.9086869629971836e-05,
      "loss": 2.9703,
      "step": 2724
    },
    {
      "epoch": 0.7546386042647466,
      "grad_norm": 2.954867124557495,
      "learning_rate": 2.902469383703409e-05,
      "loss": 3.6567,
      "step": 2725
    },
    {
      "epoch": 0.7549155358626419,
      "grad_norm": 1.0744221210479736,
      "learning_rate": 2.8962573285288695e-05,
      "loss": 2.8091,
      "step": 2726
    },
    {
      "epoch": 0.7551924674605373,
      "grad_norm": 0.9103502035140991,
      "learning_rate": 2.8900508023085172e-05,
      "loss": 2.6369,
      "step": 2727
    },
    {
      "epoch": 0.7554693990584326,
      "grad_norm": 1.605891227722168,
      "learning_rate": 2.8838498098729784e-05,
      "loss": 3.0776,
      "step": 2728
    },
    {
      "epoch": 0.7557463306563279,
      "grad_norm": 1.0748717784881592,
      "learning_rate": 2.8776543560485857e-05,
      "loss": 2.7977,
      "step": 2729
    },
    {
      "epoch": 0.7560232622542232,
      "grad_norm": 1.3585500717163086,
      "learning_rate": 2.8714644456573593e-05,
      "loss": 2.8631,
      "step": 2730
    },
    {
      "epoch": 0.7563001938521186,
      "grad_norm": 1.102792739868164,
      "learning_rate": 2.8652800835170035e-05,
      "loss": 2.6547,
      "step": 2731
    },
    {
      "epoch": 0.7565771254500139,
      "grad_norm": 1.1205133199691772,
      "learning_rate": 2.8591012744409018e-05,
      "loss": 2.8023,
      "step": 2732
    },
    {
      "epoch": 0.7568540570479092,
      "grad_norm": 1.9962104558944702,
      "learning_rate": 2.8529280232381206e-05,
      "loss": 2.9932,
      "step": 2733
    },
    {
      "epoch": 0.7571309886458045,
      "grad_norm": 2.3501553535461426,
      "learning_rate": 2.8467603347133997e-05,
      "loss": 2.86,
      "step": 2734
    },
    {
      "epoch": 0.7574079202436999,
      "grad_norm": 1.4866106510162354,
      "learning_rate": 2.8405982136671395e-05,
      "loss": 2.7031,
      "step": 2735
    },
    {
      "epoch": 0.7576848518415952,
      "grad_norm": 1.0783441066741943,
      "learning_rate": 2.8344416648954265e-05,
      "loss": 2.6901,
      "step": 2736
    },
    {
      "epoch": 0.7579617834394905,
      "grad_norm": 1.2151702642440796,
      "learning_rate": 2.82829069318999e-05,
      "loss": 2.826,
      "step": 2737
    },
    {
      "epoch": 0.7582387150373857,
      "grad_norm": 1.172462821006775,
      "learning_rate": 2.8221453033382306e-05,
      "loss": 2.7353,
      "step": 2738
    },
    {
      "epoch": 0.758515646635281,
      "grad_norm": 1.1474409103393555,
      "learning_rate": 2.816005500123203e-05,
      "loss": 2.923,
      "step": 2739
    },
    {
      "epoch": 0.7587925782331764,
      "grad_norm": 1.625393271446228,
      "learning_rate": 2.80987128832361e-05,
      "loss": 2.7295,
      "step": 2740
    },
    {
      "epoch": 0.7590695098310717,
      "grad_norm": 1.599900245666504,
      "learning_rate": 2.803742672713807e-05,
      "loss": 2.8418,
      "step": 2741
    },
    {
      "epoch": 0.759346441428967,
      "grad_norm": 1.1119273900985718,
      "learning_rate": 2.797619658063789e-05,
      "loss": 2.6902,
      "step": 2742
    },
    {
      "epoch": 0.7596233730268623,
      "grad_norm": 1.754237174987793,
      "learning_rate": 2.7915022491391995e-05,
      "loss": 2.5603,
      "step": 2743
    },
    {
      "epoch": 0.7599003046247577,
      "grad_norm": 1.3307304382324219,
      "learning_rate": 2.785390450701303e-05,
      "loss": 2.7549,
      "step": 2744
    },
    {
      "epoch": 0.760177236222653,
      "grad_norm": 1.1118719577789307,
      "learning_rate": 2.7792842675070206e-05,
      "loss": 2.8739,
      "step": 2745
    },
    {
      "epoch": 0.7604541678205483,
      "grad_norm": 1.6018906831741333,
      "learning_rate": 2.7731837043088826e-05,
      "loss": 2.8068,
      "step": 2746
    },
    {
      "epoch": 0.7607310994184436,
      "grad_norm": 1.2182482481002808,
      "learning_rate": 2.7670887658550536e-05,
      "loss": 2.6503,
      "step": 2747
    },
    {
      "epoch": 0.761008031016339,
      "grad_norm": 0.8037989735603333,
      "learning_rate": 2.7609994568893204e-05,
      "loss": 2.6132,
      "step": 2748
    },
    {
      "epoch": 0.7612849626142343,
      "grad_norm": 1.4756443500518799,
      "learning_rate": 2.7549157821510885e-05,
      "loss": 2.8708,
      "step": 2749
    },
    {
      "epoch": 0.7615618942121296,
      "grad_norm": 2.4815382957458496,
      "learning_rate": 2.7488377463753757e-05,
      "loss": 3.4011,
      "step": 2750
    },
    {
      "epoch": 0.7618388258100249,
      "grad_norm": 1.4734253883361816,
      "learning_rate": 2.7427653542928132e-05,
      "loss": 2.7257,
      "step": 2751
    },
    {
      "epoch": 0.7621157574079203,
      "grad_norm": 1.261542797088623,
      "learning_rate": 2.7366986106296434e-05,
      "loss": 2.7427,
      "step": 2752
    },
    {
      "epoch": 0.7623926890058156,
      "grad_norm": 1.117368459701538,
      "learning_rate": 2.730637520107696e-05,
      "loss": 2.8775,
      "step": 2753
    },
    {
      "epoch": 0.7626696206037109,
      "grad_norm": 1.3518387079238892,
      "learning_rate": 2.7245820874444272e-05,
      "loss": 2.5352,
      "step": 2754
    },
    {
      "epoch": 0.7629465522016062,
      "grad_norm": 5.788182258605957,
      "learning_rate": 2.7185323173528643e-05,
      "loss": 2.5623,
      "step": 2755
    },
    {
      "epoch": 0.7632234837995016,
      "grad_norm": 1.519083023071289,
      "learning_rate": 2.712488214541642e-05,
      "loss": 2.8207,
      "step": 2756
    },
    {
      "epoch": 0.7635004153973969,
      "grad_norm": 1.3019345998764038,
      "learning_rate": 2.7064497837149804e-05,
      "loss": 2.5104,
      "step": 2757
    },
    {
      "epoch": 0.7637773469952922,
      "grad_norm": 1.2609566450119019,
      "learning_rate": 2.7004170295726872e-05,
      "loss": 2.8641,
      "step": 2758
    },
    {
      "epoch": 0.7640542785931875,
      "grad_norm": 1.7451201677322388,
      "learning_rate": 2.6943899568101405e-05,
      "loss": 3.1349,
      "step": 2759
    },
    {
      "epoch": 0.7643312101910829,
      "grad_norm": 2.2196478843688965,
      "learning_rate": 2.688368570118317e-05,
      "loss": 2.6598,
      "step": 2760
    },
    {
      "epoch": 0.7646081417889781,
      "grad_norm": 0.8295835852622986,
      "learning_rate": 2.6823528741837488e-05,
      "loss": 2.8099,
      "step": 2761
    },
    {
      "epoch": 0.7648850733868734,
      "grad_norm": 1.176751732826233,
      "learning_rate": 2.6763428736885476e-05,
      "loss": 2.7665,
      "step": 2762
    },
    {
      "epoch": 0.7651620049847687,
      "grad_norm": 1.5936609506607056,
      "learning_rate": 2.6703385733103924e-05,
      "loss": 2.8111,
      "step": 2763
    },
    {
      "epoch": 0.765438936582664,
      "grad_norm": 1.1688146591186523,
      "learning_rate": 2.6643399777225232e-05,
      "loss": 2.8102,
      "step": 2764
    },
    {
      "epoch": 0.7657158681805594,
      "grad_norm": 1.168511152267456,
      "learning_rate": 2.6583470915937402e-05,
      "loss": 2.7653,
      "step": 2765
    },
    {
      "epoch": 0.7659927997784547,
      "grad_norm": 1.4523048400878906,
      "learning_rate": 2.652359919588403e-05,
      "loss": 2.6246,
      "step": 2766
    },
    {
      "epoch": 0.76626973137635,
      "grad_norm": 1.4812711477279663,
      "learning_rate": 2.6463784663664214e-05,
      "loss": 2.9475,
      "step": 2767
    },
    {
      "epoch": 0.7665466629742453,
      "grad_norm": 10.08879280090332,
      "learning_rate": 2.6404027365832473e-05,
      "loss": 2.7762,
      "step": 2768
    },
    {
      "epoch": 0.7668235945721407,
      "grad_norm": 1.3308957815170288,
      "learning_rate": 2.6344327348898958e-05,
      "loss": 2.4427,
      "step": 2769
    },
    {
      "epoch": 0.767100526170036,
      "grad_norm": 1.9766590595245361,
      "learning_rate": 2.628468465932904e-05,
      "loss": 2.39,
      "step": 2770
    },
    {
      "epoch": 0.7673774577679313,
      "grad_norm": 2.2545158863067627,
      "learning_rate": 2.622509934354359e-05,
      "loss": 2.5055,
      "step": 2771
    },
    {
      "epoch": 0.7676543893658266,
      "grad_norm": 1.3092906475067139,
      "learning_rate": 2.616557144791879e-05,
      "loss": 2.5807,
      "step": 2772
    },
    {
      "epoch": 0.767931320963722,
      "grad_norm": 1.561122179031372,
      "learning_rate": 2.6106101018786134e-05,
      "loss": 2.9317,
      "step": 2773
    },
    {
      "epoch": 0.7682082525616173,
      "grad_norm": 1.4232147932052612,
      "learning_rate": 2.6046688102432382e-05,
      "loss": 2.9531,
      "step": 2774
    },
    {
      "epoch": 0.7684851841595126,
      "grad_norm": 3.122957944869995,
      "learning_rate": 2.5987332745099558e-05,
      "loss": 3.8673,
      "step": 2775
    },
    {
      "epoch": 0.7687621157574079,
      "grad_norm": 2.0886476039886475,
      "learning_rate": 2.5928034992984872e-05,
      "loss": 2.9174,
      "step": 2776
    },
    {
      "epoch": 0.7690390473553033,
      "grad_norm": 1.1254757642745972,
      "learning_rate": 2.5868794892240612e-05,
      "loss": 2.6686,
      "step": 2777
    },
    {
      "epoch": 0.7693159789531986,
      "grad_norm": 1.8632172346115112,
      "learning_rate": 2.58096124889744e-05,
      "loss": 2.9362,
      "step": 2778
    },
    {
      "epoch": 0.7695929105510939,
      "grad_norm": 1.440072774887085,
      "learning_rate": 2.5750487829248726e-05,
      "loss": 2.4785,
      "step": 2779
    },
    {
      "epoch": 0.7698698421489892,
      "grad_norm": 1.0852787494659424,
      "learning_rate": 2.5691420959081292e-05,
      "loss": 2.6737,
      "step": 2780
    },
    {
      "epoch": 0.7701467737468846,
      "grad_norm": 1.4319770336151123,
      "learning_rate": 2.5632411924444744e-05,
      "loss": 2.6004,
      "step": 2781
    },
    {
      "epoch": 0.7704237053447799,
      "grad_norm": 1.5288673639297485,
      "learning_rate": 2.5573460771266744e-05,
      "loss": 2.8314,
      "step": 2782
    },
    {
      "epoch": 0.7707006369426752,
      "grad_norm": 1.1410452127456665,
      "learning_rate": 2.5514567545429913e-05,
      "loss": 2.6155,
      "step": 2783
    },
    {
      "epoch": 0.7709775685405704,
      "grad_norm": 1.6288542747497559,
      "learning_rate": 2.545573229277175e-05,
      "loss": 2.6698,
      "step": 2784
    },
    {
      "epoch": 0.7712545001384657,
      "grad_norm": 2.0513246059417725,
      "learning_rate": 2.5396955059084703e-05,
      "loss": 2.958,
      "step": 2785
    },
    {
      "epoch": 0.7715314317363611,
      "grad_norm": 1.474790334701538,
      "learning_rate": 2.5338235890115902e-05,
      "loss": 2.7685,
      "step": 2786
    },
    {
      "epoch": 0.7718083633342564,
      "grad_norm": 1.5853452682495117,
      "learning_rate": 2.527957483156753e-05,
      "loss": 2.7268,
      "step": 2787
    },
    {
      "epoch": 0.7720852949321517,
      "grad_norm": 1.7835679054260254,
      "learning_rate": 2.5220971929096317e-05,
      "loss": 2.7315,
      "step": 2788
    },
    {
      "epoch": 0.772362226530047,
      "grad_norm": 1.3402307033538818,
      "learning_rate": 2.5162427228313857e-05,
      "loss": 2.9819,
      "step": 2789
    },
    {
      "epoch": 0.7726391581279424,
      "grad_norm": 1.7315587997436523,
      "learning_rate": 2.5103940774786385e-05,
      "loss": 2.7618,
      "step": 2790
    },
    {
      "epoch": 0.7729160897258377,
      "grad_norm": 1.3723822832107544,
      "learning_rate": 2.504551261403487e-05,
      "loss": 2.8444,
      "step": 2791
    },
    {
      "epoch": 0.773193021323733,
      "grad_norm": 1.6342122554779053,
      "learning_rate": 2.4987142791534767e-05,
      "loss": 2.7073,
      "step": 2792
    },
    {
      "epoch": 0.7734699529216283,
      "grad_norm": 1.0601471662521362,
      "learning_rate": 2.4928831352716353e-05,
      "loss": 2.7918,
      "step": 2793
    },
    {
      "epoch": 0.7737468845195237,
      "grad_norm": 1.8162391185760498,
      "learning_rate": 2.4870578342964245e-05,
      "loss": 2.6307,
      "step": 2794
    },
    {
      "epoch": 0.774023816117419,
      "grad_norm": 1.343056559562683,
      "learning_rate": 2.481238380761769e-05,
      "loss": 2.5655,
      "step": 2795
    },
    {
      "epoch": 0.7743007477153143,
      "grad_norm": 2.555288076400757,
      "learning_rate": 2.4754247791970418e-05,
      "loss": 3.0405,
      "step": 2796
    },
    {
      "epoch": 0.7745776793132096,
      "grad_norm": 1.2910922765731812,
      "learning_rate": 2.4696170341270585e-05,
      "loss": 2.7417,
      "step": 2797
    },
    {
      "epoch": 0.774854610911105,
      "grad_norm": 1.4226361513137817,
      "learning_rate": 2.4638151500720806e-05,
      "loss": 2.7559,
      "step": 2798
    },
    {
      "epoch": 0.7751315425090003,
      "grad_norm": 1.2817487716674805,
      "learning_rate": 2.458019131547803e-05,
      "loss": 2.9169,
      "step": 2799
    },
    {
      "epoch": 0.7754084741068956,
      "grad_norm": 3.5929179191589355,
      "learning_rate": 2.452228983065361e-05,
      "loss": 3.5593,
      "step": 2800
    },
    {
      "epoch": 0.7756854057047909,
      "grad_norm": 1.4803215265274048,
      "learning_rate": 2.4464447091313104e-05,
      "loss": 2.8882,
      "step": 2801
    },
    {
      "epoch": 0.7759623373026863,
      "grad_norm": 1.2327439785003662,
      "learning_rate": 2.4406663142476525e-05,
      "loss": 2.6362,
      "step": 2802
    },
    {
      "epoch": 0.7762392689005816,
      "grad_norm": 1.368813157081604,
      "learning_rate": 2.434893802911795e-05,
      "loss": 2.8441,
      "step": 2803
    },
    {
      "epoch": 0.7765162004984769,
      "grad_norm": 1.8350450992584229,
      "learning_rate": 2.429127179616575e-05,
      "loss": 3.1696,
      "step": 2804
    },
    {
      "epoch": 0.7767931320963722,
      "grad_norm": 1.1270864009857178,
      "learning_rate": 2.4233664488502484e-05,
      "loss": 2.6051,
      "step": 2805
    },
    {
      "epoch": 0.7770700636942676,
      "grad_norm": 1.1075642108917236,
      "learning_rate": 2.417611615096479e-05,
      "loss": 2.7663,
      "step": 2806
    },
    {
      "epoch": 0.7773469952921629,
      "grad_norm": 1.2369346618652344,
      "learning_rate": 2.411862682834346e-05,
      "loss": 2.5637,
      "step": 2807
    },
    {
      "epoch": 0.7776239268900581,
      "grad_norm": 1.526141881942749,
      "learning_rate": 2.4061196565383337e-05,
      "loss": 2.6793,
      "step": 2808
    },
    {
      "epoch": 0.7779008584879534,
      "grad_norm": 1.2792106866836548,
      "learning_rate": 2.4003825406783308e-05,
      "loss": 2.6412,
      "step": 2809
    },
    {
      "epoch": 0.7781777900858488,
      "grad_norm": 1.113972783088684,
      "learning_rate": 2.394651339719618e-05,
      "loss": 2.6725,
      "step": 2810
    },
    {
      "epoch": 0.7784547216837441,
      "grad_norm": 1.1308820247650146,
      "learning_rate": 2.3889260581228878e-05,
      "loss": 2.6811,
      "step": 2811
    },
    {
      "epoch": 0.7787316532816394,
      "grad_norm": 1.0981258153915405,
      "learning_rate": 2.3832067003442093e-05,
      "loss": 2.7958,
      "step": 2812
    },
    {
      "epoch": 0.7790085848795347,
      "grad_norm": 1.6483042240142822,
      "learning_rate": 2.3774932708350505e-05,
      "loss": 3.0734,
      "step": 2813
    },
    {
      "epoch": 0.77928551647743,
      "grad_norm": 1.5144284963607788,
      "learning_rate": 2.3717857740422644e-05,
      "loss": 2.7932,
      "step": 2814
    },
    {
      "epoch": 0.7795624480753254,
      "grad_norm": 1.8899537324905396,
      "learning_rate": 2.366084214408084e-05,
      "loss": 2.9024,
      "step": 2815
    },
    {
      "epoch": 0.7798393796732207,
      "grad_norm": 1.5420624017715454,
      "learning_rate": 2.360388596370122e-05,
      "loss": 2.9661,
      "step": 2816
    },
    {
      "epoch": 0.780116311271116,
      "grad_norm": 1.57925546169281,
      "learning_rate": 2.3546989243613682e-05,
      "loss": 2.7884,
      "step": 2817
    },
    {
      "epoch": 0.7803932428690113,
      "grad_norm": 1.3874616622924805,
      "learning_rate": 2.3490152028101852e-05,
      "loss": 2.6057,
      "step": 2818
    },
    {
      "epoch": 0.7806701744669067,
      "grad_norm": 1.1913419961929321,
      "learning_rate": 2.343337436140295e-05,
      "loss": 2.6187,
      "step": 2819
    },
    {
      "epoch": 0.780947106064802,
      "grad_norm": 1.2663335800170898,
      "learning_rate": 2.337665628770803e-05,
      "loss": 2.6802,
      "step": 2820
    },
    {
      "epoch": 0.7812240376626973,
      "grad_norm": 1.302472472190857,
      "learning_rate": 2.331999785116159e-05,
      "loss": 2.9861,
      "step": 2821
    },
    {
      "epoch": 0.7815009692605926,
      "grad_norm": 1.3166072368621826,
      "learning_rate": 2.326339909586178e-05,
      "loss": 2.6539,
      "step": 2822
    },
    {
      "epoch": 0.781777900858488,
      "grad_norm": 1.3500840663909912,
      "learning_rate": 2.3206860065860302e-05,
      "loss": 2.7886,
      "step": 2823
    },
    {
      "epoch": 0.7820548324563833,
      "grad_norm": 1.730839729309082,
      "learning_rate": 2.3150380805162418e-05,
      "loss": 2.7772,
      "step": 2824
    },
    {
      "epoch": 0.7823317640542786,
      "grad_norm": 2.9924428462982178,
      "learning_rate": 2.3093961357726723e-05,
      "loss": 3.5744,
      "step": 2825
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 1.2666369676589966,
      "learning_rate": 2.3037601767465454e-05,
      "loss": 2.5735,
      "step": 2826
    },
    {
      "epoch": 0.7828856272500693,
      "grad_norm": 1.5073728561401367,
      "learning_rate": 2.2981302078244105e-05,
      "loss": 2.7513,
      "step": 2827
    },
    {
      "epoch": 0.7831625588479646,
      "grad_norm": 1.7456878423690796,
      "learning_rate": 2.292506233388162e-05,
      "loss": 2.8464,
      "step": 2828
    },
    {
      "epoch": 0.7834394904458599,
      "grad_norm": 0.926898181438446,
      "learning_rate": 2.2868882578150285e-05,
      "loss": 2.8184,
      "step": 2829
    },
    {
      "epoch": 0.7837164220437552,
      "grad_norm": 1.0987428426742554,
      "learning_rate": 2.281276285477566e-05,
      "loss": 2.6226,
      "step": 2830
    },
    {
      "epoch": 0.7839933536416505,
      "grad_norm": 1.6241536140441895,
      "learning_rate": 2.2756703207436624e-05,
      "loss": 2.7239,
      "step": 2831
    },
    {
      "epoch": 0.7842702852395458,
      "grad_norm": 0.9334229230880737,
      "learning_rate": 2.2700703679765278e-05,
      "loss": 2.7746,
      "step": 2832
    },
    {
      "epoch": 0.7845472168374411,
      "grad_norm": 1.409681797027588,
      "learning_rate": 2.2644764315346956e-05,
      "loss": 3.1124,
      "step": 2833
    },
    {
      "epoch": 0.7848241484353364,
      "grad_norm": 1.6021548509597778,
      "learning_rate": 2.258888515772005e-05,
      "loss": 2.7046,
      "step": 2834
    },
    {
      "epoch": 0.7851010800332318,
      "grad_norm": 1.2131134271621704,
      "learning_rate": 2.25330662503763e-05,
      "loss": 2.5163,
      "step": 2835
    },
    {
      "epoch": 0.7853780116311271,
      "grad_norm": 1.2493963241577148,
      "learning_rate": 2.2477307636760357e-05,
      "loss": 2.6996,
      "step": 2836
    },
    {
      "epoch": 0.7856549432290224,
      "grad_norm": 1.7190215587615967,
      "learning_rate": 2.2421609360270045e-05,
      "loss": 2.9826,
      "step": 2837
    },
    {
      "epoch": 0.7859318748269177,
      "grad_norm": 1.164171814918518,
      "learning_rate": 2.236597146425622e-05,
      "loss": 2.9398,
      "step": 2838
    },
    {
      "epoch": 0.7862088064248131,
      "grad_norm": 1.6923491954803467,
      "learning_rate": 2.2310393992022704e-05,
      "loss": 2.5764,
      "step": 2839
    },
    {
      "epoch": 0.7864857380227084,
      "grad_norm": 1.948574185371399,
      "learning_rate": 2.2254876986826323e-05,
      "loss": 2.7705,
      "step": 2840
    },
    {
      "epoch": 0.7867626696206037,
      "grad_norm": 1.9062774181365967,
      "learning_rate": 2.2199420491876845e-05,
      "loss": 3.0385,
      "step": 2841
    },
    {
      "epoch": 0.787039601218499,
      "grad_norm": 0.9570973515510559,
      "learning_rate": 2.214402455033694e-05,
      "loss": 2.9905,
      "step": 2842
    },
    {
      "epoch": 0.7873165328163944,
      "grad_norm": 1.036166787147522,
      "learning_rate": 2.2088689205322068e-05,
      "loss": 2.8339,
      "step": 2843
    },
    {
      "epoch": 0.7875934644142897,
      "grad_norm": 1.1238036155700684,
      "learning_rate": 2.2033414499900685e-05,
      "loss": 2.6468,
      "step": 2844
    },
    {
      "epoch": 0.787870396012185,
      "grad_norm": 1.448442816734314,
      "learning_rate": 2.1978200477093902e-05,
      "loss": 2.5738,
      "step": 2845
    },
    {
      "epoch": 0.7881473276100803,
      "grad_norm": 1.4662296772003174,
      "learning_rate": 2.1923047179875654e-05,
      "loss": 2.8733,
      "step": 2846
    },
    {
      "epoch": 0.7884242592079757,
      "grad_norm": 1.1000827550888062,
      "learning_rate": 2.186795465117265e-05,
      "loss": 2.461,
      "step": 2847
    },
    {
      "epoch": 0.788701190805871,
      "grad_norm": 1.3521825075149536,
      "learning_rate": 2.1812922933864256e-05,
      "loss": 2.6673,
      "step": 2848
    },
    {
      "epoch": 0.7889781224037663,
      "grad_norm": 1.9187358617782593,
      "learning_rate": 2.1757952070782504e-05,
      "loss": 2.844,
      "step": 2849
    },
    {
      "epoch": 0.7892550540016616,
      "grad_norm": 1.8662562370300293,
      "learning_rate": 2.170304210471209e-05,
      "loss": 3.4556,
      "step": 2850
    },
    {
      "epoch": 0.789531985599557,
      "grad_norm": 1.1305601596832275,
      "learning_rate": 2.1648193078390332e-05,
      "loss": 2.9509,
      "step": 2851
    },
    {
      "epoch": 0.7898089171974523,
      "grad_norm": 1.4433000087738037,
      "learning_rate": 2.1593405034506997e-05,
      "loss": 2.9256,
      "step": 2852
    },
    {
      "epoch": 0.7900858487953476,
      "grad_norm": 1.4666862487792969,
      "learning_rate": 2.1538678015704595e-05,
      "loss": 2.648,
      "step": 2853
    },
    {
      "epoch": 0.7903627803932428,
      "grad_norm": 1.5469653606414795,
      "learning_rate": 2.148401206457793e-05,
      "loss": 3.1121,
      "step": 2854
    },
    {
      "epoch": 0.7906397119911381,
      "grad_norm": 1.8728482723236084,
      "learning_rate": 2.1429407223674402e-05,
      "loss": 2.537,
      "step": 2855
    },
    {
      "epoch": 0.7909166435890335,
      "grad_norm": 0.8866031169891357,
      "learning_rate": 2.1374863535493816e-05,
      "loss": 2.609,
      "step": 2856
    },
    {
      "epoch": 0.7911935751869288,
      "grad_norm": 1.1060009002685547,
      "learning_rate": 2.132038104248839e-05,
      "loss": 2.6987,
      "step": 2857
    },
    {
      "epoch": 0.7914705067848241,
      "grad_norm": 1.1832937002182007,
      "learning_rate": 2.126595978706265e-05,
      "loss": 2.7546,
      "step": 2858
    },
    {
      "epoch": 0.7917474383827194,
      "grad_norm": 1.523499608039856,
      "learning_rate": 2.121159981157359e-05,
      "loss": 2.7163,
      "step": 2859
    },
    {
      "epoch": 0.7920243699806148,
      "grad_norm": 1.5445829629898071,
      "learning_rate": 2.1157301158330357e-05,
      "loss": 2.5706,
      "step": 2860
    },
    {
      "epoch": 0.7923013015785101,
      "grad_norm": 1.2627346515655518,
      "learning_rate": 2.1103063869594486e-05,
      "loss": 2.8291,
      "step": 2861
    },
    {
      "epoch": 0.7925782331764054,
      "grad_norm": 1.1070663928985596,
      "learning_rate": 2.104888798757969e-05,
      "loss": 2.8669,
      "step": 2862
    },
    {
      "epoch": 0.7928551647743007,
      "grad_norm": 1.612951636314392,
      "learning_rate": 2.0994773554451908e-05,
      "loss": 2.7573,
      "step": 2863
    },
    {
      "epoch": 0.7931320963721961,
      "grad_norm": 1.42770516872406,
      "learning_rate": 2.0940720612329258e-05,
      "loss": 2.535,
      "step": 2864
    },
    {
      "epoch": 0.7934090279700914,
      "grad_norm": 0.787700355052948,
      "learning_rate": 2.0886729203281995e-05,
      "loss": 2.6253,
      "step": 2865
    },
    {
      "epoch": 0.7936859595679867,
      "grad_norm": 1.5732358694076538,
      "learning_rate": 2.08327993693325e-05,
      "loss": 2.9229,
      "step": 2866
    },
    {
      "epoch": 0.793962891165882,
      "grad_norm": 0.9501615762710571,
      "learning_rate": 2.0778931152455116e-05,
      "loss": 2.5643,
      "step": 2867
    },
    {
      "epoch": 0.7942398227637774,
      "grad_norm": 0.8824294805526733,
      "learning_rate": 2.072512459457645e-05,
      "loss": 2.7563,
      "step": 2868
    },
    {
      "epoch": 0.7945167543616727,
      "grad_norm": 1.1400706768035889,
      "learning_rate": 2.067137973757489e-05,
      "loss": 2.7726,
      "step": 2869
    },
    {
      "epoch": 0.794793685959568,
      "grad_norm": 1.441460132598877,
      "learning_rate": 2.0617696623280934e-05,
      "loss": 2.8705,
      "step": 2870
    },
    {
      "epoch": 0.7950706175574633,
      "grad_norm": 1.4153046607971191,
      "learning_rate": 2.056407529347699e-05,
      "loss": 2.7678,
      "step": 2871
    },
    {
      "epoch": 0.7953475491553587,
      "grad_norm": 1.5290249586105347,
      "learning_rate": 2.0510515789897377e-05,
      "loss": 2.4862,
      "step": 2872
    },
    {
      "epoch": 0.795624480753254,
      "grad_norm": 2.0305423736572266,
      "learning_rate": 2.045701815422829e-05,
      "loss": 3.056,
      "step": 2873
    },
    {
      "epoch": 0.7959014123511493,
      "grad_norm": 1.9372893571853638,
      "learning_rate": 2.0403582428107792e-05,
      "loss": 3.0894,
      "step": 2874
    },
    {
      "epoch": 0.7961783439490446,
      "grad_norm": 3.0845742225646973,
      "learning_rate": 2.0350208653125756e-05,
      "loss": 3.8301,
      "step": 2875
    },
    {
      "epoch": 0.79645527554694,
      "grad_norm": 1.420465350151062,
      "learning_rate": 2.0296896870823766e-05,
      "loss": 2.8496,
      "step": 2876
    },
    {
      "epoch": 0.7967322071448353,
      "grad_norm": 2.17264986038208,
      "learning_rate": 2.024364712269531e-05,
      "loss": 2.9893,
      "step": 2877
    },
    {
      "epoch": 0.7970091387427305,
      "grad_norm": 1.0084600448608398,
      "learning_rate": 2.0190459450185418e-05,
      "loss": 2.6218,
      "step": 2878
    },
    {
      "epoch": 0.7972860703406258,
      "grad_norm": 1.1605966091156006,
      "learning_rate": 2.0137333894690912e-05,
      "loss": 2.449,
      "step": 2879
    },
    {
      "epoch": 0.7975630019385211,
      "grad_norm": 1.5368242263793945,
      "learning_rate": 2.0084270497560263e-05,
      "loss": 2.9994,
      "step": 2880
    },
    {
      "epoch": 0.7978399335364165,
      "grad_norm": 1.3453278541564941,
      "learning_rate": 2.0031269300093535e-05,
      "loss": 2.8973,
      "step": 2881
    },
    {
      "epoch": 0.7981168651343118,
      "grad_norm": 1.261345624923706,
      "learning_rate": 1.9978330343542384e-05,
      "loss": 2.8822,
      "step": 2882
    },
    {
      "epoch": 0.7983937967322071,
      "grad_norm": 1.3080271482467651,
      "learning_rate": 1.992545366911004e-05,
      "loss": 2.8124,
      "step": 2883
    },
    {
      "epoch": 0.7986707283301024,
      "grad_norm": 1.1540530920028687,
      "learning_rate": 1.987263931795126e-05,
      "loss": 2.8198,
      "step": 2884
    },
    {
      "epoch": 0.7989476599279978,
      "grad_norm": 1.3719029426574707,
      "learning_rate": 1.9819887331172204e-05,
      "loss": 2.8129,
      "step": 2885
    },
    {
      "epoch": 0.7992245915258931,
      "grad_norm": 1.1547625064849854,
      "learning_rate": 1.976719774983069e-05,
      "loss": 2.9531,
      "step": 2886
    },
    {
      "epoch": 0.7995015231237884,
      "grad_norm": 1.4991424083709717,
      "learning_rate": 1.9714570614935756e-05,
      "loss": 2.9222,
      "step": 2887
    },
    {
      "epoch": 0.7997784547216837,
      "grad_norm": 1.1141403913497925,
      "learning_rate": 1.966200596744794e-05,
      "loss": 2.6606,
      "step": 2888
    },
    {
      "epoch": 0.8000553863195791,
      "grad_norm": 1.5983209609985352,
      "learning_rate": 1.9609503848279144e-05,
      "loss": 2.6946,
      "step": 2889
    },
    {
      "epoch": 0.8003323179174744,
      "grad_norm": 1.5049333572387695,
      "learning_rate": 1.9557064298292615e-05,
      "loss": 2.9702,
      "step": 2890
    },
    {
      "epoch": 0.8006092495153697,
      "grad_norm": 2.0525035858154297,
      "learning_rate": 1.9504687358302798e-05,
      "loss": 2.9155,
      "step": 2891
    },
    {
      "epoch": 0.800886181113265,
      "grad_norm": 1.595201015472412,
      "learning_rate": 1.945237306907558e-05,
      "loss": 2.6907,
      "step": 2892
    },
    {
      "epoch": 0.8011631127111604,
      "grad_norm": 1.6689269542694092,
      "learning_rate": 1.9400121471327924e-05,
      "loss": 3.053,
      "step": 2893
    },
    {
      "epoch": 0.8014400443090557,
      "grad_norm": 1.5648447275161743,
      "learning_rate": 1.9347932605728093e-05,
      "loss": 2.494,
      "step": 2894
    },
    {
      "epoch": 0.801716975906951,
      "grad_norm": 1.5639302730560303,
      "learning_rate": 1.9295806512895498e-05,
      "loss": 2.7028,
      "step": 2895
    },
    {
      "epoch": 0.8019939075048463,
      "grad_norm": 0.7483145594596863,
      "learning_rate": 1.924374323340068e-05,
      "loss": 2.8071,
      "step": 2896
    },
    {
      "epoch": 0.8022708391027417,
      "grad_norm": 1.3213709592819214,
      "learning_rate": 1.9191742807765323e-05,
      "loss": 2.8306,
      "step": 2897
    },
    {
      "epoch": 0.802547770700637,
      "grad_norm": 1.1647886037826538,
      "learning_rate": 1.9139805276462165e-05,
      "loss": 2.6671,
      "step": 2898
    },
    {
      "epoch": 0.8028247022985323,
      "grad_norm": 1.4693671464920044,
      "learning_rate": 1.9087930679915023e-05,
      "loss": 2.9852,
      "step": 2899
    },
    {
      "epoch": 0.8031016338964276,
      "grad_norm": 4.712059497833252,
      "learning_rate": 1.9036119058498637e-05,
      "loss": 4.4702,
      "step": 2900
    },
    {
      "epoch": 0.8033785654943228,
      "grad_norm": 1.3639471530914307,
      "learning_rate": 1.89843704525389e-05,
      "loss": 2.7426,
      "step": 2901
    },
    {
      "epoch": 0.8036554970922182,
      "grad_norm": 0.9283391237258911,
      "learning_rate": 1.893268490231249e-05,
      "loss": 2.9628,
      "step": 2902
    },
    {
      "epoch": 0.8039324286901135,
      "grad_norm": 1.0291162729263306,
      "learning_rate": 1.8881062448047104e-05,
      "loss": 2.773,
      "step": 2903
    },
    {
      "epoch": 0.8042093602880088,
      "grad_norm": 1.4412510395050049,
      "learning_rate": 1.882950312992131e-05,
      "loss": 2.6427,
      "step": 2904
    },
    {
      "epoch": 0.8044862918859041,
      "grad_norm": 1.680640697479248,
      "learning_rate": 1.877800698806452e-05,
      "loss": 2.6781,
      "step": 2905
    },
    {
      "epoch": 0.8047632234837995,
      "grad_norm": 1.6135386228561401,
      "learning_rate": 1.8726574062557012e-05,
      "loss": 3.0742,
      "step": 2906
    },
    {
      "epoch": 0.8050401550816948,
      "grad_norm": 1.0270028114318848,
      "learning_rate": 1.8675204393429814e-05,
      "loss": 2.7521,
      "step": 2907
    },
    {
      "epoch": 0.8053170866795901,
      "grad_norm": 1.7050650119781494,
      "learning_rate": 1.8623898020664786e-05,
      "loss": 2.9526,
      "step": 2908
    },
    {
      "epoch": 0.8055940182774854,
      "grad_norm": 1.2581156492233276,
      "learning_rate": 1.8572654984194392e-05,
      "loss": 2.8206,
      "step": 2909
    },
    {
      "epoch": 0.8058709498753808,
      "grad_norm": 1.4937968254089355,
      "learning_rate": 1.8521475323901994e-05,
      "loss": 2.7404,
      "step": 2910
    },
    {
      "epoch": 0.8061478814732761,
      "grad_norm": 2.0960092544555664,
      "learning_rate": 1.847035907962146e-05,
      "loss": 2.7561,
      "step": 2911
    },
    {
      "epoch": 0.8064248130711714,
      "grad_norm": 1.0080113410949707,
      "learning_rate": 1.8419306291137372e-05,
      "loss": 2.8846,
      "step": 2912
    },
    {
      "epoch": 0.8067017446690667,
      "grad_norm": 1.9581778049468994,
      "learning_rate": 1.836831699818492e-05,
      "loss": 2.7467,
      "step": 2913
    },
    {
      "epoch": 0.8069786762669621,
      "grad_norm": 1.0130141973495483,
      "learning_rate": 1.8317391240449876e-05,
      "loss": 2.8305,
      "step": 2914
    },
    {
      "epoch": 0.8072556078648574,
      "grad_norm": 1.2986866235733032,
      "learning_rate": 1.8266529057568548e-05,
      "loss": 2.6439,
      "step": 2915
    },
    {
      "epoch": 0.8075325394627527,
      "grad_norm": 1.231287956237793,
      "learning_rate": 1.821573048912778e-05,
      "loss": 2.935,
      "step": 2916
    },
    {
      "epoch": 0.807809471060648,
      "grad_norm": 0.8357569575309753,
      "learning_rate": 1.81649955746649e-05,
      "loss": 2.8332,
      "step": 2917
    },
    {
      "epoch": 0.8080864026585434,
      "grad_norm": 1.57582426071167,
      "learning_rate": 1.8114324353667632e-05,
      "loss": 2.7864,
      "step": 2918
    },
    {
      "epoch": 0.8083633342564387,
      "grad_norm": 1.5003948211669922,
      "learning_rate": 1.8063716865574266e-05,
      "loss": 2.5028,
      "step": 2919
    },
    {
      "epoch": 0.808640265854334,
      "grad_norm": 1.723107099533081,
      "learning_rate": 1.8013173149773323e-05,
      "loss": 2.9521,
      "step": 2920
    },
    {
      "epoch": 0.8089171974522293,
      "grad_norm": 1.3525755405426025,
      "learning_rate": 1.796269324560381e-05,
      "loss": 2.8175,
      "step": 2921
    },
    {
      "epoch": 0.8091941290501247,
      "grad_norm": 1.3799983263015747,
      "learning_rate": 1.791227719235502e-05,
      "loss": 2.5512,
      "step": 2922
    },
    {
      "epoch": 0.80947106064802,
      "grad_norm": 1.5028530359268188,
      "learning_rate": 1.7861925029266592e-05,
      "loss": 3.0429,
      "step": 2923
    },
    {
      "epoch": 0.8097479922459152,
      "grad_norm": 1.3339258432388306,
      "learning_rate": 1.781163679552831e-05,
      "loss": 3.0033,
      "step": 2924
    },
    {
      "epoch": 0.8100249238438105,
      "grad_norm": 4.6099677085876465,
      "learning_rate": 1.7761412530280398e-05,
      "loss": 4.3239,
      "step": 2925
    },
    {
      "epoch": 0.8103018554417059,
      "grad_norm": 2.1620473861694336,
      "learning_rate": 1.7711252272613122e-05,
      "loss": 3.0301,
      "step": 2926
    },
    {
      "epoch": 0.8105787870396012,
      "grad_norm": 1.1942365169525146,
      "learning_rate": 1.7661156061566985e-05,
      "loss": 2.5074,
      "step": 2927
    },
    {
      "epoch": 0.8108557186374965,
      "grad_norm": 1.6597949266433716,
      "learning_rate": 1.7611123936132734e-05,
      "loss": 2.7833,
      "step": 2928
    },
    {
      "epoch": 0.8111326502353918,
      "grad_norm": 1.9225363731384277,
      "learning_rate": 1.7561155935251094e-05,
      "loss": 2.7084,
      "step": 2929
    },
    {
      "epoch": 0.8114095818332872,
      "grad_norm": 1.8110785484313965,
      "learning_rate": 1.7511252097812947e-05,
      "loss": 2.7287,
      "step": 2930
    },
    {
      "epoch": 0.8116865134311825,
      "grad_norm": 1.7865407466888428,
      "learning_rate": 1.7461412462659233e-05,
      "loss": 2.9899,
      "step": 2931
    },
    {
      "epoch": 0.8119634450290778,
      "grad_norm": 1.4619214534759521,
      "learning_rate": 1.7411637068580953e-05,
      "loss": 2.5232,
      "step": 2932
    },
    {
      "epoch": 0.8122403766269731,
      "grad_norm": 1.4430086612701416,
      "learning_rate": 1.7361925954319003e-05,
      "loss": 2.8457,
      "step": 2933
    },
    {
      "epoch": 0.8125173082248685,
      "grad_norm": 1.6865593194961548,
      "learning_rate": 1.7312279158564415e-05,
      "loss": 2.3381,
      "step": 2934
    },
    {
      "epoch": 0.8127942398227638,
      "grad_norm": 0.8854038715362549,
      "learning_rate": 1.7262696719958004e-05,
      "loss": 2.8799,
      "step": 2935
    },
    {
      "epoch": 0.8130711714206591,
      "grad_norm": 1.9693397283554077,
      "learning_rate": 1.721317867709057e-05,
      "loss": 2.6524,
      "step": 2936
    },
    {
      "epoch": 0.8133481030185544,
      "grad_norm": 0.8433536291122437,
      "learning_rate": 1.7163725068502788e-05,
      "loss": 2.5193,
      "step": 2937
    },
    {
      "epoch": 0.8136250346164497,
      "grad_norm": 0.9649327397346497,
      "learning_rate": 1.7114335932685165e-05,
      "loss": 2.8103,
      "step": 2938
    },
    {
      "epoch": 0.8139019662143451,
      "grad_norm": 1.7423204183578491,
      "learning_rate": 1.706501130807806e-05,
      "loss": 3.0139,
      "step": 2939
    },
    {
      "epoch": 0.8141788978122404,
      "grad_norm": 1.8019976615905762,
      "learning_rate": 1.7015751233071585e-05,
      "loss": 2.6158,
      "step": 2940
    },
    {
      "epoch": 0.8144558294101357,
      "grad_norm": 1.1565155982971191,
      "learning_rate": 1.6966555746005662e-05,
      "loss": 2.5206,
      "step": 2941
    },
    {
      "epoch": 0.814732761008031,
      "grad_norm": 1.6771178245544434,
      "learning_rate": 1.691742488516983e-05,
      "loss": 2.6546,
      "step": 2942
    },
    {
      "epoch": 0.8150096926059264,
      "grad_norm": 1.364692211151123,
      "learning_rate": 1.6868358688803497e-05,
      "loss": 2.8094,
      "step": 2943
    },
    {
      "epoch": 0.8152866242038217,
      "grad_norm": 0.8680933117866516,
      "learning_rate": 1.6819357195095597e-05,
      "loss": 2.4831,
      "step": 2944
    },
    {
      "epoch": 0.815563555801717,
      "grad_norm": 1.2125277519226074,
      "learning_rate": 1.6770420442184764e-05,
      "loss": 2.6179,
      "step": 2945
    },
    {
      "epoch": 0.8158404873996123,
      "grad_norm": 1.1408050060272217,
      "learning_rate": 1.6721548468159264e-05,
      "loss": 2.6832,
      "step": 2946
    },
    {
      "epoch": 0.8161174189975077,
      "grad_norm": 1.2853754758834839,
      "learning_rate": 1.6672741311056894e-05,
      "loss": 2.6696,
      "step": 2947
    },
    {
      "epoch": 0.8163943505954029,
      "grad_norm": 1.2306122779846191,
      "learning_rate": 1.6623999008865053e-05,
      "loss": 2.7143,
      "step": 2948
    },
    {
      "epoch": 0.8166712821932982,
      "grad_norm": 2.095325469970703,
      "learning_rate": 1.657532159952062e-05,
      "loss": 3.2119,
      "step": 2949
    },
    {
      "epoch": 0.8169482137911935,
      "grad_norm": 3.535750150680542,
      "learning_rate": 1.652670912091001e-05,
      "loss": 3.7123,
      "step": 2950
    },
    {
      "epoch": 0.8172251453890889,
      "grad_norm": 1.0357232093811035,
      "learning_rate": 1.6478161610869026e-05,
      "loss": 2.7336,
      "step": 2951
    },
    {
      "epoch": 0.8175020769869842,
      "grad_norm": 0.9846844673156738,
      "learning_rate": 1.6429679107183016e-05,
      "loss": 2.5957,
      "step": 2952
    },
    {
      "epoch": 0.8177790085848795,
      "grad_norm": 1.018471121788025,
      "learning_rate": 1.638126164758663e-05,
      "loss": 2.674,
      "step": 2953
    },
    {
      "epoch": 0.8180559401827748,
      "grad_norm": 1.5538946390151978,
      "learning_rate": 1.6332909269763953e-05,
      "loss": 2.734,
      "step": 2954
    },
    {
      "epoch": 0.8183328717806702,
      "grad_norm": 1.7519913911819458,
      "learning_rate": 1.628462201134838e-05,
      "loss": 3.0673,
      "step": 2955
    },
    {
      "epoch": 0.8186098033785655,
      "grad_norm": 1.0945311784744263,
      "learning_rate": 1.6236399909922684e-05,
      "loss": 2.7443,
      "step": 2956
    },
    {
      "epoch": 0.8188867349764608,
      "grad_norm": 1.7660750150680542,
      "learning_rate": 1.6188243003018798e-05,
      "loss": 2.7689,
      "step": 2957
    },
    {
      "epoch": 0.8191636665743561,
      "grad_norm": 1.089036226272583,
      "learning_rate": 1.6140151328118115e-05,
      "loss": 2.5295,
      "step": 2958
    },
    {
      "epoch": 0.8194405981722515,
      "grad_norm": 1.1123430728912354,
      "learning_rate": 1.609212492265103e-05,
      "loss": 2.7192,
      "step": 2959
    },
    {
      "epoch": 0.8197175297701468,
      "grad_norm": 1.407126784324646,
      "learning_rate": 1.6044163823997283e-05,
      "loss": 2.821,
      "step": 2960
    },
    {
      "epoch": 0.8199944613680421,
      "grad_norm": 1.7229863405227661,
      "learning_rate": 1.59962680694858e-05,
      "loss": 2.6649,
      "step": 2961
    },
    {
      "epoch": 0.8202713929659374,
      "grad_norm": 1.345894455909729,
      "learning_rate": 1.5948437696394513e-05,
      "loss": 2.77,
      "step": 2962
    },
    {
      "epoch": 0.8205483245638328,
      "grad_norm": 1.4527201652526855,
      "learning_rate": 1.5900672741950607e-05,
      "loss": 2.879,
      "step": 2963
    },
    {
      "epoch": 0.8208252561617281,
      "grad_norm": 1.544024109840393,
      "learning_rate": 1.585297324333027e-05,
      "loss": 2.778,
      "step": 2964
    },
    {
      "epoch": 0.8211021877596234,
      "grad_norm": 1.425732970237732,
      "learning_rate": 1.5805339237658813e-05,
      "loss": 2.8205,
      "step": 2965
    },
    {
      "epoch": 0.8213791193575187,
      "grad_norm": 1.687116265296936,
      "learning_rate": 1.5757770762010438e-05,
      "loss": 2.9889,
      "step": 2966
    },
    {
      "epoch": 0.821656050955414,
      "grad_norm": 0.8597702383995056,
      "learning_rate": 1.5710267853408534e-05,
      "loss": 2.5228,
      "step": 2967
    },
    {
      "epoch": 0.8219329825533094,
      "grad_norm": 1.3650617599487305,
      "learning_rate": 1.566283054882528e-05,
      "loss": 2.4924,
      "step": 2968
    },
    {
      "epoch": 0.8222099141512047,
      "grad_norm": 1.316955327987671,
      "learning_rate": 1.561545888518192e-05,
      "loss": 2.7898,
      "step": 2969
    },
    {
      "epoch": 0.8224868457491,
      "grad_norm": 1.252253770828247,
      "learning_rate": 1.556815289934854e-05,
      "loss": 2.5309,
      "step": 2970
    },
    {
      "epoch": 0.8227637773469952,
      "grad_norm": 1.389169692993164,
      "learning_rate": 1.5520912628144145e-05,
      "loss": 2.5478,
      "step": 2971
    },
    {
      "epoch": 0.8230407089448906,
      "grad_norm": 1.2352834939956665,
      "learning_rate": 1.5473738108336566e-05,
      "loss": 2.7469,
      "step": 2972
    },
    {
      "epoch": 0.8233176405427859,
      "grad_norm": 1.3856765031814575,
      "learning_rate": 1.5426629376642475e-05,
      "loss": 2.6795,
      "step": 2973
    },
    {
      "epoch": 0.8235945721406812,
      "grad_norm": 2.619903802871704,
      "learning_rate": 1.537958646972737e-05,
      "loss": 3.0275,
      "step": 2974
    },
    {
      "epoch": 0.8238715037385765,
      "grad_norm": 3.9109251499176025,
      "learning_rate": 1.533260942420539e-05,
      "loss": 4.043,
      "step": 2975
    },
    {
      "epoch": 0.8241484353364719,
      "grad_norm": 1.729552984237671,
      "learning_rate": 1.5285698276639615e-05,
      "loss": 2.5781,
      "step": 2976
    },
    {
      "epoch": 0.8244253669343672,
      "grad_norm": 1.358718991279602,
      "learning_rate": 1.5238853063541658e-05,
      "loss": 2.4212,
      "step": 2977
    },
    {
      "epoch": 0.8247022985322625,
      "grad_norm": 0.9492407441139221,
      "learning_rate": 1.5192073821371889e-05,
      "loss": 2.7406,
      "step": 2978
    },
    {
      "epoch": 0.8249792301301578,
      "grad_norm": 1.7622709274291992,
      "learning_rate": 1.5145360586539336e-05,
      "loss": 2.8144,
      "step": 2979
    },
    {
      "epoch": 0.8252561617280532,
      "grad_norm": 1.3219594955444336,
      "learning_rate": 1.5098713395401643e-05,
      "loss": 2.7726,
      "step": 2980
    },
    {
      "epoch": 0.8255330933259485,
      "grad_norm": 1.4270638227462769,
      "learning_rate": 1.5052132284265042e-05,
      "loss": 2.5905,
      "step": 2981
    },
    {
      "epoch": 0.8258100249238438,
      "grad_norm": 1.323522925376892,
      "learning_rate": 1.5005617289384343e-05,
      "loss": 2.7863,
      "step": 2982
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 1.531290054321289,
      "learning_rate": 1.4959168446962935e-05,
      "loss": 2.9003,
      "step": 2983
    },
    {
      "epoch": 0.8263638881196345,
      "grad_norm": 1.3229002952575684,
      "learning_rate": 1.4912785793152583e-05,
      "loss": 2.7109,
      "step": 2984
    },
    {
      "epoch": 0.8266408197175298,
      "grad_norm": 1.4431756734848022,
      "learning_rate": 1.4866469364053749e-05,
      "loss": 2.8354,
      "step": 2985
    },
    {
      "epoch": 0.8269177513154251,
      "grad_norm": 1.6073932647705078,
      "learning_rate": 1.4820219195715146e-05,
      "loss": 2.4679,
      "step": 2986
    },
    {
      "epoch": 0.8271946829133204,
      "grad_norm": 1.714503526687622,
      "learning_rate": 1.4774035324134039e-05,
      "loss": 2.5459,
      "step": 2987
    },
    {
      "epoch": 0.8274716145112158,
      "grad_norm": 1.2907015085220337,
      "learning_rate": 1.4727917785256051e-05,
      "loss": 2.8861,
      "step": 2988
    },
    {
      "epoch": 0.8277485461091111,
      "grad_norm": 1.1760333776474,
      "learning_rate": 1.4681866614975227e-05,
      "loss": 2.6618,
      "step": 2989
    },
    {
      "epoch": 0.8280254777070064,
      "grad_norm": 1.5860620737075806,
      "learning_rate": 1.4635881849133826e-05,
      "loss": 2.8644,
      "step": 2990
    },
    {
      "epoch": 0.8283024093049017,
      "grad_norm": 1.7209746837615967,
      "learning_rate": 1.4589963523522577e-05,
      "loss": 2.994,
      "step": 2991
    },
    {
      "epoch": 0.8285793409027971,
      "grad_norm": 1.2861586809158325,
      "learning_rate": 1.4544111673880445e-05,
      "loss": 2.7188,
      "step": 2992
    },
    {
      "epoch": 0.8288562725006924,
      "grad_norm": 0.8272273540496826,
      "learning_rate": 1.4498326335894574e-05,
      "loss": 2.7651,
      "step": 2993
    },
    {
      "epoch": 0.8291332040985876,
      "grad_norm": 1.729760766029358,
      "learning_rate": 1.4452607545200492e-05,
      "loss": 3.0048,
      "step": 2994
    },
    {
      "epoch": 0.8294101356964829,
      "grad_norm": 1.0698480606079102,
      "learning_rate": 1.440695533738179e-05,
      "loss": 2.8259,
      "step": 2995
    },
    {
      "epoch": 0.8296870672943782,
      "grad_norm": 1.9651565551757812,
      "learning_rate": 1.4361369747970311e-05,
      "loss": 3.0114,
      "step": 2996
    },
    {
      "epoch": 0.8299639988922736,
      "grad_norm": 0.9492764472961426,
      "learning_rate": 1.4315850812446042e-05,
      "loss": 2.6462,
      "step": 2997
    },
    {
      "epoch": 0.8302409304901689,
      "grad_norm": 1.9277658462524414,
      "learning_rate": 1.4270398566237098e-05,
      "loss": 3.0011,
      "step": 2998
    },
    {
      "epoch": 0.8305178620880642,
      "grad_norm": 1.873567819595337,
      "learning_rate": 1.4225013044719615e-05,
      "loss": 3.2477,
      "step": 2999
    },
    {
      "epoch": 0.8307947936859595,
      "grad_norm": 3.1596426963806152,
      "learning_rate": 1.4179694283217937e-05,
      "loss": 3.7095,
      "step": 3000
    },
    {
      "epoch": 0.8310717252838549,
      "grad_norm": 1.6815263032913208,
      "learning_rate": 1.4134442317004304e-05,
      "loss": 2.6958,
      "step": 3001
    },
    {
      "epoch": 0.8313486568817502,
      "grad_norm": 2.4771888256073,
      "learning_rate": 1.4089257181299042e-05,
      "loss": 3.3317,
      "step": 3002
    },
    {
      "epoch": 0.8316255884796455,
      "grad_norm": 1.9468337297439575,
      "learning_rate": 1.4044138911270466e-05,
      "loss": 2.4712,
      "step": 3003
    },
    {
      "epoch": 0.8319025200775408,
      "grad_norm": 2.0584003925323486,
      "learning_rate": 1.3999087542034817e-05,
      "loss": 2.9929,
      "step": 3004
    },
    {
      "epoch": 0.8321794516754362,
      "grad_norm": 0.7015234231948853,
      "learning_rate": 1.3954103108656291e-05,
      "loss": 2.6684,
      "step": 3005
    },
    {
      "epoch": 0.8324563832733315,
      "grad_norm": 1.70437753200531,
      "learning_rate": 1.3909185646146983e-05,
      "loss": 2.883,
      "step": 3006
    },
    {
      "epoch": 0.8327333148712268,
      "grad_norm": 1.496725082397461,
      "learning_rate": 1.386433518946686e-05,
      "loss": 2.7889,
      "step": 3007
    },
    {
      "epoch": 0.8330102464691221,
      "grad_norm": 1.208085298538208,
      "learning_rate": 1.3819551773523688e-05,
      "loss": 2.6361,
      "step": 3008
    },
    {
      "epoch": 0.8332871780670175,
      "grad_norm": 0.9877991676330566,
      "learning_rate": 1.3774835433173172e-05,
      "loss": 2.6832,
      "step": 3009
    },
    {
      "epoch": 0.8335641096649128,
      "grad_norm": 1.1782431602478027,
      "learning_rate": 1.3730186203218697e-05,
      "loss": 2.543,
      "step": 3010
    },
    {
      "epoch": 0.8338410412628081,
      "grad_norm": 0.740976870059967,
      "learning_rate": 1.3685604118411455e-05,
      "loss": 2.6201,
      "step": 3011
    },
    {
      "epoch": 0.8341179728607034,
      "grad_norm": 1.768426775932312,
      "learning_rate": 1.3641089213450397e-05,
      "loss": 2.7182,
      "step": 3012
    },
    {
      "epoch": 0.8343949044585988,
      "grad_norm": 1.4269012212753296,
      "learning_rate": 1.3596641522982157e-05,
      "loss": 3.1137,
      "step": 3013
    },
    {
      "epoch": 0.8346718360564941,
      "grad_norm": 1.229787826538086,
      "learning_rate": 1.3552261081601091e-05,
      "loss": 2.8571,
      "step": 3014
    },
    {
      "epoch": 0.8349487676543894,
      "grad_norm": 1.51630437374115,
      "learning_rate": 1.3507947923849163e-05,
      "loss": 2.8817,
      "step": 3015
    },
    {
      "epoch": 0.8352256992522847,
      "grad_norm": 1.804336428642273,
      "learning_rate": 1.3463702084216046e-05,
      "loss": 2.7904,
      "step": 3016
    },
    {
      "epoch": 0.83550263085018,
      "grad_norm": 1.476218819618225,
      "learning_rate": 1.3419523597138884e-05,
      "loss": 2.6541,
      "step": 3017
    },
    {
      "epoch": 0.8357795624480753,
      "grad_norm": 0.7786409854888916,
      "learning_rate": 1.3375412497002593e-05,
      "loss": 2.7245,
      "step": 3018
    },
    {
      "epoch": 0.8360564940459706,
      "grad_norm": 0.9623320698738098,
      "learning_rate": 1.3331368818139445e-05,
      "loss": 2.3649,
      "step": 3019
    },
    {
      "epoch": 0.8363334256438659,
      "grad_norm": 0.8559485673904419,
      "learning_rate": 1.3287392594829384e-05,
      "loss": 2.4915,
      "step": 3020
    },
    {
      "epoch": 0.8366103572417612,
      "grad_norm": 1.2838448286056519,
      "learning_rate": 1.3243483861299721e-05,
      "loss": 2.7953,
      "step": 3021
    },
    {
      "epoch": 0.8368872888396566,
      "grad_norm": 1.485473871231079,
      "learning_rate": 1.3199642651725408e-05,
      "loss": 2.5919,
      "step": 3022
    },
    {
      "epoch": 0.8371642204375519,
      "grad_norm": 1.3391149044036865,
      "learning_rate": 1.315586900022867e-05,
      "loss": 2.6176,
      "step": 3023
    },
    {
      "epoch": 0.8374411520354472,
      "grad_norm": 1.5078611373901367,
      "learning_rate": 1.3112162940879225e-05,
      "loss": 2.8531,
      "step": 3024
    },
    {
      "epoch": 0.8377180836333425,
      "grad_norm": 1.8650858402252197,
      "learning_rate": 1.3068524507694258e-05,
      "loss": 3.1421,
      "step": 3025
    },
    {
      "epoch": 0.8379950152312379,
      "grad_norm": 1.3443483114242554,
      "learning_rate": 1.3024953734638168e-05,
      "loss": 2.6515,
      "step": 3026
    },
    {
      "epoch": 0.8382719468291332,
      "grad_norm": 1.4175198078155518,
      "learning_rate": 1.2981450655622796e-05,
      "loss": 2.8623,
      "step": 3027
    },
    {
      "epoch": 0.8385488784270285,
      "grad_norm": 1.2949321269989014,
      "learning_rate": 1.2938015304507279e-05,
      "loss": 2.8468,
      "step": 3028
    },
    {
      "epoch": 0.8388258100249238,
      "grad_norm": 1.6689027547836304,
      "learning_rate": 1.289464771509804e-05,
      "loss": 2.8973,
      "step": 3029
    },
    {
      "epoch": 0.8391027416228192,
      "grad_norm": 1.496212124824524,
      "learning_rate": 1.2851347921148693e-05,
      "loss": 2.9497,
      "step": 3030
    },
    {
      "epoch": 0.8393796732207145,
      "grad_norm": 1.646315097808838,
      "learning_rate": 1.2808115956360233e-05,
      "loss": 2.9308,
      "step": 3031
    },
    {
      "epoch": 0.8396566048186098,
      "grad_norm": 1.179675579071045,
      "learning_rate": 1.2764951854380714e-05,
      "loss": 2.7608,
      "step": 3032
    },
    {
      "epoch": 0.8399335364165051,
      "grad_norm": 0.8177387118339539,
      "learning_rate": 1.2721855648805448e-05,
      "loss": 2.673,
      "step": 3033
    },
    {
      "epoch": 0.8402104680144005,
      "grad_norm": 0.940900444984436,
      "learning_rate": 1.2678827373176894e-05,
      "loss": 2.6349,
      "step": 3034
    },
    {
      "epoch": 0.8404873996122958,
      "grad_norm": 2.1662635803222656,
      "learning_rate": 1.2635867060984619e-05,
      "loss": 2.5167,
      "step": 3035
    },
    {
      "epoch": 0.8407643312101911,
      "grad_norm": 1.6904385089874268,
      "learning_rate": 1.259297474566532e-05,
      "loss": 2.9469,
      "step": 3036
    },
    {
      "epoch": 0.8410412628080864,
      "grad_norm": 1.4025242328643799,
      "learning_rate": 1.2550150460602761e-05,
      "loss": 2.9757,
      "step": 3037
    },
    {
      "epoch": 0.8413181944059818,
      "grad_norm": 1.0857107639312744,
      "learning_rate": 1.2507394239127757e-05,
      "loss": 2.819,
      "step": 3038
    },
    {
      "epoch": 0.8415951260038771,
      "grad_norm": 1.5865957736968994,
      "learning_rate": 1.2464706114518088e-05,
      "loss": 3.0157,
      "step": 3039
    },
    {
      "epoch": 0.8418720576017724,
      "grad_norm": 0.9632927775382996,
      "learning_rate": 1.2422086119998688e-05,
      "loss": 2.6684,
      "step": 3040
    },
    {
      "epoch": 0.8421489891996676,
      "grad_norm": 1.4581258296966553,
      "learning_rate": 1.237953428874129e-05,
      "loss": 2.651,
      "step": 3041
    },
    {
      "epoch": 0.842425920797563,
      "grad_norm": 1.2029368877410889,
      "learning_rate": 1.2337050653864679e-05,
      "loss": 2.6975,
      "step": 3042
    },
    {
      "epoch": 0.8427028523954583,
      "grad_norm": 1.4727216958999634,
      "learning_rate": 1.2294635248434528e-05,
      "loss": 3.2892,
      "step": 3043
    },
    {
      "epoch": 0.8429797839933536,
      "grad_norm": 1.3219698667526245,
      "learning_rate": 1.2252288105463405e-05,
      "loss": 2.426,
      "step": 3044
    },
    {
      "epoch": 0.8432567155912489,
      "grad_norm": 1.317419409751892,
      "learning_rate": 1.2210009257910771e-05,
      "loss": 2.6523,
      "step": 3045
    },
    {
      "epoch": 0.8435336471891443,
      "grad_norm": 1.3205360174179077,
      "learning_rate": 1.21677987386829e-05,
      "loss": 2.7689,
      "step": 3046
    },
    {
      "epoch": 0.8438105787870396,
      "grad_norm": 1.6884167194366455,
      "learning_rate": 1.2125656580632938e-05,
      "loss": 2.8025,
      "step": 3047
    },
    {
      "epoch": 0.8440875103849349,
      "grad_norm": 1.3840197324752808,
      "learning_rate": 1.2083582816560701e-05,
      "loss": 2.8889,
      "step": 3048
    },
    {
      "epoch": 0.8443644419828302,
      "grad_norm": 2.7933177947998047,
      "learning_rate": 1.2041577479212963e-05,
      "loss": 3.0994,
      "step": 3049
    },
    {
      "epoch": 0.8446413735807256,
      "grad_norm": 3.1562044620513916,
      "learning_rate": 1.1999640601283069e-05,
      "loss": 3.9769,
      "step": 3050
    },
    {
      "epoch": 0.8449183051786209,
      "grad_norm": 1.0793648958206177,
      "learning_rate": 1.1957772215411156e-05,
      "loss": 2.9159,
      "step": 3051
    },
    {
      "epoch": 0.8451952367765162,
      "grad_norm": 1.145857334136963,
      "learning_rate": 1.1915972354184058e-05,
      "loss": 2.5923,
      "step": 3052
    },
    {
      "epoch": 0.8454721683744115,
      "grad_norm": 2.3265607357025146,
      "learning_rate": 1.1874241050135282e-05,
      "loss": 2.9068,
      "step": 3053
    },
    {
      "epoch": 0.8457490999723069,
      "grad_norm": 1.6332247257232666,
      "learning_rate": 1.1832578335744882e-05,
      "loss": 2.8704,
      "step": 3054
    },
    {
      "epoch": 0.8460260315702022,
      "grad_norm": 1.4872264862060547,
      "learning_rate": 1.1790984243439674e-05,
      "loss": 3.0174,
      "step": 3055
    },
    {
      "epoch": 0.8463029631680975,
      "grad_norm": 1.0704503059387207,
      "learning_rate": 1.1749458805592983e-05,
      "loss": 2.6838,
      "step": 3056
    },
    {
      "epoch": 0.8465798947659928,
      "grad_norm": 1.7933567762374878,
      "learning_rate": 1.1708002054524647e-05,
      "loss": 2.9514,
      "step": 3057
    },
    {
      "epoch": 0.8468568263638881,
      "grad_norm": 1.73945152759552,
      "learning_rate": 1.16666140225012e-05,
      "loss": 2.7772,
      "step": 3058
    },
    {
      "epoch": 0.8471337579617835,
      "grad_norm": 1.2374337911605835,
      "learning_rate": 1.1625294741735526e-05,
      "loss": 2.9232,
      "step": 3059
    },
    {
      "epoch": 0.8474106895596788,
      "grad_norm": 1.5635979175567627,
      "learning_rate": 1.1584044244387083e-05,
      "loss": 2.8568,
      "step": 3060
    },
    {
      "epoch": 0.8476876211575741,
      "grad_norm": 1.1509644985198975,
      "learning_rate": 1.1542862562561784e-05,
      "loss": 2.7165,
      "step": 3061
    },
    {
      "epoch": 0.8479645527554694,
      "grad_norm": 1.016371250152588,
      "learning_rate": 1.1501749728311994e-05,
      "loss": 2.7675,
      "step": 3062
    },
    {
      "epoch": 0.8482414843533648,
      "grad_norm": 1.336298942565918,
      "learning_rate": 1.1460705773636427e-05,
      "loss": 2.7798,
      "step": 3063
    },
    {
      "epoch": 0.84851841595126,
      "grad_norm": 1.0788416862487793,
      "learning_rate": 1.1419730730480305e-05,
      "loss": 2.6508,
      "step": 3064
    },
    {
      "epoch": 0.8487953475491553,
      "grad_norm": 1.292082667350769,
      "learning_rate": 1.1378824630735086e-05,
      "loss": 2.7736,
      "step": 3065
    },
    {
      "epoch": 0.8490722791470506,
      "grad_norm": 0.9430363774299622,
      "learning_rate": 1.1337987506238645e-05,
      "loss": 2.7468,
      "step": 3066
    },
    {
      "epoch": 0.849349210744946,
      "grad_norm": 1.2281850576400757,
      "learning_rate": 1.129721938877516e-05,
      "loss": 2.9235,
      "step": 3067
    },
    {
      "epoch": 0.8496261423428413,
      "grad_norm": 1.5270968675613403,
      "learning_rate": 1.1256520310075102e-05,
      "loss": 3.0424,
      "step": 3068
    },
    {
      "epoch": 0.8499030739407366,
      "grad_norm": 1.4194164276123047,
      "learning_rate": 1.1215890301815201e-05,
      "loss": 2.8131,
      "step": 3069
    },
    {
      "epoch": 0.8501800055386319,
      "grad_norm": 0.9814782738685608,
      "learning_rate": 1.1175329395618417e-05,
      "loss": 2.6454,
      "step": 3070
    },
    {
      "epoch": 0.8504569371365273,
      "grad_norm": 1.30820894241333,
      "learning_rate": 1.1134837623053961e-05,
      "loss": 2.7237,
      "step": 3071
    },
    {
      "epoch": 0.8507338687344226,
      "grad_norm": 1.776144027709961,
      "learning_rate": 1.1094415015637161e-05,
      "loss": 2.6911,
      "step": 3072
    },
    {
      "epoch": 0.8510108003323179,
      "grad_norm": 1.0699905157089233,
      "learning_rate": 1.1054061604829636e-05,
      "loss": 2.6947,
      "step": 3073
    },
    {
      "epoch": 0.8512877319302132,
      "grad_norm": 0.7591848969459534,
      "learning_rate": 1.101377742203903e-05,
      "loss": 2.8729,
      "step": 3074
    },
    {
      "epoch": 0.8515646635281086,
      "grad_norm": 2.7936182022094727,
      "learning_rate": 1.0973562498619172e-05,
      "loss": 3.5067,
      "step": 3075
    },
    {
      "epoch": 0.8518415951260039,
      "grad_norm": 0.8582979440689087,
      "learning_rate": 1.093341686586995e-05,
      "loss": 2.8096,
      "step": 3076
    },
    {
      "epoch": 0.8521185267238992,
      "grad_norm": 1.1236815452575684,
      "learning_rate": 1.0893340555037357e-05,
      "loss": 2.8091,
      "step": 3077
    },
    {
      "epoch": 0.8523954583217945,
      "grad_norm": 0.9093992114067078,
      "learning_rate": 1.085333359731341e-05,
      "loss": 2.7521,
      "step": 3078
    },
    {
      "epoch": 0.8526723899196899,
      "grad_norm": 1.0349127054214478,
      "learning_rate": 1.0813396023836142e-05,
      "loss": 2.7961,
      "step": 3079
    },
    {
      "epoch": 0.8529493215175852,
      "grad_norm": 1.7463757991790771,
      "learning_rate": 1.0773527865689625e-05,
      "loss": 2.8264,
      "step": 3080
    },
    {
      "epoch": 0.8532262531154805,
      "grad_norm": 1.2333402633666992,
      "learning_rate": 1.0733729153903794e-05,
      "loss": 2.744,
      "step": 3081
    },
    {
      "epoch": 0.8535031847133758,
      "grad_norm": 1.2324150800704956,
      "learning_rate": 1.06939999194547e-05,
      "loss": 2.5561,
      "step": 3082
    },
    {
      "epoch": 0.8537801163112712,
      "grad_norm": 1.3192819356918335,
      "learning_rate": 1.0654340193264179e-05,
      "loss": 2.8713,
      "step": 3083
    },
    {
      "epoch": 0.8540570479091665,
      "grad_norm": 1.1821292638778687,
      "learning_rate": 1.0614750006200014e-05,
      "loss": 2.6635,
      "step": 3084
    },
    {
      "epoch": 0.8543339795070618,
      "grad_norm": 1.5261622667312622,
      "learning_rate": 1.0575229389075891e-05,
      "loss": 2.5562,
      "step": 3085
    },
    {
      "epoch": 0.8546109111049571,
      "grad_norm": 0.770987868309021,
      "learning_rate": 1.0535778372651317e-05,
      "loss": 2.7767,
      "step": 3086
    },
    {
      "epoch": 0.8548878427028523,
      "grad_norm": 1.4389278888702393,
      "learning_rate": 1.0496396987631606e-05,
      "loss": 3.2229,
      "step": 3087
    },
    {
      "epoch": 0.8551647743007477,
      "grad_norm": 1.6177245378494263,
      "learning_rate": 1.0457085264667954e-05,
      "loss": 3.2858,
      "step": 3088
    },
    {
      "epoch": 0.855441705898643,
      "grad_norm": 1.5438570976257324,
      "learning_rate": 1.0417843234357282e-05,
      "loss": 2.7003,
      "step": 3089
    },
    {
      "epoch": 0.8557186374965383,
      "grad_norm": 1.566727638244629,
      "learning_rate": 1.0378670927242228e-05,
      "loss": 3.1135,
      "step": 3090
    },
    {
      "epoch": 0.8559955690944336,
      "grad_norm": 1.2584772109985352,
      "learning_rate": 1.0339568373811292e-05,
      "loss": 2.7638,
      "step": 3091
    },
    {
      "epoch": 0.856272500692329,
      "grad_norm": 1.2752087116241455,
      "learning_rate": 1.030053560449854e-05,
      "loss": 2.704,
      "step": 3092
    },
    {
      "epoch": 0.8565494322902243,
      "grad_norm": 1.7779330015182495,
      "learning_rate": 1.0261572649683803e-05,
      "loss": 2.5948,
      "step": 3093
    },
    {
      "epoch": 0.8568263638881196,
      "grad_norm": 1.5309356451034546,
      "learning_rate": 1.022267953969257e-05,
      "loss": 2.9297,
      "step": 3094
    },
    {
      "epoch": 0.8571032954860149,
      "grad_norm": 1.2379956245422363,
      "learning_rate": 1.0183856304795969e-05,
      "loss": 2.6137,
      "step": 3095
    },
    {
      "epoch": 0.8573802270839103,
      "grad_norm": 0.926654040813446,
      "learning_rate": 1.0145102975210675e-05,
      "loss": 2.8791,
      "step": 3096
    },
    {
      "epoch": 0.8576571586818056,
      "grad_norm": 1.6225202083587646,
      "learning_rate": 1.0106419581099092e-05,
      "loss": 2.8463,
      "step": 3097
    },
    {
      "epoch": 0.8579340902797009,
      "grad_norm": 1.2629817724227905,
      "learning_rate": 1.0067806152569048e-05,
      "loss": 3.2293,
      "step": 3098
    },
    {
      "epoch": 0.8582110218775962,
      "grad_norm": 1.8424757719039917,
      "learning_rate": 1.0029262719674015e-05,
      "loss": 2.7858,
      "step": 3099
    },
    {
      "epoch": 0.8584879534754916,
      "grad_norm": 3.2015435695648193,
      "learning_rate": 9.99078931241293e-06,
      "loss": 3.8198,
      "step": 3100
    },
    {
      "epoch": 0.8587648850733869,
      "grad_norm": 0.9091971516609192,
      "learning_rate": 9.952385960730248e-06,
      "loss": 2.6804,
      "step": 3101
    },
    {
      "epoch": 0.8590418166712822,
      "grad_norm": 1.7424923181533813,
      "learning_rate": 9.914052694515908e-06,
      "loss": 2.8593,
      "step": 3102
    },
    {
      "epoch": 0.8593187482691775,
      "grad_norm": 1.7614586353302002,
      "learning_rate": 9.875789543605296e-06,
      "loss": 2.7757,
      "step": 3103
    },
    {
      "epoch": 0.8595956798670729,
      "grad_norm": 1.0697795152664185,
      "learning_rate": 9.837596537779237e-06,
      "loss": 2.7479,
      "step": 3104
    },
    {
      "epoch": 0.8598726114649682,
      "grad_norm": 0.9285095930099487,
      "learning_rate": 9.799473706763884e-06,
      "loss": 2.6441,
      "step": 3105
    },
    {
      "epoch": 0.8601495430628635,
      "grad_norm": 1.5014824867248535,
      "learning_rate": 9.761421080230915e-06,
      "loss": 2.9274,
      "step": 3106
    },
    {
      "epoch": 0.8604264746607588,
      "grad_norm": 1.2792730331420898,
      "learning_rate": 9.723438687797226e-06,
      "loss": 2.7602,
      "step": 3107
    },
    {
      "epoch": 0.8607034062586542,
      "grad_norm": 1.130605936050415,
      "learning_rate": 9.685526559025115e-06,
      "loss": 2.5661,
      "step": 3108
    },
    {
      "epoch": 0.8609803378565495,
      "grad_norm": 1.2102701663970947,
      "learning_rate": 9.647684723422213e-06,
      "loss": 2.8934,
      "step": 3109
    },
    {
      "epoch": 0.8612572694544448,
      "grad_norm": 1.3663723468780518,
      "learning_rate": 9.60991321044139e-06,
      "loss": 2.5983,
      "step": 3110
    },
    {
      "epoch": 0.86153420105234,
      "grad_norm": 1.541733980178833,
      "learning_rate": 9.572212049480811e-06,
      "loss": 2.7917,
      "step": 3111
    },
    {
      "epoch": 0.8618111326502353,
      "grad_norm": 1.1841416358947754,
      "learning_rate": 9.53458126988388e-06,
      "loss": 2.6935,
      "step": 3112
    },
    {
      "epoch": 0.8620880642481307,
      "grad_norm": 1.7382968664169312,
      "learning_rate": 9.497020900939246e-06,
      "loss": 2.7467,
      "step": 3113
    },
    {
      "epoch": 0.862364995846026,
      "grad_norm": 1.642300009727478,
      "learning_rate": 9.459530971880681e-06,
      "loss": 2.6757,
      "step": 3114
    },
    {
      "epoch": 0.8626419274439213,
      "grad_norm": 2.0042853355407715,
      "learning_rate": 9.422111511887244e-06,
      "loss": 3.1407,
      "step": 3115
    },
    {
      "epoch": 0.8629188590418166,
      "grad_norm": 1.2511892318725586,
      "learning_rate": 9.384762550083037e-06,
      "loss": 3.0645,
      "step": 3116
    },
    {
      "epoch": 0.863195790639712,
      "grad_norm": 1.4200023412704468,
      "learning_rate": 9.347484115537364e-06,
      "loss": 2.8308,
      "step": 3117
    },
    {
      "epoch": 0.8634727222376073,
      "grad_norm": 1.5993705987930298,
      "learning_rate": 9.310276237264615e-06,
      "loss": 2.7952,
      "step": 3118
    },
    {
      "epoch": 0.8637496538355026,
      "grad_norm": 1.4115570783615112,
      "learning_rate": 9.27313894422428e-06,
      "loss": 2.5305,
      "step": 3119
    },
    {
      "epoch": 0.8640265854333979,
      "grad_norm": 0.9652894139289856,
      "learning_rate": 9.23607226532084e-06,
      "loss": 2.5732,
      "step": 3120
    },
    {
      "epoch": 0.8643035170312933,
      "grad_norm": 1.4319514036178589,
      "learning_rate": 9.199076229403924e-06,
      "loss": 2.8168,
      "step": 3121
    },
    {
      "epoch": 0.8645804486291886,
      "grad_norm": 1.3451462984085083,
      "learning_rate": 9.162150865268127e-06,
      "loss": 2.7569,
      "step": 3122
    },
    {
      "epoch": 0.8648573802270839,
      "grad_norm": 2.326890707015991,
      "learning_rate": 9.12529620165299e-06,
      "loss": 3.0164,
      "step": 3123
    },
    {
      "epoch": 0.8651343118249792,
      "grad_norm": 2.1529388427734375,
      "learning_rate": 9.088512267243143e-06,
      "loss": 3.2552,
      "step": 3124
    },
    {
      "epoch": 0.8654112434228746,
      "grad_norm": 5.216430187225342,
      "learning_rate": 9.051799090668046e-06,
      "loss": 4.5724,
      "step": 3125
    },
    {
      "epoch": 0.8656881750207699,
      "grad_norm": 1.3277052640914917,
      "learning_rate": 9.015156700502159e-06,
      "loss": 2.7651,
      "step": 3126
    },
    {
      "epoch": 0.8659651066186652,
      "grad_norm": 1.582177758216858,
      "learning_rate": 8.978585125264816e-06,
      "loss": 2.8223,
      "step": 3127
    },
    {
      "epoch": 0.8662420382165605,
      "grad_norm": 1.732556700706482,
      "learning_rate": 8.94208439342028e-06,
      "loss": 2.858,
      "step": 3128
    },
    {
      "epoch": 0.8665189698144559,
      "grad_norm": 1.2172104120254517,
      "learning_rate": 8.905654533377583e-06,
      "loss": 2.6793,
      "step": 3129
    },
    {
      "epoch": 0.8667959014123512,
      "grad_norm": 1.3058675527572632,
      "learning_rate": 8.869295573490733e-06,
      "loss": 2.6897,
      "step": 3130
    },
    {
      "epoch": 0.8670728330102465,
      "grad_norm": 1.4295732975006104,
      "learning_rate": 8.833007542058402e-06,
      "loss": 2.9477,
      "step": 3131
    },
    {
      "epoch": 0.8673497646081418,
      "grad_norm": 1.7179218530654907,
      "learning_rate": 8.796790467324179e-06,
      "loss": 2.6237,
      "step": 3132
    },
    {
      "epoch": 0.8676266962060372,
      "grad_norm": 1.7100985050201416,
      "learning_rate": 8.760644377476368e-06,
      "loss": 2.709,
      "step": 3133
    },
    {
      "epoch": 0.8679036278039324,
      "grad_norm": 1.3927521705627441,
      "learning_rate": 8.724569300648034e-06,
      "loss": 2.8626,
      "step": 3134
    },
    {
      "epoch": 0.8681805594018277,
      "grad_norm": 1.7715344429016113,
      "learning_rate": 8.688565264916982e-06,
      "loss": 2.8555,
      "step": 3135
    },
    {
      "epoch": 0.868457490999723,
      "grad_norm": 1.1118872165679932,
      "learning_rate": 8.652632298305707e-06,
      "loss": 2.5965,
      "step": 3136
    },
    {
      "epoch": 0.8687344225976183,
      "grad_norm": 1.287585735321045,
      "learning_rate": 8.616770428781419e-06,
      "loss": 3.0108,
      "step": 3137
    },
    {
      "epoch": 0.8690113541955137,
      "grad_norm": 1.5482505559921265,
      "learning_rate": 8.58097968425593e-06,
      "loss": 2.7597,
      "step": 3138
    },
    {
      "epoch": 0.869288285793409,
      "grad_norm": 1.3987518548965454,
      "learning_rate": 8.545260092585805e-06,
      "loss": 2.7233,
      "step": 3139
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 1.880951166152954,
      "learning_rate": 8.509611681572106e-06,
      "loss": 2.9494,
      "step": 3140
    },
    {
      "epoch": 0.8698421489891996,
      "grad_norm": 1.631243348121643,
      "learning_rate": 8.474034478960557e-06,
      "loss": 2.8259,
      "step": 3141
    },
    {
      "epoch": 0.870119080587095,
      "grad_norm": 1.595776081085205,
      "learning_rate": 8.438528512441467e-06,
      "loss": 2.7464,
      "step": 3142
    },
    {
      "epoch": 0.8703960121849903,
      "grad_norm": 1.6883162260055542,
      "learning_rate": 8.403093809649675e-06,
      "loss": 2.6377,
      "step": 3143
    },
    {
      "epoch": 0.8706729437828856,
      "grad_norm": 1.1031101942062378,
      "learning_rate": 8.367730398164574e-06,
      "loss": 2.5506,
      "step": 3144
    },
    {
      "epoch": 0.870949875380781,
      "grad_norm": 1.3834490776062012,
      "learning_rate": 8.33243830551006e-06,
      "loss": 2.6796,
      "step": 3145
    },
    {
      "epoch": 0.8712268069786763,
      "grad_norm": 1.4311162233352661,
      "learning_rate": 8.297217559154535e-06,
      "loss": 2.5472,
      "step": 3146
    },
    {
      "epoch": 0.8715037385765716,
      "grad_norm": 1.4056702852249146,
      "learning_rate": 8.262068186510808e-06,
      "loss": 2.5875,
      "step": 3147
    },
    {
      "epoch": 0.8717806701744669,
      "grad_norm": 1.9589285850524902,
      "learning_rate": 8.226990214936247e-06,
      "loss": 2.8412,
      "step": 3148
    },
    {
      "epoch": 0.8720576017723622,
      "grad_norm": 2.5369226932525635,
      "learning_rate": 8.19198367173255e-06,
      "loss": 3.3846,
      "step": 3149
    },
    {
      "epoch": 0.8723345333702576,
      "grad_norm": 2.807080030441284,
      "learning_rate": 8.157048584145865e-06,
      "loss": 3.6268,
      "step": 3150
    },
    {
      "epoch": 0.8726114649681529,
      "grad_norm": 1.582241177558899,
      "learning_rate": 8.12218497936672e-06,
      "loss": 2.9072,
      "step": 3151
    },
    {
      "epoch": 0.8728883965660482,
      "grad_norm": 1.4820480346679688,
      "learning_rate": 8.08739288453001e-06,
      "loss": 2.7435,
      "step": 3152
    },
    {
      "epoch": 0.8731653281639435,
      "grad_norm": 1.1095006465911865,
      "learning_rate": 8.05267232671495e-06,
      "loss": 2.6025,
      "step": 3153
    },
    {
      "epoch": 0.8734422597618389,
      "grad_norm": 1.2409456968307495,
      "learning_rate": 8.018023332945112e-06,
      "loss": 2.8085,
      "step": 3154
    },
    {
      "epoch": 0.8737191913597342,
      "grad_norm": 1.187393307685852,
      "learning_rate": 7.98344593018836e-06,
      "loss": 2.6978,
      "step": 3155
    },
    {
      "epoch": 0.8739961229576295,
      "grad_norm": 1.2349612712860107,
      "learning_rate": 7.948940145356775e-06,
      "loss": 2.936,
      "step": 3156
    },
    {
      "epoch": 0.8742730545555247,
      "grad_norm": 1.1934999227523804,
      "learning_rate": 7.914506005306832e-06,
      "loss": 2.7566,
      "step": 3157
    },
    {
      "epoch": 0.8745499861534201,
      "grad_norm": 0.7864140868186951,
      "learning_rate": 7.880143536839091e-06,
      "loss": 2.829,
      "step": 3158
    },
    {
      "epoch": 0.8748269177513154,
      "grad_norm": 1.670947551727295,
      "learning_rate": 7.845852766698426e-06,
      "loss": 2.8919,
      "step": 3159
    },
    {
      "epoch": 0.8751038493492107,
      "grad_norm": 1.2872366905212402,
      "learning_rate": 7.811633721573885e-06,
      "loss": 2.9856,
      "step": 3160
    },
    {
      "epoch": 0.875380780947106,
      "grad_norm": 1.2436797618865967,
      "learning_rate": 7.777486428098691e-06,
      "loss": 2.947,
      "step": 3161
    },
    {
      "epoch": 0.8756577125450014,
      "grad_norm": 1.3402278423309326,
      "learning_rate": 7.74341091285018e-06,
      "loss": 2.7753,
      "step": 3162
    },
    {
      "epoch": 0.8759346441428967,
      "grad_norm": 1.3854641914367676,
      "learning_rate": 7.709407202349917e-06,
      "loss": 2.7943,
      "step": 3163
    },
    {
      "epoch": 0.876211575740792,
      "grad_norm": 1.118638277053833,
      "learning_rate": 7.675475323063475e-06,
      "loss": 2.763,
      "step": 3164
    },
    {
      "epoch": 0.8764885073386873,
      "grad_norm": 1.0722171068191528,
      "learning_rate": 7.641615301400585e-06,
      "loss": 2.9056,
      "step": 3165
    },
    {
      "epoch": 0.8767654389365827,
      "grad_norm": 2.5819013118743896,
      "learning_rate": 7.607827163715042e-06,
      "loss": 2.7059,
      "step": 3166
    },
    {
      "epoch": 0.877042370534478,
      "grad_norm": 1.32217538356781,
      "learning_rate": 7.574110936304657e-06,
      "loss": 2.8119,
      "step": 3167
    },
    {
      "epoch": 0.8773193021323733,
      "grad_norm": 1.9732376337051392,
      "learning_rate": 7.540466645411326e-06,
      "loss": 2.4895,
      "step": 3168
    },
    {
      "epoch": 0.8775962337302686,
      "grad_norm": 0.826056957244873,
      "learning_rate": 7.5068943172209025e-06,
      "loss": 2.5382,
      "step": 3169
    },
    {
      "epoch": 0.877873165328164,
      "grad_norm": 1.436314582824707,
      "learning_rate": 7.473393977863297e-06,
      "loss": 2.7361,
      "step": 3170
    },
    {
      "epoch": 0.8781500969260593,
      "grad_norm": 1.0173486471176147,
      "learning_rate": 7.439965653412273e-06,
      "loss": 2.6879,
      "step": 3171
    },
    {
      "epoch": 0.8784270285239546,
      "grad_norm": 1.8656055927276611,
      "learning_rate": 7.406609369885708e-06,
      "loss": 2.7196,
      "step": 3172
    },
    {
      "epoch": 0.8787039601218499,
      "grad_norm": 1.4444622993469238,
      "learning_rate": 7.373325153245259e-06,
      "loss": 2.8648,
      "step": 3173
    },
    {
      "epoch": 0.8789808917197452,
      "grad_norm": 1.5233651399612427,
      "learning_rate": 7.340113029396567e-06,
      "loss": 2.8779,
      "step": 3174
    },
    {
      "epoch": 0.8792578233176406,
      "grad_norm": 1.651416301727295,
      "learning_rate": 7.306973024189145e-06,
      "loss": 3.2433,
      "step": 3175
    },
    {
      "epoch": 0.8795347549155359,
      "grad_norm": 1.8997817039489746,
      "learning_rate": 7.273905163416395e-06,
      "loss": 3.0981,
      "step": 3176
    },
    {
      "epoch": 0.8798116865134312,
      "grad_norm": 0.88230299949646,
      "learning_rate": 7.240909472815538e-06,
      "loss": 2.6861,
      "step": 3177
    },
    {
      "epoch": 0.8800886181113265,
      "grad_norm": 2.475217819213867,
      "learning_rate": 7.207985978067644e-06,
      "loss": 2.6416,
      "step": 3178
    },
    {
      "epoch": 0.8803655497092219,
      "grad_norm": 1.2994290590286255,
      "learning_rate": 7.175134704797592e-06,
      "loss": 2.7834,
      "step": 3179
    },
    {
      "epoch": 0.8806424813071172,
      "grad_norm": 1.7192754745483398,
      "learning_rate": 7.142355678574009e-06,
      "loss": 2.8564,
      "step": 3180
    },
    {
      "epoch": 0.8809194129050124,
      "grad_norm": 1.0855594873428345,
      "learning_rate": 7.109648924909373e-06,
      "loss": 2.6875,
      "step": 3181
    },
    {
      "epoch": 0.8811963445029077,
      "grad_norm": 1.334720492362976,
      "learning_rate": 7.077014469259813e-06,
      "loss": 2.6114,
      "step": 3182
    },
    {
      "epoch": 0.8814732761008031,
      "grad_norm": 1.205359935760498,
      "learning_rate": 7.044452337025265e-06,
      "loss": 2.7585,
      "step": 3183
    },
    {
      "epoch": 0.8817502076986984,
      "grad_norm": 1.6648601293563843,
      "learning_rate": 7.011962553549345e-06,
      "loss": 2.8264,
      "step": 3184
    },
    {
      "epoch": 0.8820271392965937,
      "grad_norm": 0.624901294708252,
      "learning_rate": 6.979545144119348e-06,
      "loss": 2.7539,
      "step": 3185
    },
    {
      "epoch": 0.882304070894489,
      "grad_norm": 1.6340166330337524,
      "learning_rate": 6.9472001339662695e-06,
      "loss": 2.8868,
      "step": 3186
    },
    {
      "epoch": 0.8825810024923844,
      "grad_norm": 1.4930720329284668,
      "learning_rate": 6.914927548264727e-06,
      "loss": 2.7927,
      "step": 3187
    },
    {
      "epoch": 0.8828579340902797,
      "grad_norm": 1.087523102760315,
      "learning_rate": 6.8827274121329944e-06,
      "loss": 2.7179,
      "step": 3188
    },
    {
      "epoch": 0.883134865688175,
      "grad_norm": 1.7288202047348022,
      "learning_rate": 6.8505997506329024e-06,
      "loss": 2.9288,
      "step": 3189
    },
    {
      "epoch": 0.8834117972860703,
      "grad_norm": 1.4685242176055908,
      "learning_rate": 6.818544588769959e-06,
      "loss": 2.745,
      "step": 3190
    },
    {
      "epoch": 0.8836887288839657,
      "grad_norm": 1.3567365407943726,
      "learning_rate": 6.786561951493176e-06,
      "loss": 2.9668,
      "step": 3191
    },
    {
      "epoch": 0.883965660481861,
      "grad_norm": 1.338721752166748,
      "learning_rate": 6.754651863695138e-06,
      "loss": 2.517,
      "step": 3192
    },
    {
      "epoch": 0.8842425920797563,
      "grad_norm": 1.0524381399154663,
      "learning_rate": 6.72281435021197e-06,
      "loss": 2.6162,
      "step": 3193
    },
    {
      "epoch": 0.8845195236776516,
      "grad_norm": 2.0696678161621094,
      "learning_rate": 6.691049435823327e-06,
      "loss": 2.6048,
      "step": 3194
    },
    {
      "epoch": 0.884796455275547,
      "grad_norm": 1.5093910694122314,
      "learning_rate": 6.659357145252287e-06,
      "loss": 2.8598,
      "step": 3195
    },
    {
      "epoch": 0.8850733868734423,
      "grad_norm": 0.9233435988426208,
      "learning_rate": 6.627737503165532e-06,
      "loss": 2.7454,
      "step": 3196
    },
    {
      "epoch": 0.8853503184713376,
      "grad_norm": 1.0932697057724,
      "learning_rate": 6.5961905341730635e-06,
      "loss": 2.6687,
      "step": 3197
    },
    {
      "epoch": 0.8856272500692329,
      "grad_norm": 1.8618035316467285,
      "learning_rate": 6.5647162628283965e-06,
      "loss": 2.8007,
      "step": 3198
    },
    {
      "epoch": 0.8859041816671283,
      "grad_norm": 2.263721466064453,
      "learning_rate": 6.533314713628458e-06,
      "loss": 3.4359,
      "step": 3199
    },
    {
      "epoch": 0.8861811132650236,
      "grad_norm": 3.9213695526123047,
      "learning_rate": 6.5019859110135635e-06,
      "loss": 4.21,
      "step": 3200
    },
    {
      "epoch": 0.8864580448629189,
      "grad_norm": 1.4590137004852295,
      "learning_rate": 6.470729879367399e-06,
      "loss": 2.9189,
      "step": 3201
    },
    {
      "epoch": 0.8867349764608142,
      "grad_norm": 1.1353839635849,
      "learning_rate": 6.439546643017047e-06,
      "loss": 2.6915,
      "step": 3202
    },
    {
      "epoch": 0.8870119080587096,
      "grad_norm": 1.0528885126113892,
      "learning_rate": 6.4084362262329055e-06,
      "loss": 2.6218,
      "step": 3203
    },
    {
      "epoch": 0.8872888396566048,
      "grad_norm": 1.8522472381591797,
      "learning_rate": 6.377398653228661e-06,
      "loss": 2.6753,
      "step": 3204
    },
    {
      "epoch": 0.8875657712545001,
      "grad_norm": 0.7795437574386597,
      "learning_rate": 6.34643394816139e-06,
      "loss": 2.7333,
      "step": 3205
    },
    {
      "epoch": 0.8878427028523954,
      "grad_norm": 1.7405046224594116,
      "learning_rate": 6.315542135131381e-06,
      "loss": 2.9016,
      "step": 3206
    },
    {
      "epoch": 0.8881196344502907,
      "grad_norm": 0.8221628069877625,
      "learning_rate": 6.284723238182233e-06,
      "loss": 2.8126,
      "step": 3207
    },
    {
      "epoch": 0.8883965660481861,
      "grad_norm": 1.2659366130828857,
      "learning_rate": 6.253977281300749e-06,
      "loss": 2.8429,
      "step": 3208
    },
    {
      "epoch": 0.8886734976460814,
      "grad_norm": 1.5786124467849731,
      "learning_rate": 6.22330428841702e-06,
      "loss": 2.8225,
      "step": 3209
    },
    {
      "epoch": 0.8889504292439767,
      "grad_norm": 1.8932335376739502,
      "learning_rate": 6.1927042834042935e-06,
      "loss": 3.0667,
      "step": 3210
    },
    {
      "epoch": 0.889227360841872,
      "grad_norm": 1.5503121614456177,
      "learning_rate": 6.1621772900790405e-06,
      "loss": 2.7404,
      "step": 3211
    },
    {
      "epoch": 0.8895042924397674,
      "grad_norm": 1.1766709089279175,
      "learning_rate": 6.131723332200923e-06,
      "loss": 2.7821,
      "step": 3212
    },
    {
      "epoch": 0.8897812240376627,
      "grad_norm": 1.92313814163208,
      "learning_rate": 6.101342433472679e-06,
      "loss": 2.7026,
      "step": 3213
    },
    {
      "epoch": 0.890058155635558,
      "grad_norm": 1.6753193140029907,
      "learning_rate": 6.071034617540294e-06,
      "loss": 2.9872,
      "step": 3214
    },
    {
      "epoch": 0.8903350872334533,
      "grad_norm": 0.690635085105896,
      "learning_rate": 6.040799907992778e-06,
      "loss": 2.8038,
      "step": 3215
    },
    {
      "epoch": 0.8906120188313487,
      "grad_norm": 0.9176428914070129,
      "learning_rate": 6.010638328362283e-06,
      "loss": 2.8574,
      "step": 3216
    },
    {
      "epoch": 0.890888950429244,
      "grad_norm": 1.0934897661209106,
      "learning_rate": 5.980549902124055e-06,
      "loss": 2.855,
      "step": 3217
    },
    {
      "epoch": 0.8911658820271393,
      "grad_norm": 0.8238447904586792,
      "learning_rate": 5.950534652696382e-06,
      "loss": 2.7193,
      "step": 3218
    },
    {
      "epoch": 0.8914428136250346,
      "grad_norm": 1.7612361907958984,
      "learning_rate": 5.9205926034406e-06,
      "loss": 2.7353,
      "step": 3219
    },
    {
      "epoch": 0.89171974522293,
      "grad_norm": 1.9202384948730469,
      "learning_rate": 5.890723777661078e-06,
      "loss": 2.6319,
      "step": 3220
    },
    {
      "epoch": 0.8919966768208253,
      "grad_norm": 1.6076610088348389,
      "learning_rate": 5.8609281986051975e-06,
      "loss": 2.529,
      "step": 3221
    },
    {
      "epoch": 0.8922736084187206,
      "grad_norm": 1.4485604763031006,
      "learning_rate": 5.831205889463287e-06,
      "loss": 2.9042,
      "step": 3222
    },
    {
      "epoch": 0.8925505400166159,
      "grad_norm": 1.318128228187561,
      "learning_rate": 5.801556873368741e-06,
      "loss": 2.9158,
      "step": 3223
    },
    {
      "epoch": 0.8928274716145113,
      "grad_norm": 1.570043921470642,
      "learning_rate": 5.771981173397811e-06,
      "loss": 3.2793,
      "step": 3224
    },
    {
      "epoch": 0.8931044032124066,
      "grad_norm": 2.3622982501983643,
      "learning_rate": 5.74247881256974e-06,
      "loss": 3.4243,
      "step": 3225
    },
    {
      "epoch": 0.8933813348103019,
      "grad_norm": 1.2357228994369507,
      "learning_rate": 5.713049813846683e-06,
      "loss": 2.8305,
      "step": 3226
    },
    {
      "epoch": 0.8936582664081971,
      "grad_norm": 1.3328142166137695,
      "learning_rate": 5.683694200133705e-06,
      "loss": 2.5218,
      "step": 3227
    },
    {
      "epoch": 0.8939351980060924,
      "grad_norm": 1.4435487985610962,
      "learning_rate": 5.654411994278685e-06,
      "loss": 2.6991,
      "step": 3228
    },
    {
      "epoch": 0.8942121296039878,
      "grad_norm": 1.5753285884857178,
      "learning_rate": 5.625203219072495e-06,
      "loss": 2.7081,
      "step": 3229
    },
    {
      "epoch": 0.8944890612018831,
      "grad_norm": 2.0387070178985596,
      "learning_rate": 5.596067897248725e-06,
      "loss": 2.5834,
      "step": 3230
    },
    {
      "epoch": 0.8947659927997784,
      "grad_norm": 1.1547656059265137,
      "learning_rate": 5.5670060514838805e-06,
      "loss": 2.8424,
      "step": 3231
    },
    {
      "epoch": 0.8950429243976737,
      "grad_norm": 1.4619250297546387,
      "learning_rate": 5.538017704397236e-06,
      "loss": 2.8464,
      "step": 3232
    },
    {
      "epoch": 0.8953198559955691,
      "grad_norm": 0.9576718807220459,
      "learning_rate": 5.509102878550887e-06,
      "loss": 2.725,
      "step": 3233
    },
    {
      "epoch": 0.8955967875934644,
      "grad_norm": 1.6311956644058228,
      "learning_rate": 5.480261596449698e-06,
      "loss": 2.8379,
      "step": 3234
    },
    {
      "epoch": 0.8958737191913597,
      "grad_norm": 1.448899269104004,
      "learning_rate": 5.451493880541292e-06,
      "loss": 2.5566,
      "step": 3235
    },
    {
      "epoch": 0.896150650789255,
      "grad_norm": 1.109121322631836,
      "learning_rate": 5.422799753216023e-06,
      "loss": 2.6754,
      "step": 3236
    },
    {
      "epoch": 0.8964275823871504,
      "grad_norm": 1.8016066551208496,
      "learning_rate": 5.394179236806973e-06,
      "loss": 2.9337,
      "step": 3237
    },
    {
      "epoch": 0.8967045139850457,
      "grad_norm": 1.277727484703064,
      "learning_rate": 5.365632353589967e-06,
      "loss": 2.7793,
      "step": 3238
    },
    {
      "epoch": 0.896981445582941,
      "grad_norm": 1.4791587591171265,
      "learning_rate": 5.337159125783453e-06,
      "loss": 2.8146,
      "step": 3239
    },
    {
      "epoch": 0.8972583771808363,
      "grad_norm": 1.8600664138793945,
      "learning_rate": 5.308759575548617e-06,
      "loss": 3.0441,
      "step": 3240
    },
    {
      "epoch": 0.8975353087787317,
      "grad_norm": 1.9175560474395752,
      "learning_rate": 5.280433724989264e-06,
      "loss": 3.1218,
      "step": 3241
    },
    {
      "epoch": 0.897812240376627,
      "grad_norm": 1.1490317583084106,
      "learning_rate": 5.252181596151862e-06,
      "loss": 2.7404,
      "step": 3242
    },
    {
      "epoch": 0.8980891719745223,
      "grad_norm": 1.471359372138977,
      "learning_rate": 5.224003211025452e-06,
      "loss": 2.5608,
      "step": 3243
    },
    {
      "epoch": 0.8983661035724176,
      "grad_norm": 1.593123435974121,
      "learning_rate": 5.195898591541748e-06,
      "loss": 2.506,
      "step": 3244
    },
    {
      "epoch": 0.898643035170313,
      "grad_norm": 1.383209466934204,
      "learning_rate": 5.167867759575007e-06,
      "loss": 2.5702,
      "step": 3245
    },
    {
      "epoch": 0.8989199667682083,
      "grad_norm": 1.4914791584014893,
      "learning_rate": 5.139910736942044e-06,
      "loss": 3.1071,
      "step": 3246
    },
    {
      "epoch": 0.8991968983661036,
      "grad_norm": 1.4775235652923584,
      "learning_rate": 5.112027545402276e-06,
      "loss": 2.6725,
      "step": 3247
    },
    {
      "epoch": 0.8994738299639989,
      "grad_norm": 0.9750004410743713,
      "learning_rate": 5.084218206657609e-06,
      "loss": 2.5507,
      "step": 3248
    },
    {
      "epoch": 0.8997507615618943,
      "grad_norm": 1.4835078716278076,
      "learning_rate": 5.056482742352486e-06,
      "loss": 2.9729,
      "step": 3249
    },
    {
      "epoch": 0.9000276931597895,
      "grad_norm": 3.6488654613494873,
      "learning_rate": 5.028821174073861e-06,
      "loss": 4.0052,
      "step": 3250
    },
    {
      "epoch": 0.9003046247576848,
      "grad_norm": 1.391434907913208,
      "learning_rate": 5.001233523351156e-06,
      "loss": 2.7086,
      "step": 3251
    },
    {
      "epoch": 0.9005815563555801,
      "grad_norm": 1.7492971420288086,
      "learning_rate": 4.973719811656286e-06,
      "loss": 3.0276,
      "step": 3252
    },
    {
      "epoch": 0.9008584879534755,
      "grad_norm": 1.4157118797302246,
      "learning_rate": 4.9462800604035964e-06,
      "loss": 2.6521,
      "step": 3253
    },
    {
      "epoch": 0.9011354195513708,
      "grad_norm": 0.8659539222717285,
      "learning_rate": 4.9189142909498945e-06,
      "loss": 3.0122,
      "step": 3254
    },
    {
      "epoch": 0.9014123511492661,
      "grad_norm": 1.3970427513122559,
      "learning_rate": 4.891622524594342e-06,
      "loss": 2.6576,
      "step": 3255
    },
    {
      "epoch": 0.9016892827471614,
      "grad_norm": 1.1650993824005127,
      "learning_rate": 4.864404782578591e-06,
      "loss": 2.6377,
      "step": 3256
    },
    {
      "epoch": 0.9019662143450567,
      "grad_norm": 1.8018288612365723,
      "learning_rate": 4.83726108608662e-06,
      "loss": 2.9516,
      "step": 3257
    },
    {
      "epoch": 0.9022431459429521,
      "grad_norm": 1.2030545473098755,
      "learning_rate": 4.810191456244784e-06,
      "loss": 2.6211,
      "step": 3258
    },
    {
      "epoch": 0.9025200775408474,
      "grad_norm": 1.2440369129180908,
      "learning_rate": 4.783195914121818e-06,
      "loss": 2.7023,
      "step": 3259
    },
    {
      "epoch": 0.9027970091387427,
      "grad_norm": 1.9231184720993042,
      "learning_rate": 4.756274480728773e-06,
      "loss": 2.9374,
      "step": 3260
    },
    {
      "epoch": 0.903073940736638,
      "grad_norm": 1.1988775730133057,
      "learning_rate": 4.729427177018986e-06,
      "loss": 2.9222,
      "step": 3261
    },
    {
      "epoch": 0.9033508723345334,
      "grad_norm": 1.8029758930206299,
      "learning_rate": 4.7026540238881976e-06,
      "loss": 2.9991,
      "step": 3262
    },
    {
      "epoch": 0.9036278039324287,
      "grad_norm": 1.3029099702835083,
      "learning_rate": 4.6759550421743295e-06,
      "loss": 2.8718,
      "step": 3263
    },
    {
      "epoch": 0.903904735530324,
      "grad_norm": 1.1065994501113892,
      "learning_rate": 4.649330252657613e-06,
      "loss": 2.7638,
      "step": 3264
    },
    {
      "epoch": 0.9041816671282193,
      "grad_norm": 1.46235191822052,
      "learning_rate": 4.622779676060562e-06,
      "loss": 2.6686,
      "step": 3265
    },
    {
      "epoch": 0.9044585987261147,
      "grad_norm": 0.9452422857284546,
      "learning_rate": 4.596303333047891e-06,
      "loss": 2.6912,
      "step": 3266
    },
    {
      "epoch": 0.90473553032401,
      "grad_norm": 1.7004419565200806,
      "learning_rate": 4.56990124422656e-06,
      "loss": 2.913,
      "step": 3267
    },
    {
      "epoch": 0.9050124619219053,
      "grad_norm": 1.664442777633667,
      "learning_rate": 4.54357343014572e-06,
      "loss": 2.8846,
      "step": 3268
    },
    {
      "epoch": 0.9052893935198006,
      "grad_norm": 1.6478939056396484,
      "learning_rate": 4.517319911296747e-06,
      "loss": 2.6547,
      "step": 3269
    },
    {
      "epoch": 0.905566325117696,
      "grad_norm": 0.8395318984985352,
      "learning_rate": 4.491140708113117e-06,
      "loss": 2.5089,
      "step": 3270
    },
    {
      "epoch": 0.9058432567155913,
      "grad_norm": 1.541677474975586,
      "learning_rate": 4.465035840970577e-06,
      "loss": 2.8615,
      "step": 3271
    },
    {
      "epoch": 0.9061201883134866,
      "grad_norm": 1.2489078044891357,
      "learning_rate": 4.439005330186896e-06,
      "loss": 2.9873,
      "step": 3272
    },
    {
      "epoch": 0.9063971199113819,
      "grad_norm": 1.431944727897644,
      "learning_rate": 4.413049196022057e-06,
      "loss": 2.8534,
      "step": 3273
    },
    {
      "epoch": 0.9066740515092772,
      "grad_norm": 1.1621243953704834,
      "learning_rate": 4.387167458678121e-06,
      "loss": 2.9404,
      "step": 3274
    },
    {
      "epoch": 0.9069509831071725,
      "grad_norm": 2.5406911373138428,
      "learning_rate": 4.361360138299242e-06,
      "loss": 3.4224,
      "step": 3275
    },
    {
      "epoch": 0.9072279147050678,
      "grad_norm": 1.2161071300506592,
      "learning_rate": 4.335627254971675e-06,
      "loss": 2.9378,
      "step": 3276
    },
    {
      "epoch": 0.9075048463029631,
      "grad_norm": 1.362507700920105,
      "learning_rate": 4.3099688287237204e-06,
      "loss": 2.7757,
      "step": 3277
    },
    {
      "epoch": 0.9077817779008585,
      "grad_norm": 1.1091976165771484,
      "learning_rate": 4.284384879525749e-06,
      "loss": 2.7355,
      "step": 3278
    },
    {
      "epoch": 0.9080587094987538,
      "grad_norm": 1.3528761863708496,
      "learning_rate": 4.2588754272900985e-06,
      "loss": 2.7222,
      "step": 3279
    },
    {
      "epoch": 0.9083356410966491,
      "grad_norm": 1.0884292125701904,
      "learning_rate": 4.233440491871232e-06,
      "loss": 2.7229,
      "step": 3280
    },
    {
      "epoch": 0.9086125726945444,
      "grad_norm": 1.2718223333358765,
      "learning_rate": 4.208080093065536e-06,
      "loss": 2.839,
      "step": 3281
    },
    {
      "epoch": 0.9088895042924398,
      "grad_norm": 1.9365698099136353,
      "learning_rate": 4.1827942506113995e-06,
      "loss": 2.8317,
      "step": 3282
    },
    {
      "epoch": 0.9091664358903351,
      "grad_norm": 1.0091325044631958,
      "learning_rate": 4.15758298418919e-06,
      "loss": 2.8692,
      "step": 3283
    },
    {
      "epoch": 0.9094433674882304,
      "grad_norm": 1.2271143198013306,
      "learning_rate": 4.132446313421246e-06,
      "loss": 2.6124,
      "step": 3284
    },
    {
      "epoch": 0.9097202990861257,
      "grad_norm": 1.6683666706085205,
      "learning_rate": 4.107384257871816e-06,
      "loss": 3.0068,
      "step": 3285
    },
    {
      "epoch": 0.909997230684021,
      "grad_norm": 0.7507969737052917,
      "learning_rate": 4.082396837047109e-06,
      "loss": 2.7824,
      "step": 3286
    },
    {
      "epoch": 0.9102741622819164,
      "grad_norm": 1.477518916130066,
      "learning_rate": 4.057484070395213e-06,
      "loss": 3.1318,
      "step": 3287
    },
    {
      "epoch": 0.9105510938798117,
      "grad_norm": 1.1982351541519165,
      "learning_rate": 4.0326459773061045e-06,
      "loss": 2.8544,
      "step": 3288
    },
    {
      "epoch": 0.910828025477707,
      "grad_norm": 1.218491554260254,
      "learning_rate": 4.00788257711171e-06,
      "loss": 2.7816,
      "step": 3289
    },
    {
      "epoch": 0.9111049570756024,
      "grad_norm": 1.1615190505981445,
      "learning_rate": 3.9831938890857346e-06,
      "loss": 2.8993,
      "step": 3290
    },
    {
      "epoch": 0.9113818886734977,
      "grad_norm": 1.3096328973770142,
      "learning_rate": 3.958579932443773e-06,
      "loss": 2.8446,
      "step": 3291
    },
    {
      "epoch": 0.911658820271393,
      "grad_norm": 1.5708379745483398,
      "learning_rate": 3.934040726343258e-06,
      "loss": 2.8288,
      "step": 3292
    },
    {
      "epoch": 0.9119357518692883,
      "grad_norm": 1.3243839740753174,
      "learning_rate": 3.909576289883454e-06,
      "loss": 2.856,
      "step": 3293
    },
    {
      "epoch": 0.9122126834671836,
      "grad_norm": 1.4913750886917114,
      "learning_rate": 3.885186642105376e-06,
      "loss": 2.6688,
      "step": 3294
    },
    {
      "epoch": 0.912489615065079,
      "grad_norm": 1.4231692552566528,
      "learning_rate": 3.8608718019919275e-06,
      "loss": 2.7812,
      "step": 3295
    },
    {
      "epoch": 0.9127665466629743,
      "grad_norm": 1.3118404150009155,
      "learning_rate": 3.836631788467671e-06,
      "loss": 2.483,
      "step": 3296
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 1.285291314125061,
      "learning_rate": 3.8124666203990135e-06,
      "loss": 2.5784,
      "step": 3297
    },
    {
      "epoch": 0.9133204098587648,
      "grad_norm": 1.377078652381897,
      "learning_rate": 3.788376316594089e-06,
      "loss": 2.8021,
      "step": 3298
    },
    {
      "epoch": 0.9135973414566602,
      "grad_norm": 1.9740731716156006,
      "learning_rate": 3.7643608958027543e-06,
      "loss": 2.8414,
      "step": 3299
    },
    {
      "epoch": 0.9138742730545555,
      "grad_norm": 4.069131374359131,
      "learning_rate": 3.74042037671658e-06,
      "loss": 4.0822,
      "step": 3300
    },
    {
      "epoch": 0.9141512046524508,
      "grad_norm": 1.1507346630096436,
      "learning_rate": 3.7165547779688616e-06,
      "loss": 2.5579,
      "step": 3301
    },
    {
      "epoch": 0.9144281362503461,
      "grad_norm": 1.3568830490112305,
      "learning_rate": 3.692764118134573e-06,
      "loss": 2.8035,
      "step": 3302
    },
    {
      "epoch": 0.9147050678482415,
      "grad_norm": 1.8859748840332031,
      "learning_rate": 3.669048415730314e-06,
      "loss": 2.5274,
      "step": 3303
    },
    {
      "epoch": 0.9149819994461368,
      "grad_norm": 1.5772771835327148,
      "learning_rate": 3.6454076892144418e-06,
      "loss": 2.6752,
      "step": 3304
    },
    {
      "epoch": 0.9152589310440321,
      "grad_norm": 0.878659725189209,
      "learning_rate": 3.62184195698686e-06,
      "loss": 2.6252,
      "step": 3305
    },
    {
      "epoch": 0.9155358626419274,
      "grad_norm": 1.0205022096633911,
      "learning_rate": 3.598351237389175e-06,
      "loss": 2.5583,
      "step": 3306
    },
    {
      "epoch": 0.9158127942398228,
      "grad_norm": 1.0380001068115234,
      "learning_rate": 3.5749355487045722e-06,
      "loss": 2.7153,
      "step": 3307
    },
    {
      "epoch": 0.9160897258377181,
      "grad_norm": 1.053176760673523,
      "learning_rate": 3.5515949091578514e-06,
      "loss": 2.5989,
      "step": 3308
    },
    {
      "epoch": 0.9163666574356134,
      "grad_norm": 1.5257909297943115,
      "learning_rate": 3.5283293369154036e-06,
      "loss": 2.8998,
      "step": 3309
    },
    {
      "epoch": 0.9166435890335087,
      "grad_norm": 1.6158723831176758,
      "learning_rate": 3.5051388500851766e-06,
      "loss": 2.8025,
      "step": 3310
    },
    {
      "epoch": 0.9169205206314041,
      "grad_norm": 0.8846004009246826,
      "learning_rate": 3.4820234667166996e-06,
      "loss": 2.7201,
      "step": 3311
    },
    {
      "epoch": 0.9171974522292994,
      "grad_norm": 1.761206865310669,
      "learning_rate": 3.458983204801014e-06,
      "loss": 2.8465,
      "step": 3312
    },
    {
      "epoch": 0.9174743838271947,
      "grad_norm": 1.3853119611740112,
      "learning_rate": 3.436018082270753e-06,
      "loss": 2.8522,
      "step": 3313
    },
    {
      "epoch": 0.91775131542509,
      "grad_norm": 1.5413858890533447,
      "learning_rate": 3.4131281170000083e-06,
      "loss": 2.688,
      "step": 3314
    },
    {
      "epoch": 0.9180282470229854,
      "grad_norm": 1.460121512413025,
      "learning_rate": 3.3903133268043952e-06,
      "loss": 2.881,
      "step": 3315
    },
    {
      "epoch": 0.9183051786208807,
      "grad_norm": 1.2880120277404785,
      "learning_rate": 3.3675737294410425e-06,
      "loss": 3.0607,
      "step": 3316
    },
    {
      "epoch": 0.918582110218776,
      "grad_norm": 1.3540161848068237,
      "learning_rate": 3.3449093426085154e-06,
      "loss": 2.8574,
      "step": 3317
    },
    {
      "epoch": 0.9188590418166713,
      "grad_norm": 1.5042144060134888,
      "learning_rate": 3.3223201839468587e-06,
      "loss": 3.0026,
      "step": 3318
    },
    {
      "epoch": 0.9191359734145667,
      "grad_norm": 1.1813321113586426,
      "learning_rate": 3.2998062710375864e-06,
      "loss": 2.7569,
      "step": 3319
    },
    {
      "epoch": 0.9194129050124619,
      "grad_norm": 2.0243005752563477,
      "learning_rate": 3.2773676214036374e-06,
      "loss": 2.8165,
      "step": 3320
    },
    {
      "epoch": 0.9196898366103572,
      "grad_norm": 1.6620597839355469,
      "learning_rate": 3.2550042525093196e-06,
      "loss": 2.6096,
      "step": 3321
    },
    {
      "epoch": 0.9199667682082525,
      "grad_norm": 1.7305388450622559,
      "learning_rate": 3.232716181760442e-06,
      "loss": 2.7571,
      "step": 3322
    },
    {
      "epoch": 0.9202436998061478,
      "grad_norm": 1.4501975774765015,
      "learning_rate": 3.2105034265041302e-06,
      "loss": 2.744,
      "step": 3323
    },
    {
      "epoch": 0.9205206314040432,
      "grad_norm": 1.5626119375228882,
      "learning_rate": 3.188366004028931e-06,
      "loss": 2.8442,
      "step": 3324
    },
    {
      "epoch": 0.9207975630019385,
      "grad_norm": 4.155477523803711,
      "learning_rate": 3.16630393156474e-06,
      "loss": 4.1207,
      "step": 3325
    },
    {
      "epoch": 0.9210744945998338,
      "grad_norm": 2.5009517669677734,
      "learning_rate": 3.1443172262828223e-06,
      "loss": 3.1146,
      "step": 3326
    },
    {
      "epoch": 0.9213514261977291,
      "grad_norm": 1.2323756217956543,
      "learning_rate": 3.1224059052957556e-06,
      "loss": 2.6249,
      "step": 3327
    },
    {
      "epoch": 0.9216283577956245,
      "grad_norm": 1.3302520513534546,
      "learning_rate": 3.1005699856574978e-06,
      "loss": 2.5258,
      "step": 3328
    },
    {
      "epoch": 0.9219052893935198,
      "grad_norm": 1.6659871339797974,
      "learning_rate": 3.0788094843632655e-06,
      "loss": 2.7854,
      "step": 3329
    },
    {
      "epoch": 0.9221822209914151,
      "grad_norm": 1.27170991897583,
      "learning_rate": 3.0571244183495775e-06,
      "loss": 2.5841,
      "step": 3330
    },
    {
      "epoch": 0.9224591525893104,
      "grad_norm": 1.3297101259231567,
      "learning_rate": 3.0355148044943105e-06,
      "loss": 2.8057,
      "step": 3331
    },
    {
      "epoch": 0.9227360841872058,
      "grad_norm": 1.2992569208145142,
      "learning_rate": 3.0139806596165334e-06,
      "loss": 2.5827,
      "step": 3332
    },
    {
      "epoch": 0.9230130157851011,
      "grad_norm": 1.3038079738616943,
      "learning_rate": 2.9925220004766054e-06,
      "loss": 2.7893,
      "step": 3333
    },
    {
      "epoch": 0.9232899473829964,
      "grad_norm": 1.5119189023971558,
      "learning_rate": 2.9711388437761445e-06,
      "loss": 2.4833,
      "step": 3334
    },
    {
      "epoch": 0.9235668789808917,
      "grad_norm": 1.4106827974319458,
      "learning_rate": 2.9498312061580047e-06,
      "loss": 2.6852,
      "step": 3335
    },
    {
      "epoch": 0.9238438105787871,
      "grad_norm": 1.4682044982910156,
      "learning_rate": 2.9285991042062313e-06,
      "loss": 2.7403,
      "step": 3336
    },
    {
      "epoch": 0.9241207421766824,
      "grad_norm": 1.1259634494781494,
      "learning_rate": 2.907442554446138e-06,
      "loss": 2.8119,
      "step": 3337
    },
    {
      "epoch": 0.9243976737745777,
      "grad_norm": 1.324251651763916,
      "learning_rate": 2.886361573344165e-06,
      "loss": 2.9538,
      "step": 3338
    },
    {
      "epoch": 0.924674605372473,
      "grad_norm": 1.2294390201568604,
      "learning_rate": 2.8653561773079764e-06,
      "loss": 2.7142,
      "step": 3339
    },
    {
      "epoch": 0.9249515369703684,
      "grad_norm": 1.514574646949768,
      "learning_rate": 2.844426382686416e-06,
      "loss": 2.6999,
      "step": 3340
    },
    {
      "epoch": 0.9252284685682637,
      "grad_norm": 2.112339735031128,
      "learning_rate": 2.8235722057694534e-06,
      "loss": 2.8516,
      "step": 3341
    },
    {
      "epoch": 0.925505400166159,
      "grad_norm": 1.1753405332565308,
      "learning_rate": 2.802793662788239e-06,
      "loss": 2.7128,
      "step": 3342
    },
    {
      "epoch": 0.9257823317640543,
      "grad_norm": 1.5247375965118408,
      "learning_rate": 2.7820907699150246e-06,
      "loss": 2.8626,
      "step": 3343
    },
    {
      "epoch": 0.9260592633619495,
      "grad_norm": 1.4732311964035034,
      "learning_rate": 2.7614635432632097e-06,
      "loss": 2.6998,
      "step": 3344
    },
    {
      "epoch": 0.9263361949598449,
      "grad_norm": 1.4961967468261719,
      "learning_rate": 2.740911998887252e-06,
      "loss": 2.6971,
      "step": 3345
    },
    {
      "epoch": 0.9266131265577402,
      "grad_norm": 1.5658138990402222,
      "learning_rate": 2.7204361527827903e-06,
      "loss": 2.8843,
      "step": 3346
    },
    {
      "epoch": 0.9268900581556355,
      "grad_norm": 1.5200576782226562,
      "learning_rate": 2.700036020886465e-06,
      "loss": 2.7221,
      "step": 3347
    },
    {
      "epoch": 0.9271669897535308,
      "grad_norm": 1.2744560241699219,
      "learning_rate": 2.679711619076031e-06,
      "loss": 2.4677,
      "step": 3348
    },
    {
      "epoch": 0.9274439213514262,
      "grad_norm": 1.6359666585922241,
      "learning_rate": 2.6594629631702783e-06,
      "loss": 3.1598,
      "step": 3349
    },
    {
      "epoch": 0.9277208529493215,
      "grad_norm": 2.137948751449585,
      "learning_rate": 2.639290068929057e-06,
      "loss": 2.9673,
      "step": 3350
    },
    {
      "epoch": 0.9279977845472168,
      "grad_norm": 1.848804235458374,
      "learning_rate": 2.619192952053251e-06,
      "loss": 2.8939,
      "step": 3351
    },
    {
      "epoch": 0.9282747161451121,
      "grad_norm": 1.4783754348754883,
      "learning_rate": 2.599171628184749e-06,
      "loss": 2.8871,
      "step": 3352
    },
    {
      "epoch": 0.9285516477430075,
      "grad_norm": 1.6100101470947266,
      "learning_rate": 2.5792261129064854e-06,
      "loss": 2.758,
      "step": 3353
    },
    {
      "epoch": 0.9288285793409028,
      "grad_norm": 1.2014853954315186,
      "learning_rate": 2.5593564217423314e-06,
      "loss": 2.6645,
      "step": 3354
    },
    {
      "epoch": 0.9291055109387981,
      "grad_norm": 1.5369969606399536,
      "learning_rate": 2.5395625701572167e-06,
      "loss": 2.8224,
      "step": 3355
    },
    {
      "epoch": 0.9293824425366934,
      "grad_norm": 1.3187780380249023,
      "learning_rate": 2.519844573556984e-06,
      "loss": 3.0235,
      "step": 3356
    },
    {
      "epoch": 0.9296593741345888,
      "grad_norm": 1.7744249105453491,
      "learning_rate": 2.500202447288458e-06,
      "loss": 2.8146,
      "step": 3357
    },
    {
      "epoch": 0.9299363057324841,
      "grad_norm": 1.2549601793289185,
      "learning_rate": 2.4806362066394195e-06,
      "loss": 2.6661,
      "step": 3358
    },
    {
      "epoch": 0.9302132373303794,
      "grad_norm": 1.2055906057357788,
      "learning_rate": 2.461145866838599e-06,
      "loss": 2.7666,
      "step": 3359
    },
    {
      "epoch": 0.9304901689282747,
      "grad_norm": 1.3747657537460327,
      "learning_rate": 2.4417314430555837e-06,
      "loss": 2.7744,
      "step": 3360
    },
    {
      "epoch": 0.9307671005261701,
      "grad_norm": 1.4448826313018799,
      "learning_rate": 2.4223929504009646e-06,
      "loss": 2.8195,
      "step": 3361
    },
    {
      "epoch": 0.9310440321240654,
      "grad_norm": 1.956508755683899,
      "learning_rate": 2.4031304039261905e-06,
      "loss": 2.7472,
      "step": 3362
    },
    {
      "epoch": 0.9313209637219607,
      "grad_norm": 1.444071650505066,
      "learning_rate": 2.3839438186235684e-06,
      "loss": 2.8139,
      "step": 3363
    },
    {
      "epoch": 0.931597895319856,
      "grad_norm": 1.4014155864715576,
      "learning_rate": 2.364833209426376e-06,
      "loss": 3.0162,
      "step": 3364
    },
    {
      "epoch": 0.9318748269177514,
      "grad_norm": 1.24951171875,
      "learning_rate": 2.3457985912086476e-06,
      "loss": 2.7421,
      "step": 3365
    },
    {
      "epoch": 0.9321517585156467,
      "grad_norm": 1.0533653497695923,
      "learning_rate": 2.3268399787853557e-06,
      "loss": 2.7673,
      "step": 3366
    },
    {
      "epoch": 0.9324286901135419,
      "grad_norm": 0.9333207607269287,
      "learning_rate": 2.307957386912263e-06,
      "loss": 2.517,
      "step": 3367
    },
    {
      "epoch": 0.9327056217114372,
      "grad_norm": 2.0277838706970215,
      "learning_rate": 2.289150830286013e-06,
      "loss": 2.9124,
      "step": 3368
    },
    {
      "epoch": 0.9329825533093326,
      "grad_norm": 1.2337300777435303,
      "learning_rate": 2.270420323544009e-06,
      "loss": 2.6665,
      "step": 3369
    },
    {
      "epoch": 0.9332594849072279,
      "grad_norm": 1.0412017107009888,
      "learning_rate": 2.251765881264534e-06,
      "loss": 2.511,
      "step": 3370
    },
    {
      "epoch": 0.9335364165051232,
      "grad_norm": 1.042595624923706,
      "learning_rate": 2.2331875179666175e-06,
      "loss": 2.778,
      "step": 3371
    },
    {
      "epoch": 0.9338133481030185,
      "grad_norm": 1.1678199768066406,
      "learning_rate": 2.214685248110071e-06,
      "loss": 2.5503,
      "step": 3372
    },
    {
      "epoch": 0.9340902797009139,
      "grad_norm": 2.1503164768218994,
      "learning_rate": 2.1962590860955314e-06,
      "loss": 3.1102,
      "step": 3373
    },
    {
      "epoch": 0.9343672112988092,
      "grad_norm": 1.0278449058532715,
      "learning_rate": 2.177909046264348e-06,
      "loss": 2.8729,
      "step": 3374
    },
    {
      "epoch": 0.9346441428967045,
      "grad_norm": 2.8905749320983887,
      "learning_rate": 2.1596351428986637e-06,
      "loss": 3.2821,
      "step": 3375
    },
    {
      "epoch": 0.9349210744945998,
      "grad_norm": 0.8825234770774841,
      "learning_rate": 2.1414373902213236e-06,
      "loss": 2.5416,
      "step": 3376
    },
    {
      "epoch": 0.9351980060924951,
      "grad_norm": 1.6184552907943726,
      "learning_rate": 2.123315802395942e-06,
      "loss": 2.6308,
      "step": 3377
    },
    {
      "epoch": 0.9354749376903905,
      "grad_norm": 1.0278406143188477,
      "learning_rate": 2.1052703935268147e-06,
      "loss": 2.6233,
      "step": 3378
    },
    {
      "epoch": 0.9357518692882858,
      "grad_norm": 0.7804761528968811,
      "learning_rate": 2.0873011776589957e-06,
      "loss": 2.7131,
      "step": 3379
    },
    {
      "epoch": 0.9360288008861811,
      "grad_norm": 1.2199102640151978,
      "learning_rate": 2.069408168778164e-06,
      "loss": 2.6202,
      "step": 3380
    },
    {
      "epoch": 0.9363057324840764,
      "grad_norm": 1.2146902084350586,
      "learning_rate": 2.051591380810769e-06,
      "loss": 2.7762,
      "step": 3381
    },
    {
      "epoch": 0.9365826640819718,
      "grad_norm": 1.7390179634094238,
      "learning_rate": 2.0338508276238734e-06,
      "loss": 2.8942,
      "step": 3382
    },
    {
      "epoch": 0.9368595956798671,
      "grad_norm": 0.6415250897407532,
      "learning_rate": 2.016186523025232e-06,
      "loss": 2.7395,
      "step": 3383
    },
    {
      "epoch": 0.9371365272777624,
      "grad_norm": 1.0076696872711182,
      "learning_rate": 1.998598480763247e-06,
      "loss": 2.8295,
      "step": 3384
    },
    {
      "epoch": 0.9374134588756577,
      "grad_norm": 1.5434540510177612,
      "learning_rate": 1.9810867145269805e-06,
      "loss": 2.86,
      "step": 3385
    },
    {
      "epoch": 0.9376903904735531,
      "grad_norm": 1.3561028242111206,
      "learning_rate": 1.963651237946107e-06,
      "loss": 3.063,
      "step": 3386
    },
    {
      "epoch": 0.9379673220714484,
      "grad_norm": 1.0174596309661865,
      "learning_rate": 1.9462920645909155e-06,
      "loss": 2.6779,
      "step": 3387
    },
    {
      "epoch": 0.9382442536693437,
      "grad_norm": 2.0525801181793213,
      "learning_rate": 1.929009207972354e-06,
      "loss": 2.6454,
      "step": 3388
    },
    {
      "epoch": 0.938521185267239,
      "grad_norm": 0.9554057717323303,
      "learning_rate": 1.911802681541919e-06,
      "loss": 2.5937,
      "step": 3389
    },
    {
      "epoch": 0.9387981168651343,
      "grad_norm": 1.5000263452529907,
      "learning_rate": 1.8946724986917187e-06,
      "loss": 3.0528,
      "step": 3390
    },
    {
      "epoch": 0.9390750484630296,
      "grad_norm": 1.3776977062225342,
      "learning_rate": 1.8776186727544554e-06,
      "loss": 2.8474,
      "step": 3391
    },
    {
      "epoch": 0.9393519800609249,
      "grad_norm": 1.2192586660385132,
      "learning_rate": 1.8606412170033783e-06,
      "loss": 2.9783,
      "step": 3392
    },
    {
      "epoch": 0.9396289116588202,
      "grad_norm": 1.9527432918548584,
      "learning_rate": 1.843740144652295e-06,
      "loss": 2.9976,
      "step": 3393
    },
    {
      "epoch": 0.9399058432567156,
      "grad_norm": 1.2904397249221802,
      "learning_rate": 1.8269154688556056e-06,
      "loss": 2.7887,
      "step": 3394
    },
    {
      "epoch": 0.9401827748546109,
      "grad_norm": 1.648537278175354,
      "learning_rate": 1.8101672027082018e-06,
      "loss": 2.5434,
      "step": 3395
    },
    {
      "epoch": 0.9404597064525062,
      "grad_norm": 0.8858701586723328,
      "learning_rate": 1.7934953592455117e-06,
      "loss": 2.6729,
      "step": 3396
    },
    {
      "epoch": 0.9407366380504015,
      "grad_norm": 2.0341806411743164,
      "learning_rate": 1.7768999514435226e-06,
      "loss": 2.6933,
      "step": 3397
    },
    {
      "epoch": 0.9410135696482969,
      "grad_norm": 1.5567057132720947,
      "learning_rate": 1.7603809922186687e-06,
      "loss": 2.9675,
      "step": 3398
    },
    {
      "epoch": 0.9412905012461922,
      "grad_norm": 1.515378713607788,
      "learning_rate": 1.7439384944279213e-06,
      "loss": 2.8218,
      "step": 3399
    },
    {
      "epoch": 0.9415674328440875,
      "grad_norm": 3.610714912414551,
      "learning_rate": 1.7275724708687435e-06,
      "loss": 3.9697,
      "step": 3400
    },
    {
      "epoch": 0.9418443644419828,
      "grad_norm": 1.809895634651184,
      "learning_rate": 1.711282934279068e-06,
      "loss": 2.6432,
      "step": 3401
    },
    {
      "epoch": 0.9421212960398782,
      "grad_norm": 2.0301547050476074,
      "learning_rate": 1.6950698973372869e-06,
      "loss": 2.9055,
      "step": 3402
    },
    {
      "epoch": 0.9423982276377735,
      "grad_norm": 1.4767478704452515,
      "learning_rate": 1.6789333726622725e-06,
      "loss": 2.6788,
      "step": 3403
    },
    {
      "epoch": 0.9426751592356688,
      "grad_norm": 1.59397554397583,
      "learning_rate": 1.6628733728133227e-06,
      "loss": 2.9936,
      "step": 3404
    },
    {
      "epoch": 0.9429520908335641,
      "grad_norm": 1.12782621383667,
      "learning_rate": 1.6468899102901946e-06,
      "loss": 2.9682,
      "step": 3405
    },
    {
      "epoch": 0.9432290224314595,
      "grad_norm": 2.1295478343963623,
      "learning_rate": 1.6309829975330592e-06,
      "loss": 3.0926,
      "step": 3406
    },
    {
      "epoch": 0.9435059540293548,
      "grad_norm": 2.399533987045288,
      "learning_rate": 1.6151526469225243e-06,
      "loss": 2.8278,
      "step": 3407
    },
    {
      "epoch": 0.9437828856272501,
      "grad_norm": 4.179904460906982,
      "learning_rate": 1.5993988707796005e-06,
      "loss": 3.0627,
      "step": 3408
    },
    {
      "epoch": 0.9440598172251454,
      "grad_norm": 1.0878640413284302,
      "learning_rate": 1.5837216813656908e-06,
      "loss": 2.7063,
      "step": 3409
    },
    {
      "epoch": 0.9443367488230408,
      "grad_norm": 1.21199631690979,
      "learning_rate": 1.5681210908826127e-06,
      "loss": 2.6272,
      "step": 3410
    },
    {
      "epoch": 0.9446136804209361,
      "grad_norm": 1.9922666549682617,
      "learning_rate": 1.5525971114725203e-06,
      "loss": 2.8212,
      "step": 3411
    },
    {
      "epoch": 0.9448906120188314,
      "grad_norm": 1.2448945045471191,
      "learning_rate": 1.5371497552180037e-06,
      "loss": 2.7175,
      "step": 3412
    },
    {
      "epoch": 0.9451675436167266,
      "grad_norm": 1.281944751739502,
      "learning_rate": 1.521779034141968e-06,
      "loss": 2.7586,
      "step": 3413
    },
    {
      "epoch": 0.9454444752146219,
      "grad_norm": 1.6437686681747437,
      "learning_rate": 1.506484960207677e-06,
      "loss": 2.8117,
      "step": 3414
    },
    {
      "epoch": 0.9457214068125173,
      "grad_norm": 1.3451311588287354,
      "learning_rate": 1.491267545318753e-06,
      "loss": 2.6612,
      "step": 3415
    },
    {
      "epoch": 0.9459983384104126,
      "grad_norm": 1.2874051332473755,
      "learning_rate": 1.4761268013191553e-06,
      "loss": 2.743,
      "step": 3416
    },
    {
      "epoch": 0.9462752700083079,
      "grad_norm": 1.589040994644165,
      "learning_rate": 1.4610627399931687e-06,
      "loss": 2.9678,
      "step": 3417
    },
    {
      "epoch": 0.9465522016062032,
      "grad_norm": 1.0552315711975098,
      "learning_rate": 1.4460753730653587e-06,
      "loss": 2.7698,
      "step": 3418
    },
    {
      "epoch": 0.9468291332040986,
      "grad_norm": 1.3050696849822998,
      "learning_rate": 1.4311647122006721e-06,
      "loss": 2.6855,
      "step": 3419
    },
    {
      "epoch": 0.9471060648019939,
      "grad_norm": 1.5626434087753296,
      "learning_rate": 1.4163307690042593e-06,
      "loss": 2.9976,
      "step": 3420
    },
    {
      "epoch": 0.9473829963998892,
      "grad_norm": 1.091338872909546,
      "learning_rate": 1.4015735550216514e-06,
      "loss": 2.6805,
      "step": 3421
    },
    {
      "epoch": 0.9476599279977845,
      "grad_norm": 1.2689846754074097,
      "learning_rate": 1.3868930817385939e-06,
      "loss": 2.5436,
      "step": 3422
    },
    {
      "epoch": 0.9479368595956799,
      "grad_norm": 1.3194886445999146,
      "learning_rate": 1.372289360581147e-06,
      "loss": 2.5677,
      "step": 3423
    },
    {
      "epoch": 0.9482137911935752,
      "grad_norm": 0.955805242061615,
      "learning_rate": 1.3577624029155966e-06,
      "loss": 2.7816,
      "step": 3424
    },
    {
      "epoch": 0.9484907227914705,
      "grad_norm": 3.8523759841918945,
      "learning_rate": 1.3433122200485316e-06,
      "loss": 4.0054,
      "step": 3425
    },
    {
      "epoch": 0.9487676543893658,
      "grad_norm": 1.4586676359176636,
      "learning_rate": 1.328938823226722e-06,
      "loss": 2.631,
      "step": 3426
    },
    {
      "epoch": 0.9490445859872612,
      "grad_norm": 1.378494381904602,
      "learning_rate": 1.314642223637208e-06,
      "loss": 2.685,
      "step": 3427
    },
    {
      "epoch": 0.9493215175851565,
      "grad_norm": 1.3128551244735718,
      "learning_rate": 1.3004224324073e-06,
      "loss": 2.5419,
      "step": 3428
    },
    {
      "epoch": 0.9495984491830518,
      "grad_norm": 1.3833677768707275,
      "learning_rate": 1.2862794606044337e-06,
      "loss": 2.7505,
      "step": 3429
    },
    {
      "epoch": 0.9498753807809471,
      "grad_norm": 1.1600041389465332,
      "learning_rate": 1.2722133192363372e-06,
      "loss": 2.6111,
      "step": 3430
    },
    {
      "epoch": 0.9501523123788425,
      "grad_norm": 1.772107481956482,
      "learning_rate": 1.2582240192508866e-06,
      "loss": 3.0549,
      "step": 3431
    },
    {
      "epoch": 0.9504292439767378,
      "grad_norm": 0.8794018030166626,
      "learning_rate": 1.2443115715362054e-06,
      "loss": 2.8327,
      "step": 3432
    },
    {
      "epoch": 0.9507061755746331,
      "grad_norm": 1.4521026611328125,
      "learning_rate": 1.2304759869205207e-06,
      "loss": 2.719,
      "step": 3433
    },
    {
      "epoch": 0.9509831071725284,
      "grad_norm": 1.116992473602295,
      "learning_rate": 1.216717276172341e-06,
      "loss": 2.8102,
      "step": 3434
    },
    {
      "epoch": 0.9512600387704238,
      "grad_norm": 0.8220580220222473,
      "learning_rate": 1.203035450000245e-06,
      "loss": 2.6858,
      "step": 3435
    },
    {
      "epoch": 0.9515369703683191,
      "grad_norm": 1.5655560493469238,
      "learning_rate": 1.1894305190530252e-06,
      "loss": 2.7253,
      "step": 3436
    },
    {
      "epoch": 0.9518139019662143,
      "grad_norm": 0.8458607792854309,
      "learning_rate": 1.1759024939196117e-06,
      "loss": 2.9535,
      "step": 3437
    },
    {
      "epoch": 0.9520908335641096,
      "grad_norm": 1.0517743825912476,
      "learning_rate": 1.162451385129082e-06,
      "loss": 2.7594,
      "step": 3438
    },
    {
      "epoch": 0.9523677651620049,
      "grad_norm": 1.4435462951660156,
      "learning_rate": 1.1490772031506392e-06,
      "loss": 2.7204,
      "step": 3439
    },
    {
      "epoch": 0.9526446967599003,
      "grad_norm": 1.073880910873413,
      "learning_rate": 1.1357799583936235e-06,
      "loss": 3.0155,
      "step": 3440
    },
    {
      "epoch": 0.9529216283577956,
      "grad_norm": 1.1127359867095947,
      "learning_rate": 1.1225596612075006e-06,
      "loss": 2.7524,
      "step": 3441
    },
    {
      "epoch": 0.9531985599556909,
      "grad_norm": 1.126267910003662,
      "learning_rate": 1.1094163218817955e-06,
      "loss": 2.8228,
      "step": 3442
    },
    {
      "epoch": 0.9534754915535862,
      "grad_norm": 1.1283173561096191,
      "learning_rate": 1.0963499506462249e-06,
      "loss": 2.6293,
      "step": 3443
    },
    {
      "epoch": 0.9537524231514816,
      "grad_norm": 1.1441411972045898,
      "learning_rate": 1.0833605576705096e-06,
      "loss": 2.8343,
      "step": 3444
    },
    {
      "epoch": 0.9540293547493769,
      "grad_norm": 1.3150324821472168,
      "learning_rate": 1.0704481530645293e-06,
      "loss": 2.6825,
      "step": 3445
    },
    {
      "epoch": 0.9543062863472722,
      "grad_norm": 1.40483558177948,
      "learning_rate": 1.0576127468781783e-06,
      "loss": 2.7361,
      "step": 3446
    },
    {
      "epoch": 0.9545832179451675,
      "grad_norm": 1.6336510181427002,
      "learning_rate": 1.044854349101476e-06,
      "loss": 2.6316,
      "step": 3447
    },
    {
      "epoch": 0.9548601495430629,
      "grad_norm": 1.2103753089904785,
      "learning_rate": 1.0321729696644911e-06,
      "loss": 2.8156,
      "step": 3448
    },
    {
      "epoch": 0.9551370811409582,
      "grad_norm": 1.1275039911270142,
      "learning_rate": 1.0195686184373166e-06,
      "loss": 2.8522,
      "step": 3449
    },
    {
      "epoch": 0.9554140127388535,
      "grad_norm": 3.2600536346435547,
      "learning_rate": 1.0070413052301275e-06,
      "loss": 3.5911,
      "step": 3450
    },
    {
      "epoch": 0.9556909443367488,
      "grad_norm": 1.913200855255127,
      "learning_rate": 9.945910397931246e-07,
      "loss": 2.8252,
      "step": 3451
    },
    {
      "epoch": 0.9559678759346442,
      "grad_norm": 1.0066627264022827,
      "learning_rate": 9.822178318165564e-07,
      "loss": 2.6377,
      "step": 3452
    },
    {
      "epoch": 0.9562448075325395,
      "grad_norm": 1.3210686445236206,
      "learning_rate": 9.699216909306753e-07,
      "loss": 2.7325,
      "step": 3453
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 1.6365430355072021,
      "learning_rate": 9.577026267057476e-07,
      "loss": 2.4892,
      "step": 3454
    },
    {
      "epoch": 0.9567986707283301,
      "grad_norm": 1.5990711450576782,
      "learning_rate": 9.455606486520885e-07,
      "loss": 2.7405,
      "step": 3455
    },
    {
      "epoch": 0.9570756023262255,
      "grad_norm": 1.4092278480529785,
      "learning_rate": 9.334957662199717e-07,
      "loss": 2.7796,
      "step": 3456
    },
    {
      "epoch": 0.9573525339241208,
      "grad_norm": 1.3785223960876465,
      "learning_rate": 9.215079887996858e-07,
      "loss": 2.9916,
      "step": 3457
    },
    {
      "epoch": 0.9576294655220161,
      "grad_norm": 1.4673542976379395,
      "learning_rate": 9.095973257215117e-07,
      "loss": 2.6861,
      "step": 3458
    },
    {
      "epoch": 0.9579063971199114,
      "grad_norm": 1.6413600444793701,
      "learning_rate": 8.97763786255712e-07,
      "loss": 2.5797,
      "step": 3459
    },
    {
      "epoch": 0.9581833287178066,
      "grad_norm": 1.5023274421691895,
      "learning_rate": 8.860073796124857e-07,
      "loss": 2.6855,
      "step": 3460
    },
    {
      "epoch": 0.958460260315702,
      "grad_norm": 1.5544533729553223,
      "learning_rate": 8.74328114942069e-07,
      "loss": 2.6957,
      "step": 3461
    },
    {
      "epoch": 0.9587371919135973,
      "grad_norm": 1.3404407501220703,
      "learning_rate": 8.627260013345795e-07,
      "loss": 2.8544,
      "step": 3462
    },
    {
      "epoch": 0.9590141235114926,
      "grad_norm": 1.093775749206543,
      "learning_rate": 8.512010478201493e-07,
      "loss": 2.7508,
      "step": 3463
    },
    {
      "epoch": 0.959291055109388,
      "grad_norm": 1.5489881038665771,
      "learning_rate": 8.397532633688254e-07,
      "loss": 2.8989,
      "step": 3464
    },
    {
      "epoch": 0.9595679867072833,
      "grad_norm": 1.147836685180664,
      "learning_rate": 8.283826568905917e-07,
      "loss": 2.9515,
      "step": 3465
    },
    {
      "epoch": 0.9598449183051786,
      "grad_norm": 1.5323371887207031,
      "learning_rate": 8.170892372353577e-07,
      "loss": 2.8796,
      "step": 3466
    },
    {
      "epoch": 0.9601218499030739,
      "grad_norm": 1.594389796257019,
      "learning_rate": 8.058730131930037e-07,
      "loss": 3.1553,
      "step": 3467
    },
    {
      "epoch": 0.9603987815009692,
      "grad_norm": 2.070909261703491,
      "learning_rate": 7.947339934932574e-07,
      "loss": 2.5193,
      "step": 3468
    },
    {
      "epoch": 0.9606757130988646,
      "grad_norm": 17.432004928588867,
      "learning_rate": 7.836721868058061e-07,
      "loss": 2.8423,
      "step": 3469
    },
    {
      "epoch": 0.9609526446967599,
      "grad_norm": 1.2177340984344482,
      "learning_rate": 7.726876017402296e-07,
      "loss": 2.6421,
      "step": 3470
    },
    {
      "epoch": 0.9612295762946552,
      "grad_norm": 1.468483567237854,
      "learning_rate": 7.617802468460001e-07,
      "loss": 2.6394,
      "step": 3471
    },
    {
      "epoch": 0.9615065078925505,
      "grad_norm": 0.8232468366622925,
      "learning_rate": 7.509501306124823e-07,
      "loss": 2.5258,
      "step": 3472
    },
    {
      "epoch": 0.9617834394904459,
      "grad_norm": 1.6739842891693115,
      "learning_rate": 7.401972614689335e-07,
      "loss": 3.0923,
      "step": 3473
    },
    {
      "epoch": 0.9620603710883412,
      "grad_norm": 2.3070802688598633,
      "learning_rate": 7.295216477844702e-07,
      "loss": 3.124,
      "step": 3474
    },
    {
      "epoch": 0.9623373026862365,
      "grad_norm": 3.783268928527832,
      "learning_rate": 7.189232978680793e-07,
      "loss": 3.9758,
      "step": 3475
    },
    {
      "epoch": 0.9626142342841318,
      "grad_norm": 1.2166013717651367,
      "learning_rate": 7.084022199686513e-07,
      "loss": 2.8477,
      "step": 3476
    },
    {
      "epoch": 0.9628911658820272,
      "grad_norm": 1.5633941888809204,
      "learning_rate": 6.979584222748803e-07,
      "loss": 2.8795,
      "step": 3477
    },
    {
      "epoch": 0.9631680974799225,
      "grad_norm": 1.97480046749115,
      "learning_rate": 6.87591912915353e-07,
      "loss": 2.8306,
      "step": 3478
    },
    {
      "epoch": 0.9634450290778178,
      "grad_norm": 2.1172704696655273,
      "learning_rate": 6.773026999584708e-07,
      "loss": 3.0142,
      "step": 3479
    },
    {
      "epoch": 0.9637219606757131,
      "grad_norm": 1.2924878597259521,
      "learning_rate": 6.670907914125058e-07,
      "loss": 2.8441,
      "step": 3480
    },
    {
      "epoch": 0.9639988922736085,
      "grad_norm": 1.5648305416107178,
      "learning_rate": 6.569561952255332e-07,
      "loss": 2.705,
      "step": 3481
    },
    {
      "epoch": 0.9642758238715038,
      "grad_norm": 1.0812360048294067,
      "learning_rate": 6.46898919285488e-07,
      "loss": 2.5468,
      "step": 3482
    },
    {
      "epoch": 0.964552755469399,
      "grad_norm": 0.9468382000923157,
      "learning_rate": 6.369189714200863e-07,
      "loss": 2.4775,
      "step": 3483
    },
    {
      "epoch": 0.9648296870672943,
      "grad_norm": 0.991826057434082,
      "learning_rate": 6.270163593968703e-07,
      "loss": 2.7163,
      "step": 3484
    },
    {
      "epoch": 0.9651066186651897,
      "grad_norm": 0.9778124094009399,
      "learning_rate": 6.171910909232193e-07,
      "loss": 2.9867,
      "step": 3485
    },
    {
      "epoch": 0.965383550263085,
      "grad_norm": 1.0859044790267944,
      "learning_rate": 6.074431736462938e-07,
      "loss": 3.0588,
      "step": 3486
    },
    {
      "epoch": 0.9656604818609803,
      "grad_norm": 1.5992988348007202,
      "learning_rate": 5.97772615153025e-07,
      "loss": 2.6721,
      "step": 3487
    },
    {
      "epoch": 0.9659374134588756,
      "grad_norm": 1.050748586654663,
      "learning_rate": 5.88179422970192e-07,
      "loss": 2.6772,
      "step": 3488
    },
    {
      "epoch": 0.966214345056771,
      "grad_norm": 1.10953950881958,
      "learning_rate": 5.786636045643112e-07,
      "loss": 2.8636,
      "step": 3489
    },
    {
      "epoch": 0.9664912766546663,
      "grad_norm": 0.9930030703544617,
      "learning_rate": 5.692251673416804e-07,
      "loss": 2.722,
      "step": 3490
    },
    {
      "epoch": 0.9667682082525616,
      "grad_norm": 1.4072052240371704,
      "learning_rate": 5.598641186484011e-07,
      "loss": 2.6757,
      "step": 3491
    },
    {
      "epoch": 0.9670451398504569,
      "grad_norm": 1.5075595378875732,
      "learning_rate": 5.505804657703228e-07,
      "loss": 2.6082,
      "step": 3492
    },
    {
      "epoch": 0.9673220714483522,
      "grad_norm": 1.1784552335739136,
      "learning_rate": 5.413742159330548e-07,
      "loss": 2.6231,
      "step": 3493
    },
    {
      "epoch": 0.9675990030462476,
      "grad_norm": 1.228947401046753,
      "learning_rate": 5.322453763019653e-07,
      "loss": 2.5158,
      "step": 3494
    },
    {
      "epoch": 0.9678759346441429,
      "grad_norm": 1.220169186592102,
      "learning_rate": 5.231939539821706e-07,
      "loss": 2.9059,
      "step": 3495
    },
    {
      "epoch": 0.9681528662420382,
      "grad_norm": 1.6655526161193848,
      "learning_rate": 5.142199560185357e-07,
      "loss": 2.9041,
      "step": 3496
    },
    {
      "epoch": 0.9684297978399335,
      "grad_norm": 1.830000400543213,
      "learning_rate": 5.053233893956621e-07,
      "loss": 2.8374,
      "step": 3497
    },
    {
      "epoch": 0.9687067294378289,
      "grad_norm": 1.6736990213394165,
      "learning_rate": 4.96504261037889e-07,
      "loss": 3.0461,
      "step": 3498
    },
    {
      "epoch": 0.9689836610357242,
      "grad_norm": 2.2138006687164307,
      "learning_rate": 4.877625778092809e-07,
      "loss": 3.2083,
      "step": 3499
    },
    {
      "epoch": 0.9692605926336195,
      "grad_norm": 3.2736976146698,
      "learning_rate": 4.790983465136401e-07,
      "loss": 3.673,
      "step": 3500
    },
    {
      "epoch": 0.9695375242315148,
      "grad_norm": 1.161478042602539,
      "learning_rate": 4.70511573894461e-07,
      "loss": 2.6695,
      "step": 3501
    },
    {
      "epoch": 0.9698144558294102,
      "grad_norm": 2.052083730697632,
      "learning_rate": 4.6200226663495326e-07,
      "loss": 2.6304,
      "step": 3502
    },
    {
      "epoch": 0.9700913874273055,
      "grad_norm": 1.8516638278961182,
      "learning_rate": 4.535704313580635e-07,
      "loss": 2.6462,
      "step": 3503
    },
    {
      "epoch": 0.9703683190252008,
      "grad_norm": 1.1855838298797607,
      "learning_rate": 4.4521607462640893e-07,
      "loss": 2.729,
      "step": 3504
    },
    {
      "epoch": 0.9706452506230961,
      "grad_norm": 1.260607361793518,
      "learning_rate": 4.3693920294232136e-07,
      "loss": 2.6698,
      "step": 3505
    },
    {
      "epoch": 0.9709221822209915,
      "grad_norm": 1.1047613620758057,
      "learning_rate": 4.2873982274781453e-07,
      "loss": 2.8026,
      "step": 3506
    },
    {
      "epoch": 0.9711991138188867,
      "grad_norm": 1.4178338050842285,
      "learning_rate": 4.2061794042460577e-07,
      "loss": 2.8202,
      "step": 3507
    },
    {
      "epoch": 0.971476045416782,
      "grad_norm": 1.8993240594863892,
      "learning_rate": 4.1257356229407184e-07,
      "loss": 2.9319,
      "step": 3508
    },
    {
      "epoch": 0.9717529770146773,
      "grad_norm": 1.7039676904678345,
      "learning_rate": 4.046066946172822e-07,
      "loss": 2.7073,
      "step": 3509
    },
    {
      "epoch": 0.9720299086125727,
      "grad_norm": 1.7674676179885864,
      "learning_rate": 3.967173435949656e-07,
      "loss": 2.7071,
      "step": 3510
    },
    {
      "epoch": 0.972306840210468,
      "grad_norm": 1.4723281860351562,
      "learning_rate": 3.889055153675547e-07,
      "loss": 2.6833,
      "step": 3511
    },
    {
      "epoch": 0.9725837718083633,
      "grad_norm": 1.910642147064209,
      "learning_rate": 3.81171216015086e-07,
      "loss": 3.0199,
      "step": 3512
    },
    {
      "epoch": 0.9728607034062586,
      "grad_norm": 0.9734626412391663,
      "learning_rate": 3.735144515572997e-07,
      "loss": 2.6637,
      "step": 3513
    },
    {
      "epoch": 0.973137635004154,
      "grad_norm": 1.7201459407806396,
      "learning_rate": 3.659352279535733e-07,
      "loss": 2.7869,
      "step": 3514
    },
    {
      "epoch": 0.9734145666020493,
      "grad_norm": 1.0668485164642334,
      "learning_rate": 3.5843355110294354e-07,
      "loss": 2.6127,
      "step": 3515
    },
    {
      "epoch": 0.9736914981999446,
      "grad_norm": 1.1746799945831299,
      "learning_rate": 3.510094268440844e-07,
      "loss": 2.762,
      "step": 3516
    },
    {
      "epoch": 0.9739684297978399,
      "grad_norm": 0.9444898962974548,
      "learning_rate": 3.4366286095528497e-07,
      "loss": 2.6246,
      "step": 3517
    },
    {
      "epoch": 0.9742453613957353,
      "grad_norm": 1.3913249969482422,
      "learning_rate": 3.3639385915451573e-07,
      "loss": 2.5367,
      "step": 3518
    },
    {
      "epoch": 0.9745222929936306,
      "grad_norm": 1.1144284009933472,
      "learning_rate": 3.292024270993399e-07,
      "loss": 2.727,
      "step": 3519
    },
    {
      "epoch": 0.9747992245915259,
      "grad_norm": 1.6452025175094604,
      "learning_rate": 3.220885703869803e-07,
      "loss": 3.2008,
      "step": 3520
    },
    {
      "epoch": 0.9750761561894212,
      "grad_norm": 0.894197940826416,
      "learning_rate": 3.1505229455424115e-07,
      "loss": 2.5495,
      "step": 3521
    },
    {
      "epoch": 0.9753530877873166,
      "grad_norm": 0.7937556505203247,
      "learning_rate": 3.0809360507757514e-07,
      "loss": 2.5862,
      "step": 3522
    },
    {
      "epoch": 0.9756300193852119,
      "grad_norm": 0.7896092534065247,
      "learning_rate": 3.0121250737304985e-07,
      "loss": 2.7862,
      "step": 3523
    },
    {
      "epoch": 0.9759069509831072,
      "grad_norm": 1.9969873428344727,
      "learning_rate": 2.9440900679631457e-07,
      "loss": 3.0699,
      "step": 3524
    },
    {
      "epoch": 0.9761838825810025,
      "grad_norm": 3.0025217533111572,
      "learning_rate": 2.876831086426557e-07,
      "loss": 3.7546,
      "step": 3525
    },
    {
      "epoch": 0.9764608141788979,
      "grad_norm": 1.1754149198532104,
      "learning_rate": 2.8103481814693023e-07,
      "loss": 2.7908,
      "step": 3526
    },
    {
      "epoch": 0.9767377457767932,
      "grad_norm": 0.6472751498222351,
      "learning_rate": 2.7446414048361015e-07,
      "loss": 2.5569,
      "step": 3527
    },
    {
      "epoch": 0.9770146773746885,
      "grad_norm": 1.319361686706543,
      "learning_rate": 2.6797108076677127e-07,
      "loss": 2.7662,
      "step": 3528
    },
    {
      "epoch": 0.9772916089725838,
      "grad_norm": 1.3683019876480103,
      "learning_rate": 2.615556440500377e-07,
      "loss": 2.8489,
      "step": 3529
    },
    {
      "epoch": 0.977568540570479,
      "grad_norm": 0.9338417053222656,
      "learning_rate": 2.55217835326671e-07,
      "loss": 2.7165,
      "step": 3530
    },
    {
      "epoch": 0.9778454721683744,
      "grad_norm": 1.454641580581665,
      "learning_rate": 2.4895765952948067e-07,
      "loss": 2.7523,
      "step": 3531
    },
    {
      "epoch": 0.9781224037662697,
      "grad_norm": 0.9676294326782227,
      "learning_rate": 2.42775121530836e-07,
      "loss": 2.7136,
      "step": 3532
    },
    {
      "epoch": 0.978399335364165,
      "grad_norm": 1.5383201837539673,
      "learning_rate": 2.3667022614273226e-07,
      "loss": 2.5843,
      "step": 3533
    },
    {
      "epoch": 0.9786762669620603,
      "grad_norm": 2.314225673675537,
      "learning_rate": 2.306429781166908e-07,
      "loss": 2.9224,
      "step": 3534
    },
    {
      "epoch": 0.9789531985599557,
      "grad_norm": 1.3681161403656006,
      "learning_rate": 2.2469338214382574e-07,
      "loss": 2.8437,
      "step": 3535
    },
    {
      "epoch": 0.979230130157851,
      "grad_norm": 0.9051282405853271,
      "learning_rate": 2.1882144285477746e-07,
      "loss": 2.7259,
      "step": 3536
    },
    {
      "epoch": 0.9795070617557463,
      "grad_norm": 1.2935221195220947,
      "learning_rate": 2.1302716481979012e-07,
      "loss": 2.7093,
      "step": 3537
    },
    {
      "epoch": 0.9797839933536416,
      "grad_norm": 2.2106070518493652,
      "learning_rate": 2.0731055254863408e-07,
      "loss": 3.1981,
      "step": 3538
    },
    {
      "epoch": 0.980060924951537,
      "grad_norm": 1.511698603630066,
      "learning_rate": 2.016716104906391e-07,
      "loss": 2.6531,
      "step": 3539
    },
    {
      "epoch": 0.9803378565494323,
      "grad_norm": 0.9122217297554016,
      "learning_rate": 1.9611034303468335e-07,
      "loss": 2.718,
      "step": 3540
    },
    {
      "epoch": 0.9806147881473276,
      "grad_norm": 1.3138456344604492,
      "learning_rate": 1.906267545091711e-07,
      "loss": 2.8714,
      "step": 3541
    },
    {
      "epoch": 0.9808917197452229,
      "grad_norm": 2.0339131355285645,
      "learning_rate": 1.8522084918208836e-07,
      "loss": 2.7607,
      "step": 3542
    },
    {
      "epoch": 0.9811686513431183,
      "grad_norm": 1.3019803762435913,
      "learning_rate": 1.7989263126093614e-07,
      "loss": 2.7719,
      "step": 3543
    },
    {
      "epoch": 0.9814455829410136,
      "grad_norm": 1.5723626613616943,
      "learning_rate": 1.7464210489273047e-07,
      "loss": 2.6476,
      "step": 3544
    },
    {
      "epoch": 0.9817225145389089,
      "grad_norm": 1.4210187196731567,
      "learning_rate": 1.694692741640469e-07,
      "loss": 2.8297,
      "step": 3545
    },
    {
      "epoch": 0.9819994461368042,
      "grad_norm": 2.068538188934326,
      "learning_rate": 1.6437414310098708e-07,
      "loss": 2.5935,
      "step": 3546
    },
    {
      "epoch": 0.9822763777346996,
      "grad_norm": 1.3514223098754883,
      "learning_rate": 1.5935671566916777e-07,
      "loss": 2.7994,
      "step": 3547
    },
    {
      "epoch": 0.9825533093325949,
      "grad_norm": 2.3161251544952393,
      "learning_rate": 1.5441699577372072e-07,
      "loss": 2.5894,
      "step": 3548
    },
    {
      "epoch": 0.9828302409304902,
      "grad_norm": 1.861956000328064,
      "learning_rate": 1.4955498725932604e-07,
      "loss": 3.2641,
      "step": 3549
    },
    {
      "epoch": 0.9831071725283855,
      "grad_norm": 3.413618564605713,
      "learning_rate": 1.4477069391014563e-07,
      "loss": 3.7602,
      "step": 3550
    },
    {
      "epoch": 0.9833841041262809,
      "grad_norm": 1.23565673828125,
      "learning_rate": 1.4006411944988973e-07,
      "loss": 2.8523,
      "step": 3551
    },
    {
      "epoch": 0.9836610357241762,
      "grad_norm": 1.245562195777893,
      "learning_rate": 1.3543526754173918e-07,
      "loss": 2.7221,
      "step": 3552
    },
    {
      "epoch": 0.9839379673220714,
      "grad_norm": 0.939725935459137,
      "learning_rate": 1.3088414178840104e-07,
      "loss": 2.708,
      "step": 3553
    },
    {
      "epoch": 0.9842148989199667,
      "grad_norm": 1.6134532690048218,
      "learning_rate": 1.2641074573209733e-07,
      "loss": 2.704,
      "step": 3554
    },
    {
      "epoch": 0.984491830517862,
      "grad_norm": 1.680004358291626,
      "learning_rate": 1.2201508285454298e-07,
      "loss": 3.0606,
      "step": 3555
    },
    {
      "epoch": 0.9847687621157574,
      "grad_norm": 1.0474539995193481,
      "learning_rate": 1.176971565769458e-07,
      "loss": 2.686,
      "step": 3556
    },
    {
      "epoch": 0.9850456937136527,
      "grad_norm": 1.3566621541976929,
      "learning_rate": 1.1345697026001745e-07,
      "loss": 2.6762,
      "step": 3557
    },
    {
      "epoch": 0.985322625311548,
      "grad_norm": 1.2527090311050415,
      "learning_rate": 1.092945272039625e-07,
      "loss": 2.7056,
      "step": 3558
    },
    {
      "epoch": 0.9855995569094433,
      "grad_norm": 1.1411058902740479,
      "learning_rate": 1.0520983064847833e-07,
      "loss": 2.6102,
      "step": 3559
    },
    {
      "epoch": 0.9858764885073387,
      "grad_norm": 1.2734065055847168,
      "learning_rate": 1.0120288377274412e-07,
      "loss": 2.9378,
      "step": 3560
    },
    {
      "epoch": 0.986153420105234,
      "grad_norm": 1.2071735858917236,
      "learning_rate": 9.727368969542072e-08,
      "loss": 2.6635,
      "step": 3561
    },
    {
      "epoch": 0.9864303517031293,
      "grad_norm": 1.6888424158096313,
      "learning_rate": 9.342225147467299e-08,
      "loss": 2.9836,
      "step": 3562
    },
    {
      "epoch": 0.9867072833010246,
      "grad_norm": 1.2385547161102295,
      "learning_rate": 8.964857210814748e-08,
      "loss": 2.8505,
      "step": 3563
    },
    {
      "epoch": 0.98698421489892,
      "grad_norm": 1.72456955909729,
      "learning_rate": 8.595265453292811e-08,
      "loss": 2.7266,
      "step": 3564
    },
    {
      "epoch": 0.9872611464968153,
      "grad_norm": 2.967880964279175,
      "learning_rate": 8.233450162562495e-08,
      "loss": 3.0941,
      "step": 3565
    },
    {
      "epoch": 0.9875380780947106,
      "grad_norm": 1.703004240989685,
      "learning_rate": 7.87941162023076e-08,
      "loss": 3.0268,
      "step": 3566
    },
    {
      "epoch": 0.9878150096926059,
      "grad_norm": 1.254521131515503,
      "learning_rate": 7.533150101849407e-08,
      "loss": 2.6175,
      "step": 3567
    },
    {
      "epoch": 0.9880919412905013,
      "grad_norm": 1.4605085849761963,
      "learning_rate": 7.194665876920636e-08,
      "loss": 2.9556,
      "step": 3568
    },
    {
      "epoch": 0.9883688728883966,
      "grad_norm": 1.6051185131072998,
      "learning_rate": 6.86395920889149e-08,
      "loss": 2.8252,
      "step": 3569
    },
    {
      "epoch": 0.9886458044862919,
      "grad_norm": 1.8399105072021484,
      "learning_rate": 6.541030355156075e-08,
      "loss": 2.5048,
      "step": 3570
    },
    {
      "epoch": 0.9889227360841872,
      "grad_norm": 1.030032992362976,
      "learning_rate": 6.225879567055559e-08,
      "loss": 2.5485,
      "step": 3571
    },
    {
      "epoch": 0.9891996676820826,
      "grad_norm": 1.6225028038024902,
      "learning_rate": 5.918507089877068e-08,
      "loss": 2.7388,
      "step": 3572
    },
    {
      "epoch": 0.9894765992799779,
      "grad_norm": 1.342437505722046,
      "learning_rate": 5.6189131628525684e-08,
      "loss": 2.5809,
      "step": 3573
    },
    {
      "epoch": 0.9897535308778732,
      "grad_norm": 1.1848745346069336,
      "learning_rate": 5.327098019159982e-08,
      "loss": 2.746,
      "step": 3574
    },
    {
      "epoch": 0.9900304624757685,
      "grad_norm": 2.597025156021118,
      "learning_rate": 5.043061885925404e-08,
      "loss": 3.5433,
      "step": 3575
    },
    {
      "epoch": 0.9903073940736639,
      "grad_norm": 1.5176355838775635,
      "learning_rate": 4.7668049842175544e-08,
      "loss": 2.7927,
      "step": 3576
    },
    {
      "epoch": 0.9905843256715591,
      "grad_norm": 1.9392883777618408,
      "learning_rate": 4.498327529051105e-08,
      "loss": 2.9864,
      "step": 3577
    },
    {
      "epoch": 0.9908612572694544,
      "grad_norm": 1.6624773740768433,
      "learning_rate": 4.237629729387793e-08,
      "loss": 2.6176,
      "step": 3578
    },
    {
      "epoch": 0.9911381888673497,
      "grad_norm": 1.1208992004394531,
      "learning_rate": 3.9847117881308685e-08,
      "loss": 2.6707,
      "step": 3579
    },
    {
      "epoch": 0.991415120465245,
      "grad_norm": 1.0683854818344116,
      "learning_rate": 3.739573902131754e-08,
      "loss": 2.5527,
      "step": 3580
    },
    {
      "epoch": 0.9916920520631404,
      "grad_norm": 2.0292575359344482,
      "learning_rate": 3.502216262184499e-08,
      "loss": 2.7259,
      "step": 3581
    },
    {
      "epoch": 0.9919689836610357,
      "grad_norm": 0.834223210811615,
      "learning_rate": 3.272639053029103e-08,
      "loss": 2.5719,
      "step": 3582
    },
    {
      "epoch": 0.992245915258931,
      "grad_norm": 1.624928593635559,
      "learning_rate": 3.05084245334708e-08,
      "loss": 2.9334,
      "step": 3583
    },
    {
      "epoch": 0.9925228468568263,
      "grad_norm": 1.3256527185440063,
      "learning_rate": 2.8368266357681194e-08,
      "loss": 2.6071,
      "step": 3584
    },
    {
      "epoch": 0.9927997784547217,
      "grad_norm": 1.2522574663162231,
      "learning_rate": 2.6305917668645318e-08,
      "loss": 2.7054,
      "step": 3585
    },
    {
      "epoch": 0.993076710052617,
      "grad_norm": 1.1217739582061768,
      "learning_rate": 2.432138007150142e-08,
      "loss": 3.0231,
      "step": 3586
    },
    {
      "epoch": 0.9933536416505123,
      "grad_norm": 2.0638465881347656,
      "learning_rate": 2.2414655110858383e-08,
      "loss": 3.0328,
      "step": 3587
    },
    {
      "epoch": 0.9936305732484076,
      "grad_norm": 1.2661341428756714,
      "learning_rate": 2.0585744270740226e-08,
      "loss": 2.5907,
      "step": 3588
    },
    {
      "epoch": 0.993907504846303,
      "grad_norm": 1.2835946083068848,
      "learning_rate": 1.8834648974630497e-08,
      "loss": 2.7709,
      "step": 3589
    },
    {
      "epoch": 0.9941844364441983,
      "grad_norm": 2.156313180923462,
      "learning_rate": 1.7161370585427883e-08,
      "loss": 2.6873,
      "step": 3590
    },
    {
      "epoch": 0.9944613680420936,
      "grad_norm": 1.453169345855713,
      "learning_rate": 1.55659104054795e-08,
      "loss": 2.6949,
      "step": 3591
    },
    {
      "epoch": 0.9947382996399889,
      "grad_norm": 1.0746440887451172,
      "learning_rate": 1.4048269676536497e-08,
      "loss": 2.8744,
      "step": 3592
    },
    {
      "epoch": 0.9950152312378843,
      "grad_norm": 1.2922847270965576,
      "learning_rate": 1.2608449579820658e-08,
      "loss": 2.6111,
      "step": 3593
    },
    {
      "epoch": 0.9952921628357796,
      "grad_norm": 1.3704134225845337,
      "learning_rate": 1.12464512359578e-08,
      "loss": 2.803,
      "step": 3594
    },
    {
      "epoch": 0.9955690944336749,
      "grad_norm": 2.0052356719970703,
      "learning_rate": 9.96227570501107e-09,
      "loss": 2.5901,
      "step": 3595
    },
    {
      "epoch": 0.9958460260315702,
      "grad_norm": 1.3504281044006348,
      "learning_rate": 8.755923986480952e-09,
      "loss": 2.5067,
      "step": 3596
    },
    {
      "epoch": 0.9961229576294656,
      "grad_norm": 1.0229085683822632,
      "learning_rate": 7.627397019294158e-09,
      "loss": 2.6755,
      "step": 3597
    },
    {
      "epoch": 0.9963998892273609,
      "grad_norm": 1.2293682098388672,
      "learning_rate": 6.576695681792533e-09,
      "loss": 2.8801,
      "step": 3598
    },
    {
      "epoch": 0.9966768208252562,
      "grad_norm": 1.2077336311340332,
      "learning_rate": 5.603820791755254e-09,
      "loss": 3.0605,
      "step": 3599
    },
    {
      "epoch": 0.9969537524231514,
      "grad_norm": 4.453784465789795,
      "learning_rate": 4.708773106376629e-09,
      "loss": 4.3675,
      "step": 3600
    },
    {
      "epoch": 0.9972306840210468,
      "grad_norm": 1.1345912218093872,
      "learning_rate": 3.891553322299401e-09,
      "loss": 2.6014,
      "step": 3601
    },
    {
      "epoch": 0.9975076156189421,
      "grad_norm": 1.3901625871658325,
      "learning_rate": 3.152162075581444e-09,
      "loss": 2.6007,
      "step": 3602
    },
    {
      "epoch": 0.9977845472168374,
      "grad_norm": 1.705886960029602,
      "learning_rate": 2.4905999416957593e-09,
      "loss": 2.8874,
      "step": 3603
    },
    {
      "epoch": 0.9980614788147327,
      "grad_norm": 1.9912961721420288,
      "learning_rate": 1.9068674355415815e-09,
      "loss": 2.6562,
      "step": 3604
    },
    {
      "epoch": 0.998338410412628,
      "grad_norm": 1.2312753200531006,
      "learning_rate": 1.4009650114554796e-09,
      "loss": 2.5166,
      "step": 3605
    },
    {
      "epoch": 0.9986153420105234,
      "grad_norm": 0.9925754070281982,
      "learning_rate": 9.728930631780487e-10,
      "loss": 2.7784,
      "step": 3606
    },
    {
      "epoch": 0.9988922736084187,
      "grad_norm": 1.4208216667175293,
      "learning_rate": 6.226519238983209e-10,
      "loss": 2.6608,
      "step": 3607
    },
    {
      "epoch": 0.999169205206314,
      "grad_norm": 1.1245412826538086,
      "learning_rate": 3.502418662093554e-10,
      "loss": 2.5002,
      "step": 3608
    },
    {
      "epoch": 0.9994461368042094,
      "grad_norm": 1.6515275239944458,
      "learning_rate": 1.5566310213044333e-10,
      "loss": 2.5429,
      "step": 3609
    },
    {
      "epoch": 0.9997230684021047,
      "grad_norm": 1.8428101539611816,
      "learning_rate": 3.891578310710742e-11,
      "loss": 2.9925,
      "step": 3610
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.526169776916504,
      "learning_rate": 0.0,
      "loss": 3.4422,
      "step": 3611
    }
  ],
  "logging_steps": 1,
  "max_steps": 3611,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 903,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.452436247864672e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}