{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.4999794145497962,
  "eval_steps": 759,
  "global_step": 1518,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003293672032607353,
      "grad_norm": 3.2412221431732178,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 3.3662,
      "step": 1
    },
    {
      "epoch": 0.0003293672032607353,
      "eval_loss": 3.95210599899292,
      "eval_runtime": 812.7321,
      "eval_samples_per_second": 3.146,
      "eval_steps_per_second": 1.574,
      "step": 1
    },
    {
      "epoch": 0.0006587344065214706,
      "grad_norm": 3.8850836753845215,
      "learning_rate": 4.000000000000001e-06,
      "loss": 3.2364,
      "step": 2
    },
    {
      "epoch": 0.000988101609782206,
      "grad_norm": 3.7942073345184326,
      "learning_rate": 6e-06,
      "loss": 3.3337,
      "step": 3
    },
    {
      "epoch": 0.0013174688130429412,
      "grad_norm": 4.045947074890137,
      "learning_rate": 8.000000000000001e-06,
      "loss": 3.523,
      "step": 4
    },
    {
      "epoch": 0.0016468360163036766,
      "grad_norm": 3.6181905269622803,
      "learning_rate": 1e-05,
      "loss": 3.1772,
      "step": 5
    },
    {
      "epoch": 0.001976203219564412,
      "grad_norm": 4.814149379730225,
      "learning_rate": 1.2e-05,
      "loss": 3.871,
      "step": 6
    },
    {
      "epoch": 0.002305570422825147,
      "grad_norm": 4.0820112228393555,
      "learning_rate": 1.4000000000000001e-05,
      "loss": 3.7199,
      "step": 7
    },
    {
      "epoch": 0.0026349376260858823,
      "grad_norm": 4.483249187469482,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 3.5852,
      "step": 8
    },
    {
      "epoch": 0.002964304829346618,
      "grad_norm": 5.03192138671875,
      "learning_rate": 1.8e-05,
      "loss": 3.5165,
      "step": 9
    },
    {
      "epoch": 0.003293672032607353,
      "grad_norm": 4.41070556640625,
      "learning_rate": 2e-05,
      "loss": 3.1522,
      "step": 10
    },
    {
      "epoch": 0.0036230392358680883,
      "grad_norm": 4.54289436340332,
      "learning_rate": 2.2000000000000003e-05,
      "loss": 3.113,
      "step": 11
    },
    {
      "epoch": 0.003952406439128824,
      "grad_norm": 5.175803184509277,
      "learning_rate": 2.4e-05,
      "loss": 3.0833,
      "step": 12
    },
    {
      "epoch": 0.004281773642389559,
      "grad_norm": 5.162367820739746,
      "learning_rate": 2.6000000000000002e-05,
      "loss": 3.2495,
      "step": 13
    },
    {
      "epoch": 0.004611140845650294,
      "grad_norm": 5.13967227935791,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 3.1866,
      "step": 14
    },
    {
      "epoch": 0.0049405080489110294,
      "grad_norm": 4.819941520690918,
      "learning_rate": 3e-05,
      "loss": 2.8483,
      "step": 15
    },
    {
      "epoch": 0.005269875252171765,
      "grad_norm": 4.200347900390625,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 3.0112,
      "step": 16
    },
    {
      "epoch": 0.005599242455432501,
      "grad_norm": 3.748765230178833,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 2.4482,
      "step": 17
    },
    {
      "epoch": 0.005928609658693236,
      "grad_norm": 4.706761360168457,
      "learning_rate": 3.6e-05,
      "loss": 2.7336,
      "step": 18
    },
    {
      "epoch": 0.006257976861953971,
      "grad_norm": 4.122259616851807,
      "learning_rate": 3.8e-05,
      "loss": 2.2935,
      "step": 19
    },
    {
      "epoch": 0.006587344065214706,
      "grad_norm": 3.4756200313568115,
      "learning_rate": 4e-05,
      "loss": 2.4558,
      "step": 20
    },
    {
      "epoch": 0.006916711268475441,
      "grad_norm": 4.478485107421875,
      "learning_rate": 4.2e-05,
      "loss": 2.7799,
      "step": 21
    },
    {
      "epoch": 0.007246078471736177,
      "grad_norm": 3.8455657958984375,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 2.3758,
      "step": 22
    },
    {
      "epoch": 0.007575445674996912,
      "grad_norm": 4.439505100250244,
      "learning_rate": 4.600000000000001e-05,
      "loss": 2.4312,
      "step": 23
    },
    {
      "epoch": 0.007904812878257648,
      "grad_norm": 5.45050048828125,
      "learning_rate": 4.8e-05,
      "loss": 2.3483,
      "step": 24
    },
    {
      "epoch": 0.008234180081518383,
      "grad_norm": 6.179609298706055,
      "learning_rate": 5e-05,
      "loss": 2.1845,
      "step": 25
    },
    {
      "epoch": 0.008563547284779118,
      "grad_norm": 3.980731964111328,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 2.7766,
      "step": 26
    },
    {
      "epoch": 0.008892914488039853,
      "grad_norm": 3.381983995437622,
      "learning_rate": 5.4000000000000005e-05,
      "loss": 2.5084,
      "step": 27
    },
    {
      "epoch": 0.009222281691300589,
      "grad_norm": 3.6727781295776367,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 2.8683,
      "step": 28
    },
    {
      "epoch": 0.009551648894561324,
      "grad_norm": 3.6398210525512695,
      "learning_rate": 5.8e-05,
      "loss": 2.4222,
      "step": 29
    },
    {
      "epoch": 0.009881016097822059,
      "grad_norm": 3.005326986312866,
      "learning_rate": 6e-05,
      "loss": 2.5306,
      "step": 30
    },
    {
      "epoch": 0.010210383301082794,
      "grad_norm": 3.8456501960754395,
      "learning_rate": 6.2e-05,
      "loss": 2.7113,
      "step": 31
    },
    {
      "epoch": 0.01053975050434353,
      "grad_norm": 2.880403518676758,
      "learning_rate": 6.400000000000001e-05,
      "loss": 2.3046,
      "step": 32
    },
    {
      "epoch": 0.010869117707604266,
      "grad_norm": 3.180576801300049,
      "learning_rate": 6.6e-05,
      "loss": 2.5024,
      "step": 33
    },
    {
      "epoch": 0.011198484910865001,
      "grad_norm": 3.2660038471221924,
      "learning_rate": 6.800000000000001e-05,
      "loss": 2.6287,
      "step": 34
    },
    {
      "epoch": 0.011527852114125737,
      "grad_norm": 2.8470723628997803,
      "learning_rate": 7e-05,
      "loss": 2.3198,
      "step": 35
    },
    {
      "epoch": 0.011857219317386472,
      "grad_norm": 3.0491487979888916,
      "learning_rate": 7.2e-05,
      "loss": 2.3468,
      "step": 36
    },
    {
      "epoch": 0.012186586520647207,
      "grad_norm": 3.558762788772583,
      "learning_rate": 7.4e-05,
      "loss": 2.5952,
      "step": 37
    },
    {
      "epoch": 0.012515953723907942,
      "grad_norm": 3.0420889854431152,
      "learning_rate": 7.6e-05,
      "loss": 2.2754,
      "step": 38
    },
    {
      "epoch": 0.012845320927168677,
      "grad_norm": 2.9328525066375732,
      "learning_rate": 7.800000000000001e-05,
      "loss": 1.9192,
      "step": 39
    },
    {
      "epoch": 0.013174688130429412,
      "grad_norm": 2.9751060009002686,
      "learning_rate": 8e-05,
      "loss": 2.4325,
      "step": 40
    },
    {
      "epoch": 0.013504055333690148,
      "grad_norm": 2.924984931945801,
      "learning_rate": 8.2e-05,
      "loss": 2.376,
      "step": 41
    },
    {
      "epoch": 0.013833422536950883,
      "grad_norm": 2.9344444274902344,
      "learning_rate": 8.4e-05,
      "loss": 2.3072,
      "step": 42
    },
    {
      "epoch": 0.014162789740211618,
      "grad_norm": 3.2519760131835938,
      "learning_rate": 8.6e-05,
      "loss": 2.3017,
      "step": 43
    },
    {
      "epoch": 0.014492156943472353,
      "grad_norm": 2.9699273109436035,
      "learning_rate": 8.800000000000001e-05,
      "loss": 2.2801,
      "step": 44
    },
    {
      "epoch": 0.014821524146733088,
      "grad_norm": 3.125183343887329,
      "learning_rate": 9e-05,
      "loss": 2.1891,
      "step": 45
    },
    {
      "epoch": 0.015150891349993824,
      "grad_norm": 4.185096740722656,
      "learning_rate": 9.200000000000001e-05,
      "loss": 2.3958,
      "step": 46
    },
    {
      "epoch": 0.01548025855325456,
      "grad_norm": 4.01360559463501,
      "learning_rate": 9.4e-05,
      "loss": 2.3923,
      "step": 47
    },
    {
      "epoch": 0.015809625756515296,
      "grad_norm": 3.355820417404175,
      "learning_rate": 9.6e-05,
      "loss": 2.0905,
      "step": 48
    },
    {
      "epoch": 0.01613899295977603,
      "grad_norm": 4.308192729949951,
      "learning_rate": 9.8e-05,
      "loss": 2.0585,
      "step": 49
    },
    {
      "epoch": 0.016468360163036766,
      "grad_norm": 5.098552227020264,
      "learning_rate": 0.0001,
      "loss": 2.7164,
      "step": 50
    },
    {
      "epoch": 0.0167977273662975,
      "grad_norm": 2.8389906883239746,
      "learning_rate": 9.999997232675378e-05,
      "loss": 2.7739,
      "step": 51
    },
    {
      "epoch": 0.017127094569558236,
      "grad_norm": 3.0663294792175293,
      "learning_rate": 9.999988930704576e-05,
      "loss": 2.7123,
      "step": 52
    },
    {
      "epoch": 0.01745646177281897,
      "grad_norm": 3.3522698879241943,
      "learning_rate": 9.99997509409678e-05,
      "loss": 2.4191,
      "step": 53
    },
    {
      "epoch": 0.017785828976079707,
      "grad_norm": 2.815250873565674,
      "learning_rate": 9.999955722867312e-05,
      "loss": 2.6329,
      "step": 54
    },
    {
      "epoch": 0.018115196179340442,
      "grad_norm": 2.9915289878845215,
      "learning_rate": 9.999930817037609e-05,
      "loss": 2.4786,
      "step": 55
    },
    {
      "epoch": 0.018444563382601177,
      "grad_norm": 3.284151554107666,
      "learning_rate": 9.999900376635243e-05,
      "loss": 2.4249,
      "step": 56
    },
    {
      "epoch": 0.018773930585861912,
      "grad_norm": 2.4534027576446533,
      "learning_rate": 9.999864401693908e-05,
      "loss": 2.356,
      "step": 57
    },
    {
      "epoch": 0.019103297789122647,
      "grad_norm": 2.522468328475952,
      "learning_rate": 9.999822892253428e-05,
      "loss": 2.3852,
      "step": 58
    },
    {
      "epoch": 0.019432664992383383,
      "grad_norm": 2.9176650047302246,
      "learning_rate": 9.999775848359749e-05,
      "loss": 2.2109,
      "step": 59
    },
    {
      "epoch": 0.019762032195644118,
      "grad_norm": 3.167917013168335,
      "learning_rate": 9.999723270064945e-05,
      "loss": 2.3511,
      "step": 60
    },
    {
      "epoch": 0.020091399398904853,
      "grad_norm": 2.6770145893096924,
      "learning_rate": 9.999665157427218e-05,
      "loss": 2.6865,
      "step": 61
    },
    {
      "epoch": 0.020420766602165588,
      "grad_norm": 2.729907989501953,
      "learning_rate": 9.999601510510895e-05,
      "loss": 2.3012,
      "step": 62
    },
    {
      "epoch": 0.020750133805426323,
      "grad_norm": 2.7215871810913086,
      "learning_rate": 9.999532329386425e-05,
      "loss": 2.0514,
      "step": 63
    },
    {
      "epoch": 0.02107950100868706,
      "grad_norm": 2.7751474380493164,
      "learning_rate": 9.999457614130391e-05,
      "loss": 2.4867,
      "step": 64
    },
    {
      "epoch": 0.021408868211947794,
      "grad_norm": 2.4375815391540527,
      "learning_rate": 9.999377364825496e-05,
      "loss": 2.2029,
      "step": 65
    },
    {
      "epoch": 0.021738235415208532,
      "grad_norm": 2.7066309452056885,
      "learning_rate": 9.99929158156057e-05,
      "loss": 2.1274,
      "step": 66
    },
    {
      "epoch": 0.022067602618469268,
      "grad_norm": 3.297457695007324,
      "learning_rate": 9.999200264430569e-05,
      "loss": 2.5863,
      "step": 67
    },
    {
      "epoch": 0.022396969821730003,
      "grad_norm": 3.2771711349487305,
      "learning_rate": 9.999103413536574e-05,
      "loss": 2.334,
      "step": 68
    },
    {
      "epoch": 0.022726337024990738,
      "grad_norm": 2.9507834911346436,
      "learning_rate": 9.999001028985795e-05,
      "loss": 2.3783,
      "step": 69
    },
    {
      "epoch": 0.023055704228251473,
      "grad_norm": 2.6401610374450684,
      "learning_rate": 9.998893110891562e-05,
      "loss": 1.9256,
      "step": 70
    },
    {
      "epoch": 0.023385071431512208,
      "grad_norm": 3.0901267528533936,
      "learning_rate": 9.998779659373332e-05,
      "loss": 2.1796,
      "step": 71
    },
    {
      "epoch": 0.023714438634772943,
      "grad_norm": 3.132500171661377,
      "learning_rate": 9.99866067455669e-05,
      "loss": 2.2996,
      "step": 72
    },
    {
      "epoch": 0.02404380583803368,
      "grad_norm": 3.1304476261138916,
      "learning_rate": 9.998536156573343e-05,
      "loss": 2.1264,
      "step": 73
    },
    {
      "epoch": 0.024373173041294414,
      "grad_norm": 3.315488576889038,
      "learning_rate": 9.998406105561125e-05,
      "loss": 1.984,
      "step": 74
    },
    {
      "epoch": 0.02470254024455515,
      "grad_norm": 7.0925774574279785,
      "learning_rate": 9.99827052166399e-05,
      "loss": 2.533,
      "step": 75
    },
    {
      "epoch": 0.025031907447815884,
      "grad_norm": 2.485670328140259,
      "learning_rate": 9.998129405032022e-05,
      "loss": 2.4111,
      "step": 76
    },
    {
      "epoch": 0.02536127465107662,
      "grad_norm": 2.9346237182617188,
      "learning_rate": 9.997982755821428e-05,
      "loss": 2.4954,
      "step": 77
    },
    {
      "epoch": 0.025690641854337355,
      "grad_norm": 2.6853690147399902,
      "learning_rate": 9.997830574194538e-05,
      "loss": 2.7849,
      "step": 78
    },
    {
      "epoch": 0.02602000905759809,
      "grad_norm": 2.614497661590576,
      "learning_rate": 9.997672860319804e-05,
      "loss": 2.7054,
      "step": 79
    },
    {
      "epoch": 0.026349376260858825,
      "grad_norm": 2.476630687713623,
      "learning_rate": 9.997509614371807e-05,
      "loss": 2.3267,
      "step": 80
    },
    {
      "epoch": 0.02667874346411956,
      "grad_norm": 2.7469708919525146,
      "learning_rate": 9.997340836531249e-05,
      "loss": 2.3416,
      "step": 81
    },
    {
      "epoch": 0.027008110667380295,
      "grad_norm": 2.3859927654266357,
      "learning_rate": 9.997166526984954e-05,
      "loss": 2.1941,
      "step": 82
    },
    {
      "epoch": 0.02733747787064103,
      "grad_norm": 2.3550686836242676,
      "learning_rate": 9.996986685925868e-05,
      "loss": 2.2731,
      "step": 83
    },
    {
      "epoch": 0.027666845073901766,
      "grad_norm": 2.6245596408843994,
      "learning_rate": 9.996801313553068e-05,
      "loss": 2.1934,
      "step": 84
    },
    {
      "epoch": 0.0279962122771625,
      "grad_norm": 2.337979793548584,
      "learning_rate": 9.996610410071742e-05,
      "loss": 2.0485,
      "step": 85
    },
    {
      "epoch": 0.028325579480423236,
      "grad_norm": 2.2919275760650635,
      "learning_rate": 9.996413975693214e-05,
      "loss": 2.1097,
      "step": 86
    },
    {
      "epoch": 0.02865494668368397,
      "grad_norm": 2.4434707164764404,
      "learning_rate": 9.996212010634917e-05,
      "loss": 2.1181,
      "step": 87
    },
    {
      "epoch": 0.028984313886944706,
      "grad_norm": 2.2074220180511475,
      "learning_rate": 9.996004515120414e-05,
      "loss": 2.1341,
      "step": 88
    },
    {
      "epoch": 0.02931368109020544,
      "grad_norm": 2.670886993408203,
      "learning_rate": 9.995791489379388e-05,
      "loss": 2.2701,
      "step": 89
    },
    {
      "epoch": 0.029643048293466177,
      "grad_norm": 3.0483765602111816,
      "learning_rate": 9.995572933647643e-05,
      "loss": 2.2896,
      "step": 90
    },
    {
      "epoch": 0.029972415496726912,
      "grad_norm": 2.74815034866333,
      "learning_rate": 9.995348848167107e-05,
      "loss": 2.3094,
      "step": 91
    },
    {
      "epoch": 0.030301782699987647,
      "grad_norm": 3.188274621963501,
      "learning_rate": 9.995119233185825e-05,
      "loss": 2.3274,
      "step": 92
    },
    {
      "epoch": 0.030631149903248386,
      "grad_norm": 2.603101968765259,
      "learning_rate": 9.994884088957966e-05,
      "loss": 2.0735,
      "step": 93
    },
    {
      "epoch": 0.03096051710650912,
      "grad_norm": 2.7947723865509033,
      "learning_rate": 9.994643415743817e-05,
      "loss": 2.571,
      "step": 94
    },
    {
      "epoch": 0.03128988430976985,
      "grad_norm": 2.8341128826141357,
      "learning_rate": 9.994397213809786e-05,
      "loss": 2.4703,
      "step": 95
    },
    {
      "epoch": 0.03161925151303059,
      "grad_norm": 2.8162853717803955,
      "learning_rate": 9.994145483428403e-05,
      "loss": 2.2555,
      "step": 96
    },
    {
      "epoch": 0.03194861871629132,
      "grad_norm": 2.9116039276123047,
      "learning_rate": 9.993888224878313e-05,
      "loss": 2.0821,
      "step": 97
    },
    {
      "epoch": 0.03227798591955206,
      "grad_norm": 3.1359169483184814,
      "learning_rate": 9.993625438444287e-05,
      "loss": 1.8411,
      "step": 98
    },
    {
      "epoch": 0.03260735312281279,
      "grad_norm": 3.9783849716186523,
      "learning_rate": 9.993357124417209e-05,
      "loss": 2.3985,
      "step": 99
    },
    {
      "epoch": 0.03293672032607353,
      "grad_norm": 3.3506076335906982,
      "learning_rate": 9.993083283094084e-05,
      "loss": 2.2946,
      "step": 100
    },
    {
      "epoch": 0.033266087529334264,
      "grad_norm": 2.503321409225464,
      "learning_rate": 9.992803914778034e-05,
      "loss": 2.6129,
      "step": 101
    },
    {
      "epoch": 0.033595454732595,
      "grad_norm": 3.6384737491607666,
      "learning_rate": 9.992519019778301e-05,
      "loss": 2.6609,
      "step": 102
    },
    {
      "epoch": 0.033924821935855734,
      "grad_norm": 2.3811521530151367,
      "learning_rate": 9.992228598410244e-05,
      "loss": 2.3603,
      "step": 103
    },
    {
      "epoch": 0.03425418913911647,
      "grad_norm": 2.4915661811828613,
      "learning_rate": 9.991932650995341e-05,
      "loss": 2.2988,
      "step": 104
    },
    {
      "epoch": 0.034583556342377204,
      "grad_norm": 2.7037079334259033,
      "learning_rate": 9.991631177861182e-05,
      "loss": 2.4169,
      "step": 105
    },
    {
      "epoch": 0.03491292354563794,
      "grad_norm": 2.6235880851745605,
      "learning_rate": 9.991324179341478e-05,
      "loss": 2.2463,
      "step": 106
    },
    {
      "epoch": 0.03524229074889868,
      "grad_norm": 2.3382320404052734,
      "learning_rate": 9.991011655776055e-05,
      "loss": 2.2194,
      "step": 107
    },
    {
      "epoch": 0.03557165795215941,
      "grad_norm": 2.669034957885742,
      "learning_rate": 9.990693607510853e-05,
      "loss": 2.4065,
      "step": 108
    },
    {
      "epoch": 0.03590102515542015,
      "grad_norm": 2.219449281692505,
      "learning_rate": 9.990370034897931e-05,
      "loss": 2.0945,
      "step": 109
    },
    {
      "epoch": 0.036230392358680884,
      "grad_norm": 3.07869291305542,
      "learning_rate": 9.99004093829546e-05,
      "loss": 2.3285,
      "step": 110
    },
    {
      "epoch": 0.03655975956194162,
      "grad_norm": 2.4732134342193604,
      "learning_rate": 9.989706318067728e-05,
      "loss": 2.0391,
      "step": 111
    },
    {
      "epoch": 0.036889126765202354,
      "grad_norm": 2.9737696647644043,
      "learning_rate": 9.989366174585135e-05,
      "loss": 2.0482,
      "step": 112
    },
    {
      "epoch": 0.03721849396846309,
      "grad_norm": 2.8496570587158203,
      "learning_rate": 9.989020508224197e-05,
      "loss": 2.3701,
      "step": 113
    },
    {
      "epoch": 0.037547861171723824,
      "grad_norm": 3.108743190765381,
      "learning_rate": 9.98866931936754e-05,
      "loss": 2.4347,
      "step": 114
    },
    {
      "epoch": 0.03787722837498456,
      "grad_norm": 2.642160177230835,
      "learning_rate": 9.988312608403909e-05,
      "loss": 2.3828,
      "step": 115
    },
    {
      "epoch": 0.038206595578245295,
      "grad_norm": 2.7258148193359375,
      "learning_rate": 9.987950375728157e-05,
      "loss": 2.0533,
      "step": 116
    },
    {
      "epoch": 0.038535962781506033,
      "grad_norm": 2.6295080184936523,
      "learning_rate": 9.98758262174125e-05,
      "loss": 2.3752,
      "step": 117
    },
    {
      "epoch": 0.038865329984766765,
      "grad_norm": 2.905092477798462,
      "learning_rate": 9.987209346850263e-05,
      "loss": 2.031,
      "step": 118
    },
    {
      "epoch": 0.039194697188027504,
      "grad_norm": 2.780714988708496,
      "learning_rate": 9.986830551468388e-05,
      "loss": 2.0643,
      "step": 119
    },
    {
      "epoch": 0.039524064391288236,
      "grad_norm": 2.858346939086914,
      "learning_rate": 9.986446236014925e-05,
      "loss": 2.3293,
      "step": 120
    },
    {
      "epoch": 0.039853431594548974,
      "grad_norm": 3.3452649116516113,
      "learning_rate": 9.986056400915284e-05,
      "loss": 2.4871,
      "step": 121
    },
    {
      "epoch": 0.040182798797809706,
      "grad_norm": 2.921239137649536,
      "learning_rate": 9.985661046600984e-05,
      "loss": 2.3667,
      "step": 122
    },
    {
      "epoch": 0.040512166001070445,
      "grad_norm": 3.3637163639068604,
      "learning_rate": 9.985260173509656e-05,
      "loss": 2.2286,
      "step": 123
    },
    {
      "epoch": 0.040841533204331176,
      "grad_norm": 2.8042445182800293,
      "learning_rate": 9.984853782085035e-05,
      "loss": 1.9868,
      "step": 124
    },
    {
      "epoch": 0.041170900407591915,
      "grad_norm": 3.1596014499664307,
      "learning_rate": 9.984441872776973e-05,
      "loss": 1.9366,
      "step": 125
    },
    {
      "epoch": 0.04150026761085265,
      "grad_norm": 2.2009477615356445,
      "learning_rate": 9.984024446041423e-05,
      "loss": 2.7554,
      "step": 126
    },
    {
      "epoch": 0.041829634814113385,
      "grad_norm": 2.5528178215026855,
      "learning_rate": 9.983601502340443e-05,
      "loss": 2.4577,
      "step": 127
    },
    {
      "epoch": 0.04215900201737412,
      "grad_norm": 2.445089101791382,
      "learning_rate": 9.983173042142208e-05,
      "loss": 2.3525,
      "step": 128
    },
    {
      "epoch": 0.042488369220634856,
      "grad_norm": 2.8235416412353516,
      "learning_rate": 9.98273906592099e-05,
      "loss": 2.6086,
      "step": 129
    },
    {
      "epoch": 0.04281773642389559,
      "grad_norm": 2.569711446762085,
      "learning_rate": 9.98229957415717e-05,
      "loss": 2.4181,
      "step": 130
    },
    {
      "epoch": 0.043147103627156326,
      "grad_norm": 2.627051591873169,
      "learning_rate": 9.981854567337237e-05,
      "loss": 2.307,
      "step": 131
    },
    {
      "epoch": 0.043476470830417065,
      "grad_norm": 2.5236432552337646,
      "learning_rate": 9.98140404595378e-05,
      "loss": 2.5002,
      "step": 132
    },
    {
      "epoch": 0.043805838033677796,
      "grad_norm": 2.83263897895813,
      "learning_rate": 9.980948010505493e-05,
      "loss": 2.6664,
      "step": 133
    },
    {
      "epoch": 0.044135205236938535,
      "grad_norm": 2.404651641845703,
      "learning_rate": 9.98048646149718e-05,
      "loss": 2.4746,
      "step": 134
    },
    {
      "epoch": 0.04446457244019927,
      "grad_norm": 2.768198013305664,
      "learning_rate": 9.980019399439741e-05,
      "loss": 2.3371,
      "step": 135
    },
    {
      "epoch": 0.044793939643460005,
      "grad_norm": 3.2502036094665527,
      "learning_rate": 9.97954682485018e-05,
      "loss": 2.7268,
      "step": 136
    },
    {
      "epoch": 0.04512330684672074,
      "grad_norm": 2.356938362121582,
      "learning_rate": 9.979068738251605e-05,
      "loss": 2.271,
      "step": 137
    },
    {
      "epoch": 0.045452674049981476,
      "grad_norm": 2.790741205215454,
      "learning_rate": 9.978585140173225e-05,
      "loss": 2.5781,
      "step": 138
    },
    {
      "epoch": 0.04578204125324221,
      "grad_norm": 3.1573290824890137,
      "learning_rate": 9.978096031150346e-05,
      "loss": 2.3545,
      "step": 139
    },
    {
      "epoch": 0.046111408456502946,
      "grad_norm": 2.6217617988586426,
      "learning_rate": 9.977601411724382e-05,
      "loss": 2.2068,
      "step": 140
    },
    {
      "epoch": 0.04644077565976368,
      "grad_norm": 3.0412278175354004,
      "learning_rate": 9.977101282442839e-05,
      "loss": 2.386,
      "step": 141
    },
    {
      "epoch": 0.046770142863024416,
      "grad_norm": 3.4128334522247314,
      "learning_rate": 9.976595643859326e-05,
      "loss": 2.5365,
      "step": 142
    },
    {
      "epoch": 0.04709951006628515,
      "grad_norm": 3.5312652587890625,
      "learning_rate": 9.976084496533547e-05,
      "loss": 2.2243,
      "step": 143
    },
    {
      "epoch": 0.04742887726954589,
      "grad_norm": 2.8533828258514404,
      "learning_rate": 9.97556784103131e-05,
      "loss": 2.2667,
      "step": 144
    },
    {
      "epoch": 0.04775824447280662,
      "grad_norm": 3.081562042236328,
      "learning_rate": 9.975045677924515e-05,
      "loss": 2.3508,
      "step": 145
    },
    {
      "epoch": 0.04808761167606736,
      "grad_norm": 3.2083470821380615,
      "learning_rate": 9.97451800779116e-05,
      "loss": 2.4371,
      "step": 146
    },
    {
      "epoch": 0.04841697887932809,
      "grad_norm": 3.0021450519561768,
      "learning_rate": 9.973984831215337e-05,
      "loss": 1.9932,
      "step": 147
    },
    {
      "epoch": 0.04874634608258883,
      "grad_norm": 3.146559953689575,
      "learning_rate": 9.973446148787238e-05,
      "loss": 2.0892,
      "step": 148
    },
    {
      "epoch": 0.04907571328584956,
      "grad_norm": 3.2698886394500732,
      "learning_rate": 9.972901961103145e-05,
      "loss": 1.9643,
      "step": 149
    },
    {
      "epoch": 0.0494050804891103,
      "grad_norm": 3.640223503112793,
      "learning_rate": 9.972352268765434e-05,
      "loss": 2.0784,
      "step": 150
    },
    {
      "epoch": 0.04973444769237103,
      "grad_norm": 2.1278653144836426,
      "learning_rate": 9.971797072382579e-05,
      "loss": 2.4746,
      "step": 151
    },
    {
      "epoch": 0.05006381489563177,
      "grad_norm": 2.588524341583252,
      "learning_rate": 9.971236372569142e-05,
      "loss": 2.2406,
      "step": 152
    },
    {
      "epoch": 0.0503931820988925,
      "grad_norm": 2.5750515460968018,
      "learning_rate": 9.97067016994578e-05,
      "loss": 2.254,
      "step": 153
    },
    {
      "epoch": 0.05072254930215324,
      "grad_norm": 2.47708797454834,
      "learning_rate": 9.970098465139236e-05,
      "loss": 2.2676,
      "step": 154
    },
    {
      "epoch": 0.05105191650541397,
      "grad_norm": 2.496344566345215,
      "learning_rate": 9.969521258782351e-05,
      "loss": 2.3978,
      "step": 155
    },
    {
      "epoch": 0.05138128370867471,
      "grad_norm": 2.462228775024414,
      "learning_rate": 9.968938551514048e-05,
      "loss": 2.3217,
      "step": 156
    },
    {
      "epoch": 0.05171065091193544,
      "grad_norm": 2.3593785762786865,
      "learning_rate": 9.968350343979346e-05,
      "loss": 2.0463,
      "step": 157
    },
    {
      "epoch": 0.05204001811519618,
      "grad_norm": 2.4684934616088867,
      "learning_rate": 9.967756636829348e-05,
      "loss": 2.3118,
      "step": 158
    },
    {
      "epoch": 0.05236938531845692,
      "grad_norm": 3.1452877521514893,
      "learning_rate": 9.967157430721248e-05,
      "loss": 2.3831,
      "step": 159
    },
    {
      "epoch": 0.05269875252171765,
      "grad_norm": 2.745805025100708,
      "learning_rate": 9.966552726318323e-05,
      "loss": 2.3436,
      "step": 160
    },
    {
      "epoch": 0.05302811972497839,
      "grad_norm": 2.490478038787842,
      "learning_rate": 9.965942524289941e-05,
      "loss": 2.2698,
      "step": 161
    },
    {
      "epoch": 0.05335748692823912,
      "grad_norm": 2.3748586177825928,
      "learning_rate": 9.96532682531155e-05,
      "loss": 2.4236,
      "step": 162
    },
    {
      "epoch": 0.05368685413149986,
      "grad_norm": 2.378679037094116,
      "learning_rate": 9.964705630064686e-05,
      "loss": 2.1829,
      "step": 163
    },
    {
      "epoch": 0.05401622133476059,
      "grad_norm": 2.8342976570129395,
      "learning_rate": 9.964078939236971e-05,
      "loss": 2.3079,
      "step": 164
    },
    {
      "epoch": 0.05434558853802133,
      "grad_norm": 2.9072232246398926,
      "learning_rate": 9.963446753522104e-05,
      "loss": 2.3423,
      "step": 165
    },
    {
      "epoch": 0.05467495574128206,
      "grad_norm": 2.8593156337738037,
      "learning_rate": 9.962809073619875e-05,
      "loss": 2.2235,
      "step": 166
    },
    {
      "epoch": 0.0550043229445428,
      "grad_norm": 2.832493543624878,
      "learning_rate": 9.962165900236146e-05,
      "loss": 2.2889,
      "step": 167
    },
    {
      "epoch": 0.05533369014780353,
      "grad_norm": 2.806488037109375,
      "learning_rate": 9.961517234082866e-05,
      "loss": 2.1615,
      "step": 168
    },
    {
      "epoch": 0.05566305735106427,
      "grad_norm": 3.301941394805908,
      "learning_rate": 9.960863075878067e-05,
      "loss": 2.2195,
      "step": 169
    },
    {
      "epoch": 0.055992424554325,
      "grad_norm": 2.458503484725952,
      "learning_rate": 9.960203426345851e-05,
      "loss": 2.1645,
      "step": 170
    },
    {
      "epoch": 0.05632179175758574,
      "grad_norm": 2.415736675262451,
      "learning_rate": 9.959538286216408e-05,
      "loss": 2.012,
      "step": 171
    },
    {
      "epoch": 0.05665115896084647,
      "grad_norm": 2.7939391136169434,
      "learning_rate": 9.958867656225997e-05,
      "loss": 2.3091,
      "step": 172
    },
    {
      "epoch": 0.05698052616410721,
      "grad_norm": 2.971738576889038,
      "learning_rate": 9.958191537116963e-05,
      "loss": 2.1566,
      "step": 173
    },
    {
      "epoch": 0.05730989336736794,
      "grad_norm": 3.0159671306610107,
      "learning_rate": 9.957509929637719e-05,
      "loss": 2.0143,
      "step": 174
    },
    {
      "epoch": 0.05763926057062868,
      "grad_norm": 3.120633125305176,
      "learning_rate": 9.956822834542759e-05,
      "loss": 1.8494,
      "step": 175
    },
    {
      "epoch": 0.05796862777388941,
      "grad_norm": 2.3217828273773193,
      "learning_rate": 9.956130252592646e-05,
      "loss": 2.6393,
      "step": 176
    },
    {
      "epoch": 0.05829799497715015,
      "grad_norm": 2.3948323726654053,
      "learning_rate": 9.955432184554024e-05,
      "loss": 2.6342,
      "step": 177
    },
    {
      "epoch": 0.05862736218041088,
      "grad_norm": 2.1750946044921875,
      "learning_rate": 9.9547286311996e-05,
      "loss": 2.3683,
      "step": 178
    },
    {
      "epoch": 0.05895672938367162,
      "grad_norm": 2.6148295402526855,
      "learning_rate": 9.954019593308163e-05,
      "loss": 2.5178,
      "step": 179
    },
    {
      "epoch": 0.05928609658693235,
      "grad_norm": 2.6671082973480225,
      "learning_rate": 9.953305071664566e-05,
      "loss": 2.3501,
      "step": 180
    },
    {
      "epoch": 0.05961546379019309,
      "grad_norm": 2.740058422088623,
      "learning_rate": 9.952585067059734e-05,
      "loss": 2.3677,
      "step": 181
    },
    {
      "epoch": 0.059944830993453824,
      "grad_norm": 2.2972073554992676,
      "learning_rate": 9.951859580290664e-05,
      "loss": 2.5598,
      "step": 182
    },
    {
      "epoch": 0.06027419819671456,
      "grad_norm": 2.5012662410736084,
      "learning_rate": 9.951128612160417e-05,
      "loss": 2.247,
      "step": 183
    },
    {
      "epoch": 0.060603565399975294,
      "grad_norm": 2.5461273193359375,
      "learning_rate": 9.950392163478121e-05,
      "loss": 2.4683,
      "step": 184
    },
    {
      "epoch": 0.06093293260323603,
      "grad_norm": 3.3869452476501465,
      "learning_rate": 9.949650235058978e-05,
      "loss": 2.5158,
      "step": 185
    },
    {
      "epoch": 0.06126229980649677,
      "grad_norm": 2.2931103706359863,
      "learning_rate": 9.948902827724248e-05,
      "loss": 2.2837,
      "step": 186
    },
    {
      "epoch": 0.0615916670097575,
      "grad_norm": 2.4342219829559326,
      "learning_rate": 9.94814994230126e-05,
      "loss": 2.2988,
      "step": 187
    },
    {
      "epoch": 0.06192103421301824,
      "grad_norm": 2.924483299255371,
      "learning_rate": 9.947391579623401e-05,
      "loss": 2.4679,
      "step": 188
    },
    {
      "epoch": 0.06225040141627897,
      "grad_norm": 2.381253480911255,
      "learning_rate": 9.946627740530131e-05,
      "loss": 2.0651,
      "step": 189
    },
    {
      "epoch": 0.0625797686195397,
      "grad_norm": 2.3202030658721924,
      "learning_rate": 9.945858425866962e-05,
      "loss": 2.0079,
      "step": 190
    },
    {
      "epoch": 0.06290913582280044,
      "grad_norm": 2.717766523361206,
      "learning_rate": 9.945083636485476e-05,
      "loss": 2.6815,
      "step": 191
    },
    {
      "epoch": 0.06323850302606118,
      "grad_norm": 3.0983986854553223,
      "learning_rate": 9.944303373243307e-05,
      "loss": 2.4154,
      "step": 192
    },
    {
      "epoch": 0.06356787022932192,
      "grad_norm": 2.5674819946289062,
      "learning_rate": 9.943517637004151e-05,
      "loss": 1.8935,
      "step": 193
    },
    {
      "epoch": 0.06389723743258265,
      "grad_norm": 2.8902697563171387,
      "learning_rate": 9.942726428637765e-05,
      "loss": 2.19,
      "step": 194
    },
    {
      "epoch": 0.06422660463584338,
      "grad_norm": 2.824228525161743,
      "learning_rate": 9.941929749019961e-05,
      "loss": 1.7962,
      "step": 195
    },
    {
      "epoch": 0.06455597183910412,
      "grad_norm": 2.9178125858306885,
      "learning_rate": 9.941127599032605e-05,
      "loss": 1.9707,
      "step": 196
    },
    {
      "epoch": 0.06488533904236486,
      "grad_norm": 3.112900495529175,
      "learning_rate": 9.940319979563624e-05,
      "loss": 2.1085,
      "step": 197
    },
    {
      "epoch": 0.06521470624562559,
      "grad_norm": 2.4243252277374268,
      "learning_rate": 9.939506891506993e-05,
      "loss": 1.6683,
      "step": 198
    },
    {
      "epoch": 0.06554407344888633,
      "grad_norm": 3.3545095920562744,
      "learning_rate": 9.938688335762747e-05,
      "loss": 1.9903,
      "step": 199
    },
    {
      "epoch": 0.06587344065214706,
      "grad_norm": 4.020653247833252,
      "learning_rate": 9.937864313236968e-05,
      "loss": 1.9782,
      "step": 200
    },
    {
      "epoch": 0.0662028078554078,
      "grad_norm": 2.1895785331726074,
      "learning_rate": 9.93703482484179e-05,
      "loss": 2.4232,
      "step": 201
    },
    {
      "epoch": 0.06653217505866853,
      "grad_norm": 2.779294967651367,
      "learning_rate": 9.9361998714954e-05,
      "loss": 2.4474,
      "step": 202
    },
    {
      "epoch": 0.06686154226192927,
      "grad_norm": 2.666214942932129,
      "learning_rate": 9.935359454122033e-05,
      "loss": 2.3747,
      "step": 203
    },
    {
      "epoch": 0.06719090946519,
      "grad_norm": 2.4927260875701904,
      "learning_rate": 9.93451357365197e-05,
      "loss": 2.3229,
      "step": 204
    },
    {
      "epoch": 0.06752027666845074,
      "grad_norm": 2.6281189918518066,
      "learning_rate": 9.933662231021543e-05,
      "loss": 2.3106,
      "step": 205
    },
    {
      "epoch": 0.06784964387171147,
      "grad_norm": 2.487201452255249,
      "learning_rate": 9.932805427173128e-05,
      "loss": 2.1396,
      "step": 206
    },
    {
      "epoch": 0.0681790110749722,
      "grad_norm": 2.6833136081695557,
      "learning_rate": 9.931943163055148e-05,
      "loss": 2.6855,
      "step": 207
    },
    {
      "epoch": 0.06850837827823295,
      "grad_norm": 2.670117139816284,
      "learning_rate": 9.931075439622069e-05,
      "loss": 2.0407,
      "step": 208
    },
    {
      "epoch": 0.06883774548149368,
      "grad_norm": 2.8142549991607666,
      "learning_rate": 9.930202257834397e-05,
      "loss": 2.5156,
      "step": 209
    },
    {
      "epoch": 0.06916711268475441,
      "grad_norm": 2.5977020263671875,
      "learning_rate": 9.929323618658686e-05,
      "loss": 2.2659,
      "step": 210
    },
    {
      "epoch": 0.06949647988801515,
      "grad_norm": 2.188446521759033,
      "learning_rate": 9.928439523067526e-05,
      "loss": 1.853,
      "step": 211
    },
    {
      "epoch": 0.06982584709127589,
      "grad_norm": 2.691819906234741,
      "learning_rate": 9.92754997203955e-05,
      "loss": 2.0574,
      "step": 212
    },
    {
      "epoch": 0.07015521429453662,
      "grad_norm": 2.599579334259033,
      "learning_rate": 9.926654966559427e-05,
      "loss": 2.1189,
      "step": 213
    },
    {
      "epoch": 0.07048458149779736,
      "grad_norm": 3.0365447998046875,
      "learning_rate": 9.925754507617868e-05,
      "loss": 2.2052,
      "step": 214
    },
    {
      "epoch": 0.07081394870105809,
      "grad_norm": 3.086376190185547,
      "learning_rate": 9.924848596211618e-05,
      "loss": 2.5058,
      "step": 215
    },
    {
      "epoch": 0.07114331590431883,
      "grad_norm": 3.420269012451172,
      "learning_rate": 9.923937233343453e-05,
      "loss": 2.0489,
      "step": 216
    },
    {
      "epoch": 0.07147268310757957,
      "grad_norm": 2.6292338371276855,
      "learning_rate": 9.923020420022191e-05,
      "loss": 2.4083,
      "step": 217
    },
    {
      "epoch": 0.0718020503108403,
      "grad_norm": 3.046985387802124,
      "learning_rate": 9.92209815726268e-05,
      "loss": 2.2224,
      "step": 218
    },
    {
      "epoch": 0.07213141751410103,
      "grad_norm": 3.1213648319244385,
      "learning_rate": 9.921170446085798e-05,
      "loss": 2.0798,
      "step": 219
    },
    {
      "epoch": 0.07246078471736177,
      "grad_norm": 2.71501088142395,
      "learning_rate": 9.920237287518462e-05,
      "loss": 2.3788,
      "step": 220
    },
    {
      "epoch": 0.0727901519206225,
      "grad_norm": 2.7265591621398926,
      "learning_rate": 9.919298682593605e-05,
      "loss": 2.0445,
      "step": 221
    },
    {
      "epoch": 0.07311951912388324,
      "grad_norm": 3.3277218341827393,
      "learning_rate": 9.918354632350202e-05,
      "loss": 2.1541,
      "step": 222
    },
    {
      "epoch": 0.07344888632714397,
      "grad_norm": 3.288353443145752,
      "learning_rate": 9.917405137833249e-05,
      "loss": 2.2666,
      "step": 223
    },
    {
      "epoch": 0.07377825353040471,
      "grad_norm": 3.049253463745117,
      "learning_rate": 9.916450200093771e-05,
      "loss": 1.8277,
      "step": 224
    },
    {
      "epoch": 0.07410762073366545,
      "grad_norm": 3.3985233306884766,
      "learning_rate": 9.915489820188814e-05,
      "loss": 1.9754,
      "step": 225
    },
    {
      "epoch": 0.07443698793692619,
      "grad_norm": 2.7046058177948,
      "learning_rate": 9.914523999181456e-05,
      "loss": 2.3886,
      "step": 226
    },
    {
      "epoch": 0.07476635514018691,
      "grad_norm": 2.472142219543457,
      "learning_rate": 9.91355273814079e-05,
      "loss": 2.4031,
      "step": 227
    },
    {
      "epoch": 0.07509572234344765,
      "grad_norm": 2.1829640865325928,
      "learning_rate": 9.912576038141933e-05,
      "loss": 2.34,
      "step": 228
    },
    {
      "epoch": 0.07542508954670839,
      "grad_norm": 2.4181277751922607,
      "learning_rate": 9.911593900266026e-05,
      "loss": 2.1865,
      "step": 229
    },
    {
      "epoch": 0.07575445674996913,
      "grad_norm": 2.6987533569335938,
      "learning_rate": 9.910606325600223e-05,
      "loss": 2.3855,
      "step": 230
    },
    {
      "epoch": 0.07608382395322985,
      "grad_norm": 2.5802574157714844,
      "learning_rate": 9.909613315237702e-05,
      "loss": 2.3325,
      "step": 231
    },
    {
      "epoch": 0.07641319115649059,
      "grad_norm": 2.356382369995117,
      "learning_rate": 9.90861487027766e-05,
      "loss": 2.1446,
      "step": 232
    },
    {
      "epoch": 0.07674255835975133,
      "grad_norm": 2.3594143390655518,
      "learning_rate": 9.907610991825298e-05,
      "loss": 2.204,
      "step": 233
    },
    {
      "epoch": 0.07707192556301207,
      "grad_norm": 2.47929310798645,
      "learning_rate": 9.906601680991842e-05,
      "loss": 2.2276,
      "step": 234
    },
    {
      "epoch": 0.07740129276627279,
      "grad_norm": 3.100281000137329,
      "learning_rate": 9.905586938894531e-05,
      "loss": 2.3447,
      "step": 235
    },
    {
      "epoch": 0.07773065996953353,
      "grad_norm": 2.3578147888183594,
      "learning_rate": 9.904566766656612e-05,
      "loss": 2.1765,
      "step": 236
    },
    {
      "epoch": 0.07806002717279427,
      "grad_norm": 2.6452767848968506,
      "learning_rate": 9.903541165407341e-05,
      "loss": 2.2725,
      "step": 237
    },
    {
      "epoch": 0.07838939437605501,
      "grad_norm": 2.8530819416046143,
      "learning_rate": 9.902510136281989e-05,
      "loss": 2.1286,
      "step": 238
    },
    {
      "epoch": 0.07871876157931575,
      "grad_norm": 3.382469892501831,
      "learning_rate": 9.901473680421833e-05,
      "loss": 2.5109,
      "step": 239
    },
    {
      "epoch": 0.07904812878257647,
      "grad_norm": 2.7474164962768555,
      "learning_rate": 9.900431798974158e-05,
      "loss": 2.0808,
      "step": 240
    },
    {
      "epoch": 0.07937749598583721,
      "grad_norm": 2.889378070831299,
      "learning_rate": 9.899384493092252e-05,
      "loss": 2.6418,
      "step": 241
    },
    {
      "epoch": 0.07970686318909795,
      "grad_norm": 2.788848876953125,
      "learning_rate": 9.89833176393541e-05,
      "loss": 2.4061,
      "step": 242
    },
    {
      "epoch": 0.08003623039235869,
      "grad_norm": 2.540822982788086,
      "learning_rate": 9.897273612668927e-05,
      "loss": 1.9808,
      "step": 243
    },
    {
      "epoch": 0.08036559759561941,
      "grad_norm": 2.4531843662261963,
      "learning_rate": 9.896210040464105e-05,
      "loss": 1.9014,
      "step": 244
    },
    {
      "epoch": 0.08069496479888015,
      "grad_norm": 3.2541840076446533,
      "learning_rate": 9.895141048498244e-05,
      "loss": 2.5161,
      "step": 245
    },
    {
      "epoch": 0.08102433200214089,
      "grad_norm": 2.396268367767334,
      "learning_rate": 9.89406663795464e-05,
      "loss": 1.944,
      "step": 246
    },
    {
      "epoch": 0.08135369920540163,
      "grad_norm": 3.345994710922241,
      "learning_rate": 9.892986810022594e-05,
      "loss": 2.4834,
      "step": 247
    },
    {
      "epoch": 0.08168306640866235,
      "grad_norm": 2.90889048576355,
      "learning_rate": 9.891901565897397e-05,
      "loss": 2.041,
      "step": 248
    },
    {
      "epoch": 0.08201243361192309,
      "grad_norm": 3.424887180328369,
      "learning_rate": 9.89081090678034e-05,
      "loss": 2.1515,
      "step": 249
    },
    {
      "epoch": 0.08234180081518383,
      "grad_norm": 3.129890203475952,
      "learning_rate": 9.889714833878705e-05,
      "loss": 1.5253,
      "step": 250
    },
    {
      "epoch": 0.08267116801844457,
      "grad_norm": 2.033660411834717,
      "learning_rate": 9.888613348405766e-05,
      "loss": 2.4284,
      "step": 251
    },
    {
      "epoch": 0.0830005352217053,
      "grad_norm": 2.726742744445801,
      "learning_rate": 9.887506451580794e-05,
      "loss": 2.5739,
      "step": 252
    },
    {
      "epoch": 0.08332990242496603,
      "grad_norm": 2.1317944526672363,
      "learning_rate": 9.886394144629044e-05,
      "loss": 2.3195,
      "step": 253
    },
    {
      "epoch": 0.08365926962822677,
      "grad_norm": 2.2404332160949707,
      "learning_rate": 9.885276428781763e-05,
      "loss": 2.6881,
      "step": 254
    },
    {
      "epoch": 0.08398863683148751,
      "grad_norm": 2.376636028289795,
      "learning_rate": 9.88415330527618e-05,
      "loss": 2.2499,
      "step": 255
    },
    {
      "epoch": 0.08431800403474823,
      "grad_norm": 2.24052095413208,
      "learning_rate": 9.88302477535552e-05,
      "loss": 2.4867,
      "step": 256
    },
    {
      "epoch": 0.08464737123800897,
      "grad_norm": 2.6284337043762207,
      "learning_rate": 9.881890840268981e-05,
      "loss": 2.5267,
      "step": 257
    },
    {
      "epoch": 0.08497673844126971,
      "grad_norm": 2.5630033016204834,
      "learning_rate": 9.880751501271755e-05,
      "loss": 2.3627,
      "step": 258
    },
    {
      "epoch": 0.08530610564453045,
      "grad_norm": 2.3372888565063477,
      "learning_rate": 9.879606759625004e-05,
      "loss": 2.4922,
      "step": 259
    },
    {
      "epoch": 0.08563547284779117,
      "grad_norm": 2.3916890621185303,
      "learning_rate": 9.878456616595882e-05,
      "loss": 2.1065,
      "step": 260
    },
    {
      "epoch": 0.08596484005105191,
      "grad_norm": 2.6594431400299072,
      "learning_rate": 9.877301073457515e-05,
      "loss": 2.034,
      "step": 261
    },
    {
      "epoch": 0.08629420725431265,
      "grad_norm": 2.3828513622283936,
      "learning_rate": 9.876140131489008e-05,
      "loss": 1.9415,
      "step": 262
    },
    {
      "epoch": 0.08662357445757339,
      "grad_norm": 2.558377265930176,
      "learning_rate": 9.874973791975442e-05,
      "loss": 2.1253,
      "step": 263
    },
    {
      "epoch": 0.08695294166083413,
      "grad_norm": 2.41732120513916,
      "learning_rate": 9.873802056207872e-05,
      "loss": 2.4188,
      "step": 264
    },
    {
      "epoch": 0.08728230886409485,
      "grad_norm": 2.653940439224243,
      "learning_rate": 9.872624925483329e-05,
      "loss": 2.1664,
      "step": 265
    },
    {
      "epoch": 0.08761167606735559,
      "grad_norm": 2.6138205528259277,
      "learning_rate": 9.871442401104816e-05,
      "loss": 1.9422,
      "step": 266
    },
    {
      "epoch": 0.08794104327061633,
      "grad_norm": 2.4396393299102783,
      "learning_rate": 9.870254484381299e-05,
      "loss": 2.2988,
      "step": 267
    },
    {
      "epoch": 0.08827041047387707,
      "grad_norm": 3.269818067550659,
      "learning_rate": 9.869061176627724e-05,
      "loss": 2.4621,
      "step": 268
    },
    {
      "epoch": 0.0885997776771378,
      "grad_norm": 2.802405595779419,
      "learning_rate": 9.867862479164996e-05,
      "loss": 2.0724,
      "step": 269
    },
    {
      "epoch": 0.08892914488039853,
      "grad_norm": 3.132948160171509,
      "learning_rate": 9.866658393319988e-05,
      "loss": 2.3857,
      "step": 270
    },
    {
      "epoch": 0.08925851208365927,
      "grad_norm": 3.446258544921875,
      "learning_rate": 9.865448920425541e-05,
      "loss": 2.1038,
      "step": 271
    },
    {
      "epoch": 0.08958787928692001,
      "grad_norm": 3.5878865718841553,
      "learning_rate": 9.864234061820458e-05,
      "loss": 2.4794,
      "step": 272
    },
    {
      "epoch": 0.08991724649018074,
      "grad_norm": 2.7170724868774414,
      "learning_rate": 9.863013818849499e-05,
      "loss": 2.0187,
      "step": 273
    },
    {
      "epoch": 0.09024661369344147,
      "grad_norm": 3.4046213626861572,
      "learning_rate": 9.861788192863388e-05,
      "loss": 1.9167,
      "step": 274
    },
    {
      "epoch": 0.09057598089670221,
      "grad_norm": 4.091611385345459,
      "learning_rate": 9.860557185218808e-05,
      "loss": 2.203,
      "step": 275
    },
    {
      "epoch": 0.09090534809996295,
      "grad_norm": 1.9869258403778076,
      "learning_rate": 9.859320797278397e-05,
      "loss": 2.4434,
      "step": 276
    },
    {
      "epoch": 0.09123471530322368,
      "grad_norm": 2.5179264545440674,
      "learning_rate": 9.85807903041075e-05,
      "loss": 2.2607,
      "step": 277
    },
    {
      "epoch": 0.09156408250648441,
      "grad_norm": 2.7460920810699463,
      "learning_rate": 9.856831885990416e-05,
      "loss": 2.5456,
      "step": 278
    },
    {
      "epoch": 0.09189344970974515,
      "grad_norm": 2.803849458694458,
      "learning_rate": 9.855579365397898e-05,
      "loss": 2.5656,
      "step": 279
    },
    {
      "epoch": 0.09222281691300589,
      "grad_norm": 2.4078571796417236,
      "learning_rate": 9.854321470019646e-05,
      "loss": 2.4701,
      "step": 280
    },
    {
      "epoch": 0.09255218411626662,
      "grad_norm": 2.735297679901123,
      "learning_rate": 9.853058201248063e-05,
      "loss": 2.4882,
      "step": 281
    },
    {
      "epoch": 0.09288155131952736,
      "grad_norm": 2.441884756088257,
      "learning_rate": 9.851789560481499e-05,
      "loss": 2.2741,
      "step": 282
    },
    {
      "epoch": 0.0932109185227881,
      "grad_norm": 2.480804681777954,
      "learning_rate": 9.85051554912425e-05,
      "loss": 2.2021,
      "step": 283
    },
    {
      "epoch": 0.09354028572604883,
      "grad_norm": 2.59104585647583,
      "learning_rate": 9.849236168586558e-05,
      "loss": 2.7452,
      "step": 284
    },
    {
      "epoch": 0.09386965292930956,
      "grad_norm": 2.392718553543091,
      "learning_rate": 9.847951420284605e-05,
      "loss": 2.231,
      "step": 285
    },
    {
      "epoch": 0.0941990201325703,
      "grad_norm": 2.880892753601074,
      "learning_rate": 9.84666130564052e-05,
      "loss": 2.5514,
      "step": 286
    },
    {
      "epoch": 0.09452838733583104,
      "grad_norm": 2.457305431365967,
      "learning_rate": 9.845365826082368e-05,
      "loss": 2.261,
      "step": 287
    },
    {
      "epoch": 0.09485775453909177,
      "grad_norm": 2.8255691528320312,
      "learning_rate": 9.844064983044157e-05,
      "loss": 2.4296,
      "step": 288
    },
    {
      "epoch": 0.0951871217423525,
      "grad_norm": 2.97965407371521,
      "learning_rate": 9.842758777965824e-05,
      "loss": 2.3662,
      "step": 289
    },
    {
      "epoch": 0.09551648894561324,
      "grad_norm": 2.454676389694214,
      "learning_rate": 9.841447212293249e-05,
      "loss": 2.1213,
      "step": 290
    },
    {
      "epoch": 0.09584585614887398,
      "grad_norm": 3.022413492202759,
      "learning_rate": 9.840130287478245e-05,
      "loss": 2.4408,
      "step": 291
    },
    {
      "epoch": 0.09617522335213471,
      "grad_norm": 3.0308666229248047,
      "learning_rate": 9.83880800497855e-05,
      "loss": 2.5066,
      "step": 292
    },
    {
      "epoch": 0.09650459055539545,
      "grad_norm": 2.674705743789673,
      "learning_rate": 9.837480366257844e-05,
      "loss": 2.2084,
      "step": 293
    },
    {
      "epoch": 0.09683395775865618,
      "grad_norm": 2.988152503967285,
      "learning_rate": 9.836147372785726e-05,
      "loss": 2.2093,
      "step": 294
    },
    {
      "epoch": 0.09716332496191692,
      "grad_norm": 2.502009868621826,
      "learning_rate": 9.834809026037728e-05,
      "loss": 2.259,
      "step": 295
    },
    {
      "epoch": 0.09749269216517766,
      "grad_norm": 2.590487241744995,
      "learning_rate": 9.833465327495306e-05,
      "loss": 2.1714,
      "step": 296
    },
    {
      "epoch": 0.0978220593684384,
      "grad_norm": 2.34224534034729,
      "learning_rate": 9.83211627864584e-05,
      "loss": 1.7337,
      "step": 297
    },
    {
      "epoch": 0.09815142657169912,
      "grad_norm": 2.985863447189331,
      "learning_rate": 9.83076188098263e-05,
      "loss": 2.0128,
      "step": 298
    },
    {
      "epoch": 0.09848079377495986,
      "grad_norm": 2.8273167610168457,
      "learning_rate": 9.829402136004904e-05,
      "loss": 2.0277,
      "step": 299
    },
    {
      "epoch": 0.0988101609782206,
      "grad_norm": 3.533780813217163,
      "learning_rate": 9.8280370452178e-05,
      "loss": 2.0994,
      "step": 300
    },
    {
      "epoch": 0.09913952818148133,
      "grad_norm": 2.0469958782196045,
      "learning_rate": 9.82666661013238e-05,
      "loss": 2.4583,
      "step": 301
    },
    {
      "epoch": 0.09946889538474206,
      "grad_norm": 2.3586039543151855,
      "learning_rate": 9.825290832265617e-05,
      "loss": 2.4677,
      "step": 302
    },
    {
      "epoch": 0.0997982625880028,
      "grad_norm": 2.269946813583374,
      "learning_rate": 9.823909713140404e-05,
      "loss": 2.3393,
      "step": 303
    },
    {
      "epoch": 0.10012762979126354,
      "grad_norm": 2.338125705718994,
      "learning_rate": 9.82252325428554e-05,
      "loss": 2.4787,
      "step": 304
    },
    {
      "epoch": 0.10045699699452428,
      "grad_norm": 2.1585426330566406,
      "learning_rate": 9.821131457235739e-05,
      "loss": 2.5393,
      "step": 305
    },
    {
      "epoch": 0.100786364197785,
      "grad_norm": 2.6568026542663574,
      "learning_rate": 9.819734323531624e-05,
      "loss": 2.5194,
      "step": 306
    },
    {
      "epoch": 0.10111573140104574,
      "grad_norm": 2.206839084625244,
      "learning_rate": 9.818331854719722e-05,
      "loss": 2.4154,
      "step": 307
    },
    {
      "epoch": 0.10144509860430648,
      "grad_norm": 2.444082498550415,
      "learning_rate": 9.816924052352468e-05,
      "loss": 2.2583,
      "step": 308
    },
    {
      "epoch": 0.10177446580756722,
      "grad_norm": 2.4031789302825928,
      "learning_rate": 9.815510917988202e-05,
      "loss": 2.5014,
      "step": 309
    },
    {
      "epoch": 0.10210383301082794,
      "grad_norm": 2.008598566055298,
      "learning_rate": 9.814092453191163e-05,
      "loss": 2.1755,
      "step": 310
    },
    {
      "epoch": 0.10243320021408868,
      "grad_norm": 2.6430673599243164,
      "learning_rate": 9.812668659531492e-05,
      "loss": 2.4391,
      "step": 311
    },
    {
      "epoch": 0.10276256741734942,
      "grad_norm": 2.4818711280822754,
      "learning_rate": 9.811239538585229e-05,
      "loss": 2.2518,
      "step": 312
    },
    {
      "epoch": 0.10309193462061016,
      "grad_norm": 2.733666181564331,
      "learning_rate": 9.80980509193431e-05,
      "loss": 2.3118,
      "step": 313
    },
    {
      "epoch": 0.10342130182387088,
      "grad_norm": 2.3446598052978516,
      "learning_rate": 9.808365321166568e-05,
      "loss": 2.3457,
      "step": 314
    },
    {
      "epoch": 0.10375066902713162,
      "grad_norm": 2.2961266040802,
      "learning_rate": 9.806920227875729e-05,
      "loss": 1.9795,
      "step": 315
    },
    {
      "epoch": 0.10408003623039236,
      "grad_norm": 2.767897844314575,
      "learning_rate": 9.805469813661408e-05,
      "loss": 2.2274,
      "step": 316
    },
    {
      "epoch": 0.1044094034336531,
      "grad_norm": 2.9815187454223633,
      "learning_rate": 9.804014080129111e-05,
      "loss": 2.4279,
      "step": 317
    },
    {
      "epoch": 0.10473877063691384,
      "grad_norm": 2.3294548988342285,
      "learning_rate": 9.802553028890237e-05,
      "loss": 1.9681,
      "step": 318
    },
    {
      "epoch": 0.10506813784017456,
      "grad_norm": 3.110809564590454,
      "learning_rate": 9.801086661562062e-05,
      "loss": 2.3353,
      "step": 319
    },
    {
      "epoch": 0.1053975050434353,
      "grad_norm": 2.6092398166656494,
      "learning_rate": 9.799614979767757e-05,
      "loss": 2.1682,
      "step": 320
    },
    {
      "epoch": 0.10572687224669604,
      "grad_norm": 3.033212184906006,
      "learning_rate": 9.798137985136367e-05,
      "loss": 1.9523,
      "step": 321
    },
    {
      "epoch": 0.10605623944995678,
      "grad_norm": 2.9443624019622803,
      "learning_rate": 9.79665567930282e-05,
      "loss": 2.1126,
      "step": 322
    },
    {
      "epoch": 0.1063856066532175,
      "grad_norm": 3.2337043285369873,
      "learning_rate": 9.795168063907929e-05,
      "loss": 2.1455,
      "step": 323
    },
    {
      "epoch": 0.10671497385647824,
      "grad_norm": 2.5773916244506836,
      "learning_rate": 9.793675140598377e-05,
      "loss": 1.7072,
      "step": 324
    },
    {
      "epoch": 0.10704434105973898,
      "grad_norm": 3.643908977508545,
      "learning_rate": 9.792176911026727e-05,
      "loss": 2.2543,
      "step": 325
    },
    {
      "epoch": 0.10737370826299972,
      "grad_norm": 1.8910375833511353,
      "learning_rate": 9.790673376851414e-05,
      "loss": 2.5806,
      "step": 326
    },
    {
      "epoch": 0.10770307546626044,
      "grad_norm": 2.3565833568573,
      "learning_rate": 9.789164539736741e-05,
      "loss": 2.4109,
      "step": 327
    },
    {
      "epoch": 0.10803244266952118,
      "grad_norm": 2.3572747707366943,
      "learning_rate": 9.78765040135289e-05,
      "loss": 2.3321,
      "step": 328
    },
    {
      "epoch": 0.10836180987278192,
      "grad_norm": 2.759070873260498,
      "learning_rate": 9.786130963375904e-05,
      "loss": 2.4526,
      "step": 329
    },
    {
      "epoch": 0.10869117707604266,
      "grad_norm": 2.419929027557373,
      "learning_rate": 9.784606227487693e-05,
      "loss": 2.1635,
      "step": 330
    },
    {
      "epoch": 0.10902054427930338,
      "grad_norm": 2.4334778785705566,
      "learning_rate": 9.783076195376036e-05,
      "loss": 2.5785,
      "step": 331
    },
    {
      "epoch": 0.10934991148256412,
      "grad_norm": 2.6062378883361816,
      "learning_rate": 9.781540868734567e-05,
      "loss": 2.4172,
      "step": 332
    },
    {
      "epoch": 0.10967927868582486,
      "grad_norm": 2.5069472789764404,
      "learning_rate": 9.780000249262787e-05,
      "loss": 2.3606,
      "step": 333
    },
    {
      "epoch": 0.1100086458890856,
      "grad_norm": 2.3731164932250977,
      "learning_rate": 9.778454338666053e-05,
      "loss": 2.3024,
      "step": 334
    },
    {
      "epoch": 0.11033801309234632,
      "grad_norm": 2.8083693981170654,
      "learning_rate": 9.776903138655581e-05,
      "loss": 2.426,
      "step": 335
    },
    {
      "epoch": 0.11066738029560706,
      "grad_norm": 2.4966108798980713,
      "learning_rate": 9.775346650948439e-05,
      "loss": 2.2338,
      "step": 336
    },
    {
      "epoch": 0.1109967474988678,
      "grad_norm": 2.9775607585906982,
      "learning_rate": 9.77378487726755e-05,
      "loss": 2.7963,
      "step": 337
    },
    {
      "epoch": 0.11132611470212854,
      "grad_norm": 2.692918062210083,
      "learning_rate": 9.77221781934169e-05,
      "loss": 2.4771,
      "step": 338
    },
    {
      "epoch": 0.11165548190538926,
      "grad_norm": 2.5929603576660156,
      "learning_rate": 9.770645478905481e-05,
      "loss": 2.4814,
      "step": 339
    },
    {
      "epoch": 0.11198484910865,
      "grad_norm": 2.6808626651763916,
      "learning_rate": 9.76906785769939e-05,
      "loss": 2.1842,
      "step": 340
    },
    {
      "epoch": 0.11231421631191074,
      "grad_norm": 2.846548080444336,
      "learning_rate": 9.767484957469739e-05,
      "loss": 2.1864,
      "step": 341
    },
    {
      "epoch": 0.11264358351517148,
      "grad_norm": 2.9507384300231934,
      "learning_rate": 9.765896779968685e-05,
      "loss": 2.3369,
      "step": 342
    },
    {
      "epoch": 0.11297295071843222,
      "grad_norm": 3.360044002532959,
      "learning_rate": 9.764303326954226e-05,
      "loss": 2.2568,
      "step": 343
    },
    {
      "epoch": 0.11330231792169294,
      "grad_norm": 2.5054543018341064,
      "learning_rate": 9.762704600190207e-05,
      "loss": 1.8953,
      "step": 344
    },
    {
      "epoch": 0.11363168512495368,
      "grad_norm": 3.037480592727661,
      "learning_rate": 9.761100601446304e-05,
      "loss": 2.4476,
      "step": 345
    },
    {
      "epoch": 0.11396105232821442,
      "grad_norm": 3.0009777545928955,
      "learning_rate": 9.759491332498032e-05,
      "loss": 2.3728,
      "step": 346
    },
    {
      "epoch": 0.11429041953147516,
      "grad_norm": 2.856393814086914,
      "learning_rate": 9.757876795126739e-05,
      "loss": 1.9786,
      "step": 347
    },
    {
      "epoch": 0.11461978673473588,
      "grad_norm": 3.3026115894317627,
      "learning_rate": 9.756256991119603e-05,
      "loss": 2.6015,
      "step": 348
    },
    {
      "epoch": 0.11494915393799662,
      "grad_norm": 3.5809285640716553,
      "learning_rate": 9.754631922269636e-05,
      "loss": 2.307,
      "step": 349
    },
    {
      "epoch": 0.11527852114125736,
      "grad_norm": 3.11942195892334,
      "learning_rate": 9.753001590375674e-05,
      "loss": 2.0157,
      "step": 350
    },
    {
      "epoch": 0.1156078883445181,
      "grad_norm": 2.313547372817993,
      "learning_rate": 9.75136599724238e-05,
      "loss": 2.4783,
      "step": 351
    },
    {
      "epoch": 0.11593725554777883,
      "grad_norm": 2.4683542251586914,
      "learning_rate": 9.74972514468024e-05,
      "loss": 2.4294,
      "step": 352
    },
    {
      "epoch": 0.11626662275103956,
      "grad_norm": 2.895655632019043,
      "learning_rate": 9.748079034505565e-05,
      "loss": 2.6217,
      "step": 353
    },
    {
      "epoch": 0.1165959899543003,
      "grad_norm": 2.646327495574951,
      "learning_rate": 9.746427668540481e-05,
      "loss": 2.5583,
      "step": 354
    },
    {
      "epoch": 0.11692535715756104,
      "grad_norm": 2.4664547443389893,
      "learning_rate": 9.744771048612935e-05,
      "loss": 2.4272,
      "step": 355
    },
    {
      "epoch": 0.11725472436082177,
      "grad_norm": 2.5208280086517334,
      "learning_rate": 9.743109176556689e-05,
      "loss": 2.7008,
      "step": 356
    },
    {
      "epoch": 0.1175840915640825,
      "grad_norm": 2.5895395278930664,
      "learning_rate": 9.741442054211319e-05,
      "loss": 2.4581,
      "step": 357
    },
    {
      "epoch": 0.11791345876734324,
      "grad_norm": 2.3770651817321777,
      "learning_rate": 9.739769683422214e-05,
      "loss": 2.2697,
      "step": 358
    },
    {
      "epoch": 0.11824282597060398,
      "grad_norm": 2.263289451599121,
      "learning_rate": 9.738092066040568e-05,
      "loss": 2.1457,
      "step": 359
    },
    {
      "epoch": 0.1185721931738647,
      "grad_norm": 3.5201501846313477,
      "learning_rate": 9.736409203923388e-05,
      "loss": 2.4604,
      "step": 360
    },
    {
      "epoch": 0.11890156037712545,
      "grad_norm": 2.3830199241638184,
      "learning_rate": 9.734721098933484e-05,
      "loss": 2.0263,
      "step": 361
    },
    {
      "epoch": 0.11923092758038618,
      "grad_norm": 2.431260347366333,
      "learning_rate": 9.73302775293947e-05,
      "loss": 2.2848,
      "step": 362
    },
    {
      "epoch": 0.11956029478364692,
      "grad_norm": 2.4053165912628174,
      "learning_rate": 9.73132916781576e-05,
      "loss": 2.0885,
      "step": 363
    },
    {
      "epoch": 0.11988966198690765,
      "grad_norm": 2.584679126739502,
      "learning_rate": 9.72962534544257e-05,
      "loss": 2.2592,
      "step": 364
    },
    {
      "epoch": 0.12021902919016839,
      "grad_norm": 2.6707746982574463,
      "learning_rate": 9.727916287705912e-05,
      "loss": 2.135,
      "step": 365
    },
    {
      "epoch": 0.12054839639342912,
      "grad_norm": 2.802182197570801,
      "learning_rate": 9.72620199649759e-05,
      "loss": 2.6641,
      "step": 366
    },
    {
      "epoch": 0.12087776359668986,
      "grad_norm": 3.923982620239258,
      "learning_rate": 9.724482473715207e-05,
      "loss": 2.3298,
      "step": 367
    },
    {
      "epoch": 0.12120713079995059,
      "grad_norm": 2.7437329292297363,
      "learning_rate": 9.722757721262154e-05,
      "loss": 2.2673,
      "step": 368
    },
    {
      "epoch": 0.12153649800321133,
      "grad_norm": 2.640639543533325,
      "learning_rate": 9.72102774104761e-05,
      "loss": 2.2149,
      "step": 369
    },
    {
      "epoch": 0.12186586520647207,
      "grad_norm": 3.37756609916687,
      "learning_rate": 9.719292534986543e-05,
      "loss": 2.0474,
      "step": 370
    },
    {
      "epoch": 0.1221952324097328,
      "grad_norm": 2.524691581726074,
      "learning_rate": 9.717552104999703e-05,
      "loss": 2.0271,
      "step": 371
    },
    {
      "epoch": 0.12252459961299354,
      "grad_norm": 2.769646406173706,
      "learning_rate": 9.715806453013625e-05,
      "loss": 2.0452,
      "step": 372
    },
    {
      "epoch": 0.12285396681625427,
      "grad_norm": 2.9961395263671875,
      "learning_rate": 9.714055580960622e-05,
      "loss": 1.9754,
      "step": 373
    },
    {
      "epoch": 0.123183334019515,
      "grad_norm": 3.112914800643921,
      "learning_rate": 9.712299490778786e-05,
      "loss": 2.1425,
      "step": 374
    },
    {
      "epoch": 0.12351270122277574,
      "grad_norm": 3.4256157875061035,
      "learning_rate": 9.710538184411991e-05,
      "loss": 1.9117,
      "step": 375
    },
    {
      "epoch": 0.12384206842603648,
      "grad_norm": 1.9335092306137085,
      "learning_rate": 9.708771663809872e-05,
      "loss": 2.4769,
      "step": 376
    },
    {
      "epoch": 0.12417143562929721,
      "grad_norm": 2.1666107177734375,
      "learning_rate": 9.706999930927848e-05,
      "loss": 2.2265,
      "step": 377
    },
    {
      "epoch": 0.12450080283255795,
      "grad_norm": 2.2735953330993652,
      "learning_rate": 9.7052229877271e-05,
      "loss": 2.4031,
      "step": 378
    },
    {
      "epoch": 0.12483017003581869,
      "grad_norm": 2.531383514404297,
      "learning_rate": 9.703440836174583e-05,
      "loss": 2.4251,
      "step": 379
    },
    {
      "epoch": 0.1251595372390794,
      "grad_norm": 2.608523368835449,
      "learning_rate": 9.701653478243013e-05,
      "loss": 2.4985,
      "step": 380
    },
    {
      "epoch": 0.12548890444234015,
      "grad_norm": 2.4094133377075195,
      "learning_rate": 9.699860915910868e-05,
      "loss": 1.9842,
      "step": 381
    },
    {
      "epoch": 0.1258182716456009,
      "grad_norm": 2.278822183609009,
      "learning_rate": 9.698063151162389e-05,
      "loss": 2.0549,
      "step": 382
    },
    {
      "epoch": 0.12614763884886163,
      "grad_norm": 2.65533709526062,
      "learning_rate": 9.696260185987576e-05,
      "loss": 2.4869,
      "step": 383
    },
    {
      "epoch": 0.12647700605212236,
      "grad_norm": 2.8089699745178223,
      "learning_rate": 9.694452022382186e-05,
      "loss": 2.3468,
      "step": 384
    },
    {
      "epoch": 0.1268063732553831,
      "grad_norm": 2.4362573623657227,
      "learning_rate": 9.692638662347728e-05,
      "loss": 2.5076,
      "step": 385
    },
    {
      "epoch": 0.12713574045864384,
      "grad_norm": 3.138957977294922,
      "learning_rate": 9.690820107891466e-05,
      "loss": 2.3684,
      "step": 386
    },
    {
      "epoch": 0.12746510766190455,
      "grad_norm": 3.3768868446350098,
      "learning_rate": 9.68899636102641e-05,
      "loss": 2.7286,
      "step": 387
    },
    {
      "epoch": 0.1277944748651653,
      "grad_norm": 2.5961859226226807,
      "learning_rate": 9.68716742377132e-05,
      "loss": 2.1429,
      "step": 388
    },
    {
      "epoch": 0.12812384206842603,
      "grad_norm": 2.7435123920440674,
      "learning_rate": 9.685333298150702e-05,
      "loss": 2.2378,
      "step": 389
    },
    {
      "epoch": 0.12845320927168677,
      "grad_norm": 2.207853317260742,
      "learning_rate": 9.683493986194808e-05,
      "loss": 2.0057,
      "step": 390
    },
    {
      "epoch": 0.1287825764749475,
      "grad_norm": 3.450223922729492,
      "learning_rate": 9.681649489939619e-05,
      "loss": 2.5243,
      "step": 391
    },
    {
      "epoch": 0.12911194367820825,
      "grad_norm": 2.479057788848877,
      "learning_rate": 9.67979981142687e-05,
      "loss": 1.9977,
      "step": 392
    },
    {
      "epoch": 0.12944131088146899,
      "grad_norm": 2.4462480545043945,
      "learning_rate": 9.677944952704023e-05,
      "loss": 1.9085,
      "step": 393
    },
    {
      "epoch": 0.12977067808472972,
      "grad_norm": 3.183197021484375,
      "learning_rate": 9.676084915824276e-05,
      "loss": 2.6043,
      "step": 394
    },
    {
      "epoch": 0.13010004528799043,
      "grad_norm": 2.802366256713867,
      "learning_rate": 9.674219702846561e-05,
      "loss": 2.3849,
      "step": 395
    },
    {
      "epoch": 0.13042941249125117,
      "grad_norm": 3.248178482055664,
      "learning_rate": 9.672349315835535e-05,
      "loss": 2.3824,
      "step": 396
    },
    {
      "epoch": 0.1307587796945119,
      "grad_norm": 2.677258253097534,
      "learning_rate": 9.670473756861588e-05,
      "loss": 1.8382,
      "step": 397
    },
    {
      "epoch": 0.13108814689777265,
      "grad_norm": 2.8905255794525146,
      "learning_rate": 9.668593028000831e-05,
      "loss": 1.9997,
      "step": 398
    },
    {
      "epoch": 0.1314175141010334,
      "grad_norm": 2.8859400749206543,
      "learning_rate": 9.6667071313351e-05,
      "loss": 1.904,
      "step": 399
    },
    {
      "epoch": 0.13174688130429413,
      "grad_norm": 3.3241395950317383,
      "learning_rate": 9.664816068951947e-05,
      "loss": 1.7025,
      "step": 400
    },
    {
      "epoch": 0.13207624850755487,
      "grad_norm": 2.0233919620513916,
      "learning_rate": 9.662919842944651e-05,
      "loss": 2.3296,
      "step": 401
    },
    {
      "epoch": 0.1324056157108156,
      "grad_norm": 2.4507946968078613,
      "learning_rate": 9.661018455412197e-05,
      "loss": 2.431,
      "step": 402
    },
    {
      "epoch": 0.13273498291407634,
      "grad_norm": 2.341461658477783,
      "learning_rate": 9.659111908459288e-05,
      "loss": 2.5369,
      "step": 403
    },
    {
      "epoch": 0.13306435011733705,
      "grad_norm": 2.3247382640838623,
      "learning_rate": 9.657200204196337e-05,
      "loss": 2.3336,
      "step": 404
    },
    {
      "epoch": 0.1333937173205978,
      "grad_norm": 2.6917145252227783,
      "learning_rate": 9.65528334473947e-05,
      "loss": 2.5245,
      "step": 405
    },
    {
      "epoch": 0.13372308452385853,
      "grad_norm": 2.5759096145629883,
      "learning_rate": 9.653361332210513e-05,
      "loss": 2.5367,
      "step": 406
    },
    {
      "epoch": 0.13405245172711927,
      "grad_norm": 3.123220682144165,
      "learning_rate": 9.651434168737e-05,
      "loss": 2.3026,
      "step": 407
    },
    {
      "epoch": 0.13438181893038,
      "grad_norm": 1.9661033153533936,
      "learning_rate": 9.649501856452165e-05,
      "loss": 1.9149,
      "step": 408
    },
    {
      "epoch": 0.13471118613364075,
      "grad_norm": 2.2201578617095947,
      "learning_rate": 9.647564397494944e-05,
      "loss": 2.4501,
      "step": 409
    },
    {
      "epoch": 0.1350405533369015,
      "grad_norm": 2.485238790512085,
      "learning_rate": 9.645621794009967e-05,
      "loss": 2.3879,
      "step": 410
    },
    {
      "epoch": 0.13536992054016223,
      "grad_norm": 2.429931163787842,
      "learning_rate": 9.643674048147558e-05,
      "loss": 2.3842,
      "step": 411
    },
    {
      "epoch": 0.13569928774342294,
      "grad_norm": 2.674360513687134,
      "learning_rate": 9.641721162063739e-05,
      "loss": 2.2783,
      "step": 412
    },
    {
      "epoch": 0.13602865494668367,
      "grad_norm": 2.56316876411438,
      "learning_rate": 9.639763137920214e-05,
      "loss": 2.2338,
      "step": 413
    },
    {
      "epoch": 0.1363580221499444,
      "grad_norm": 2.6116116046905518,
      "learning_rate": 9.637799977884381e-05,
      "loss": 2.2164,
      "step": 414
    },
    {
      "epoch": 0.13668738935320515,
      "grad_norm": 2.379546880722046,
      "learning_rate": 9.635831684129318e-05,
      "loss": 1.8579,
      "step": 415
    },
    {
      "epoch": 0.1370167565564659,
      "grad_norm": 2.250623941421509,
      "learning_rate": 9.63385825883379e-05,
      "loss": 1.9759,
      "step": 416
    },
    {
      "epoch": 0.13734612375972663,
      "grad_norm": 2.5346574783325195,
      "learning_rate": 9.63187970418224e-05,
      "loss": 2.1965,
      "step": 417
    },
    {
      "epoch": 0.13767549096298737,
      "grad_norm": 2.7046220302581787,
      "learning_rate": 9.62989602236479e-05,
      "loss": 2.207,
      "step": 418
    },
    {
      "epoch": 0.1380048581662481,
      "grad_norm": 2.8171615600585938,
      "learning_rate": 9.627907215577236e-05,
      "loss": 2.1648,
      "step": 419
    },
    {
      "epoch": 0.13833422536950882,
      "grad_norm": 2.6681857109069824,
      "learning_rate": 9.625913286021046e-05,
      "loss": 2.2677,
      "step": 420
    },
    {
      "epoch": 0.13866359257276956,
      "grad_norm": 2.8763694763183594,
      "learning_rate": 9.623914235903362e-05,
      "loss": 1.9661,
      "step": 421
    },
    {
      "epoch": 0.1389929597760303,
      "grad_norm": 2.8630616664886475,
      "learning_rate": 9.621910067436992e-05,
      "loss": 2.1464,
      "step": 422
    },
    {
      "epoch": 0.13932232697929103,
      "grad_norm": 3.0975148677825928,
      "learning_rate": 9.61990078284041e-05,
      "loss": 2.1959,
      "step": 423
    },
    {
      "epoch": 0.13965169418255177,
      "grad_norm": 3.8912901878356934,
      "learning_rate": 9.617886384337751e-05,
      "loss": 2.1552,
      "step": 424
    },
    {
      "epoch": 0.1399810613858125,
      "grad_norm": 3.1637001037597656,
      "learning_rate": 9.615866874158816e-05,
      "loss": 1.8698,
      "step": 425
    },
    {
      "epoch": 0.14031042858907325,
      "grad_norm": 2.0410444736480713,
      "learning_rate": 9.613842254539058e-05,
      "loss": 2.6313,
      "step": 426
    },
    {
      "epoch": 0.140639795792334,
      "grad_norm": 2.081902503967285,
      "learning_rate": 9.611812527719593e-05,
      "loss": 2.444,
      "step": 427
    },
    {
      "epoch": 0.14096916299559473,
      "grad_norm": 2.232922077178955,
      "learning_rate": 9.609777695947182e-05,
      "loss": 2.1034,
      "step": 428
    },
    {
      "epoch": 0.14129853019885544,
      "grad_norm": 2.2757458686828613,
      "learning_rate": 9.607737761474242e-05,
      "loss": 2.5483,
      "step": 429
    },
    {
      "epoch": 0.14162789740211618,
      "grad_norm": 2.3721635341644287,
      "learning_rate": 9.60569272655884e-05,
      "loss": 2.6453,
      "step": 430
    },
    {
      "epoch": 0.14195726460537691,
      "grad_norm": 2.366621971130371,
      "learning_rate": 9.603642593464683e-05,
      "loss": 2.3196,
      "step": 431
    },
    {
      "epoch": 0.14228663180863765,
      "grad_norm": 2.4534189701080322,
      "learning_rate": 9.601587364461127e-05,
      "loss": 2.4592,
      "step": 432
    },
    {
      "epoch": 0.1426159990118984,
      "grad_norm": 2.350576639175415,
      "learning_rate": 9.599527041823164e-05,
      "loss": 2.0809,
      "step": 433
    },
    {
      "epoch": 0.14294536621515913,
      "grad_norm": 2.2983639240264893,
      "learning_rate": 9.59746162783143e-05,
      "loss": 2.1881,
      "step": 434
    },
    {
      "epoch": 0.14327473341841987,
      "grad_norm": 2.5199313163757324,
      "learning_rate": 9.595391124772189e-05,
      "loss": 2.2941,
      "step": 435
    },
    {
      "epoch": 0.1436041006216806,
      "grad_norm": 2.4015860557556152,
      "learning_rate": 9.593315534937345e-05,
      "loss": 2.2748,
      "step": 436
    },
    {
      "epoch": 0.14393346782494132,
      "grad_norm": 2.675955057144165,
      "learning_rate": 9.591234860624431e-05,
      "loss": 2.1505,
      "step": 437
    },
    {
      "epoch": 0.14426283502820206,
      "grad_norm": 2.4210708141326904,
      "learning_rate": 9.589149104136605e-05,
      "loss": 2.2871,
      "step": 438
    },
    {
      "epoch": 0.1445922022314628,
      "grad_norm": 2.417851686477661,
      "learning_rate": 9.587058267782656e-05,
      "loss": 2.1308,
      "step": 439
    },
    {
      "epoch": 0.14492156943472354,
      "grad_norm": 2.456392526626587,
      "learning_rate": 9.584962353876992e-05,
      "loss": 1.8146,
      "step": 440
    },
    {
      "epoch": 0.14525093663798427,
      "grad_norm": 2.4803600311279297,
      "learning_rate": 9.582861364739642e-05,
      "loss": 2.2325,
      "step": 441
    },
    {
      "epoch": 0.145580303841245,
      "grad_norm": 2.4525747299194336,
      "learning_rate": 9.580755302696256e-05,
      "loss": 2.0824,
      "step": 442
    },
    {
      "epoch": 0.14590967104450575,
      "grad_norm": 3.1618423461914062,
      "learning_rate": 9.578644170078093e-05,
      "loss": 2.3309,
      "step": 443
    },
    {
      "epoch": 0.1462390382477665,
      "grad_norm": 3.0864148139953613,
      "learning_rate": 9.576527969222031e-05,
      "loss": 2.2085,
      "step": 444
    },
    {
      "epoch": 0.1465684054510272,
      "grad_norm": 2.7343194484710693,
      "learning_rate": 9.574406702470558e-05,
      "loss": 2.2462,
      "step": 445
    },
    {
      "epoch": 0.14689777265428794,
      "grad_norm": 2.57436466217041,
      "learning_rate": 9.572280372171763e-05,
      "loss": 2.0448,
      "step": 446
    },
    {
      "epoch": 0.14722713985754868,
      "grad_norm": 2.608705759048462,
      "learning_rate": 9.570148980679347e-05,
      "loss": 1.9546,
      "step": 447
    },
    {
      "epoch": 0.14755650706080942,
      "grad_norm": 2.5100154876708984,
      "learning_rate": 9.56801253035261e-05,
      "loss": 1.6897,
      "step": 448
    },
    {
      "epoch": 0.14788587426407016,
      "grad_norm": 3.806117296218872,
      "learning_rate": 9.565871023556455e-05,
      "loss": 1.8645,
      "step": 449
    },
    {
      "epoch": 0.1482152414673309,
      "grad_norm": 3.5076465606689453,
      "learning_rate": 9.563724462661376e-05,
      "loss": 2.2529,
      "step": 450
    },
    {
      "epoch": 0.14854460867059163,
      "grad_norm": 2.111990451812744,
      "learning_rate": 9.561572850043467e-05,
      "loss": 2.5226,
      "step": 451
    },
    {
      "epoch": 0.14887397587385237,
      "grad_norm": 2.713508367538452,
      "learning_rate": 9.559416188084416e-05,
      "loss": 2.548,
      "step": 452
    },
    {
      "epoch": 0.1492033430771131,
      "grad_norm": 2.7908146381378174,
      "learning_rate": 9.557254479171489e-05,
      "loss": 2.8369,
      "step": 453
    },
    {
      "epoch": 0.14953271028037382,
      "grad_norm": 2.3136017322540283,
      "learning_rate": 9.555087725697554e-05,
      "loss": 2.3652,
      "step": 454
    },
    {
      "epoch": 0.14986207748363456,
      "grad_norm": 2.1749308109283447,
      "learning_rate": 9.552915930061048e-05,
      "loss": 2.2855,
      "step": 455
    },
    {
      "epoch": 0.1501914446868953,
      "grad_norm": 2.302049398422241,
      "learning_rate": 9.550739094666002e-05,
      "loss": 2.3987,
      "step": 456
    },
    {
      "epoch": 0.15052081189015604,
      "grad_norm": 2.3191065788269043,
      "learning_rate": 9.548557221922017e-05,
      "loss": 2.4523,
      "step": 457
    },
    {
      "epoch": 0.15085017909341678,
      "grad_norm": 2.958138942718506,
      "learning_rate": 9.546370314244273e-05,
      "loss": 2.2964,
      "step": 458
    },
    {
      "epoch": 0.1511795462966775,
      "grad_norm": 2.8905577659606934,
      "learning_rate": 9.544178374053524e-05,
      "loss": 2.6665,
      "step": 459
    },
    {
      "epoch": 0.15150891349993825,
      "grad_norm": 2.6026418209075928,
      "learning_rate": 9.541981403776095e-05,
      "loss": 2.5692,
      "step": 460
    },
    {
      "epoch": 0.151838280703199,
      "grad_norm": 2.7903120517730713,
      "learning_rate": 9.539779405843876e-05,
      "loss": 2.439,
      "step": 461
    },
    {
      "epoch": 0.1521676479064597,
      "grad_norm": 3.0123376846313477,
      "learning_rate": 9.537572382694328e-05,
      "loss": 2.592,
      "step": 462
    },
    {
      "epoch": 0.15249701510972044,
      "grad_norm": 2.3931968212127686,
      "learning_rate": 9.535360336770467e-05,
      "loss": 2.2706,
      "step": 463
    },
    {
      "epoch": 0.15282638231298118,
      "grad_norm": 2.501070022583008,
      "learning_rate": 9.533143270520873e-05,
      "loss": 1.8545,
      "step": 464
    },
    {
      "epoch": 0.15315574951624192,
      "grad_norm": 2.836297035217285,
      "learning_rate": 9.530921186399684e-05,
      "loss": 2.3613,
      "step": 465
    },
    {
      "epoch": 0.15348511671950266,
      "grad_norm": 2.518571138381958,
      "learning_rate": 9.528694086866592e-05,
      "loss": 2.1096,
      "step": 466
    },
    {
      "epoch": 0.1538144839227634,
      "grad_norm": 2.5794365406036377,
      "learning_rate": 9.526461974386838e-05,
      "loss": 2.0714,
      "step": 467
    },
    {
      "epoch": 0.15414385112602413,
      "grad_norm": 2.955522060394287,
      "learning_rate": 9.524224851431214e-05,
      "loss": 2.1713,
      "step": 468
    },
    {
      "epoch": 0.15447321832928487,
      "grad_norm": 3.465235948562622,
      "learning_rate": 9.521982720476062e-05,
      "loss": 2.2217,
      "step": 469
    },
    {
      "epoch": 0.15480258553254558,
      "grad_norm": 2.497987985610962,
      "learning_rate": 9.519735584003257e-05,
      "loss": 1.9994,
      "step": 470
    },
    {
      "epoch": 0.15513195273580632,
      "grad_norm": 2.911043643951416,
      "learning_rate": 9.517483444500228e-05,
      "loss": 1.9883,
      "step": 471
    },
    {
      "epoch": 0.15546131993906706,
      "grad_norm": 2.7313320636749268,
      "learning_rate": 9.51522630445993e-05,
      "loss": 2.1227,
      "step": 472
    },
    {
      "epoch": 0.1557906871423278,
      "grad_norm": 3.4482412338256836,
      "learning_rate": 9.512964166380864e-05,
      "loss": 2.2148,
      "step": 473
    },
    {
      "epoch": 0.15612005434558854,
      "grad_norm": 2.664477825164795,
      "learning_rate": 9.510697032767053e-05,
      "loss": 1.7443,
      "step": 474
    },
    {
      "epoch": 0.15644942154884928,
      "grad_norm": 2.8564321994781494,
      "learning_rate": 9.508424906128058e-05,
      "loss": 1.9044,
      "step": 475
    },
    {
      "epoch": 0.15677878875211002,
      "grad_norm": 1.97623872756958,
      "learning_rate": 9.506147788978965e-05,
      "loss": 2.3548,
      "step": 476
    },
    {
      "epoch": 0.15710815595537075,
      "grad_norm": 2.3602867126464844,
      "learning_rate": 9.503865683840378e-05,
      "loss": 2.5651,
      "step": 477
    },
    {
      "epoch": 0.1574375231586315,
      "grad_norm": 2.1931259632110596,
      "learning_rate": 9.501578593238432e-05,
      "loss": 2.1644,
      "step": 478
    },
    {
      "epoch": 0.1577668903618922,
      "grad_norm": 2.284613609313965,
      "learning_rate": 9.499286519704773e-05,
      "loss": 2.2849,
      "step": 479
    },
    {
      "epoch": 0.15809625756515294,
      "grad_norm": 2.498391628265381,
      "learning_rate": 9.49698946577657e-05,
      "loss": 2.5516,
      "step": 480
    },
    {
      "epoch": 0.15842562476841368,
      "grad_norm": 2.4198145866394043,
      "learning_rate": 9.494687433996493e-05,
      "loss": 2.194,
      "step": 481
    },
    {
      "epoch": 0.15875499197167442,
      "grad_norm": 2.404466390609741,
      "learning_rate": 9.492380426912737e-05,
      "loss": 2.3409,
      "step": 482
    },
    {
      "epoch": 0.15908435917493516,
      "grad_norm": 2.326627254486084,
      "learning_rate": 9.490068447078992e-05,
      "loss": 2.5327,
      "step": 483
    },
    {
      "epoch": 0.1594137263781959,
      "grad_norm": 2.2029290199279785,
      "learning_rate": 9.487751497054461e-05,
      "loss": 2.3595,
      "step": 484
    },
    {
      "epoch": 0.15974309358145664,
      "grad_norm": 2.798891544342041,
      "learning_rate": 9.485429579403843e-05,
      "loss": 2.5281,
      "step": 485
    },
    {
      "epoch": 0.16007246078471737,
      "grad_norm": 2.421884775161743,
      "learning_rate": 9.483102696697339e-05,
      "loss": 2.3762,
      "step": 486
    },
    {
      "epoch": 0.16040182798797809,
      "grad_norm": 2.618962049484253,
      "learning_rate": 9.480770851510644e-05,
      "loss": 2.5659,
      "step": 487
    },
    {
      "epoch": 0.16073119519123882,
      "grad_norm": 2.394176721572876,
      "learning_rate": 9.478434046424948e-05,
      "loss": 2.0389,
      "step": 488
    },
    {
      "epoch": 0.16106056239449956,
      "grad_norm": 2.6122007369995117,
      "learning_rate": 9.47609228402693e-05,
      "loss": 2.2712,
      "step": 489
    },
    {
      "epoch": 0.1613899295977603,
      "grad_norm": 3.2165706157684326,
      "learning_rate": 9.473745566908756e-05,
      "loss": 2.2492,
      "step": 490
    },
    {
      "epoch": 0.16171929680102104,
      "grad_norm": 2.527129650115967,
      "learning_rate": 9.471393897668078e-05,
      "loss": 2.4401,
      "step": 491
    },
    {
      "epoch": 0.16204866400428178,
      "grad_norm": 2.758704900741577,
      "learning_rate": 9.469037278908029e-05,
      "loss": 2.2011,
      "step": 492
    },
    {
      "epoch": 0.16237803120754252,
      "grad_norm": 2.8025074005126953,
      "learning_rate": 9.46667571323722e-05,
      "loss": 2.0435,
      "step": 493
    },
    {
      "epoch": 0.16270739841080326,
      "grad_norm": 2.798211097717285,
      "learning_rate": 9.464309203269739e-05,
      "loss": 2.2279,
      "step": 494
    },
    {
      "epoch": 0.16303676561406397,
      "grad_norm": 2.942800521850586,
      "learning_rate": 9.461937751625145e-05,
      "loss": 2.082,
      "step": 495
    },
    {
      "epoch": 0.1633661328173247,
      "grad_norm": 2.7005648612976074,
      "learning_rate": 9.459561360928472e-05,
      "loss": 2.2156,
      "step": 496
    },
    {
      "epoch": 0.16369550002058544,
      "grad_norm": 2.9613757133483887,
      "learning_rate": 9.457180033810216e-05,
      "loss": 2.3678,
      "step": 497
    },
    {
      "epoch": 0.16402486722384618,
      "grad_norm": 2.888354539871216,
      "learning_rate": 9.454793772906336e-05,
      "loss": 1.9226,
      "step": 498
    },
    {
      "epoch": 0.16435423442710692,
      "grad_norm": 3.021557331085205,
      "learning_rate": 9.452402580858261e-05,
      "loss": 2.1693,
      "step": 499
    },
    {
      "epoch": 0.16468360163036766,
      "grad_norm": 3.1927671432495117,
      "learning_rate": 9.45000646031287e-05,
      "loss": 1.6268,
      "step": 500
    },
    {
      "epoch": 0.1650129688336284,
      "grad_norm": 1.9104129076004028,
      "learning_rate": 9.447605413922499e-05,
      "loss": 2.6226,
      "step": 501
    },
    {
      "epoch": 0.16534233603688914,
      "grad_norm": 1.981658935546875,
      "learning_rate": 9.44519944434494e-05,
      "loss": 2.288,
      "step": 502
    },
    {
      "epoch": 0.16567170324014988,
      "grad_norm": 2.278552532196045,
      "learning_rate": 9.442788554243431e-05,
      "loss": 2.4093,
      "step": 503
    },
    {
      "epoch": 0.1660010704434106,
      "grad_norm": 2.9332423210144043,
      "learning_rate": 9.440372746286661e-05,
      "loss": 2.5302,
      "step": 504
    },
    {
      "epoch": 0.16633043764667133,
      "grad_norm": 2.4313466548919678,
      "learning_rate": 9.437952023148757e-05,
      "loss": 2.52,
      "step": 505
    },
    {
      "epoch": 0.16665980484993206,
      "grad_norm": 2.3095078468322754,
      "learning_rate": 9.43552638750929e-05,
      "loss": 2.424,
      "step": 506
    },
    {
      "epoch": 0.1669891720531928,
      "grad_norm": 2.3015639781951904,
      "learning_rate": 9.433095842053272e-05,
      "loss": 2.5717,
      "step": 507
    },
    {
      "epoch": 0.16731853925645354,
      "grad_norm": 2.5541129112243652,
      "learning_rate": 9.43066038947114e-05,
      "loss": 2.5831,
      "step": 508
    },
    {
      "epoch": 0.16764790645971428,
      "grad_norm": 2.421571969985962,
      "learning_rate": 9.428220032458776e-05,
      "loss": 2.4438,
      "step": 509
    },
    {
      "epoch": 0.16797727366297502,
      "grad_norm": 2.5124785900115967,
      "learning_rate": 9.425774773717479e-05,
      "loss": 2.2709,
      "step": 510
    },
    {
      "epoch": 0.16830664086623576,
      "grad_norm": 2.592477560043335,
      "learning_rate": 9.423324615953982e-05,
      "loss": 2.3089,
      "step": 511
    },
    {
      "epoch": 0.16863600806949647,
      "grad_norm": 2.2433090209960938,
      "learning_rate": 9.420869561880434e-05,
      "loss": 2.318,
      "step": 512
    },
    {
      "epoch": 0.1689653752727572,
      "grad_norm": 2.6666271686553955,
      "learning_rate": 9.418409614214412e-05,
      "loss": 2.3501,
      "step": 513
    },
    {
      "epoch": 0.16929474247601795,
      "grad_norm": 2.3065149784088135,
      "learning_rate": 9.415944775678902e-05,
      "loss": 2.0432,
      "step": 514
    },
    {
      "epoch": 0.16962410967927868,
      "grad_norm": 3.0598034858703613,
      "learning_rate": 9.41347504900231e-05,
      "loss": 2.6206,
      "step": 515
    },
    {
      "epoch": 0.16995347688253942,
      "grad_norm": 2.051896095275879,
      "learning_rate": 9.411000436918449e-05,
      "loss": 1.9563,
      "step": 516
    },
    {
      "epoch": 0.17028284408580016,
      "grad_norm": 3.1122515201568604,
      "learning_rate": 9.408520942166541e-05,
      "loss": 2.2816,
      "step": 517
    },
    {
      "epoch": 0.1706122112890609,
      "grad_norm": 2.379194974899292,
      "learning_rate": 9.406036567491213e-05,
      "loss": 2.4151,
      "step": 518
    },
    {
      "epoch": 0.17094157849232164,
      "grad_norm": 2.4957690238952637,
      "learning_rate": 9.403547315642493e-05,
      "loss": 1.8377,
      "step": 519
    },
    {
      "epoch": 0.17127094569558235,
      "grad_norm": 2.6822221279144287,
      "learning_rate": 9.401053189375809e-05,
      "loss": 2.0926,
      "step": 520
    },
    {
      "epoch": 0.1716003128988431,
      "grad_norm": 2.902961492538452,
      "learning_rate": 9.398554191451983e-05,
      "loss": 2.1816,
      "step": 521
    },
    {
      "epoch": 0.17192968010210383,
      "grad_norm": 3.225004196166992,
      "learning_rate": 9.396050324637228e-05,
      "loss": 2.5233,
      "step": 522
    },
    {
      "epoch": 0.17225904730536457,
      "grad_norm": 2.8753762245178223,
      "learning_rate": 9.393541591703156e-05,
      "loss": 2.3371,
      "step": 523
    },
    {
      "epoch": 0.1725884145086253,
      "grad_norm": 3.5578978061676025,
      "learning_rate": 9.39102799542675e-05,
      "loss": 2.3127,
      "step": 524
    },
    {
      "epoch": 0.17291778171188604,
      "grad_norm": 3.2395496368408203,
      "learning_rate": 9.388509538590391e-05,
      "loss": 2.0262,
      "step": 525
    },
    {
      "epoch": 0.17324714891514678,
      "grad_norm": 2.063525438308716,
      "learning_rate": 9.385986223981833e-05,
      "loss": 2.5577,
      "step": 526
    },
    {
      "epoch": 0.17357651611840752,
      "grad_norm": 2.227280616760254,
      "learning_rate": 9.383458054394206e-05,
      "loss": 2.5892,
      "step": 527
    },
    {
      "epoch": 0.17390588332166826,
      "grad_norm": 2.1608548164367676,
      "learning_rate": 9.380925032626015e-05,
      "loss": 2.5988,
      "step": 528
    },
    {
      "epoch": 0.17423525052492897,
      "grad_norm": 2.333763360977173,
      "learning_rate": 9.378387161481142e-05,
      "loss": 2.4371,
      "step": 529
    },
    {
      "epoch": 0.1745646177281897,
      "grad_norm": 2.1032464504241943,
      "learning_rate": 9.375844443768829e-05,
      "loss": 2.1269,
      "step": 530
    },
    {
      "epoch": 0.17489398493145045,
      "grad_norm": 2.7457637786865234,
      "learning_rate": 9.373296882303688e-05,
      "loss": 2.4893,
      "step": 531
    },
    {
      "epoch": 0.17522335213471119,
      "grad_norm": 2.414435625076294,
      "learning_rate": 9.37074447990569e-05,
      "loss": 2.4437,
      "step": 532
    },
    {
      "epoch": 0.17555271933797192,
      "grad_norm": 2.7596664428710938,
      "learning_rate": 9.368187239400166e-05,
      "loss": 2.3113,
      "step": 533
    },
    {
      "epoch": 0.17588208654123266,
      "grad_norm": 2.7227630615234375,
      "learning_rate": 9.3656251636178e-05,
      "loss": 2.2728,
      "step": 534
    },
    {
      "epoch": 0.1762114537444934,
      "grad_norm": 2.8053038120269775,
      "learning_rate": 9.363058255394632e-05,
      "loss": 2.3559,
      "step": 535
    },
    {
      "epoch": 0.17654082094775414,
      "grad_norm": 2.6018600463867188,
      "learning_rate": 9.360486517572049e-05,
      "loss": 2.2176,
      "step": 536
    },
    {
      "epoch": 0.17687018815101485,
      "grad_norm": 2.3456811904907227,
      "learning_rate": 9.357909952996784e-05,
      "loss": 2.1538,
      "step": 537
    },
    {
      "epoch": 0.1771995553542756,
      "grad_norm": 2.6372413635253906,
      "learning_rate": 9.355328564520914e-05,
      "loss": 2.1687,
      "step": 538
    },
    {
      "epoch": 0.17752892255753633,
      "grad_norm": 2.4839794635772705,
      "learning_rate": 9.352742355001853e-05,
      "loss": 2.2029,
      "step": 539
    },
    {
      "epoch": 0.17785828976079707,
      "grad_norm": 2.5911648273468018,
      "learning_rate": 9.350151327302356e-05,
      "loss": 2.0988,
      "step": 540
    },
    {
      "epoch": 0.1781876569640578,
      "grad_norm": 2.3182106018066406,
      "learning_rate": 9.347555484290507e-05,
      "loss": 1.9714,
      "step": 541
    },
    {
      "epoch": 0.17851702416731854,
      "grad_norm": 3.564013957977295,
      "learning_rate": 9.344954828839722e-05,
      "loss": 2.3427,
      "step": 542
    },
    {
      "epoch": 0.17884639137057928,
      "grad_norm": 2.816439628601074,
      "learning_rate": 9.342349363828748e-05,
      "loss": 2.198,
      "step": 543
    },
    {
      "epoch": 0.17917575857384002,
      "grad_norm": 2.91499924659729,
      "learning_rate": 9.339739092141647e-05,
      "loss": 2.2565,
      "step": 544
    },
    {
      "epoch": 0.17950512577710073,
      "grad_norm": 3.363369941711426,
      "learning_rate": 9.337124016667809e-05,
      "loss": 2.1877,
      "step": 545
    },
    {
      "epoch": 0.17983449298036147,
      "grad_norm": 3.335421085357666,
      "learning_rate": 9.334504140301938e-05,
      "loss": 2.2754,
      "step": 546
    },
    {
      "epoch": 0.1801638601836222,
      "grad_norm": 3.2244529724121094,
      "learning_rate": 9.331879465944056e-05,
      "loss": 2.0735,
      "step": 547
    },
    {
      "epoch": 0.18049322738688295,
      "grad_norm": 2.5784671306610107,
      "learning_rate": 9.32924999649949e-05,
      "loss": 2.0593,
      "step": 548
    },
    {
      "epoch": 0.1808225945901437,
      "grad_norm": 3.826298236846924,
      "learning_rate": 9.326615734878878e-05,
      "loss": 2.4232,
      "step": 549
    },
    {
      "epoch": 0.18115196179340443,
      "grad_norm": 3.9787750244140625,
      "learning_rate": 9.323976683998168e-05,
      "loss": 1.9951,
      "step": 550
    },
    {
      "epoch": 0.18148132899666516,
      "grad_norm": 2.1491682529449463,
      "learning_rate": 9.321332846778599e-05,
      "loss": 2.497,
      "step": 551
    },
    {
      "epoch": 0.1818106961999259,
      "grad_norm": 2.5805752277374268,
      "learning_rate": 9.318684226146714e-05,
      "loss": 2.343,
      "step": 552
    },
    {
      "epoch": 0.18214006340318661,
      "grad_norm": 2.671565055847168,
      "learning_rate": 9.316030825034354e-05,
      "loss": 2.314,
      "step": 553
    },
    {
      "epoch": 0.18246943060644735,
      "grad_norm": 2.459749221801758,
      "learning_rate": 9.313372646378643e-05,
      "loss": 2.3503,
      "step": 554
    },
    {
      "epoch": 0.1827987978097081,
      "grad_norm": 2.3150362968444824,
      "learning_rate": 9.310709693122002e-05,
      "loss": 1.9586,
      "step": 555
    },
    {
      "epoch": 0.18312816501296883,
      "grad_norm": 2.813565254211426,
      "learning_rate": 9.308041968212131e-05,
      "loss": 2.467,
      "step": 556
    },
    {
      "epoch": 0.18345753221622957,
      "grad_norm": 2.5931906700134277,
      "learning_rate": 9.305369474602015e-05,
      "loss": 2.245,
      "step": 557
    },
    {
      "epoch": 0.1837868994194903,
      "grad_norm": 2.7060670852661133,
      "learning_rate": 9.302692215249918e-05,
      "loss": 2.7003,
      "step": 558
    },
    {
      "epoch": 0.18411626662275105,
      "grad_norm": 2.543288230895996,
      "learning_rate": 9.300010193119376e-05,
      "loss": 2.2948,
      "step": 559
    },
    {
      "epoch": 0.18444563382601178,
      "grad_norm": 2.492933750152588,
      "learning_rate": 9.297323411179202e-05,
      "loss": 2.2784,
      "step": 560
    },
    {
      "epoch": 0.18477500102927252,
      "grad_norm": 2.2065117359161377,
      "learning_rate": 9.294631872403474e-05,
      "loss": 2.2058,
      "step": 561
    },
    {
      "epoch": 0.18510436823253323,
      "grad_norm": 3.046983003616333,
      "learning_rate": 9.291935579771536e-05,
      "loss": 2.3053,
      "step": 562
    },
    {
      "epoch": 0.18543373543579397,
      "grad_norm": 2.53873610496521,
      "learning_rate": 9.289234536267996e-05,
      "loss": 2.4844,
      "step": 563
    },
    {
      "epoch": 0.1857631026390547,
      "grad_norm": 2.3958847522735596,
      "learning_rate": 9.286528744882719e-05,
      "loss": 2.0553,
      "step": 564
    },
    {
      "epoch": 0.18609246984231545,
      "grad_norm": 3.1099889278411865,
      "learning_rate": 9.283818208610826e-05,
      "loss": 2.7911,
      "step": 565
    },
    {
      "epoch": 0.1864218370455762,
      "grad_norm": 2.7768149375915527,
      "learning_rate": 9.28110293045269e-05,
      "loss": 2.3027,
      "step": 566
    },
    {
      "epoch": 0.18675120424883693,
      "grad_norm": 2.551649570465088,
      "learning_rate": 9.278382913413935e-05,
      "loss": 1.9763,
      "step": 567
    },
    {
      "epoch": 0.18708057145209767,
      "grad_norm": 2.6487247943878174,
      "learning_rate": 9.27565816050543e-05,
      "loss": 2.2145,
      "step": 568
    },
    {
      "epoch": 0.1874099386553584,
      "grad_norm": 3.50011944770813,
      "learning_rate": 9.272928674743282e-05,
      "loss": 2.0824,
      "step": 569
    },
    {
      "epoch": 0.18773930585861912,
      "grad_norm": 2.5027475357055664,
      "learning_rate": 9.270194459148841e-05,
      "loss": 2.0633,
      "step": 570
    },
    {
      "epoch": 0.18806867306187985,
      "grad_norm": 2.7622134685516357,
      "learning_rate": 9.267455516748693e-05,
      "loss": 2.4109,
      "step": 571
    },
    {
      "epoch": 0.1883980402651406,
      "grad_norm": 2.5253777503967285,
      "learning_rate": 9.264711850574657e-05,
      "loss": 1.8391,
      "step": 572
    },
    {
      "epoch": 0.18872740746840133,
      "grad_norm": 3.9980525970458984,
      "learning_rate": 9.261963463663775e-05,
      "loss": 2.4288,
      "step": 573
    },
    {
      "epoch": 0.18905677467166207,
      "grad_norm": 3.201934814453125,
      "learning_rate": 9.25921035905832e-05,
      "loss": 2.1041,
      "step": 574
    },
    {
      "epoch": 0.1893861418749228,
      "grad_norm": 3.45932674407959,
      "learning_rate": 9.256452539805787e-05,
      "loss": 1.9162,
      "step": 575
    },
    {
      "epoch": 0.18971550907818355,
      "grad_norm": 2.3997511863708496,
      "learning_rate": 9.253690008958886e-05,
      "loss": 2.7296,
      "step": 576
    },
    {
      "epoch": 0.19004487628144429,
      "grad_norm": 2.100877523422241,
      "learning_rate": 9.250922769575548e-05,
      "loss": 2.2391,
      "step": 577
    },
    {
      "epoch": 0.190374243484705,
      "grad_norm": 2.0351099967956543,
      "learning_rate": 9.248150824718911e-05,
      "loss": 2.3191,
      "step": 578
    },
    {
      "epoch": 0.19070361068796574,
      "grad_norm": 2.368957757949829,
      "learning_rate": 9.245374177457323e-05,
      "loss": 2.2142,
      "step": 579
    },
    {
      "epoch": 0.19103297789122647,
      "grad_norm": 2.49444580078125,
      "learning_rate": 9.242592830864339e-05,
      "loss": 2.3158,
      "step": 580
    },
    {
      "epoch": 0.1913623450944872,
      "grad_norm": 2.4949872493743896,
      "learning_rate": 9.239806788018714e-05,
      "loss": 2.3183,
      "step": 581
    },
    {
      "epoch": 0.19169171229774795,
      "grad_norm": 2.6433401107788086,
      "learning_rate": 9.2370160520044e-05,
      "loss": 2.4043,
      "step": 582
    },
    {
      "epoch": 0.1920210795010087,
      "grad_norm": 2.3650400638580322,
      "learning_rate": 9.23422062591055e-05,
      "loss": 2.2305,
      "step": 583
    },
    {
      "epoch": 0.19235044670426943,
      "grad_norm": 2.623833179473877,
      "learning_rate": 9.231420512831501e-05,
      "loss": 2.2516,
      "step": 584
    },
    {
      "epoch": 0.19267981390753017,
      "grad_norm": 2.3704848289489746,
      "learning_rate": 9.228615715866785e-05,
      "loss": 2.1916,
      "step": 585
    },
    {
      "epoch": 0.1930091811107909,
      "grad_norm": 2.6421825885772705,
      "learning_rate": 9.225806238121113e-05,
      "loss": 2.1705,
      "step": 586
    },
    {
      "epoch": 0.19333854831405162,
      "grad_norm": 2.507702589035034,
      "learning_rate": 9.222992082704381e-05,
      "loss": 2.2798,
      "step": 587
    },
    {
      "epoch": 0.19366791551731236,
      "grad_norm": 2.977654457092285,
      "learning_rate": 9.22017325273166e-05,
      "loss": 2.3934,
      "step": 588
    },
    {
      "epoch": 0.1939972827205731,
      "grad_norm": 2.6255760192871094,
      "learning_rate": 9.217349751323199e-05,
      "loss": 2.168,
      "step": 589
    },
    {
      "epoch": 0.19432664992383383,
      "grad_norm": 2.767040252685547,
      "learning_rate": 9.214521581604415e-05,
      "loss": 2.3227,
      "step": 590
    },
    {
      "epoch": 0.19465601712709457,
      "grad_norm": 2.74783992767334,
      "learning_rate": 9.211688746705894e-05,
      "loss": 2.5598,
      "step": 591
    },
    {
      "epoch": 0.1949853843303553,
      "grad_norm": 3.419125556945801,
      "learning_rate": 9.208851249763385e-05,
      "loss": 2.1441,
      "step": 592
    },
    {
      "epoch": 0.19531475153361605,
      "grad_norm": 2.531890392303467,
      "learning_rate": 9.206009093917798e-05,
      "loss": 2.2535,
      "step": 593
    },
    {
      "epoch": 0.1956441187368768,
      "grad_norm": 2.928637742996216,
      "learning_rate": 9.203162282315201e-05,
      "loss": 1.9981,
      "step": 594
    },
    {
      "epoch": 0.1959734859401375,
      "grad_norm": 2.740192174911499,
      "learning_rate": 9.200310818106813e-05,
      "loss": 2.1965,
      "step": 595
    },
    {
      "epoch": 0.19630285314339824,
      "grad_norm": 2.5833513736724854,
      "learning_rate": 9.197454704449007e-05,
      "loss": 1.9684,
      "step": 596
    },
    {
      "epoch": 0.19663222034665898,
      "grad_norm": 2.816075563430786,
      "learning_rate": 9.194593944503298e-05,
      "loss": 1.9281,
      "step": 597
    },
    {
      "epoch": 0.19696158754991971,
      "grad_norm": 2.4949111938476562,
      "learning_rate": 9.19172854143635e-05,
      "loss": 1.8233,
      "step": 598
    },
    {
      "epoch": 0.19729095475318045,
      "grad_norm": 3.0978989601135254,
      "learning_rate": 9.18885849841996e-05,
      "loss": 1.9407,
      "step": 599
    },
    {
      "epoch": 0.1976203219564412,
      "grad_norm": 3.714794397354126,
      "learning_rate": 9.185983818631066e-05,
      "loss": 2.082,
      "step": 600
    },
    {
      "epoch": 0.19794968915970193,
      "grad_norm": 2.1515870094299316,
      "learning_rate": 9.183104505251735e-05,
      "loss": 2.5559,
      "step": 601
    },
    {
      "epoch": 0.19827905636296267,
      "grad_norm": 2.2169995307922363,
      "learning_rate": 9.180220561469167e-05,
      "loss": 2.2574,
      "step": 602
    },
    {
      "epoch": 0.19860842356622338,
      "grad_norm": 2.327986717224121,
      "learning_rate": 9.177331990475685e-05,
      "loss": 2.4616,
      "step": 603
    },
    {
      "epoch": 0.19893779076948412,
      "grad_norm": 3.0551161766052246,
      "learning_rate": 9.174438795468734e-05,
      "loss": 2.602,
      "step": 604
    },
    {
      "epoch": 0.19926715797274486,
      "grad_norm": 2.404407024383545,
      "learning_rate": 9.171540979650879e-05,
      "loss": 2.4922,
      "step": 605
    },
    {
      "epoch": 0.1995965251760056,
      "grad_norm": 2.458751916885376,
      "learning_rate": 9.168638546229796e-05,
      "loss": 2.3752,
      "step": 606
    },
    {
      "epoch": 0.19992589237926633,
      "grad_norm": 3.044050693511963,
      "learning_rate": 9.165731498418277e-05,
      "loss": 2.6778,
      "step": 607
    },
    {
      "epoch": 0.20025525958252707,
      "grad_norm": 2.5065767765045166,
      "learning_rate": 9.162819839434223e-05,
      "loss": 2.3761,
      "step": 608
    },
    {
      "epoch": 0.2005846267857878,
      "grad_norm": 2.5272152423858643,
      "learning_rate": 9.15990357250063e-05,
      "loss": 2.2685,
      "step": 609
    },
    {
      "epoch": 0.20091399398904855,
      "grad_norm": 3.1523914337158203,
      "learning_rate": 9.156982700845606e-05,
      "loss": 2.516,
      "step": 610
    },
    {
      "epoch": 0.2012433611923093,
      "grad_norm": 2.7383854389190674,
      "learning_rate": 9.154057227702348e-05,
      "loss": 2.4093,
      "step": 611
    },
    {
      "epoch": 0.20157272839557,
      "grad_norm": 3.010639190673828,
      "learning_rate": 9.151127156309151e-05,
      "loss": 2.5929,
      "step": 612
    },
    {
      "epoch": 0.20190209559883074,
      "grad_norm": 3.279294967651367,
      "learning_rate": 9.1481924899094e-05,
      "loss": 2.266,
      "step": 613
    },
    {
      "epoch": 0.20223146280209148,
      "grad_norm": 3.147496223449707,
      "learning_rate": 9.145253231751563e-05,
      "loss": 2.1337,
      "step": 614
    },
    {
      "epoch": 0.20256083000535222,
      "grad_norm": 2.710343599319458,
      "learning_rate": 9.142309385089191e-05,
      "loss": 2.0808,
      "step": 615
    },
    {
      "epoch": 0.20289019720861295,
      "grad_norm": 2.6864521503448486,
      "learning_rate": 9.139360953180918e-05,
      "loss": 2.1494,
      "step": 616
    },
    {
      "epoch": 0.2032195644118737,
      "grad_norm": 2.7441837787628174,
      "learning_rate": 9.136407939290451e-05,
      "loss": 2.5828,
      "step": 617
    },
    {
      "epoch": 0.20354893161513443,
      "grad_norm": 3.46586537361145,
      "learning_rate": 9.13345034668657e-05,
      "loss": 2.6775,
      "step": 618
    },
    {
      "epoch": 0.20387829881839517,
      "grad_norm": 2.7498793601989746,
      "learning_rate": 9.130488178643119e-05,
      "loss": 2.0033,
      "step": 619
    },
    {
      "epoch": 0.20420766602165588,
      "grad_norm": 3.1091184616088867,
      "learning_rate": 9.127521438439015e-05,
      "loss": 2.4799,
      "step": 620
    },
    {
      "epoch": 0.20453703322491662,
      "grad_norm": 2.3706719875335693,
      "learning_rate": 9.124550129358227e-05,
      "loss": 1.8221,
      "step": 621
    },
    {
      "epoch": 0.20486640042817736,
      "grad_norm": 2.63212251663208,
      "learning_rate": 9.121574254689788e-05,
      "loss": 2.0923,
      "step": 622
    },
    {
      "epoch": 0.2051957676314381,
      "grad_norm": 3.275517225265503,
      "learning_rate": 9.118593817727782e-05,
      "loss": 2.1223,
      "step": 623
    },
    {
      "epoch": 0.20552513483469884,
      "grad_norm": 3.2868645191192627,
      "learning_rate": 9.115608821771347e-05,
      "loss": 2.1894,
      "step": 624
    },
    {
      "epoch": 0.20585450203795957,
      "grad_norm": 3.2911734580993652,
      "learning_rate": 9.112619270124658e-05,
      "loss": 2.0714,
      "step": 625
    },
    {
      "epoch": 0.2061838692412203,
      "grad_norm": 1.8351259231567383,
      "learning_rate": 9.109625166096942e-05,
      "loss": 2.5735,
      "step": 626
    },
    {
      "epoch": 0.20651323644448105,
      "grad_norm": 2.289583206176758,
      "learning_rate": 9.106626513002464e-05,
      "loss": 2.2809,
      "step": 627
    },
    {
      "epoch": 0.20684260364774176,
      "grad_norm": 2.3926987648010254,
      "learning_rate": 9.103623314160518e-05,
      "loss": 2.4775,
      "step": 628
    },
    {
      "epoch": 0.2071719708510025,
      "grad_norm": 2.8485803604125977,
      "learning_rate": 9.100615572895439e-05,
      "loss": 2.5643,
      "step": 629
    },
    {
      "epoch": 0.20750133805426324,
      "grad_norm": 2.601393461227417,
      "learning_rate": 9.097603292536583e-05,
      "loss": 2.1331,
      "step": 630
    },
    {
      "epoch": 0.20783070525752398,
      "grad_norm": 2.364741086959839,
      "learning_rate": 9.094586476418335e-05,
      "loss": 2.2653,
      "step": 631
    },
    {
      "epoch": 0.20816007246078472,
      "grad_norm": 2.307363510131836,
      "learning_rate": 9.091565127880096e-05,
      "loss": 2.8223,
      "step": 632
    },
    {
      "epoch": 0.20848943966404546,
      "grad_norm": 2.207406997680664,
      "learning_rate": 9.088539250266287e-05,
      "loss": 2.2839,
      "step": 633
    },
    {
      "epoch": 0.2088188068673062,
      "grad_norm": 2.271679401397705,
      "learning_rate": 9.085508846926345e-05,
      "loss": 2.1932,
      "step": 634
    },
    {
      "epoch": 0.20914817407056693,
      "grad_norm": 2.566964864730835,
      "learning_rate": 9.082473921214714e-05,
      "loss": 2.2745,
      "step": 635
    },
    {
      "epoch": 0.20947754127382767,
      "grad_norm": 2.6665945053100586,
      "learning_rate": 9.07943447649084e-05,
      "loss": 2.2285,
      "step": 636
    },
    {
      "epoch": 0.20980690847708838,
      "grad_norm": 2.232455253601074,
      "learning_rate": 9.07639051611918e-05,
      "loss": 2.0741,
      "step": 637
    },
    {
      "epoch": 0.21013627568034912,
      "grad_norm": 2.365143299102783,
      "learning_rate": 9.07334204346918e-05,
      "loss": 1.9137,
      "step": 638
    },
    {
      "epoch": 0.21046564288360986,
      "grad_norm": 2.6072494983673096,
      "learning_rate": 9.070289061915289e-05,
      "loss": 2.235,
      "step": 639
    },
    {
      "epoch": 0.2107950100868706,
      "grad_norm": 2.8605432510375977,
      "learning_rate": 9.06723157483694e-05,
      "loss": 2.2824,
      "step": 640
    },
    {
      "epoch": 0.21112437729013134,
      "grad_norm": 2.70151424407959,
      "learning_rate": 9.064169585618561e-05,
      "loss": 2.202,
      "step": 641
    },
    {
      "epoch": 0.21145374449339208,
      "grad_norm": 2.540966272354126,
      "learning_rate": 9.061103097649554e-05,
      "loss": 2.0908,
      "step": 642
    },
    {
      "epoch": 0.21178311169665281,
      "grad_norm": 2.4950785636901855,
      "learning_rate": 9.05803211432431e-05,
      "loss": 1.9015,
      "step": 643
    },
    {
      "epoch": 0.21211247889991355,
      "grad_norm": 2.925896406173706,
      "learning_rate": 9.054956639042194e-05,
      "loss": 2.2201,
      "step": 644
    },
    {
      "epoch": 0.21244184610317426,
      "grad_norm": 3.0082502365112305,
      "learning_rate": 9.051876675207535e-05,
      "loss": 2.3677,
      "step": 645
    },
    {
      "epoch": 0.212771213306435,
      "grad_norm": 2.554163694381714,
      "learning_rate": 9.048792226229642e-05,
      "loss": 1.9325,
      "step": 646
    },
    {
      "epoch": 0.21310058050969574,
      "grad_norm": 2.7593939304351807,
      "learning_rate": 9.04570329552278e-05,
      "loss": 2.2431,
      "step": 647
    },
    {
      "epoch": 0.21342994771295648,
      "grad_norm": 3.1389267444610596,
      "learning_rate": 9.042609886506183e-05,
      "loss": 2.2714,
      "step": 648
    },
    {
      "epoch": 0.21375931491621722,
      "grad_norm": 2.6484322547912598,
      "learning_rate": 9.039512002604034e-05,
      "loss": 2.0055,
      "step": 649
    },
    {
      "epoch": 0.21408868211947796,
      "grad_norm": 3.6394948959350586,
      "learning_rate": 9.036409647245474e-05,
      "loss": 2.2731,
      "step": 650
    },
    {
      "epoch": 0.2144180493227387,
      "grad_norm": 1.968092679977417,
      "learning_rate": 9.033302823864595e-05,
      "loss": 2.3822,
      "step": 651
    },
    {
      "epoch": 0.21474741652599943,
      "grad_norm": 2.4255237579345703,
      "learning_rate": 9.03019153590043e-05,
      "loss": 2.5761,
      "step": 652
    },
    {
      "epoch": 0.21507678372926015,
      "grad_norm": 2.2358973026275635,
      "learning_rate": 9.027075786796957e-05,
      "loss": 2.2964,
      "step": 653
    },
    {
      "epoch": 0.21540615093252088,
      "grad_norm": 3.0591135025024414,
      "learning_rate": 9.023955580003092e-05,
      "loss": 2.609,
      "step": 654
    },
    {
      "epoch": 0.21573551813578162,
      "grad_norm": 2.4640660285949707,
      "learning_rate": 9.020830918972684e-05,
      "loss": 2.4714,
      "step": 655
    },
    {
      "epoch": 0.21606488533904236,
      "grad_norm": 2.639633893966675,
      "learning_rate": 9.017701807164516e-05,
      "loss": 1.9724,
      "step": 656
    },
    {
      "epoch": 0.2163942525423031,
      "grad_norm": 2.35853910446167,
      "learning_rate": 9.014568248042292e-05,
      "loss": 2.1988,
      "step": 657
    },
    {
      "epoch": 0.21672361974556384,
      "grad_norm": 2.6157610416412354,
      "learning_rate": 9.011430245074645e-05,
      "loss": 2.4365,
      "step": 658
    },
    {
      "epoch": 0.21705298694882458,
      "grad_norm": 2.606468439102173,
      "learning_rate": 9.008287801735124e-05,
      "loss": 2.1511,
      "step": 659
    },
    {
      "epoch": 0.21738235415208532,
      "grad_norm": 2.8526015281677246,
      "learning_rate": 9.005140921502193e-05,
      "loss": 2.366,
      "step": 660
    },
    {
      "epoch": 0.21771172135534606,
      "grad_norm": 2.663635492324829,
      "learning_rate": 9.001989607859226e-05,
      "loss": 2.418,
      "step": 661
    },
    {
      "epoch": 0.21804108855860677,
      "grad_norm": 2.6935718059539795,
      "learning_rate": 8.998833864294507e-05,
      "loss": 2.327,
      "step": 662
    },
    {
      "epoch": 0.2183704557618675,
      "grad_norm": 2.840543031692505,
      "learning_rate": 8.995673694301223e-05,
      "loss": 2.106,
      "step": 663
    },
    {
      "epoch": 0.21869982296512824,
      "grad_norm": 2.640491485595703,
      "learning_rate": 8.99250910137746e-05,
      "loss": 1.9047,
      "step": 664
    },
    {
      "epoch": 0.21902919016838898,
      "grad_norm": 2.4656693935394287,
      "learning_rate": 8.989340089026203e-05,
      "loss": 1.9766,
      "step": 665
    },
    {
      "epoch": 0.21935855737164972,
      "grad_norm": 3.2263641357421875,
      "learning_rate": 8.986166660755321e-05,
      "loss": 2.4087,
      "step": 666
    },
    {
      "epoch": 0.21968792457491046,
      "grad_norm": 2.758763313293457,
      "learning_rate": 8.982988820077582e-05,
      "loss": 2.2525,
      "step": 667
    },
    {
      "epoch": 0.2200172917781712,
      "grad_norm": 2.882479429244995,
      "learning_rate": 8.979806570510631e-05,
      "loss": 2.3654,
      "step": 668
    },
    {
      "epoch": 0.22034665898143194,
      "grad_norm": 2.9043984413146973,
      "learning_rate": 8.976619915576994e-05,
      "loss": 1.992,
      "step": 669
    },
    {
      "epoch": 0.22067602618469265,
      "grad_norm": 2.918984889984131,
      "learning_rate": 8.973428858804073e-05,
      "loss": 2.2139,
      "step": 670
    },
    {
      "epoch": 0.22100539338795339,
      "grad_norm": 2.9728167057037354,
      "learning_rate": 8.970233403724146e-05,
      "loss": 2.0774,
      "step": 671
    },
    {
      "epoch": 0.22133476059121412,
      "grad_norm": 2.9786489009857178,
      "learning_rate": 8.96703355387436e-05,
      "loss": 2.1803,
      "step": 672
    },
    {
      "epoch": 0.22166412779447486,
      "grad_norm": 3.05253267288208,
      "learning_rate": 8.963829312796718e-05,
      "loss": 2.1476,
      "step": 673
    },
    {
      "epoch": 0.2219934949977356,
      "grad_norm": 2.640706777572632,
      "learning_rate": 8.960620684038097e-05,
      "loss": 2.1194,
      "step": 674
    },
    {
      "epoch": 0.22232286220099634,
      "grad_norm": 3.7444612979888916,
      "learning_rate": 8.95740767115022e-05,
      "loss": 2.1602,
      "step": 675
    },
    {
      "epoch": 0.22265222940425708,
      "grad_norm": 1.9276431798934937,
      "learning_rate": 8.95419027768967e-05,
      "loss": 2.1702,
      "step": 676
    },
    {
      "epoch": 0.22298159660751782,
      "grad_norm": 2.062800168991089,
      "learning_rate": 8.95096850721787e-05,
      "loss": 2.2015,
      "step": 677
    },
    {
      "epoch": 0.22331096381077853,
      "grad_norm": 2.143061637878418,
      "learning_rate": 8.947742363301098e-05,
      "loss": 2.4098,
      "step": 678
    },
    {
      "epoch": 0.22364033101403927,
      "grad_norm": 2.5208868980407715,
      "learning_rate": 8.944511849510469e-05,
      "loss": 2.5313,
      "step": 679
    },
    {
      "epoch": 0.2239696982173,
      "grad_norm": 2.5163257122039795,
      "learning_rate": 8.941276969421935e-05,
      "loss": 2.4371,
      "step": 680
    },
    {
      "epoch": 0.22429906542056074,
      "grad_norm": 2.5549163818359375,
      "learning_rate": 8.938037726616281e-05,
      "loss": 2.3066,
      "step": 681
    },
    {
      "epoch": 0.22462843262382148,
      "grad_norm": 2.5244922637939453,
      "learning_rate": 8.934794124679121e-05,
      "loss": 2.4772,
      "step": 682
    },
    {
      "epoch": 0.22495779982708222,
      "grad_norm": 2.399825096130371,
      "learning_rate": 8.931546167200895e-05,
      "loss": 2.5803,
      "step": 683
    },
    {
      "epoch": 0.22528716703034296,
      "grad_norm": 2.410311460494995,
      "learning_rate": 8.928293857776866e-05,
      "loss": 2.0751,
      "step": 684
    },
    {
      "epoch": 0.2256165342336037,
      "grad_norm": 2.29736590385437,
      "learning_rate": 8.925037200007109e-05,
      "loss": 2.2536,
      "step": 685
    },
    {
      "epoch": 0.22594590143686444,
      "grad_norm": 2.3026459217071533,
      "learning_rate": 8.921776197496518e-05,
      "loss": 2.0917,
      "step": 686
    },
    {
      "epoch": 0.22627526864012515,
      "grad_norm": 2.490049362182617,
      "learning_rate": 8.918510853854794e-05,
      "loss": 2.0983,
      "step": 687
    },
    {
      "epoch": 0.2266046358433859,
      "grad_norm": 3.218235969543457,
      "learning_rate": 8.915241172696441e-05,
      "loss": 2.3049,
      "step": 688
    },
    {
      "epoch": 0.22693400304664663,
      "grad_norm": 2.5197603702545166,
      "learning_rate": 8.911967157640771e-05,
      "loss": 2.1897,
      "step": 689
    },
    {
      "epoch": 0.22726337024990736,
      "grad_norm": 2.4397006034851074,
      "learning_rate": 8.908688812311884e-05,
      "loss": 1.9715,
      "step": 690
    },
    {
      "epoch": 0.2275927374531681,
      "grad_norm": 2.9002199172973633,
      "learning_rate": 8.905406140338683e-05,
      "loss": 2.2867,
      "step": 691
    },
    {
      "epoch": 0.22792210465642884,
      "grad_norm": 2.6698174476623535,
      "learning_rate": 8.902119145354852e-05,
      "loss": 2.4758,
      "step": 692
    },
    {
      "epoch": 0.22825147185968958,
      "grad_norm": 3.1235620975494385,
      "learning_rate": 8.898827830998864e-05,
      "loss": 2.1978,
      "step": 693
    },
    {
      "epoch": 0.22858083906295032,
      "grad_norm": 2.7362828254699707,
      "learning_rate": 8.895532200913976e-05,
      "loss": 2.326,
      "step": 694
    },
    {
      "epoch": 0.22891020626621103,
      "grad_norm": 2.6940107345581055,
      "learning_rate": 8.892232258748217e-05,
      "loss": 2.0774,
      "step": 695
    },
    {
      "epoch": 0.22923957346947177,
      "grad_norm": 2.963489294052124,
      "learning_rate": 8.888928008154393e-05,
      "loss": 2.2016,
      "step": 696
    },
    {
      "epoch": 0.2295689406727325,
      "grad_norm": 2.9156124591827393,
      "learning_rate": 8.885619452790078e-05,
      "loss": 2.4717,
      "step": 697
    },
    {
      "epoch": 0.22989830787599325,
      "grad_norm": 3.306898355484009,
      "learning_rate": 8.882306596317606e-05,
      "loss": 2.2373,
      "step": 698
    },
    {
      "epoch": 0.23022767507925399,
      "grad_norm": 2.8223652839660645,
      "learning_rate": 8.878989442404082e-05,
      "loss": 1.9008,
      "step": 699
    },
    {
      "epoch": 0.23055704228251472,
      "grad_norm": 2.8709568977355957,
      "learning_rate": 8.87566799472136e-05,
      "loss": 1.9383,
      "step": 700
    },
    {
      "epoch": 0.23088640948577546,
      "grad_norm": 2.175177574157715,
      "learning_rate": 8.872342256946051e-05,
      "loss": 2.3998,
      "step": 701
    },
    {
      "epoch": 0.2312157766890362,
      "grad_norm": 2.408965587615967,
      "learning_rate": 8.869012232759512e-05,
      "loss": 2.6233,
      "step": 702
    },
    {
      "epoch": 0.2315451438922969,
      "grad_norm": 2.2874746322631836,
      "learning_rate": 8.865677925847848e-05,
      "loss": 2.0586,
      "step": 703
    },
    {
      "epoch": 0.23187451109555765,
      "grad_norm": 2.436617374420166,
      "learning_rate": 8.862339339901902e-05,
      "loss": 2.324,
      "step": 704
    },
    {
      "epoch": 0.2322038782988184,
      "grad_norm": 2.8110032081604004,
      "learning_rate": 8.858996478617253e-05,
      "loss": 2.4255,
      "step": 705
    },
    {
      "epoch": 0.23253324550207913,
      "grad_norm": 2.681027889251709,
      "learning_rate": 8.855649345694216e-05,
      "loss": 2.5485,
      "step": 706
    },
    {
      "epoch": 0.23286261270533987,
      "grad_norm": 2.6605820655822754,
      "learning_rate": 8.852297944837831e-05,
      "loss": 2.5063,
      "step": 707
    },
    {
      "epoch": 0.2331919799086006,
      "grad_norm": 2.8735132217407227,
      "learning_rate": 8.848942279757864e-05,
      "loss": 2.1953,
      "step": 708
    },
    {
      "epoch": 0.23352134711186134,
      "grad_norm": 2.5370914936065674,
      "learning_rate": 8.845582354168802e-05,
      "loss": 1.9245,
      "step": 709
    },
    {
      "epoch": 0.23385071431512208,
      "grad_norm": 2.733022928237915,
      "learning_rate": 8.842218171789846e-05,
      "loss": 2.109,
      "step": 710
    },
    {
      "epoch": 0.2341800815183828,
      "grad_norm": 3.819556951522827,
      "learning_rate": 8.838849736344909e-05,
      "loss": 2.405,
      "step": 711
    },
    {
      "epoch": 0.23450944872164353,
      "grad_norm": 2.486302137374878,
      "learning_rate": 8.835477051562613e-05,
      "loss": 2.1244,
      "step": 712
    },
    {
      "epoch": 0.23483881592490427,
      "grad_norm": 3.0748279094696045,
      "learning_rate": 8.832100121176285e-05,
      "loss": 2.2095,
      "step": 713
    },
    {
      "epoch": 0.235168183128165,
      "grad_norm": 3.035090208053589,
      "learning_rate": 8.828718948923949e-05,
      "loss": 1.9369,
      "step": 714
    },
    {
      "epoch": 0.23549755033142575,
      "grad_norm": 2.62367582321167,
      "learning_rate": 8.825333538548326e-05,
      "loss": 2.1745,
      "step": 715
    },
    {
      "epoch": 0.2358269175346865,
      "grad_norm": 2.8850131034851074,
      "learning_rate": 8.821943893796826e-05,
      "loss": 2.3818,
      "step": 716
    },
    {
      "epoch": 0.23615628473794723,
      "grad_norm": 3.0183825492858887,
      "learning_rate": 8.81855001842155e-05,
      "loss": 2.2054,
      "step": 717
    },
    {
      "epoch": 0.23648565194120796,
      "grad_norm": 2.698676586151123,
      "learning_rate": 8.81515191617928e-05,
      "loss": 2.24,
      "step": 718
    },
    {
      "epoch": 0.2368150191444687,
      "grad_norm": 2.8108513355255127,
      "learning_rate": 8.811749590831475e-05,
      "loss": 2.1469,
      "step": 719
    },
    {
      "epoch": 0.2371443863477294,
      "grad_norm": 2.8320016860961914,
      "learning_rate": 8.808343046144271e-05,
      "loss": 2.1924,
      "step": 720
    },
    {
      "epoch": 0.23747375355099015,
      "grad_norm": 3.3814282417297363,
      "learning_rate": 8.804932285888477e-05,
      "loss": 2.3818,
      "step": 721
    },
    {
      "epoch": 0.2378031207542509,
      "grad_norm": 3.158623695373535,
      "learning_rate": 8.80151731383956e-05,
      "loss": 2.3962,
      "step": 722
    },
    {
      "epoch": 0.23813248795751163,
      "grad_norm": 2.743208646774292,
      "learning_rate": 8.798098133777659e-05,
      "loss": 1.779,
      "step": 723
    },
    {
      "epoch": 0.23846185516077237,
      "grad_norm": 3.1935389041900635,
      "learning_rate": 8.794674749487565e-05,
      "loss": 2.1095,
      "step": 724
    },
    {
      "epoch": 0.2387912223640331,
      "grad_norm": 3.515878438949585,
      "learning_rate": 8.791247164758722e-05,
      "loss": 1.9777,
      "step": 725
    },
    {
      "epoch": 0.23912058956729385,
      "grad_norm": 2.7194595336914062,
      "learning_rate": 8.78781538338523e-05,
      "loss": 2.3815,
      "step": 726
    },
    {
      "epoch": 0.23944995677055458,
      "grad_norm": 2.3084821701049805,
      "learning_rate": 8.784379409165828e-05,
      "loss": 2.2721,
      "step": 727
    },
    {
      "epoch": 0.2397793239738153,
      "grad_norm": 2.4673047065734863,
      "learning_rate": 8.780939245903898e-05,
      "loss": 2.1049,
      "step": 728
    },
    {
      "epoch": 0.24010869117707603,
      "grad_norm": 2.5129013061523438,
      "learning_rate": 8.77749489740746e-05,
      "loss": 2.3995,
      "step": 729
    },
    {
      "epoch": 0.24043805838033677,
      "grad_norm": 2.559809923171997,
      "learning_rate": 8.774046367489166e-05,
      "loss": 2.3422,
      "step": 730
    },
    {
      "epoch": 0.2407674255835975,
      "grad_norm": 2.3660426139831543,
      "learning_rate": 8.770593659966298e-05,
      "loss": 2.0759,
      "step": 731
    },
    {
      "epoch": 0.24109679278685825,
      "grad_norm": 2.502711057662964,
      "learning_rate": 8.767136778660759e-05,
      "loss": 2.2224,
      "step": 732
    },
    {
      "epoch": 0.241426159990119,
      "grad_norm": 2.396199941635132,
      "learning_rate": 8.763675727399075e-05,
      "loss": 2.4539,
      "step": 733
    },
    {
      "epoch": 0.24175552719337973,
      "grad_norm": 2.830613851547241,
      "learning_rate": 8.760210510012387e-05,
      "loss": 2.4058,
      "step": 734
    },
    {
      "epoch": 0.24208489439664047,
      "grad_norm": 2.6569724082946777,
      "learning_rate": 8.756741130336448e-05,
      "loss": 2.4416,
      "step": 735
    },
    {
      "epoch": 0.24241426159990118,
      "grad_norm": 3.4117109775543213,
      "learning_rate": 8.753267592211616e-05,
      "loss": 2.483,
      "step": 736
    },
    {
      "epoch": 0.24274362880316191,
      "grad_norm": 2.8380532264709473,
      "learning_rate": 8.749789899482856e-05,
      "loss": 2.4988,
      "step": 737
    },
    {
      "epoch": 0.24307299600642265,
      "grad_norm": 2.325429916381836,
      "learning_rate": 8.74630805599973e-05,
      "loss": 1.8716,
      "step": 738
    },
    {
      "epoch": 0.2434023632096834,
      "grad_norm": 2.5336060523986816,
      "learning_rate": 8.742822065616393e-05,
      "loss": 1.8746,
      "step": 739
    },
    {
      "epoch": 0.24373173041294413,
      "grad_norm": 2.6919002532958984,
      "learning_rate": 8.739331932191592e-05,
      "loss": 2.2107,
      "step": 740
    },
    {
      "epoch": 0.24406109761620487,
      "grad_norm": 3.1390397548675537,
      "learning_rate": 8.735837659588661e-05,
      "loss": 2.5521,
      "step": 741
    },
    {
      "epoch": 0.2443904648194656,
      "grad_norm": 2.4885716438293457,
      "learning_rate": 8.732339251675516e-05,
      "loss": 2.1718,
      "step": 742
    },
    {
      "epoch": 0.24471983202272635,
      "grad_norm": 2.5541298389434814,
      "learning_rate": 8.728836712324646e-05,
      "loss": 2.1153,
      "step": 743
    },
    {
      "epoch": 0.24504919922598709,
      "grad_norm": 2.7616307735443115,
      "learning_rate": 8.725330045413117e-05,
      "loss": 2.0635,
      "step": 744
    },
    {
      "epoch": 0.2453785664292478,
      "grad_norm": 2.8875701427459717,
      "learning_rate": 8.721819254822565e-05,
      "loss": 2.3036,
      "step": 745
    },
    {
      "epoch": 0.24570793363250854,
      "grad_norm": 3.5547261238098145,
      "learning_rate": 8.718304344439186e-05,
      "loss": 2.5285,
      "step": 746
    },
    {
      "epoch": 0.24603730083576927,
      "grad_norm": 3.126332998275757,
      "learning_rate": 8.714785318153742e-05,
      "loss": 2.1983,
      "step": 747
    },
    {
      "epoch": 0.24636666803903,
      "grad_norm": 2.997291088104248,
      "learning_rate": 8.711262179861547e-05,
      "loss": 2.2518,
      "step": 748
    },
    {
      "epoch": 0.24669603524229075,
      "grad_norm": 2.786508083343506,
      "learning_rate": 8.70773493346247e-05,
      "loss": 1.8481,
      "step": 749
    },
    {
      "epoch": 0.2470254024455515,
      "grad_norm": 3.188028573989868,
      "learning_rate": 8.704203582860922e-05,
      "loss": 1.929,
      "step": 750
    },
    {
      "epoch": 0.24735476964881223,
      "grad_norm": 1.924673318862915,
      "learning_rate": 8.700668131965861e-05,
      "loss": 2.319,
      "step": 751
    },
    {
      "epoch": 0.24768413685207297,
      "grad_norm": 2.163625955581665,
      "learning_rate": 8.697128584690785e-05,
      "loss": 2.2707,
      "step": 752
    },
    {
      "epoch": 0.24801350405533368,
      "grad_norm": 2.0983192920684814,
      "learning_rate": 8.693584944953723e-05,
      "loss": 2.4019,
      "step": 753
    },
    {
      "epoch": 0.24834287125859442,
      "grad_norm": 2.2360665798187256,
      "learning_rate": 8.690037216677236e-05,
      "loss": 2.5004,
      "step": 754
    },
    {
      "epoch": 0.24867223846185516,
      "grad_norm": 2.301961660385132,
      "learning_rate": 8.686485403788411e-05,
      "loss": 2.2912,
      "step": 755
    },
    {
      "epoch": 0.2490016056651159,
      "grad_norm": 2.593334436416626,
      "learning_rate": 8.682929510218855e-05,
      "loss": 2.7432,
      "step": 756
    },
    {
      "epoch": 0.24933097286837663,
      "grad_norm": 2.3250508308410645,
      "learning_rate": 8.679369539904693e-05,
      "loss": 2.525,
      "step": 757
    },
    {
      "epoch": 0.24966034007163737,
      "grad_norm": 2.4877517223358154,
      "learning_rate": 8.675805496786563e-05,
      "loss": 2.2531,
      "step": 758
    },
    {
      "epoch": 0.2499897072748981,
      "grad_norm": 2.423072099685669,
      "learning_rate": 8.672237384809609e-05,
      "loss": 2.3282,
      "step": 759
    },
    {
      "epoch": 0.2499897072748981,
      "eval_loss": 2.2632205486297607,
      "eval_runtime": 795.2953,
      "eval_samples_per_second": 3.215,
      "eval_steps_per_second": 1.608,
      "step": 759
    },
    {
      "epoch": 0.2503190744781588,
      "grad_norm": 2.326564073562622,
      "learning_rate": 8.668665207923482e-05,
      "loss": 2.0632,
      "step": 760
    },
    {
      "epoch": 0.2506484416814196,
      "grad_norm": 2.884378671646118,
      "learning_rate": 8.665088970082331e-05,
      "loss": 2.5123,
      "step": 761
    },
    {
      "epoch": 0.2509778088846803,
      "grad_norm": 2.3617143630981445,
      "learning_rate": 8.661508675244801e-05,
      "loss": 2.1108,
      "step": 762
    },
    {
      "epoch": 0.25130717608794106,
      "grad_norm": 2.4837584495544434,
      "learning_rate": 8.657924327374025e-05,
      "loss": 2.3088,
      "step": 763
    },
    {
      "epoch": 0.2516365432912018,
      "grad_norm": 2.4218530654907227,
      "learning_rate": 8.654335930437627e-05,
      "loss": 1.8391,
      "step": 764
    },
    {
      "epoch": 0.25196591049446254,
      "grad_norm": 2.5416500568389893,
      "learning_rate": 8.650743488407709e-05,
      "loss": 2.1901,
      "step": 765
    },
    {
      "epoch": 0.25229527769772325,
      "grad_norm": 2.6304333209991455,
      "learning_rate": 8.647147005260854e-05,
      "loss": 2.4257,
      "step": 766
    },
    {
      "epoch": 0.25262464490098396,
      "grad_norm": 3.04967999458313,
      "learning_rate": 8.643546484978115e-05,
      "loss": 2.1485,
      "step": 767
    },
    {
      "epoch": 0.25295401210424473,
      "grad_norm": 2.736220359802246,
      "learning_rate": 8.639941931545017e-05,
      "loss": 2.0484,
      "step": 768
    },
    {
      "epoch": 0.25328337930750544,
      "grad_norm": 2.4498915672302246,
      "learning_rate": 8.636333348951546e-05,
      "loss": 1.9845,
      "step": 769
    },
    {
      "epoch": 0.2536127465107662,
      "grad_norm": 3.074000597000122,
      "learning_rate": 8.63272074119215e-05,
      "loss": 2.0217,
      "step": 770
    },
    {
      "epoch": 0.2539421137140269,
      "grad_norm": 2.7875514030456543,
      "learning_rate": 8.629104112265735e-05,
      "loss": 2.1935,
      "step": 771
    },
    {
      "epoch": 0.2542714809172877,
      "grad_norm": 2.7638540267944336,
      "learning_rate": 8.625483466175652e-05,
      "loss": 2.1215,
      "step": 772
    },
    {
      "epoch": 0.2546008481205484,
      "grad_norm": 2.99078106880188,
      "learning_rate": 8.621858806929705e-05,
      "loss": 1.8759,
      "step": 773
    },
    {
      "epoch": 0.2549302153238091,
      "grad_norm": 3.0909841060638428,
      "learning_rate": 8.618230138540136e-05,
      "loss": 1.8021,
      "step": 774
    },
    {
      "epoch": 0.2552595825270699,
      "grad_norm": 3.437960147857666,
      "learning_rate": 8.614597465023626e-05,
      "loss": 1.9589,
      "step": 775
    },
    {
      "epoch": 0.2555889497303306,
      "grad_norm": 2.2460529804229736,
      "learning_rate": 8.61096079040129e-05,
      "loss": 2.2812,
      "step": 776
    },
    {
      "epoch": 0.25591831693359135,
      "grad_norm": 2.4062023162841797,
      "learning_rate": 8.607320118698674e-05,
      "loss": 2.1588,
      "step": 777
    },
    {
      "epoch": 0.25624768413685206,
      "grad_norm": 3.011057138442993,
      "learning_rate": 8.603675453945742e-05,
      "loss": 2.5091,
      "step": 778
    },
    {
      "epoch": 0.2565770513401128,
      "grad_norm": 2.4397964477539062,
      "learning_rate": 8.600026800176885e-05,
      "loss": 2.4397,
      "step": 779
    },
    {
      "epoch": 0.25690641854337354,
      "grad_norm": 2.548110246658325,
      "learning_rate": 8.596374161430907e-05,
      "loss": 2.4302,
      "step": 780
    },
    {
      "epoch": 0.2572357857466343,
      "grad_norm": 2.477574586868286,
      "learning_rate": 8.592717541751021e-05,
      "loss": 2.2464,
      "step": 781
    },
    {
      "epoch": 0.257565152949895,
      "grad_norm": 2.5044217109680176,
      "learning_rate": 8.58905694518485e-05,
      "loss": 2.13,
      "step": 782
    },
    {
      "epoch": 0.2578945201531557,
      "grad_norm": 2.3912570476531982,
      "learning_rate": 8.585392375784418e-05,
      "loss": 2.3024,
      "step": 783
    },
    {
      "epoch": 0.2582238873564165,
      "grad_norm": 2.51187801361084,
      "learning_rate": 8.581723837606144e-05,
      "loss": 2.3613,
      "step": 784
    },
    {
      "epoch": 0.2585532545596772,
      "grad_norm": 2.4491524696350098,
      "learning_rate": 8.578051334710844e-05,
      "loss": 2.0575,
      "step": 785
    },
    {
      "epoch": 0.25888262176293797,
      "grad_norm": 2.263449192047119,
      "learning_rate": 8.574374871163721e-05,
      "loss": 2.0833,
      "step": 786
    },
    {
      "epoch": 0.2592119889661987,
      "grad_norm": 2.714989423751831,
      "learning_rate": 8.570694451034362e-05,
      "loss": 2.2463,
      "step": 787
    },
    {
      "epoch": 0.25954135616945945,
      "grad_norm": 2.5804946422576904,
      "learning_rate": 8.567010078396735e-05,
      "loss": 2.2576,
      "step": 788
    },
    {
      "epoch": 0.25987072337272016,
      "grad_norm": 2.759988784790039,
      "learning_rate": 8.56332175732918e-05,
      "loss": 2.2353,
      "step": 789
    },
    {
      "epoch": 0.26020009057598087,
      "grad_norm": 2.70926570892334,
      "learning_rate": 8.559629491914412e-05,
      "loss": 2.0866,
      "step": 790
    },
    {
      "epoch": 0.26052945777924164,
      "grad_norm": 3.139805555343628,
      "learning_rate": 8.555933286239508e-05,
      "loss": 2.4547,
      "step": 791
    },
    {
      "epoch": 0.26085882498250235,
      "grad_norm": 2.390231132507324,
      "learning_rate": 8.552233144395907e-05,
      "loss": 1.9382,
      "step": 792
    },
    {
      "epoch": 0.2611881921857631,
      "grad_norm": 2.88167142868042,
      "learning_rate": 8.54852907047941e-05,
      "loss": 2.1961,
      "step": 793
    },
    {
      "epoch": 0.2615175593890238,
      "grad_norm": 2.3817718029022217,
      "learning_rate": 8.544821068590165e-05,
      "loss": 1.7874,
      "step": 794
    },
    {
      "epoch": 0.2618469265922846,
      "grad_norm": 2.8125686645507812,
      "learning_rate": 8.54110914283267e-05,
      "loss": 2.1755,
      "step": 795
    },
    {
      "epoch": 0.2621762937955453,
      "grad_norm": 2.719560384750366,
      "learning_rate": 8.537393297315767e-05,
      "loss": 2.0847,
      "step": 796
    },
    {
      "epoch": 0.26250566099880607,
      "grad_norm": 2.7798001766204834,
      "learning_rate": 8.533673536152638e-05,
      "loss": 1.8334,
      "step": 797
    },
    {
      "epoch": 0.2628350282020668,
      "grad_norm": 4.041043758392334,
      "learning_rate": 8.529949863460793e-05,
      "loss": 2.3578,
      "step": 798
    },
    {
      "epoch": 0.2631643954053275,
      "grad_norm": 3.4026458263397217,
      "learning_rate": 8.52622228336208e-05,
      "loss": 2.0998,
      "step": 799
    },
    {
      "epoch": 0.26349376260858826,
      "grad_norm": 4.440029144287109,
      "learning_rate": 8.522490799982669e-05,
      "loss": 1.7791,
      "step": 800
    },
    {
      "epoch": 0.26382312981184897,
      "grad_norm": 2.11838436126709,
      "learning_rate": 8.518755417453048e-05,
      "loss": 2.4855,
      "step": 801
    },
    {
      "epoch": 0.26415249701510973,
      "grad_norm": 2.5172719955444336,
      "learning_rate": 8.515016139908024e-05,
      "loss": 2.2894,
      "step": 802
    },
    {
      "epoch": 0.26448186421837044,
      "grad_norm": 2.1991279125213623,
      "learning_rate": 8.511272971486717e-05,
      "loss": 2.2683,
      "step": 803
    },
    {
      "epoch": 0.2648112314216312,
      "grad_norm": 2.425428867340088,
      "learning_rate": 8.507525916332549e-05,
      "loss": 2.2494,
      "step": 804
    },
    {
      "epoch": 0.2651405986248919,
      "grad_norm": 2.5921618938446045,
      "learning_rate": 8.503774978593248e-05,
      "loss": 2.2534,
      "step": 805
    },
    {
      "epoch": 0.2654699658281527,
      "grad_norm": 2.6209821701049805,
      "learning_rate": 8.500020162420841e-05,
      "loss": 2.3588,
      "step": 806
    },
    {
      "epoch": 0.2657993330314134,
      "grad_norm": 2.933804750442505,
      "learning_rate": 8.496261471971645e-05,
      "loss": 2.5508,
      "step": 807
    },
    {
      "epoch": 0.2661287002346741,
      "grad_norm": 2.4999735355377197,
      "learning_rate": 8.492498911406266e-05,
      "loss": 2.0871,
      "step": 808
    },
    {
      "epoch": 0.2664580674379349,
      "grad_norm": 2.6054482460021973,
      "learning_rate": 8.488732484889594e-05,
      "loss": 2.1609,
      "step": 809
    },
    {
      "epoch": 0.2667874346411956,
      "grad_norm": 2.588648557662964,
      "learning_rate": 8.4849621965908e-05,
      "loss": 2.2175,
      "step": 810
    },
    {
      "epoch": 0.26711680184445635,
      "grad_norm": 2.7105255126953125,
      "learning_rate": 8.481188050683328e-05,
      "loss": 2.1727,
      "step": 811
    },
    {
      "epoch": 0.26744616904771706,
      "grad_norm": 2.858750581741333,
      "learning_rate": 8.477410051344895e-05,
      "loss": 2.6405,
      "step": 812
    },
    {
      "epoch": 0.26777553625097783,
      "grad_norm": 2.562743663787842,
      "learning_rate": 8.473628202757478e-05,
      "loss": 2.2222,
      "step": 813
    },
    {
      "epoch": 0.26810490345423854,
      "grad_norm": 2.7561135292053223,
      "learning_rate": 8.46984250910732e-05,
      "loss": 2.3755,
      "step": 814
    },
    {
      "epoch": 0.26843427065749925,
      "grad_norm": 2.5868582725524902,
      "learning_rate": 8.466052974584918e-05,
      "loss": 2.3861,
      "step": 815
    },
    {
      "epoch": 0.26876363786076,
      "grad_norm": 2.836190938949585,
      "learning_rate": 8.46225960338502e-05,
      "loss": 2.1303,
      "step": 816
    },
    {
      "epoch": 0.26909300506402073,
      "grad_norm": 3.028157949447632,
      "learning_rate": 8.458462399706623e-05,
      "loss": 2.2734,
      "step": 817
    },
    {
      "epoch": 0.2694223722672815,
      "grad_norm": 2.804187536239624,
      "learning_rate": 8.454661367752965e-05,
      "loss": 2.1357,
      "step": 818
    },
    {
      "epoch": 0.2697517394705422,
      "grad_norm": 2.80621075630188,
      "learning_rate": 8.450856511731519e-05,
      "loss": 1.732,
      "step": 819
    },
    {
      "epoch": 0.270081106673803,
      "grad_norm": 2.5028510093688965,
      "learning_rate": 8.447047835853999e-05,
      "loss": 1.8806,
      "step": 820
    },
    {
      "epoch": 0.2704104738770637,
      "grad_norm": 2.597686529159546,
      "learning_rate": 8.443235344336337e-05,
      "loss": 2.1617,
      "step": 821
    },
    {
      "epoch": 0.27073984108032445,
      "grad_norm": 2.7407145500183105,
      "learning_rate": 8.439419041398698e-05,
      "loss": 2.157,
      "step": 822
    },
    {
      "epoch": 0.27106920828358516,
      "grad_norm": 3.096123695373535,
      "learning_rate": 8.435598931265459e-05,
      "loss": 1.9549,
      "step": 823
    },
    {
      "epoch": 0.27139857548684587,
      "grad_norm": 4.0686516761779785,
      "learning_rate": 8.431775018165211e-05,
      "loss": 2.139,
      "step": 824
    },
    {
      "epoch": 0.27172794269010664,
      "grad_norm": 2.7538039684295654,
      "learning_rate": 8.427947306330764e-05,
      "loss": 1.8598,
      "step": 825
    },
    {
      "epoch": 0.27205730989336735,
      "grad_norm": 2.3142900466918945,
      "learning_rate": 8.424115799999122e-05,
      "loss": 2.3159,
      "step": 826
    },
    {
      "epoch": 0.2723866770966281,
      "grad_norm": 2.2066521644592285,
      "learning_rate": 8.420280503411495e-05,
      "loss": 2.2287,
      "step": 827
    },
    {
      "epoch": 0.2727160442998888,
      "grad_norm": 2.188645839691162,
      "learning_rate": 8.416441420813288e-05,
      "loss": 2.2174,
      "step": 828
    },
    {
      "epoch": 0.2730454115031496,
      "grad_norm": 2.808419704437256,
      "learning_rate": 8.412598556454096e-05,
      "loss": 2.6173,
      "step": 829
    },
    {
      "epoch": 0.2733747787064103,
      "grad_norm": 2.461317300796509,
      "learning_rate": 8.408751914587699e-05,
      "loss": 2.5954,
      "step": 830
    },
    {
      "epoch": 0.27370414590967107,
      "grad_norm": 2.9252431392669678,
      "learning_rate": 8.40490149947206e-05,
      "loss": 2.4617,
      "step": 831
    },
    {
      "epoch": 0.2740335131129318,
      "grad_norm": 2.5286619663238525,
      "learning_rate": 8.401047315369319e-05,
      "loss": 2.5215,
      "step": 832
    },
    {
      "epoch": 0.2743628803161925,
      "grad_norm": 2.650164842605591,
      "learning_rate": 8.397189366545786e-05,
      "loss": 2.1433,
      "step": 833
    },
    {
      "epoch": 0.27469224751945326,
      "grad_norm": 2.2741010189056396,
      "learning_rate": 8.393327657271944e-05,
      "loss": 2.0067,
      "step": 834
    },
    {
      "epoch": 0.27502161472271397,
      "grad_norm": 2.3971540927886963,
      "learning_rate": 8.389462191822428e-05,
      "loss": 2.2778,
      "step": 835
    },
    {
      "epoch": 0.27535098192597474,
      "grad_norm": 2.9199283123016357,
      "learning_rate": 8.385592974476042e-05,
      "loss": 2.375,
      "step": 836
    },
    {
      "epoch": 0.27568034912923545,
      "grad_norm": 2.3536462783813477,
      "learning_rate": 8.381720009515736e-05,
      "loss": 2.2465,
      "step": 837
    },
    {
      "epoch": 0.2760097163324962,
      "grad_norm": 2.7381389141082764,
      "learning_rate": 8.377843301228611e-05,
      "loss": 2.201,
      "step": 838
    },
    {
      "epoch": 0.2763390835357569,
      "grad_norm": 2.795403003692627,
      "learning_rate": 8.373962853905912e-05,
      "loss": 2.1858,
      "step": 839
    },
    {
      "epoch": 0.27666845073901764,
      "grad_norm": 2.549487829208374,
      "learning_rate": 8.37007867184302e-05,
      "loss": 2.0456,
      "step": 840
    },
    {
      "epoch": 0.2769978179422784,
      "grad_norm": 2.419311046600342,
      "learning_rate": 8.366190759339453e-05,
      "loss": 1.8163,
      "step": 841
    },
    {
      "epoch": 0.2773271851455391,
      "grad_norm": 2.8960893154144287,
      "learning_rate": 8.362299120698858e-05,
      "loss": 2.0316,
      "step": 842
    },
    {
      "epoch": 0.2776565523487999,
      "grad_norm": 2.440534830093384,
      "learning_rate": 8.358403760229006e-05,
      "loss": 1.8316,
      "step": 843
    },
    {
      "epoch": 0.2779859195520606,
      "grad_norm": 2.9307198524475098,
      "learning_rate": 8.354504682241786e-05,
      "loss": 2.5626,
      "step": 844
    },
    {
      "epoch": 0.27831528675532136,
      "grad_norm": 2.7770323753356934,
      "learning_rate": 8.350601891053207e-05,
      "loss": 2.2409,
      "step": 845
    },
    {
      "epoch": 0.27864465395858207,
      "grad_norm": 2.7087655067443848,
      "learning_rate": 8.346695390983382e-05,
      "loss": 1.9896,
      "step": 846
    },
    {
      "epoch": 0.27897402116184283,
      "grad_norm": 2.673274517059326,
      "learning_rate": 8.342785186356534e-05,
      "loss": 2.0128,
      "step": 847
    },
    {
      "epoch": 0.27930338836510354,
      "grad_norm": 3.0754430294036865,
      "learning_rate": 8.338871281500984e-05,
      "loss": 2.008,
      "step": 848
    },
    {
      "epoch": 0.27963275556836426,
      "grad_norm": 2.8769547939300537,
      "learning_rate": 8.334953680749152e-05,
      "loss": 1.8857,
      "step": 849
    },
    {
      "epoch": 0.279962122771625,
      "grad_norm": 3.0530593395233154,
      "learning_rate": 8.331032388437546e-05,
      "loss": 1.967,
      "step": 850
    },
    {
      "epoch": 0.28029148997488573,
      "grad_norm": 1.8923742771148682,
      "learning_rate": 8.32710740890676e-05,
      "loss": 2.4918,
      "step": 851
    },
    {
      "epoch": 0.2806208571781465,
      "grad_norm": 2.0591697692871094,
      "learning_rate": 8.323178746501475e-05,
      "loss": 2.2662,
      "step": 852
    },
    {
      "epoch": 0.2809502243814072,
      "grad_norm": 2.4539167881011963,
      "learning_rate": 8.319246405570441e-05,
      "loss": 2.4903,
      "step": 853
    },
    {
      "epoch": 0.281279591584668,
      "grad_norm": 2.1013600826263428,
      "learning_rate": 8.315310390466487e-05,
      "loss": 2.1026,
      "step": 854
    },
    {
      "epoch": 0.2816089587879287,
      "grad_norm": 2.0880558490753174,
      "learning_rate": 8.311370705546501e-05,
      "loss": 2.3886,
      "step": 855
    },
    {
      "epoch": 0.28193832599118945,
      "grad_norm": 2.4675230979919434,
      "learning_rate": 8.307427355171443e-05,
      "loss": 2.2145,
      "step": 856
    },
    {
      "epoch": 0.28226769319445016,
      "grad_norm": 2.3811771869659424,
      "learning_rate": 8.303480343706321e-05,
      "loss": 2.6376,
      "step": 857
    },
    {
      "epoch": 0.2825970603977109,
      "grad_norm": 2.4288272857666016,
      "learning_rate": 8.299529675520201e-05,
      "loss": 2.2381,
      "step": 858
    },
    {
      "epoch": 0.28292642760097164,
      "grad_norm": 2.4424099922180176,
      "learning_rate": 8.295575354986196e-05,
      "loss": 2.4726,
      "step": 859
    },
    {
      "epoch": 0.28325579480423235,
      "grad_norm": 2.6503515243530273,
      "learning_rate": 8.291617386481463e-05,
      "loss": 2.4256,
      "step": 860
    },
    {
      "epoch": 0.2835851620074931,
      "grad_norm": 2.2516067028045654,
      "learning_rate": 8.287655774387193e-05,
      "loss": 1.9415,
      "step": 861
    },
    {
      "epoch": 0.28391452921075383,
      "grad_norm": 2.435704231262207,
      "learning_rate": 8.283690523088613e-05,
      "loss": 2.0356,
      "step": 862
    },
    {
      "epoch": 0.2842438964140146,
      "grad_norm": 2.6750681400299072,
      "learning_rate": 8.279721636974978e-05,
      "loss": 2.2815,
      "step": 863
    },
    {
      "epoch": 0.2845732636172753,
      "grad_norm": 3.6107234954833984,
      "learning_rate": 8.275749120439569e-05,
      "loss": 2.2028,
      "step": 864
    },
    {
      "epoch": 0.284902630820536,
      "grad_norm": 2.5893361568450928,
      "learning_rate": 8.27177297787968e-05,
      "loss": 2.1027,
      "step": 865
    },
    {
      "epoch": 0.2852319980237968,
      "grad_norm": 2.4202561378479004,
      "learning_rate": 8.267793213696624e-05,
      "loss": 2.0077,
      "step": 866
    },
    {
      "epoch": 0.2855613652270575,
      "grad_norm": 2.8321430683135986,
      "learning_rate": 8.263809832295719e-05,
      "loss": 1.9374,
      "step": 867
    },
    {
      "epoch": 0.28589073243031826,
      "grad_norm": 2.830946683883667,
      "learning_rate": 8.25982283808629e-05,
      "loss": 2.3701,
      "step": 868
    },
    {
      "epoch": 0.286220099633579,
      "grad_norm": 2.626882791519165,
      "learning_rate": 8.255832235481659e-05,
      "loss": 2.2236,
      "step": 869
    },
    {
      "epoch": 0.28654946683683974,
      "grad_norm": 2.398036003112793,
      "learning_rate": 8.251838028899143e-05,
      "loss": 2.2344,
      "step": 870
    },
    {
      "epoch": 0.28687883404010045,
      "grad_norm": 3.660982608795166,
      "learning_rate": 8.247840222760052e-05,
      "loss": 2.2156,
      "step": 871
    },
    {
      "epoch": 0.2872082012433612,
      "grad_norm": 3.3050460815429688,
      "learning_rate": 8.243838821489671e-05,
      "loss": 2.0017,
      "step": 872
    },
    {
      "epoch": 0.2875375684466219,
      "grad_norm": 3.096376895904541,
      "learning_rate": 8.239833829517276e-05,
      "loss": 1.9194,
      "step": 873
    },
    {
      "epoch": 0.28786693564988264,
      "grad_norm": 2.761716365814209,
      "learning_rate": 8.235825251276108e-05,
      "loss": 1.9807,
      "step": 874
    },
    {
      "epoch": 0.2881963028531434,
      "grad_norm": 3.867349624633789,
      "learning_rate": 8.231813091203385e-05,
      "loss": 1.9222,
      "step": 875
    },
    {
      "epoch": 0.2885256700564041,
      "grad_norm": 2.335017442703247,
      "learning_rate": 8.227797353740286e-05,
      "loss": 2.4247,
      "step": 876
    },
    {
      "epoch": 0.2888550372596649,
      "grad_norm": 2.4474287033081055,
      "learning_rate": 8.223778043331948e-05,
      "loss": 2.6098,
      "step": 877
    },
    {
      "epoch": 0.2891844044629256,
      "grad_norm": 2.22808575630188,
      "learning_rate": 8.219755164427469e-05,
      "loss": 2.2784,
      "step": 878
    },
    {
      "epoch": 0.28951377166618636,
      "grad_norm": 2.265848398208618,
      "learning_rate": 8.215728721479892e-05,
      "loss": 2.2767,
      "step": 879
    },
    {
      "epoch": 0.28984313886944707,
      "grad_norm": 2.5640437602996826,
      "learning_rate": 8.211698718946208e-05,
      "loss": 1.9147,
      "step": 880
    },
    {
      "epoch": 0.29017250607270784,
      "grad_norm": 2.4126553535461426,
      "learning_rate": 8.207665161287345e-05,
      "loss": 2.2837,
      "step": 881
    },
    {
      "epoch": 0.29050187327596855,
      "grad_norm": 2.4454050064086914,
      "learning_rate": 8.20362805296817e-05,
      "loss": 2.3148,
      "step": 882
    },
    {
      "epoch": 0.29083124047922926,
      "grad_norm": 2.776961088180542,
      "learning_rate": 8.19958739845748e-05,
      "loss": 2.7321,
      "step": 883
    },
    {
      "epoch": 0.29116060768249,
      "grad_norm": 2.721742630004883,
      "learning_rate": 8.195543202227993e-05,
      "loss": 2.3516,
      "step": 884
    },
    {
      "epoch": 0.29148997488575074,
      "grad_norm": 2.417945384979248,
      "learning_rate": 8.191495468756354e-05,
      "loss": 2.0815,
      "step": 885
    },
    {
      "epoch": 0.2918193420890115,
      "grad_norm": 2.7946279048919678,
      "learning_rate": 8.187444202523116e-05,
      "loss": 2.5238,
      "step": 886
    },
    {
      "epoch": 0.2921487092922722,
      "grad_norm": 2.2200636863708496,
      "learning_rate": 8.183389408012752e-05,
      "loss": 2.2116,
      "step": 887
    },
    {
      "epoch": 0.292478076495533,
      "grad_norm": 2.7158875465393066,
      "learning_rate": 8.179331089713629e-05,
      "loss": 2.2724,
      "step": 888
    },
    {
      "epoch": 0.2928074436987937,
      "grad_norm": 3.0811691284179688,
      "learning_rate": 8.175269252118023e-05,
      "loss": 2.481,
      "step": 889
    },
    {
      "epoch": 0.2931368109020544,
      "grad_norm": 2.898715019226074,
      "learning_rate": 8.171203899722105e-05,
      "loss": 1.9759,
      "step": 890
    },
    {
      "epoch": 0.29346617810531517,
      "grad_norm": 3.110511541366577,
      "learning_rate": 8.167135037025933e-05,
      "loss": 2.3635,
      "step": 891
    },
    {
      "epoch": 0.2937955453085759,
      "grad_norm": 2.4876153469085693,
      "learning_rate": 8.163062668533454e-05,
      "loss": 2.2224,
      "step": 892
    },
    {
      "epoch": 0.29412491251183664,
      "grad_norm": 2.962139844894409,
      "learning_rate": 8.158986798752492e-05,
      "loss": 2.1494,
      "step": 893
    },
    {
      "epoch": 0.29445427971509736,
      "grad_norm": 2.6857833862304688,
      "learning_rate": 8.154907432194751e-05,
      "loss": 1.9066,
      "step": 894
    },
    {
      "epoch": 0.2947836469183581,
      "grad_norm": 3.1264843940734863,
      "learning_rate": 8.150824573375804e-05,
      "loss": 2.1503,
      "step": 895
    },
    {
      "epoch": 0.29511301412161883,
      "grad_norm": 3.0480618476867676,
      "learning_rate": 8.146738226815087e-05,
      "loss": 2.2263,
      "step": 896
    },
    {
      "epoch": 0.2954423813248796,
      "grad_norm": 2.6517231464385986,
      "learning_rate": 8.142648397035899e-05,
      "loss": 1.98,
      "step": 897
    },
    {
      "epoch": 0.2957717485281403,
      "grad_norm": 2.765357255935669,
      "learning_rate": 8.138555088565398e-05,
      "loss": 1.8292,
      "step": 898
    },
    {
      "epoch": 0.296101115731401,
      "grad_norm": 2.788534641265869,
      "learning_rate": 8.134458305934587e-05,
      "loss": 1.7272,
      "step": 899
    },
    {
      "epoch": 0.2964304829346618,
      "grad_norm": 3.0059924125671387,
      "learning_rate": 8.130358053678315e-05,
      "loss": 2.0136,
      "step": 900
    },
    {
      "epoch": 0.2967598501379225,
      "grad_norm": 2.0941505432128906,
      "learning_rate": 8.126254336335279e-05,
      "loss": 2.5732,
      "step": 901
    },
    {
      "epoch": 0.29708921734118326,
      "grad_norm": 2.0665740966796875,
      "learning_rate": 8.122147158448002e-05,
      "loss": 2.4054,
      "step": 902
    },
    {
      "epoch": 0.297418584544444,
      "grad_norm": 2.2430355548858643,
      "learning_rate": 8.118036524562841e-05,
      "loss": 2.4054,
      "step": 903
    },
    {
      "epoch": 0.29774795174770474,
      "grad_norm": 2.2699286937713623,
      "learning_rate": 8.113922439229982e-05,
      "loss": 2.1162,
      "step": 904
    },
    {
      "epoch": 0.29807731895096545,
      "grad_norm": 2.1671202182769775,
      "learning_rate": 8.109804907003429e-05,
      "loss": 2.3838,
      "step": 905
    },
    {
      "epoch": 0.2984066861542262,
      "grad_norm": 2.175811529159546,
      "learning_rate": 8.105683932441e-05,
      "loss": 2.0848,
      "step": 906
    },
    {
      "epoch": 0.29873605335748693,
      "grad_norm": 3.168586492538452,
      "learning_rate": 8.101559520104323e-05,
      "loss": 2.4891,
      "step": 907
    },
    {
      "epoch": 0.29906542056074764,
      "grad_norm": 2.39568829536438,
      "learning_rate": 8.097431674558838e-05,
      "loss": 2.3179,
      "step": 908
    },
    {
      "epoch": 0.2993947877640084,
      "grad_norm": 2.7230722904205322,
      "learning_rate": 8.093300400373775e-05,
      "loss": 2.2789,
      "step": 909
    },
    {
      "epoch": 0.2997241549672691,
      "grad_norm": 2.686748504638672,
      "learning_rate": 8.08916570212217e-05,
      "loss": 2.1697,
      "step": 910
    },
    {
      "epoch": 0.3000535221705299,
      "grad_norm": 2.3742809295654297,
      "learning_rate": 8.08502758438084e-05,
      "loss": 1.9485,
      "step": 911
    },
    {
      "epoch": 0.3003828893737906,
      "grad_norm": 2.234506607055664,
      "learning_rate": 8.080886051730391e-05,
      "loss": 2.2139,
      "step": 912
    },
    {
      "epoch": 0.30071225657705136,
      "grad_norm": 2.438096284866333,
      "learning_rate": 8.076741108755212e-05,
      "loss": 2.338,
      "step": 913
    },
    {
      "epoch": 0.3010416237803121,
      "grad_norm": 2.3559389114379883,
      "learning_rate": 8.072592760043463e-05,
      "loss": 2.3608,
      "step": 914
    },
    {
      "epoch": 0.3013709909835728,
      "grad_norm": 2.8507134914398193,
      "learning_rate": 8.068441010187073e-05,
      "loss": 2.3359,
      "step": 915
    },
    {
      "epoch": 0.30170035818683355,
      "grad_norm": 3.0053601264953613,
      "learning_rate": 8.06428586378174e-05,
      "loss": 2.2598,
      "step": 916
    },
    {
      "epoch": 0.30202972539009426,
      "grad_norm": 2.6545512676239014,
      "learning_rate": 8.06012732542692e-05,
      "loss": 2.2021,
      "step": 917
    },
    {
      "epoch": 0.302359092593355,
      "grad_norm": 2.2311315536499023,
      "learning_rate": 8.05596539972582e-05,
      "loss": 2.0849,
      "step": 918
    },
    {
      "epoch": 0.30268845979661574,
      "grad_norm": 3.3348686695098877,
      "learning_rate": 8.051800091285404e-05,
      "loss": 2.2344,
      "step": 919
    },
    {
      "epoch": 0.3030178269998765,
      "grad_norm": 2.89279842376709,
      "learning_rate": 8.047631404716374e-05,
      "loss": 2.5125,
      "step": 920
    },
    {
      "epoch": 0.3033471942031372,
      "grad_norm": 2.6193151473999023,
      "learning_rate": 8.043459344633173e-05,
      "loss": 1.7652,
      "step": 921
    },
    {
      "epoch": 0.303676561406398,
      "grad_norm": 2.6484382152557373,
      "learning_rate": 8.039283915653979e-05,
      "loss": 1.9555,
      "step": 922
    },
    {
      "epoch": 0.3040059286096587,
      "grad_norm": 2.553636312484741,
      "learning_rate": 8.035105122400701e-05,
      "loss": 1.8701,
      "step": 923
    },
    {
      "epoch": 0.3043352958129194,
      "grad_norm": 3.712156057357788,
      "learning_rate": 8.030922969498968e-05,
      "loss": 2.2537,
      "step": 924
    },
    {
      "epoch": 0.30466466301618017,
      "grad_norm": 3.272002696990967,
      "learning_rate": 8.026737461578132e-05,
      "loss": 2.1644,
      "step": 925
    },
    {
      "epoch": 0.3049940302194409,
      "grad_norm": 2.4612069129943848,
      "learning_rate": 8.022548603271252e-05,
      "loss": 2.4379,
      "step": 926
    },
    {
      "epoch": 0.30532339742270165,
      "grad_norm": 3.1458165645599365,
      "learning_rate": 8.018356399215104e-05,
      "loss": 2.6069,
      "step": 927
    },
    {
      "epoch": 0.30565276462596236,
      "grad_norm": 2.0101094245910645,
      "learning_rate": 8.014160854050164e-05,
      "loss": 2.185,
      "step": 928
    },
    {
      "epoch": 0.3059821318292231,
      "grad_norm": 2.1659107208251953,
      "learning_rate": 8.009961972420607e-05,
      "loss": 2.1568,
      "step": 929
    },
    {
      "epoch": 0.30631149903248384,
      "grad_norm": 2.243806838989258,
      "learning_rate": 8.005759758974296e-05,
      "loss": 2.4728,
      "step": 930
    },
    {
      "epoch": 0.3066408662357446,
      "grad_norm": 2.525348424911499,
      "learning_rate": 8.001554218362791e-05,
      "loss": 2.6265,
      "step": 931
    },
    {
      "epoch": 0.3069702334390053,
      "grad_norm": 2.3284335136413574,
      "learning_rate": 7.997345355241328e-05,
      "loss": 2.3675,
      "step": 932
    },
    {
      "epoch": 0.307299600642266,
      "grad_norm": 2.4215810298919678,
      "learning_rate": 7.993133174268826e-05,
      "loss": 2.1221,
      "step": 933
    },
    {
      "epoch": 0.3076289678455268,
      "grad_norm": 2.8312551975250244,
      "learning_rate": 7.988917680107871e-05,
      "loss": 2.5939,
      "step": 934
    },
    {
      "epoch": 0.3079583350487875,
      "grad_norm": 2.665449857711792,
      "learning_rate": 7.984698877424718e-05,
      "loss": 2.24,
      "step": 935
    },
    {
      "epoch": 0.30828770225204827,
      "grad_norm": 2.5410232543945312,
      "learning_rate": 7.980476770889289e-05,
      "loss": 2.2135,
      "step": 936
    },
    {
      "epoch": 0.308617069455309,
      "grad_norm": 2.556459426879883,
      "learning_rate": 7.976251365175158e-05,
      "loss": 2.2969,
      "step": 937
    },
    {
      "epoch": 0.30894643665856975,
      "grad_norm": 3.4073662757873535,
      "learning_rate": 7.972022664959554e-05,
      "loss": 2.3814,
      "step": 938
    },
    {
      "epoch": 0.30927580386183046,
      "grad_norm": 2.888021230697632,
      "learning_rate": 7.96779067492335e-05,
      "loss": 2.4569,
      "step": 939
    },
    {
      "epoch": 0.30960517106509117,
      "grad_norm": 2.819286584854126,
      "learning_rate": 7.963555399751063e-05,
      "loss": 2.1128,
      "step": 940
    },
    {
      "epoch": 0.30993453826835193,
      "grad_norm": 2.760483741760254,
      "learning_rate": 7.959316844130846e-05,
      "loss": 2.4026,
      "step": 941
    },
    {
      "epoch": 0.31026390547161264,
      "grad_norm": 2.6997532844543457,
      "learning_rate": 7.95507501275448e-05,
      "loss": 2.0134,
      "step": 942
    },
    {
      "epoch": 0.3105932726748734,
      "grad_norm": 2.7757112979888916,
      "learning_rate": 7.950829910317379e-05,
      "loss": 2.207,
      "step": 943
    },
    {
      "epoch": 0.3109226398781341,
      "grad_norm": 2.5524275302886963,
      "learning_rate": 7.946581541518569e-05,
      "loss": 1.7721,
      "step": 944
    },
    {
      "epoch": 0.3112520070813949,
      "grad_norm": 2.4421706199645996,
      "learning_rate": 7.942329911060703e-05,
      "loss": 2.1257,
      "step": 945
    },
    {
      "epoch": 0.3115813742846556,
      "grad_norm": 2.7249844074249268,
      "learning_rate": 7.938075023650029e-05,
      "loss": 1.9599,
      "step": 946
    },
    {
      "epoch": 0.31191074148791637,
      "grad_norm": 2.9589931964874268,
      "learning_rate": 7.933816883996415e-05,
      "loss": 2.3046,
      "step": 947
    },
    {
      "epoch": 0.3122401086911771,
      "grad_norm": 2.614107608795166,
      "learning_rate": 7.92955549681332e-05,
      "loss": 1.8874,
      "step": 948
    },
    {
      "epoch": 0.3125694758944378,
      "grad_norm": 3.2241110801696777,
      "learning_rate": 7.925290866817802e-05,
      "loss": 2.3943,
      "step": 949
    },
    {
      "epoch": 0.31289884309769855,
      "grad_norm": 3.045088291168213,
      "learning_rate": 7.921022998730507e-05,
      "loss": 1.961,
      "step": 950
    },
    {
      "epoch": 0.31322821030095926,
      "grad_norm": 2.1225154399871826,
      "learning_rate": 7.916751897275665e-05,
      "loss": 2.749,
      "step": 951
    },
    {
      "epoch": 0.31355757750422003,
      "grad_norm": 2.060499668121338,
      "learning_rate": 7.912477567181086e-05,
      "loss": 2.2708,
      "step": 952
    },
    {
      "epoch": 0.31388694470748074,
      "grad_norm": 2.4853827953338623,
      "learning_rate": 7.908200013178156e-05,
      "loss": 2.2059,
      "step": 953
    },
    {
      "epoch": 0.3142163119107415,
      "grad_norm": 2.317662477493286,
      "learning_rate": 7.903919240001824e-05,
      "loss": 2.4268,
      "step": 954
    },
    {
      "epoch": 0.3145456791140022,
      "grad_norm": 2.5555713176727295,
      "learning_rate": 7.899635252390606e-05,
      "loss": 2.4417,
      "step": 955
    },
    {
      "epoch": 0.314875046317263,
      "grad_norm": 2.6387650966644287,
      "learning_rate": 7.895348055086577e-05,
      "loss": 2.3655,
      "step": 956
    },
    {
      "epoch": 0.3152044135205237,
      "grad_norm": 2.4570939540863037,
      "learning_rate": 7.891057652835361e-05,
      "loss": 2.4014,
      "step": 957
    },
    {
      "epoch": 0.3155337807237844,
      "grad_norm": 2.4972894191741943,
      "learning_rate": 7.886764050386135e-05,
      "loss": 2.3593,
      "step": 958
    },
    {
      "epoch": 0.3158631479270452,
      "grad_norm": 2.601262092590332,
      "learning_rate": 7.882467252491617e-05,
      "loss": 2.4721,
      "step": 959
    },
    {
      "epoch": 0.3161925151303059,
      "grad_norm": 2.5913026332855225,
      "learning_rate": 7.878167263908056e-05,
      "loss": 2.2818,
      "step": 960
    },
    {
      "epoch": 0.31652188233356665,
      "grad_norm": 2.6688802242279053,
      "learning_rate": 7.873864089395243e-05,
      "loss": 2.1604,
      "step": 961
    },
    {
      "epoch": 0.31685124953682736,
      "grad_norm": 2.4998068809509277,
      "learning_rate": 7.869557733716488e-05,
      "loss": 1.95,
      "step": 962
    },
    {
      "epoch": 0.31718061674008813,
      "grad_norm": 2.5691847801208496,
      "learning_rate": 7.865248201638623e-05,
      "loss": 2.1274,
      "step": 963
    },
    {
      "epoch": 0.31750998394334884,
      "grad_norm": 2.2748234272003174,
      "learning_rate": 7.860935497932e-05,
      "loss": 1.9988,
      "step": 964
    },
    {
      "epoch": 0.31783935114660955,
      "grad_norm": 2.8164100646972656,
      "learning_rate": 7.856619627370479e-05,
      "loss": 2.3755,
      "step": 965
    },
    {
      "epoch": 0.3181687183498703,
      "grad_norm": 2.8128836154937744,
      "learning_rate": 7.852300594731425e-05,
      "loss": 2.1516,
      "step": 966
    },
    {
      "epoch": 0.318498085553131,
      "grad_norm": 2.7594070434570312,
      "learning_rate": 7.847978404795704e-05,
      "loss": 2.177,
      "step": 967
    },
    {
      "epoch": 0.3188274527563918,
      "grad_norm": 2.687586784362793,
      "learning_rate": 7.843653062347679e-05,
      "loss": 2.4126,
      "step": 968
    },
    {
      "epoch": 0.3191568199596525,
      "grad_norm": 2.610460042953491,
      "learning_rate": 7.8393245721752e-05,
      "loss": 2.3253,
      "step": 969
    },
    {
      "epoch": 0.31948618716291327,
      "grad_norm": 2.525682210922241,
      "learning_rate": 7.8349929390696e-05,
      "loss": 1.9292,
      "step": 970
    },
    {
      "epoch": 0.319815554366174,
      "grad_norm": 3.1786725521087646,
      "learning_rate": 7.830658167825696e-05,
      "loss": 1.956,
      "step": 971
    },
    {
      "epoch": 0.32014492156943475,
      "grad_norm": 2.809406280517578,
      "learning_rate": 7.826320263241771e-05,
      "loss": 2.1892,
      "step": 972
    },
    {
      "epoch": 0.32047428877269546,
      "grad_norm": 3.148411750793457,
      "learning_rate": 7.821979230119587e-05,
      "loss": 2.0155,
      "step": 973
    },
    {
      "epoch": 0.32080365597595617,
      "grad_norm": 3.3248815536499023,
      "learning_rate": 7.81763507326436e-05,
      "loss": 1.7539,
      "step": 974
    },
    {
      "epoch": 0.32113302317921694,
      "grad_norm": 3.2047150135040283,
      "learning_rate": 7.813287797484768e-05,
      "loss": 2.028,
      "step": 975
    },
    {
      "epoch": 0.32146239038247765,
      "grad_norm": 2.169050455093384,
      "learning_rate": 7.808937407592938e-05,
      "loss": 2.4696,
      "step": 976
    },
    {
      "epoch": 0.3217917575857384,
      "grad_norm": 2.0645432472229004,
      "learning_rate": 7.804583908404448e-05,
      "loss": 2.1029,
      "step": 977
    },
    {
      "epoch": 0.3221211247889991,
      "grad_norm": 2.496066093444824,
      "learning_rate": 7.800227304738317e-05,
      "loss": 2.3193,
      "step": 978
    },
    {
      "epoch": 0.3224504919922599,
      "grad_norm": 2.2872400283813477,
      "learning_rate": 7.795867601416998e-05,
      "loss": 2.4506,
      "step": 979
    },
    {
      "epoch": 0.3227798591955206,
      "grad_norm": 2.6444544792175293,
      "learning_rate": 7.791504803266377e-05,
      "loss": 2.818,
      "step": 980
    },
    {
      "epoch": 0.32310922639878137,
      "grad_norm": 2.2178843021392822,
      "learning_rate": 7.787138915115768e-05,
      "loss": 1.9535,
      "step": 981
    },
    {
      "epoch": 0.3234385936020421,
      "grad_norm": 2.528449535369873,
      "learning_rate": 7.782769941797899e-05,
      "loss": 2.1525,
      "step": 982
    },
    {
      "epoch": 0.3237679608053028,
      "grad_norm": 2.597266435623169,
      "learning_rate": 7.778397888148921e-05,
      "loss": 2.4295,
      "step": 983
    },
    {
      "epoch": 0.32409732800856356,
      "grad_norm": 2.5413691997528076,
      "learning_rate": 7.774022759008386e-05,
      "loss": 2.3796,
      "step": 984
    },
    {
      "epoch": 0.32442669521182427,
      "grad_norm": 2.5218312740325928,
      "learning_rate": 7.76964455921926e-05,
      "loss": 2.4834,
      "step": 985
    },
    {
      "epoch": 0.32475606241508503,
      "grad_norm": 2.420651435852051,
      "learning_rate": 7.7652632936279e-05,
      "loss": 2.1914,
      "step": 986
    },
    {
      "epoch": 0.32508542961834574,
      "grad_norm": 2.723442792892456,
      "learning_rate": 7.760878967084059e-05,
      "loss": 2.5173,
      "step": 987
    },
    {
      "epoch": 0.3254147968216065,
      "grad_norm": 2.400263786315918,
      "learning_rate": 7.756491584440882e-05,
      "loss": 2.1829,
      "step": 988
    },
    {
      "epoch": 0.3257441640248672,
      "grad_norm": 2.4888710975646973,
      "learning_rate": 7.75210115055489e-05,
      "loss": 2.1607,
      "step": 989
    },
    {
      "epoch": 0.32607353122812793,
      "grad_norm": 2.6729047298431396,
      "learning_rate": 7.747707670285989e-05,
      "loss": 2.3433,
      "step": 990
    },
    {
      "epoch": 0.3264028984313887,
      "grad_norm": 2.9809768199920654,
      "learning_rate": 7.743311148497452e-05,
      "loss": 2.3124,
      "step": 991
    },
    {
      "epoch": 0.3267322656346494,
      "grad_norm": 2.287365198135376,
      "learning_rate": 7.73891159005592e-05,
      "loss": 2.0576,
      "step": 992
    },
    {
      "epoch": 0.3270616328379102,
      "grad_norm": 2.4659500122070312,
      "learning_rate": 7.734508999831394e-05,
      "loss": 2.0228,
      "step": 993
    },
    {
      "epoch": 0.3273910000411709,
      "grad_norm": 2.66711163520813,
      "learning_rate": 7.730103382697236e-05,
      "loss": 2.1277,
      "step": 994
    },
    {
      "epoch": 0.32772036724443165,
      "grad_norm": 2.902860641479492,
      "learning_rate": 7.725694743530153e-05,
      "loss": 2.3815,
      "step": 995
    },
    {
      "epoch": 0.32804973444769236,
      "grad_norm": 2.8413193225860596,
      "learning_rate": 7.721283087210199e-05,
      "loss": 2.2505,
      "step": 996
    },
    {
      "epoch": 0.32837910165095313,
      "grad_norm": 3.3534953594207764,
      "learning_rate": 7.716868418620768e-05,
      "loss": 2.2123,
      "step": 997
    },
    {
      "epoch": 0.32870846885421384,
      "grad_norm": 2.6357085704803467,
      "learning_rate": 7.71245074264859e-05,
      "loss": 1.8754,
      "step": 998
    },
    {
      "epoch": 0.32903783605747455,
      "grad_norm": 3.168962001800537,
      "learning_rate": 7.70803006418372e-05,
      "loss": 1.9192,
      "step": 999
    },
    {
      "epoch": 0.3293672032607353,
      "grad_norm": 3.3432531356811523,
      "learning_rate": 7.703606388119542e-05,
      "loss": 1.7519,
      "step": 1000
    },
    {
      "epoch": 0.32969657046399603,
      "grad_norm": 1.857453465461731,
      "learning_rate": 7.699179719352752e-05,
      "loss": 2.5652,
      "step": 1001
    },
    {
      "epoch": 0.3300259376672568,
      "grad_norm": 1.9321167469024658,
      "learning_rate": 7.694750062783363e-05,
      "loss": 2.1067,
      "step": 1002
    },
    {
      "epoch": 0.3303553048705175,
      "grad_norm": 2.123394250869751,
      "learning_rate": 7.690317423314696e-05,
      "loss": 2.311,
      "step": 1003
    },
    {
      "epoch": 0.3306846720737783,
      "grad_norm": 2.345137357711792,
      "learning_rate": 7.685881805853369e-05,
      "loss": 2.1725,
      "step": 1004
    },
    {
      "epoch": 0.331014039277039,
      "grad_norm": 2.3156449794769287,
      "learning_rate": 7.6814432153093e-05,
      "loss": 2.3528,
      "step": 1005
    },
    {
      "epoch": 0.33134340648029975,
      "grad_norm": 2.4214887619018555,
      "learning_rate": 7.6770016565957e-05,
      "loss": 2.0471,
      "step": 1006
    },
    {
      "epoch": 0.33167277368356046,
      "grad_norm": 2.3343491554260254,
      "learning_rate": 7.672557134629059e-05,
      "loss": 2.1914,
      "step": 1007
    },
    {
      "epoch": 0.3320021408868212,
      "grad_norm": 2.713235378265381,
      "learning_rate": 7.668109654329154e-05,
      "loss": 2.4468,
      "step": 1008
    },
    {
      "epoch": 0.33233150809008194,
      "grad_norm": 3.0459039211273193,
      "learning_rate": 7.663659220619033e-05,
      "loss": 2.4859,
      "step": 1009
    },
    {
      "epoch": 0.33266087529334265,
      "grad_norm": 2.7493438720703125,
      "learning_rate": 7.659205838425013e-05,
      "loss": 2.4392,
      "step": 1010
    },
    {
      "epoch": 0.3329902424966034,
      "grad_norm": 2.3039515018463135,
      "learning_rate": 7.654749512676676e-05,
      "loss": 2.0348,
      "step": 1011
    },
    {
      "epoch": 0.33331960969986413,
      "grad_norm": 2.667750597000122,
      "learning_rate": 7.650290248306863e-05,
      "loss": 1.949,
      "step": 1012
    },
    {
      "epoch": 0.3336489769031249,
      "grad_norm": 2.9373788833618164,
      "learning_rate": 7.645828050251665e-05,
      "loss": 2.591,
      "step": 1013
    },
    {
      "epoch": 0.3339783441063856,
      "grad_norm": 3.202880620956421,
      "learning_rate": 7.641362923450424e-05,
      "loss": 2.5421,
      "step": 1014
    },
    {
      "epoch": 0.3343077113096463,
      "grad_norm": 2.4246268272399902,
      "learning_rate": 7.636894872845722e-05,
      "loss": 2.1465,
      "step": 1015
    },
    {
      "epoch": 0.3346370785129071,
      "grad_norm": 2.732734203338623,
      "learning_rate": 7.632423903383374e-05,
      "loss": 2.1029,
      "step": 1016
    },
    {
      "epoch": 0.3349664457161678,
      "grad_norm": 3.3760523796081543,
      "learning_rate": 7.627950020012434e-05,
      "loss": 2.0353,
      "step": 1017
    },
    {
      "epoch": 0.33529581291942856,
      "grad_norm": 3.0070247650146484,
      "learning_rate": 7.623473227685176e-05,
      "loss": 2.365,
      "step": 1018
    },
    {
      "epoch": 0.33562518012268927,
      "grad_norm": 2.8173720836639404,
      "learning_rate": 7.618993531357094e-05,
      "loss": 2.0214,
      "step": 1019
    },
    {
      "epoch": 0.33595454732595004,
      "grad_norm": 2.5577118396759033,
      "learning_rate": 7.614510935986898e-05,
      "loss": 1.8253,
      "step": 1020
    },
    {
      "epoch": 0.33628391452921075,
      "grad_norm": 2.8958218097686768,
      "learning_rate": 7.610025446536509e-05,
      "loss": 2.2123,
      "step": 1021
    },
    {
      "epoch": 0.3366132817324715,
      "grad_norm": 3.0846564769744873,
      "learning_rate": 7.605537067971045e-05,
      "loss": 2.0833,
      "step": 1022
    },
    {
      "epoch": 0.3369426489357322,
      "grad_norm": 3.0642452239990234,
      "learning_rate": 7.601045805258828e-05,
      "loss": 1.8784,
      "step": 1023
    },
    {
      "epoch": 0.33727201613899294,
      "grad_norm": 4.532246112823486,
      "learning_rate": 7.596551663371372e-05,
      "loss": 2.4327,
      "step": 1024
    },
    {
      "epoch": 0.3376013833422537,
      "grad_norm": 3.3755955696105957,
      "learning_rate": 7.592054647283375e-05,
      "loss": 1.7789,
      "step": 1025
    },
    {
      "epoch": 0.3379307505455144,
      "grad_norm": 1.943467140197754,
      "learning_rate": 7.587554761972718e-05,
      "loss": 2.3184,
      "step": 1026
    },
    {
      "epoch": 0.3382601177487752,
      "grad_norm": 2.520038604736328,
      "learning_rate": 7.583052012420461e-05,
      "loss": 2.2305,
      "step": 1027
    },
    {
      "epoch": 0.3385894849520359,
      "grad_norm": 2.3161160945892334,
      "learning_rate": 7.57854640361083e-05,
      "loss": 2.2228,
      "step": 1028
    },
    {
      "epoch": 0.33891885215529666,
      "grad_norm": 2.4692294597625732,
      "learning_rate": 7.574037940531218e-05,
      "loss": 2.251,
      "step": 1029
    },
    {
      "epoch": 0.33924821935855737,
      "grad_norm": 2.4752914905548096,
      "learning_rate": 7.569526628172177e-05,
      "loss": 2.2454,
      "step": 1030
    },
    {
      "epoch": 0.33957758656181813,
      "grad_norm": 2.6334524154663086,
      "learning_rate": 7.565012471527416e-05,
      "loss": 2.5675,
      "step": 1031
    },
    {
      "epoch": 0.33990695376507885,
      "grad_norm": 2.3727474212646484,
      "learning_rate": 7.560495475593785e-05,
      "loss": 2.2345,
      "step": 1032
    },
    {
      "epoch": 0.34023632096833956,
      "grad_norm": 2.6379683017730713,
      "learning_rate": 7.555975645371285e-05,
      "loss": 2.4907,
      "step": 1033
    },
    {
      "epoch": 0.3405656881716003,
      "grad_norm": 2.5622193813323975,
      "learning_rate": 7.55145298586305e-05,
      "loss": 2.3077,
      "step": 1034
    },
    {
      "epoch": 0.34089505537486103,
      "grad_norm": 2.888123035430908,
      "learning_rate": 7.546927502075348e-05,
      "loss": 2.2664,
      "step": 1035
    },
    {
      "epoch": 0.3412244225781218,
      "grad_norm": 2.4946439266204834,
      "learning_rate": 7.542399199017568e-05,
      "loss": 2.3239,
      "step": 1036
    },
    {
      "epoch": 0.3415537897813825,
      "grad_norm": 2.6942648887634277,
      "learning_rate": 7.53786808170223e-05,
      "loss": 2.501,
      "step": 1037
    },
    {
      "epoch": 0.3418831569846433,
      "grad_norm": 2.577432632446289,
      "learning_rate": 7.53333415514496e-05,
      "loss": 2.3515,
      "step": 1038
    },
    {
      "epoch": 0.342212524187904,
      "grad_norm": 2.2709529399871826,
      "learning_rate": 7.528797424364496e-05,
      "loss": 1.9878,
      "step": 1039
    },
    {
      "epoch": 0.3425418913911647,
      "grad_norm": 2.2214181423187256,
      "learning_rate": 7.524257894382681e-05,
      "loss": 1.8414,
      "step": 1040
    },
    {
      "epoch": 0.34287125859442547,
      "grad_norm": 2.8622872829437256,
      "learning_rate": 7.519715570224457e-05,
      "loss": 1.9751,
      "step": 1041
    },
    {
      "epoch": 0.3432006257976862,
      "grad_norm": 2.7308106422424316,
      "learning_rate": 7.515170456917857e-05,
      "loss": 2.1777,
      "step": 1042
    },
    {
      "epoch": 0.34352999300094694,
      "grad_norm": 3.008892297744751,
      "learning_rate": 7.510622559494002e-05,
      "loss": 2.1712,
      "step": 1043
    },
    {
      "epoch": 0.34385936020420765,
      "grad_norm": 2.538496732711792,
      "learning_rate": 7.5060718829871e-05,
      "loss": 1.5176,
      "step": 1044
    },
    {
      "epoch": 0.3441887274074684,
      "grad_norm": 2.9815316200256348,
      "learning_rate": 7.501518432434424e-05,
      "loss": 1.9269,
      "step": 1045
    },
    {
      "epoch": 0.34451809461072913,
      "grad_norm": 2.786571741104126,
      "learning_rate": 7.49696221287633e-05,
      "loss": 2.0934,
      "step": 1046
    },
    {
      "epoch": 0.3448474618139899,
      "grad_norm": 2.8320257663726807,
      "learning_rate": 7.49240322935623e-05,
      "loss": 1.7772,
      "step": 1047
    },
    {
      "epoch": 0.3451768290172506,
      "grad_norm": 3.3367862701416016,
      "learning_rate": 7.487841486920599e-05,
      "loss": 2.2513,
      "step": 1048
    },
    {
      "epoch": 0.3455061962205113,
      "grad_norm": 3.6575772762298584,
      "learning_rate": 7.48327699061897e-05,
      "loss": 2.2954,
      "step": 1049
    },
    {
      "epoch": 0.3458355634237721,
      "grad_norm": 3.485363721847534,
      "learning_rate": 7.478709745503913e-05,
      "loss": 1.9509,
      "step": 1050
    },
    {
      "epoch": 0.3461649306270328,
      "grad_norm": 2.551144599914551,
      "learning_rate": 7.474139756631056e-05,
      "loss": 2.5046,
      "step": 1051
    },
    {
      "epoch": 0.34649429783029356,
      "grad_norm": 2.3824918270111084,
      "learning_rate": 7.46956702905905e-05,
      "loss": 2.3288,
      "step": 1052
    },
    {
      "epoch": 0.3468236650335543,
      "grad_norm": 2.3063457012176514,
      "learning_rate": 7.464991567849586e-05,
      "loss": 2.2688,
      "step": 1053
    },
    {
      "epoch": 0.34715303223681504,
      "grad_norm": 2.4683382511138916,
      "learning_rate": 7.460413378067379e-05,
      "loss": 2.4154,
      "step": 1054
    },
    {
      "epoch": 0.34748239944007575,
      "grad_norm": 2.572463035583496,
      "learning_rate": 7.455832464780162e-05,
      "loss": 2.4033,
      "step": 1055
    },
    {
      "epoch": 0.3478117666433365,
      "grad_norm": 2.575000762939453,
      "learning_rate": 7.451248833058687e-05,
      "loss": 2.2997,
      "step": 1056
    },
    {
      "epoch": 0.34814113384659723,
      "grad_norm": 2.226249933242798,
      "learning_rate": 7.446662487976713e-05,
      "loss": 2.1283,
      "step": 1057
    },
    {
      "epoch": 0.34847050104985794,
      "grad_norm": 2.5053114891052246,
      "learning_rate": 7.442073434610997e-05,
      "loss": 2.1099,
      "step": 1058
    },
    {
      "epoch": 0.3487998682531187,
      "grad_norm": 2.71102237701416,
      "learning_rate": 7.437481678041307e-05,
      "loss": 2.3652,
      "step": 1059
    },
    {
      "epoch": 0.3491292354563794,
      "grad_norm": 3.036839723587036,
      "learning_rate": 7.43288722335039e-05,
      "loss": 1.9926,
      "step": 1060
    },
    {
      "epoch": 0.3494586026596402,
      "grad_norm": 2.7319066524505615,
      "learning_rate": 7.428290075623987e-05,
      "loss": 2.5434,
      "step": 1061
    },
    {
      "epoch": 0.3497879698629009,
      "grad_norm": 2.671752691268921,
      "learning_rate": 7.423690239950818e-05,
      "loss": 2.1477,
      "step": 1062
    },
    {
      "epoch": 0.35011733706616166,
      "grad_norm": 3.032055616378784,
      "learning_rate": 7.419087721422576e-05,
      "loss": 2.3655,
      "step": 1063
    },
    {
      "epoch": 0.35044670426942237,
      "grad_norm": 2.8573625087738037,
      "learning_rate": 7.414482525133928e-05,
      "loss": 2.6337,
      "step": 1064
    },
    {
      "epoch": 0.3507760714726831,
      "grad_norm": 2.5970046520233154,
      "learning_rate": 7.409874656182506e-05,
      "loss": 1.9232,
      "step": 1065
    },
    {
      "epoch": 0.35110543867594385,
      "grad_norm": 2.5858652591705322,
      "learning_rate": 7.405264119668894e-05,
      "loss": 2.4204,
      "step": 1066
    },
    {
      "epoch": 0.35143480587920456,
      "grad_norm": 3.2637603282928467,
      "learning_rate": 7.400650920696633e-05,
      "loss": 2.6369,
      "step": 1067
    },
    {
      "epoch": 0.3517641730824653,
      "grad_norm": 2.568941831588745,
      "learning_rate": 7.396035064372214e-05,
      "loss": 2.2192,
      "step": 1068
    },
    {
      "epoch": 0.35209354028572604,
      "grad_norm": 3.0660107135772705,
      "learning_rate": 7.39141655580506e-05,
      "loss": 2.1387,
      "step": 1069
    },
    {
      "epoch": 0.3524229074889868,
      "grad_norm": 3.537260055541992,
      "learning_rate": 7.386795400107539e-05,
      "loss": 2.1265,
      "step": 1070
    },
    {
      "epoch": 0.3527522746922475,
      "grad_norm": 2.7157890796661377,
      "learning_rate": 7.382171602394948e-05,
      "loss": 1.7636,
      "step": 1071
    },
    {
      "epoch": 0.3530816418955083,
      "grad_norm": 3.227700710296631,
      "learning_rate": 7.377545167785506e-05,
      "loss": 2.1963,
      "step": 1072
    },
    {
      "epoch": 0.353411009098769,
      "grad_norm": 2.9469199180603027,
      "learning_rate": 7.372916101400349e-05,
      "loss": 1.8702,
      "step": 1073
    },
    {
      "epoch": 0.3537403763020297,
      "grad_norm": 2.9933621883392334,
      "learning_rate": 7.368284408363531e-05,
      "loss": 1.6875,
      "step": 1074
    },
    {
      "epoch": 0.35406974350529047,
      "grad_norm": 3.143301248550415,
      "learning_rate": 7.363650093802012e-05,
      "loss": 1.8475,
      "step": 1075
    },
    {
      "epoch": 0.3543991107085512,
      "grad_norm": 1.9010746479034424,
      "learning_rate": 7.35901316284565e-05,
      "loss": 2.4547,
      "step": 1076
    },
    {
      "epoch": 0.35472847791181195,
      "grad_norm": 1.9302737712860107,
      "learning_rate": 7.354373620627205e-05,
      "loss": 2.2665,
      "step": 1077
    },
    {
      "epoch": 0.35505784511507266,
      "grad_norm": 2.267845392227173,
      "learning_rate": 7.349731472282325e-05,
      "loss": 2.3649,
      "step": 1078
    },
    {
      "epoch": 0.3553872123183334,
      "grad_norm": 2.5052695274353027,
      "learning_rate": 7.345086722949539e-05,
      "loss": 2.3261,
      "step": 1079
    },
    {
      "epoch": 0.35571657952159413,
      "grad_norm": 2.4212067127227783,
      "learning_rate": 7.340439377770263e-05,
      "loss": 2.1668,
      "step": 1080
    },
    {
      "epoch": 0.35604594672485484,
      "grad_norm": 2.1286873817443848,
      "learning_rate": 7.335789441888781e-05,
      "loss": 2.1269,
      "step": 1081
    },
    {
      "epoch": 0.3563753139281156,
      "grad_norm": 2.6201021671295166,
      "learning_rate": 7.331136920452244e-05,
      "loss": 2.498,
      "step": 1082
    },
    {
      "epoch": 0.3567046811313763,
      "grad_norm": 2.8522377014160156,
      "learning_rate": 7.326481818610668e-05,
      "loss": 2.2456,
      "step": 1083
    },
    {
      "epoch": 0.3570340483346371,
      "grad_norm": 2.749220609664917,
      "learning_rate": 7.321824141516926e-05,
      "loss": 2.2296,
      "step": 1084
    },
    {
      "epoch": 0.3573634155378978,
      "grad_norm": 2.6423492431640625,
      "learning_rate": 7.317163894326735e-05,
      "loss": 2.0497,
      "step": 1085
    },
    {
      "epoch": 0.35769278274115857,
      "grad_norm": 2.9978692531585693,
      "learning_rate": 7.312501082198666e-05,
      "loss": 2.4605,
      "step": 1086
    },
    {
      "epoch": 0.3580221499444193,
      "grad_norm": 2.5553252696990967,
      "learning_rate": 7.307835710294125e-05,
      "loss": 1.9969,
      "step": 1087
    },
    {
      "epoch": 0.35835151714768004,
      "grad_norm": 3.3244998455047607,
      "learning_rate": 7.303167783777349e-05,
      "loss": 2.2522,
      "step": 1088
    },
    {
      "epoch": 0.35868088435094075,
      "grad_norm": 2.7964744567871094,
      "learning_rate": 7.298497307815406e-05,
      "loss": 2.2721,
      "step": 1089
    },
    {
      "epoch": 0.35901025155420146,
      "grad_norm": 2.972792625427246,
      "learning_rate": 7.293824287578185e-05,
      "loss": 2.2599,
      "step": 1090
    },
    {
      "epoch": 0.35933961875746223,
      "grad_norm": 2.953064441680908,
      "learning_rate": 7.289148728238392e-05,
      "loss": 2.4102,
      "step": 1091
    },
    {
      "epoch": 0.35966898596072294,
      "grad_norm": 3.144014835357666,
      "learning_rate": 7.284470634971544e-05,
      "loss": 2.4154,
      "step": 1092
    },
    {
      "epoch": 0.3599983531639837,
      "grad_norm": 2.5227506160736084,
      "learning_rate": 7.279790012955961e-05,
      "loss": 1.9305,
      "step": 1093
    },
    {
      "epoch": 0.3603277203672444,
      "grad_norm": 3.1171419620513916,
      "learning_rate": 7.275106867372762e-05,
      "loss": 2.2716,
      "step": 1094
    },
    {
      "epoch": 0.3606570875705052,
      "grad_norm": 2.6330418586730957,
      "learning_rate": 7.270421203405863e-05,
      "loss": 1.8731,
      "step": 1095
    },
    {
      "epoch": 0.3609864547737659,
      "grad_norm": 2.5206286907196045,
      "learning_rate": 7.265733026241966e-05,
      "loss": 1.8349,
      "step": 1096
    },
    {
      "epoch": 0.36131582197702666,
      "grad_norm": 2.8334505558013916,
      "learning_rate": 7.261042341070552e-05,
      "loss": 1.4885,
      "step": 1097
    },
    {
      "epoch": 0.3616451891802874,
      "grad_norm": 3.205961227416992,
      "learning_rate": 7.256349153083881e-05,
      "loss": 1.9328,
      "step": 1098
    },
    {
      "epoch": 0.3619745563835481,
      "grad_norm": 3.352240562438965,
      "learning_rate": 7.251653467476983e-05,
      "loss": 2.0997,
      "step": 1099
    },
    {
      "epoch": 0.36230392358680885,
      "grad_norm": 3.4709959030151367,
      "learning_rate": 7.246955289447653e-05,
      "loss": 1.7248,
      "step": 1100
    },
    {
      "epoch": 0.36263329079006956,
      "grad_norm": 2.071342945098877,
      "learning_rate": 7.242254624196443e-05,
      "loss": 2.2652,
      "step": 1101
    },
    {
      "epoch": 0.36296265799333033,
      "grad_norm": 2.2652060985565186,
      "learning_rate": 7.237551476926661e-05,
      "loss": 2.3592,
      "step": 1102
    },
    {
      "epoch": 0.36329202519659104,
      "grad_norm": 2.8345634937286377,
      "learning_rate": 7.232845852844361e-05,
      "loss": 2.6346,
      "step": 1103
    },
    {
      "epoch": 0.3636213923998518,
      "grad_norm": 2.3296942710876465,
      "learning_rate": 7.228137757158338e-05,
      "loss": 2.0973,
      "step": 1104
    },
    {
      "epoch": 0.3639507596031125,
      "grad_norm": 2.4769644737243652,
      "learning_rate": 7.223427195080126e-05,
      "loss": 2.2648,
      "step": 1105
    },
    {
      "epoch": 0.36428012680637323,
      "grad_norm": 2.285839080810547,
      "learning_rate": 7.218714171823984e-05,
      "loss": 2.1648,
      "step": 1106
    },
    {
      "epoch": 0.364609494009634,
      "grad_norm": 2.404283046722412,
      "learning_rate": 7.2139986926069e-05,
      "loss": 2.3887,
      "step": 1107
    },
    {
      "epoch": 0.3649388612128947,
      "grad_norm": 2.4443814754486084,
      "learning_rate": 7.209280762648576e-05,
      "loss": 2.227,
      "step": 1108
    },
    {
      "epoch": 0.36526822841615547,
      "grad_norm": 2.4258129596710205,
      "learning_rate": 7.204560387171432e-05,
      "loss": 2.2972,
      "step": 1109
    },
    {
      "epoch": 0.3655975956194162,
      "grad_norm": 2.630659341812134,
      "learning_rate": 7.199837571400591e-05,
      "loss": 2.3323,
      "step": 1110
    },
    {
      "epoch": 0.36592696282267695,
      "grad_norm": 2.332148313522339,
      "learning_rate": 7.195112320563881e-05,
      "loss": 2.0955,
      "step": 1111
    },
    {
      "epoch": 0.36625633002593766,
      "grad_norm": 2.656656503677368,
      "learning_rate": 7.190384639891822e-05,
      "loss": 2.016,
      "step": 1112
    },
    {
      "epoch": 0.3665856972291984,
      "grad_norm": 2.663341760635376,
      "learning_rate": 7.185654534617623e-05,
      "loss": 2.3012,
      "step": 1113
    },
    {
      "epoch": 0.36691506443245914,
      "grad_norm": 2.5622754096984863,
      "learning_rate": 7.180922009977181e-05,
      "loss": 1.9749,
      "step": 1114
    },
    {
      "epoch": 0.36724443163571985,
      "grad_norm": 2.865852117538452,
      "learning_rate": 7.176187071209069e-05,
      "loss": 1.648,
      "step": 1115
    },
    {
      "epoch": 0.3675737988389806,
      "grad_norm": 2.8726089000701904,
      "learning_rate": 7.171449723554531e-05,
      "loss": 2.0493,
      "step": 1116
    },
    {
      "epoch": 0.3679031660422413,
      "grad_norm": 2.720703125,
      "learning_rate": 7.166709972257478e-05,
      "loss": 2.3402,
      "step": 1117
    },
    {
      "epoch": 0.3682325332455021,
      "grad_norm": 3.307445764541626,
      "learning_rate": 7.161967822564483e-05,
      "loss": 1.8724,
      "step": 1118
    },
    {
      "epoch": 0.3685619004487628,
      "grad_norm": 3.013970375061035,
      "learning_rate": 7.157223279724775e-05,
      "loss": 2.1087,
      "step": 1119
    },
    {
      "epoch": 0.36889126765202357,
      "grad_norm": 2.9190618991851807,
      "learning_rate": 7.152476348990224e-05,
      "loss": 2.1204,
      "step": 1120
    },
    {
      "epoch": 0.3692206348552843,
      "grad_norm": 2.860560178756714,
      "learning_rate": 7.147727035615355e-05,
      "loss": 2.0234,
      "step": 1121
    },
    {
      "epoch": 0.36955000205854505,
      "grad_norm": 3.1333353519439697,
      "learning_rate": 7.142975344857325e-05,
      "loss": 2.1354,
      "step": 1122
    },
    {
      "epoch": 0.36987936926180576,
      "grad_norm": 3.1250181198120117,
      "learning_rate": 7.138221281975919e-05,
      "loss": 2.2142,
      "step": 1123
    },
    {
      "epoch": 0.37020873646506647,
      "grad_norm": 3.049356460571289,
      "learning_rate": 7.133464852233553e-05,
      "loss": 1.9174,
      "step": 1124
    },
    {
      "epoch": 0.37053810366832723,
      "grad_norm": 3.432244062423706,
      "learning_rate": 7.12870606089526e-05,
      "loss": 1.4808,
      "step": 1125
    },
    {
      "epoch": 0.37086747087158795,
      "grad_norm": 1.7517294883728027,
      "learning_rate": 7.123944913228688e-05,
      "loss": 2.5009,
      "step": 1126
    },
    {
      "epoch": 0.3711968380748487,
      "grad_norm": 2.627819299697876,
      "learning_rate": 7.119181414504095e-05,
      "loss": 2.4364,
      "step": 1127
    },
    {
      "epoch": 0.3715262052781094,
      "grad_norm": 2.8850955963134766,
      "learning_rate": 7.11441556999434e-05,
      "loss": 2.4496,
      "step": 1128
    },
    {
      "epoch": 0.3718555724813702,
      "grad_norm": 2.320338726043701,
      "learning_rate": 7.109647384974876e-05,
      "loss": 1.9873,
      "step": 1129
    },
    {
      "epoch": 0.3721849396846309,
      "grad_norm": 2.7479662895202637,
      "learning_rate": 7.104876864723751e-05,
      "loss": 2.4404,
      "step": 1130
    },
    {
      "epoch": 0.3725143068878916,
      "grad_norm": 2.618581533432007,
      "learning_rate": 7.100104014521598e-05,
      "loss": 2.0846,
      "step": 1131
    },
    {
      "epoch": 0.3728436740911524,
      "grad_norm": 3.1042213439941406,
      "learning_rate": 7.095328839651625e-05,
      "loss": 2.6769,
      "step": 1132
    },
    {
      "epoch": 0.3731730412944131,
      "grad_norm": 2.402482271194458,
      "learning_rate": 7.090551345399616e-05,
      "loss": 2.1385,
      "step": 1133
    },
    {
      "epoch": 0.37350240849767385,
      "grad_norm": 2.716562271118164,
      "learning_rate": 7.085771537053923e-05,
      "loss": 2.2691,
      "step": 1134
    },
    {
      "epoch": 0.37383177570093457,
      "grad_norm": 2.5292835235595703,
      "learning_rate": 7.080989419905456e-05,
      "loss": 2.0973,
      "step": 1135
    },
    {
      "epoch": 0.37416114290419533,
      "grad_norm": 2.4309535026550293,
      "learning_rate": 7.076204999247686e-05,
      "loss": 2.2923,
      "step": 1136
    },
    {
      "epoch": 0.37449051010745604,
      "grad_norm": 2.3152859210968018,
      "learning_rate": 7.071418280376629e-05,
      "loss": 1.898,
      "step": 1137
    },
    {
      "epoch": 0.3748198773107168,
      "grad_norm": 2.7684056758880615,
      "learning_rate": 7.06662926859085e-05,
      "loss": 2.192,
      "step": 1138
    },
    {
      "epoch": 0.3751492445139775,
      "grad_norm": 2.886624813079834,
      "learning_rate": 7.061837969191445e-05,
      "loss": 2.2443,
      "step": 1139
    },
    {
      "epoch": 0.37547861171723823,
      "grad_norm": 2.6988983154296875,
      "learning_rate": 7.05704438748205e-05,
      "loss": 2.0046,
      "step": 1140
    },
    {
      "epoch": 0.375807978920499,
      "grad_norm": 2.849220037460327,
      "learning_rate": 7.05224852876882e-05,
      "loss": 2.2765,
      "step": 1141
    },
    {
      "epoch": 0.3761373461237597,
      "grad_norm": 2.494020462036133,
      "learning_rate": 7.047450398360438e-05,
      "loss": 2.1473,
      "step": 1142
    },
    {
      "epoch": 0.3764667133270205,
      "grad_norm": 2.90225887298584,
      "learning_rate": 7.042650001568097e-05,
      "loss": 2.3484,
      "step": 1143
    },
    {
      "epoch": 0.3767960805302812,
      "grad_norm": 2.873544692993164,
      "learning_rate": 7.037847343705496e-05,
      "loss": 2.0771,
      "step": 1144
    },
    {
      "epoch": 0.37712544773354195,
      "grad_norm": 2.2210004329681396,
      "learning_rate": 7.033042430088844e-05,
      "loss": 1.6013,
      "step": 1145
    },
    {
      "epoch": 0.37745481493680266,
      "grad_norm": 2.627661943435669,
      "learning_rate": 7.028235266036841e-05,
      "loss": 2.0004,
      "step": 1146
    },
    {
      "epoch": 0.37778418214006343,
      "grad_norm": 2.7000410556793213,
      "learning_rate": 7.023425856870683e-05,
      "loss": 1.8592,
      "step": 1147
    },
    {
      "epoch": 0.37811354934332414,
      "grad_norm": 3.242762327194214,
      "learning_rate": 7.018614207914047e-05,
      "loss": 2.1389,
      "step": 1148
    },
    {
      "epoch": 0.37844291654658485,
      "grad_norm": 3.2626426219940186,
      "learning_rate": 7.013800324493089e-05,
      "loss": 2.0661,
      "step": 1149
    },
    {
      "epoch": 0.3787722837498456,
      "grad_norm": 4.664367198944092,
      "learning_rate": 7.008984211936446e-05,
      "loss": 2.0999,
      "step": 1150
    },
    {
      "epoch": 0.37910165095310633,
      "grad_norm": 1.8425556421279907,
      "learning_rate": 7.00416587557521e-05,
      "loss": 2.4092,
      "step": 1151
    },
    {
      "epoch": 0.3794310181563671,
      "grad_norm": 2.1894047260284424,
      "learning_rate": 6.999345320742945e-05,
      "loss": 2.4542,
      "step": 1152
    },
    {
      "epoch": 0.3797603853596278,
      "grad_norm": 2.095588207244873,
      "learning_rate": 6.994522552775666e-05,
      "loss": 2.1205,
      "step": 1153
    },
    {
      "epoch": 0.38008975256288857,
      "grad_norm": 2.427074909210205,
      "learning_rate": 6.98969757701184e-05,
      "loss": 2.2412,
      "step": 1154
    },
    {
      "epoch": 0.3804191197661493,
      "grad_norm": 2.322599411010742,
      "learning_rate": 6.984870398792374e-05,
      "loss": 2.0839,
      "step": 1155
    },
    {
      "epoch": 0.38074848696941,
      "grad_norm": 2.316253900527954,
      "learning_rate": 6.980041023460619e-05,
      "loss": 2.2265,
      "step": 1156
    },
    {
      "epoch": 0.38107785417267076,
      "grad_norm": 2.33427095413208,
      "learning_rate": 6.975209456362353e-05,
      "loss": 2.3765,
      "step": 1157
    },
    {
      "epoch": 0.38140722137593147,
      "grad_norm": 2.23207426071167,
      "learning_rate": 6.97037570284578e-05,
      "loss": 1.9018,
      "step": 1158
    },
    {
      "epoch": 0.38173658857919224,
      "grad_norm": 2.3566267490386963,
      "learning_rate": 6.965539768261531e-05,
      "loss": 2.1916,
      "step": 1159
    },
    {
      "epoch": 0.38206595578245295,
      "grad_norm": 2.377549409866333,
      "learning_rate": 6.960701657962641e-05,
      "loss": 2.4266,
      "step": 1160
    },
    {
      "epoch": 0.3823953229857137,
      "grad_norm": 2.494011163711548,
      "learning_rate": 6.955861377304564e-05,
      "loss": 2.1707,
      "step": 1161
    },
    {
      "epoch": 0.3827246901889744,
      "grad_norm": 2.6809515953063965,
      "learning_rate": 6.951018931645146e-05,
      "loss": 2.3089,
      "step": 1162
    },
    {
      "epoch": 0.3830540573922352,
      "grad_norm": 2.6300032138824463,
      "learning_rate": 6.946174326344637e-05,
      "loss": 2.1554,
      "step": 1163
    },
    {
      "epoch": 0.3833834245954959,
      "grad_norm": 2.4809701442718506,
      "learning_rate": 6.941327566765675e-05,
      "loss": 2.1591,
      "step": 1164
    },
    {
      "epoch": 0.3837127917987566,
      "grad_norm": 2.6750636100769043,
      "learning_rate": 6.936478658273285e-05,
      "loss": 2.3151,
      "step": 1165
    },
    {
      "epoch": 0.3840421590020174,
      "grad_norm": 2.421334981918335,
      "learning_rate": 6.931627606234865e-05,
      "loss": 1.9933,
      "step": 1166
    },
    {
      "epoch": 0.3843715262052781,
      "grad_norm": 2.758087158203125,
      "learning_rate": 6.92677441602019e-05,
      "loss": 2.2134,
      "step": 1167
    },
    {
      "epoch": 0.38470089340853886,
      "grad_norm": 3.3581409454345703,
      "learning_rate": 6.921919093001402e-05,
      "loss": 2.3097,
      "step": 1168
    },
    {
      "epoch": 0.38503026061179957,
      "grad_norm": 3.3593056201934814,
      "learning_rate": 6.917061642553005e-05,
      "loss": 1.9195,
      "step": 1169
    },
    {
      "epoch": 0.38535962781506033,
      "grad_norm": 2.783876895904541,
      "learning_rate": 6.91220207005185e-05,
      "loss": 1.8634,
      "step": 1170
    },
    {
      "epoch": 0.38568899501832105,
      "grad_norm": 2.8132247924804688,
      "learning_rate": 6.907340380877149e-05,
      "loss": 2.1477,
      "step": 1171
    },
    {
      "epoch": 0.3860183622215818,
      "grad_norm": 2.878632068634033,
      "learning_rate": 6.902476580410449e-05,
      "loss": 2.254,
      "step": 1172
    },
    {
      "epoch": 0.3863477294248425,
      "grad_norm": 3.4279277324676514,
      "learning_rate": 6.897610674035634e-05,
      "loss": 1.95,
      "step": 1173
    },
    {
      "epoch": 0.38667709662810323,
      "grad_norm": 3.4778835773468018,
      "learning_rate": 6.892742667138923e-05,
      "loss": 1.8198,
      "step": 1174
    },
    {
      "epoch": 0.387006463831364,
      "grad_norm": 3.2819323539733887,
      "learning_rate": 6.887872565108859e-05,
      "loss": 1.8538,
      "step": 1175
    },
    {
      "epoch": 0.3873358310346247,
      "grad_norm": 1.8319649696350098,
      "learning_rate": 6.883000373336299e-05,
      "loss": 2.4531,
      "step": 1176
    },
    {
      "epoch": 0.3876651982378855,
      "grad_norm": 2.054316520690918,
      "learning_rate": 6.878126097214421e-05,
      "loss": 1.9968,
      "step": 1177
    },
    {
      "epoch": 0.3879945654411462,
      "grad_norm": 2.3420443534851074,
      "learning_rate": 6.873249742138709e-05,
      "loss": 2.2451,
      "step": 1178
    },
    {
      "epoch": 0.38832393264440696,
      "grad_norm": 1.8880903720855713,
      "learning_rate": 6.868371313506941e-05,
      "loss": 2.3786,
      "step": 1179
    },
    {
      "epoch": 0.38865329984766767,
      "grad_norm": 2.393928289413452,
      "learning_rate": 6.863490816719196e-05,
      "loss": 2.1902,
      "step": 1180
    },
    {
      "epoch": 0.3889826670509284,
      "grad_norm": 2.3493289947509766,
      "learning_rate": 6.858608257177846e-05,
      "loss": 2.053,
      "step": 1181
    },
    {
      "epoch": 0.38931203425418914,
      "grad_norm": 2.4995625019073486,
      "learning_rate": 6.853723640287535e-05,
      "loss": 2.3704,
      "step": 1182
    },
    {
      "epoch": 0.38964140145744985,
      "grad_norm": 2.5445573329925537,
      "learning_rate": 6.848836971455197e-05,
      "loss": 1.9575,
      "step": 1183
    },
    {
      "epoch": 0.3899707686607106,
      "grad_norm": 2.5418543815612793,
      "learning_rate": 6.84394825609003e-05,
      "loss": 2.4032,
      "step": 1184
    },
    {
      "epoch": 0.39030013586397133,
      "grad_norm": 3.205739736557007,
      "learning_rate": 6.839057499603497e-05,
      "loss": 2.6626,
      "step": 1185
    },
    {
      "epoch": 0.3906295030672321,
      "grad_norm": 2.529710531234741,
      "learning_rate": 6.834164707409326e-05,
      "loss": 2.1196,
      "step": 1186
    },
    {
      "epoch": 0.3909588702704928,
      "grad_norm": 2.9174511432647705,
      "learning_rate": 6.829269884923491e-05,
      "loss": 2.4384,
      "step": 1187
    },
    {
      "epoch": 0.3912882374737536,
      "grad_norm": 2.5731706619262695,
      "learning_rate": 6.82437303756422e-05,
      "loss": 2.21,
      "step": 1188
    },
    {
      "epoch": 0.3916176046770143,
      "grad_norm": 3.202188014984131,
      "learning_rate": 6.819474170751978e-05,
      "loss": 2.5139,
      "step": 1189
    },
    {
      "epoch": 0.391946971880275,
      "grad_norm": 2.6559245586395264,
      "learning_rate": 6.814573289909466e-05,
      "loss": 2.3044,
      "step": 1190
    },
    {
      "epoch": 0.39227633908353576,
      "grad_norm": 3.1062824726104736,
      "learning_rate": 6.809670400461618e-05,
      "loss": 2.0021,
      "step": 1191
    },
    {
      "epoch": 0.3926057062867965,
      "grad_norm": 2.513209342956543,
      "learning_rate": 6.804765507835587e-05,
      "loss": 1.8991,
      "step": 1192
    },
    {
      "epoch": 0.39293507349005724,
      "grad_norm": 2.8009233474731445,
      "learning_rate": 6.799858617460744e-05,
      "loss": 2.3419,
      "step": 1193
    },
    {
      "epoch": 0.39326444069331795,
      "grad_norm": 2.8235421180725098,
      "learning_rate": 6.794949734768674e-05,
      "loss": 2.2565,
      "step": 1194
    },
    {
      "epoch": 0.3935938078965787,
      "grad_norm": 2.455665111541748,
      "learning_rate": 6.790038865193167e-05,
      "loss": 1.9415,
      "step": 1195
    },
    {
      "epoch": 0.39392317509983943,
      "grad_norm": 2.5987555980682373,
      "learning_rate": 6.785126014170207e-05,
      "loss": 1.8636,
      "step": 1196
    },
    {
      "epoch": 0.3942525423031002,
      "grad_norm": 2.7285478115081787,
      "learning_rate": 6.780211187137981e-05,
      "loss": 2.0818,
      "step": 1197
    },
    {
      "epoch": 0.3945819095063609,
      "grad_norm": 3.124101161956787,
      "learning_rate": 6.775294389536852e-05,
      "loss": 2.1855,
      "step": 1198
    },
    {
      "epoch": 0.3949112767096216,
      "grad_norm": 2.466757297515869,
      "learning_rate": 6.770375626809373e-05,
      "loss": 1.6696,
      "step": 1199
    },
    {
      "epoch": 0.3952406439128824,
      "grad_norm": 3.0805959701538086,
      "learning_rate": 6.76545490440027e-05,
      "loss": 2.1575,
      "step": 1200
    },
    {
      "epoch": 0.3955700111161431,
      "grad_norm": 2.310868978500366,
      "learning_rate": 6.760532227756435e-05,
      "loss": 2.3745,
      "step": 1201
    },
    {
      "epoch": 0.39589937831940386,
      "grad_norm": 2.0340662002563477,
      "learning_rate": 6.755607602326928e-05,
      "loss": 2.2478,
      "step": 1202
    },
    {
      "epoch": 0.39622874552266457,
      "grad_norm": 2.500797748565674,
      "learning_rate": 6.750681033562964e-05,
      "loss": 2.593,
      "step": 1203
    },
    {
      "epoch": 0.39655811272592534,
      "grad_norm": 2.07173228263855,
      "learning_rate": 6.745752526917907e-05,
      "loss": 2.3154,
      "step": 1204
    },
    {
      "epoch": 0.39688747992918605,
      "grad_norm": 2.0684971809387207,
      "learning_rate": 6.74082208784727e-05,
      "loss": 2.0726,
      "step": 1205
    },
    {
      "epoch": 0.39721684713244676,
      "grad_norm": 2.2718148231506348,
      "learning_rate": 6.735889721808703e-05,
      "loss": 2.2801,
      "step": 1206
    },
    {
      "epoch": 0.3975462143357075,
      "grad_norm": 2.2261784076690674,
      "learning_rate": 6.730955434261986e-05,
      "loss": 2.1202,
      "step": 1207
    },
    {
      "epoch": 0.39787558153896824,
      "grad_norm": 2.9758946895599365,
      "learning_rate": 6.726019230669034e-05,
      "loss": 2.4945,
      "step": 1208
    },
    {
      "epoch": 0.398204948742229,
      "grad_norm": 2.5161709785461426,
      "learning_rate": 6.721081116493874e-05,
      "loss": 2.1904,
      "step": 1209
    },
    {
      "epoch": 0.3985343159454897,
      "grad_norm": 2.922379493713379,
      "learning_rate": 6.716141097202657e-05,
      "loss": 2.5583,
      "step": 1210
    },
    {
      "epoch": 0.3988636831487505,
      "grad_norm": 2.4563610553741455,
      "learning_rate": 6.711199178263632e-05,
      "loss": 2.355,
      "step": 1211
    },
    {
      "epoch": 0.3991930503520112,
      "grad_norm": 3.078686237335205,
      "learning_rate": 6.706255365147161e-05,
      "loss": 2.604,
      "step": 1212
    },
    {
      "epoch": 0.39952241755527196,
      "grad_norm": 2.5945472717285156,
      "learning_rate": 6.701309663325696e-05,
      "loss": 2.3826,
      "step": 1213
    },
    {
      "epoch": 0.39985178475853267,
      "grad_norm": 2.8002431392669678,
      "learning_rate": 6.696362078273781e-05,
      "loss": 2.1889,
      "step": 1214
    },
    {
      "epoch": 0.4001811519617934,
      "grad_norm": 2.4396965503692627,
      "learning_rate": 6.69141261546805e-05,
      "loss": 1.9122,
      "step": 1215
    },
    {
      "epoch": 0.40051051916505415,
      "grad_norm": 2.8927268981933594,
      "learning_rate": 6.686461280387203e-05,
      "loss": 2.2138,
      "step": 1216
    },
    {
      "epoch": 0.40083988636831486,
      "grad_norm": 2.633314847946167,
      "learning_rate": 6.681508078512031e-05,
      "loss": 1.7317,
      "step": 1217
    },
    {
      "epoch": 0.4011692535715756,
      "grad_norm": 3.017258882522583,
      "learning_rate": 6.676553015325372e-05,
      "loss": 2.1284,
      "step": 1218
    },
    {
      "epoch": 0.40149862077483633,
      "grad_norm": 2.7072062492370605,
      "learning_rate": 6.67159609631214e-05,
      "loss": 1.8943,
      "step": 1219
    },
    {
      "epoch": 0.4018279879780971,
      "grad_norm": 2.961946487426758,
      "learning_rate": 6.666637326959293e-05,
      "loss": 2.2615,
      "step": 1220
    },
    {
      "epoch": 0.4021573551813578,
      "grad_norm": 2.648956775665283,
      "learning_rate": 6.661676712755842e-05,
      "loss": 1.9888,
      "step": 1221
    },
    {
      "epoch": 0.4024867223846186,
      "grad_norm": 2.841571092605591,
      "learning_rate": 6.656714259192839e-05,
      "loss": 2.2434,
      "step": 1222
    },
    {
      "epoch": 0.4028160895878793,
      "grad_norm": 2.6502773761749268,
      "learning_rate": 6.65174997176337e-05,
      "loss": 1.9564,
      "step": 1223
    },
    {
      "epoch": 0.40314545679114,
      "grad_norm": 2.589456796646118,
      "learning_rate": 6.646783855962555e-05,
      "loss": 1.664,
      "step": 1224
    },
    {
      "epoch": 0.40347482399440077,
      "grad_norm": 2.849358320236206,
      "learning_rate": 6.641815917287535e-05,
      "loss": 1.6624,
      "step": 1225
    },
    {
      "epoch": 0.4038041911976615,
      "grad_norm": 2.1147572994232178,
      "learning_rate": 6.63684616123747e-05,
      "loss": 2.2939,
      "step": 1226
    },
    {
      "epoch": 0.40413355840092224,
      "grad_norm": 2.6644229888916016,
      "learning_rate": 6.631874593313531e-05,
      "loss": 2.2582,
      "step": 1227
    },
    {
      "epoch": 0.40446292560418295,
      "grad_norm": 2.094411849975586,
      "learning_rate": 6.626901219018895e-05,
      "loss": 2.3132,
      "step": 1228
    },
    {
      "epoch": 0.4047922928074437,
      "grad_norm": 2.631877899169922,
      "learning_rate": 6.621926043858739e-05,
      "loss": 2.1575,
      "step": 1229
    },
    {
      "epoch": 0.40512166001070443,
      "grad_norm": 2.610651731491089,
      "learning_rate": 6.616949073340232e-05,
      "loss": 2.3285,
      "step": 1230
    },
    {
      "epoch": 0.40545102721396514,
      "grad_norm": 2.4871439933776855,
      "learning_rate": 6.611970312972531e-05,
      "loss": 2.0776,
      "step": 1231
    },
    {
      "epoch": 0.4057803944172259,
      "grad_norm": 2.7471463680267334,
      "learning_rate": 6.606989768266776e-05,
      "loss": 2.5548,
      "step": 1232
    },
    {
      "epoch": 0.4061097616204866,
      "grad_norm": 2.746669292449951,
      "learning_rate": 6.602007444736077e-05,
      "loss": 2.2351,
      "step": 1233
    },
    {
      "epoch": 0.4064391288237474,
      "grad_norm": 2.907914161682129,
      "learning_rate": 6.597023347895524e-05,
      "loss": 2.7394,
      "step": 1234
    },
    {
      "epoch": 0.4067684960270081,
      "grad_norm": 2.477755308151245,
      "learning_rate": 6.592037483262156e-05,
      "loss": 2.0522,
      "step": 1235
    },
    {
      "epoch": 0.40709786323026886,
      "grad_norm": 2.537019968032837,
      "learning_rate": 6.587049856354977e-05,
      "loss": 2.2179,
      "step": 1236
    },
    {
      "epoch": 0.4074272304335296,
      "grad_norm": 2.399907350540161,
      "learning_rate": 6.582060472694939e-05,
      "loss": 2.2994,
      "step": 1237
    },
    {
      "epoch": 0.40775659763679034,
      "grad_norm": 2.9592201709747314,
      "learning_rate": 6.577069337804944e-05,
      "loss": 2.246,
      "step": 1238
    },
    {
      "epoch": 0.40808596484005105,
      "grad_norm": 2.831124782562256,
      "learning_rate": 6.572076457209822e-05,
      "loss": 2.0093,
      "step": 1239
    },
    {
      "epoch": 0.40841533204331176,
      "grad_norm": 2.620274066925049,
      "learning_rate": 6.567081836436346e-05,
      "loss": 1.8906,
      "step": 1240
    },
    {
      "epoch": 0.40874469924657253,
      "grad_norm": 2.5152428150177,
      "learning_rate": 6.562085481013211e-05,
      "loss": 2.2982,
      "step": 1241
    },
    {
      "epoch": 0.40907406644983324,
      "grad_norm": 3.2875349521636963,
      "learning_rate": 6.55708739647103e-05,
      "loss": 2.4582,
      "step": 1242
    },
    {
      "epoch": 0.409403433653094,
      "grad_norm": 2.650205612182617,
      "learning_rate": 6.552087588342332e-05,
      "loss": 2.1196,
      "step": 1243
    },
    {
      "epoch": 0.4097328008563547,
      "grad_norm": 2.7873198986053467,
      "learning_rate": 6.547086062161555e-05,
      "loss": 1.9238,
      "step": 1244
    },
    {
      "epoch": 0.4100621680596155,
      "grad_norm": 2.823976993560791,
      "learning_rate": 6.542082823465037e-05,
      "loss": 1.9929,
      "step": 1245
    },
    {
      "epoch": 0.4103915352628762,
      "grad_norm": 3.1395912170410156,
      "learning_rate": 6.537077877791011e-05,
      "loss": 2.3549,
      "step": 1246
    },
    {
      "epoch": 0.41072090246613696,
      "grad_norm": 2.8609724044799805,
      "learning_rate": 6.532071230679604e-05,
      "loss": 2.1222,
      "step": 1247
    },
    {
      "epoch": 0.41105026966939767,
      "grad_norm": 2.8728225231170654,
      "learning_rate": 6.527062887672819e-05,
      "loss": 2.1451,
      "step": 1248
    },
    {
      "epoch": 0.4113796368726584,
      "grad_norm": 3.3821842670440674,
      "learning_rate": 6.522052854314544e-05,
      "loss": 2.3055,
      "step": 1249
    },
    {
      "epoch": 0.41170900407591915,
      "grad_norm": 3.682586431503296,
      "learning_rate": 6.517041136150534e-05,
      "loss": 1.9983,
      "step": 1250
    },
    {
      "epoch": 0.41203837127917986,
      "grad_norm": 2.3784186840057373,
      "learning_rate": 6.512027738728407e-05,
      "loss": 2.4458,
      "step": 1251
    },
    {
      "epoch": 0.4123677384824406,
      "grad_norm": 2.384232997894287,
      "learning_rate": 6.507012667597643e-05,
      "loss": 2.5749,
      "step": 1252
    },
    {
      "epoch": 0.41269710568570134,
      "grad_norm": 2.3909189701080322,
      "learning_rate": 6.501995928309577e-05,
      "loss": 2.3923,
      "step": 1253
    },
    {
      "epoch": 0.4130264728889621,
      "grad_norm": 2.478942394256592,
      "learning_rate": 6.496977526417383e-05,
      "loss": 2.5673,
      "step": 1254
    },
    {
      "epoch": 0.4133558400922228,
      "grad_norm": 2.6806704998016357,
      "learning_rate": 6.491957467476081e-05,
      "loss": 2.1562,
      "step": 1255
    },
    {
      "epoch": 0.4136852072954835,
      "grad_norm": 2.777129888534546,
      "learning_rate": 6.486935757042529e-05,
      "loss": 2.3145,
      "step": 1256
    },
    {
      "epoch": 0.4140145744987443,
      "grad_norm": 2.46478533744812,
      "learning_rate": 6.481912400675402e-05,
      "loss": 2.1051,
      "step": 1257
    },
    {
      "epoch": 0.414343941702005,
      "grad_norm": 2.9172468185424805,
      "learning_rate": 6.476887403935204e-05,
      "loss": 2.32,
      "step": 1258
    },
    {
      "epoch": 0.41467330890526577,
      "grad_norm": 2.7249600887298584,
      "learning_rate": 6.471860772384256e-05,
      "loss": 2.2816,
      "step": 1259
    },
    {
      "epoch": 0.4150026761085265,
      "grad_norm": 2.6038973331451416,
      "learning_rate": 6.466832511586687e-05,
      "loss": 2.1756,
      "step": 1260
    },
    {
      "epoch": 0.41533204331178725,
      "grad_norm": 2.6487905979156494,
      "learning_rate": 6.461802627108426e-05,
      "loss": 2.6021,
      "step": 1261
    },
    {
      "epoch": 0.41566141051504796,
      "grad_norm": 2.545063018798828,
      "learning_rate": 6.456771124517205e-05,
      "loss": 2.0434,
      "step": 1262
    },
    {
      "epoch": 0.4159907777183087,
      "grad_norm": 2.7668240070343018,
      "learning_rate": 6.451738009382542e-05,
      "loss": 2.3299,
      "step": 1263
    },
    {
      "epoch": 0.41632014492156943,
      "grad_norm": 2.4981627464294434,
      "learning_rate": 6.446703287275745e-05,
      "loss": 1.9392,
      "step": 1264
    },
    {
      "epoch": 0.41664951212483015,
      "grad_norm": 2.7257003784179688,
      "learning_rate": 6.441666963769897e-05,
      "loss": 2.0421,
      "step": 1265
    },
    {
      "epoch": 0.4169788793280909,
      "grad_norm": 2.8627986907958984,
      "learning_rate": 6.436629044439854e-05,
      "loss": 2.3554,
      "step": 1266
    },
    {
      "epoch": 0.4173082465313516,
      "grad_norm": 2.81748628616333,
      "learning_rate": 6.43158953486224e-05,
      "loss": 1.9084,
      "step": 1267
    },
    {
      "epoch": 0.4176376137346124,
      "grad_norm": 2.941399335861206,
      "learning_rate": 6.426548440615438e-05,
      "loss": 2.0619,
      "step": 1268
    },
    {
      "epoch": 0.4179669809378731,
      "grad_norm": 3.05411434173584,
      "learning_rate": 6.421505767279588e-05,
      "loss": 2.1444,
      "step": 1269
    },
    {
      "epoch": 0.41829634814113387,
      "grad_norm": 2.318077564239502,
      "learning_rate": 6.416461520436571e-05,
      "loss": 1.7884,
      "step": 1270
    },
    {
      "epoch": 0.4186257153443946,
      "grad_norm": 2.866882562637329,
      "learning_rate": 6.411415705670021e-05,
      "loss": 2.0699,
      "step": 1271
    },
    {
      "epoch": 0.41895508254765534,
      "grad_norm": 2.7729244232177734,
      "learning_rate": 6.406368328565295e-05,
      "loss": 2.2929,
      "step": 1272
    },
    {
      "epoch": 0.41928444975091606,
      "grad_norm": 3.3135483264923096,
      "learning_rate": 6.401319394709489e-05,
      "loss": 2.4658,
      "step": 1273
    },
    {
      "epoch": 0.41961381695417677,
      "grad_norm": 3.1503915786743164,
      "learning_rate": 6.396268909691414e-05,
      "loss": 2.117,
      "step": 1274
    },
    {
      "epoch": 0.41994318415743753,
      "grad_norm": 2.9462268352508545,
      "learning_rate": 6.391216879101608e-05,
      "loss": 2.0571,
      "step": 1275
    },
    {
      "epoch": 0.42027255136069824,
      "grad_norm": 2.186408042907715,
      "learning_rate": 6.386163308532314e-05,
      "loss": 2.5956,
      "step": 1276
    },
    {
      "epoch": 0.420601918563959,
      "grad_norm": 2.4059898853302,
      "learning_rate": 6.381108203577476e-05,
      "loss": 2.4601,
      "step": 1277
    },
    {
      "epoch": 0.4209312857672197,
      "grad_norm": 2.4408154487609863,
      "learning_rate": 6.376051569832742e-05,
      "loss": 2.537,
      "step": 1278
    },
    {
      "epoch": 0.4212606529704805,
      "grad_norm": 2.3048956394195557,
      "learning_rate": 6.370993412895454e-05,
      "loss": 2.4092,
      "step": 1279
    },
    {
      "epoch": 0.4215900201737412,
      "grad_norm": 2.4109179973602295,
      "learning_rate": 6.365933738364634e-05,
      "loss": 2.2523,
      "step": 1280
    },
    {
      "epoch": 0.4219193873770019,
      "grad_norm": 2.6136887073516846,
      "learning_rate": 6.360872551840988e-05,
      "loss": 2.0975,
      "step": 1281
    },
    {
      "epoch": 0.4222487545802627,
      "grad_norm": 2.697741985321045,
      "learning_rate": 6.355809858926893e-05,
      "loss": 2.5818,
      "step": 1282
    },
    {
      "epoch": 0.4225781217835234,
      "grad_norm": 2.3792057037353516,
      "learning_rate": 6.350745665226396e-05,
      "loss": 2.1916,
      "step": 1283
    },
    {
      "epoch": 0.42290748898678415,
      "grad_norm": 2.742786169052124,
      "learning_rate": 6.345679976345205e-05,
      "loss": 2.5287,
      "step": 1284
    },
    {
      "epoch": 0.42323685619004486,
      "grad_norm": 2.550384998321533,
      "learning_rate": 6.34061279789068e-05,
      "loss": 2.3782,
      "step": 1285
    },
    {
      "epoch": 0.42356622339330563,
      "grad_norm": 3.0929136276245117,
      "learning_rate": 6.335544135471834e-05,
      "loss": 2.3501,
      "step": 1286
    },
    {
      "epoch": 0.42389559059656634,
      "grad_norm": 2.920880079269409,
      "learning_rate": 6.330473994699318e-05,
      "loss": 2.3128,
      "step": 1287
    },
    {
      "epoch": 0.4242249577998271,
      "grad_norm": 2.258695125579834,
      "learning_rate": 6.325402381185426e-05,
      "loss": 2.0221,
      "step": 1288
    },
    {
      "epoch": 0.4245543250030878,
      "grad_norm": 2.6346912384033203,
      "learning_rate": 6.320329300544076e-05,
      "loss": 2.2641,
      "step": 1289
    },
    {
      "epoch": 0.42488369220634853,
      "grad_norm": 3.050056219100952,
      "learning_rate": 6.315254758390814e-05,
      "loss": 2.4125,
      "step": 1290
    },
    {
      "epoch": 0.4252130594096093,
      "grad_norm": 2.530167818069458,
      "learning_rate": 6.3101787603428e-05,
      "loss": 2.0573,
      "step": 1291
    },
    {
      "epoch": 0.42554242661287,
      "grad_norm": 3.0110995769500732,
      "learning_rate": 6.305101312018809e-05,
      "loss": 2.6722,
      "step": 1292
    },
    {
      "epoch": 0.4258717938161308,
      "grad_norm": 2.6352243423461914,
      "learning_rate": 6.300022419039219e-05,
      "loss": 2.0293,
      "step": 1293
    },
    {
      "epoch": 0.4262011610193915,
      "grad_norm": 2.864758253097534,
      "learning_rate": 6.294942087026011e-05,
      "loss": 1.9841,
      "step": 1294
    },
    {
      "epoch": 0.42653052822265225,
      "grad_norm": 3.216989517211914,
      "learning_rate": 6.289860321602754e-05,
      "loss": 2.005,
      "step": 1295
    },
    {
      "epoch": 0.42685989542591296,
      "grad_norm": 3.134946584701538,
      "learning_rate": 6.284777128394603e-05,
      "loss": 2.3214,
      "step": 1296
    },
    {
      "epoch": 0.4271892626291737,
      "grad_norm": 3.151312828063965,
      "learning_rate": 6.279692513028304e-05,
      "loss": 2.2082,
      "step": 1297
    },
    {
      "epoch": 0.42751862983243444,
      "grad_norm": 2.858823776245117,
      "learning_rate": 6.274606481132163e-05,
      "loss": 2.0439,
      "step": 1298
    },
    {
      "epoch": 0.42784799703569515,
      "grad_norm": 2.6305992603302,
      "learning_rate": 6.269519038336062e-05,
      "loss": 1.7326,
      "step": 1299
    },
    {
      "epoch": 0.4281773642389559,
      "grad_norm": 4.182598114013672,
      "learning_rate": 6.264430190271444e-05,
      "loss": 1.8915,
      "step": 1300
    },
    {
      "epoch": 0.4285067314422166,
      "grad_norm": 2.144503116607666,
      "learning_rate": 6.259339942571307e-05,
      "loss": 2.3162,
      "step": 1301
    },
    {
      "epoch": 0.4288360986454774,
      "grad_norm": 2.138836622238159,
      "learning_rate": 6.254248300870198e-05,
      "loss": 2.3545,
      "step": 1302
    },
    {
      "epoch": 0.4291654658487381,
      "grad_norm": 2.6747050285339355,
      "learning_rate": 6.249155270804206e-05,
      "loss": 2.6766,
      "step": 1303
    },
    {
      "epoch": 0.42949483305199887,
      "grad_norm": 2.2941064834594727,
      "learning_rate": 6.24406085801096e-05,
      "loss": 2.6815,
      "step": 1304
    },
    {
      "epoch": 0.4298242002552596,
      "grad_norm": 2.450798988342285,
      "learning_rate": 6.238965068129616e-05,
      "loss": 2.3887,
      "step": 1305
    },
    {
      "epoch": 0.4301535674585203,
      "grad_norm": 2.7236177921295166,
      "learning_rate": 6.233867906800856e-05,
      "loss": 2.6746,
      "step": 1306
    },
    {
      "epoch": 0.43048293466178106,
      "grad_norm": 2.221660852432251,
      "learning_rate": 6.22876937966688e-05,
      "loss": 2.4207,
      "step": 1307
    },
    {
      "epoch": 0.43081230186504177,
      "grad_norm": 2.9190986156463623,
      "learning_rate": 6.2236694923714e-05,
      "loss": 2.5129,
      "step": 1308
    },
    {
      "epoch": 0.43114166906830254,
      "grad_norm": 2.183797597885132,
      "learning_rate": 6.218568250559634e-05,
      "loss": 2.209,
      "step": 1309
    },
    {
      "epoch": 0.43147103627156325,
      "grad_norm": 2.7680306434631348,
      "learning_rate": 6.2134656598783e-05,
      "loss": 1.9334,
      "step": 1310
    },
    {
      "epoch": 0.431800403474824,
      "grad_norm": 2.6052422523498535,
      "learning_rate": 6.208361725975605e-05,
      "loss": 2.2873,
      "step": 1311
    },
    {
      "epoch": 0.4321297706780847,
      "grad_norm": 2.873652696609497,
      "learning_rate": 6.203256454501248e-05,
      "loss": 2.5024,
      "step": 1312
    },
    {
      "epoch": 0.4324591378813455,
      "grad_norm": 3.202440023422241,
      "learning_rate": 6.198149851106407e-05,
      "loss": 2.2169,
      "step": 1313
    },
    {
      "epoch": 0.4327885050846062,
      "grad_norm": 2.7076804637908936,
      "learning_rate": 6.19304192144373e-05,
      "loss": 2.0037,
      "step": 1314
    },
    {
      "epoch": 0.4331178722878669,
      "grad_norm": 3.2877442836761475,
      "learning_rate": 6.187932671167342e-05,
      "loss": 2.6063,
      "step": 1315
    },
    {
      "epoch": 0.4334472394911277,
      "grad_norm": 2.6223089694976807,
      "learning_rate": 6.18282210593282e-05,
      "loss": 1.959,
      "step": 1316
    },
    {
      "epoch": 0.4337766066943884,
      "grad_norm": 3.001115083694458,
      "learning_rate": 6.177710231397203e-05,
      "loss": 2.1598,
      "step": 1317
    },
    {
      "epoch": 0.43410597389764916,
      "grad_norm": 2.4432201385498047,
      "learning_rate": 6.172597053218978e-05,
      "loss": 1.7752,
      "step": 1318
    },
    {
      "epoch": 0.43443534110090987,
      "grad_norm": 2.4193572998046875,
      "learning_rate": 6.167482577058075e-05,
      "loss": 2.0442,
      "step": 1319
    },
    {
      "epoch": 0.43476470830417063,
      "grad_norm": 2.7738735675811768,
      "learning_rate": 6.162366808575857e-05,
      "loss": 2.1012,
      "step": 1320
    },
    {
      "epoch": 0.43509407550743134,
      "grad_norm": 2.398259401321411,
      "learning_rate": 6.157249753435124e-05,
      "loss": 1.641,
      "step": 1321
    },
    {
      "epoch": 0.4354234427106921,
      "grad_norm": 2.803773880004883,
      "learning_rate": 6.152131417300098e-05,
      "loss": 2.1607,
      "step": 1322
    },
    {
      "epoch": 0.4357528099139528,
      "grad_norm": 3.041407346725464,
      "learning_rate": 6.147011805836414e-05,
      "loss": 1.8865,
      "step": 1323
    },
    {
      "epoch": 0.43608217711721353,
      "grad_norm": 4.000000476837158,
      "learning_rate": 6.141890924711126e-05,
      "loss": 1.8523,
      "step": 1324
    },
    {
      "epoch": 0.4364115443204743,
      "grad_norm": 3.650296688079834,
      "learning_rate": 6.136768779592691e-05,
      "loss": 2.2463,
      "step": 1325
    },
    {
      "epoch": 0.436740911523735,
      "grad_norm": 2.288025140762329,
      "learning_rate": 6.13164537615096e-05,
      "loss": 2.5091,
      "step": 1326
    },
    {
      "epoch": 0.4370702787269958,
      "grad_norm": 2.0283560752868652,
      "learning_rate": 6.126520720057186e-05,
      "loss": 2.2552,
      "step": 1327
    },
    {
      "epoch": 0.4373996459302565,
      "grad_norm": 2.147888422012329,
      "learning_rate": 6.121394816984e-05,
      "loss": 2.3754,
      "step": 1328
    },
    {
      "epoch": 0.43772901313351725,
      "grad_norm": 2.3866569995880127,
      "learning_rate": 6.11626767260542e-05,
      "loss": 2.2485,
      "step": 1329
    },
    {
      "epoch": 0.43805838033677796,
      "grad_norm": 2.4724600315093994,
      "learning_rate": 6.111139292596834e-05,
      "loss": 2.3486,
      "step": 1330
    },
    {
      "epoch": 0.4383877475400387,
      "grad_norm": 2.1698498725891113,
      "learning_rate": 6.106009682634997e-05,
      "loss": 2.2673,
      "step": 1331
    },
    {
      "epoch": 0.43871711474329944,
      "grad_norm": 2.1860690116882324,
      "learning_rate": 6.100878848398032e-05,
      "loss": 2.1531,
      "step": 1332
    },
    {
      "epoch": 0.43904648194656015,
      "grad_norm": 2.3432462215423584,
      "learning_rate": 6.095746795565408e-05,
      "loss": 2.3907,
      "step": 1333
    },
    {
      "epoch": 0.4393758491498209,
      "grad_norm": 3.00425386428833,
      "learning_rate": 6.090613529817949e-05,
      "loss": 2.5884,
      "step": 1334
    },
    {
      "epoch": 0.43970521635308163,
      "grad_norm": 3.2894668579101562,
      "learning_rate": 6.085479056837821e-05,
      "loss": 2.2589,
      "step": 1335
    },
    {
      "epoch": 0.4400345835563424,
      "grad_norm": 2.9489943981170654,
      "learning_rate": 6.0803433823085244e-05,
      "loss": 2.2962,
      "step": 1336
    },
    {
      "epoch": 0.4403639507596031,
      "grad_norm": 3.29754638671875,
      "learning_rate": 6.075206511914891e-05,
      "loss": 2.4655,
      "step": 1337
    },
    {
      "epoch": 0.4406933179628639,
      "grad_norm": 2.559438705444336,
      "learning_rate": 6.070068451343074e-05,
      "loss": 2.2637,
      "step": 1338
    },
    {
      "epoch": 0.4410226851661246,
      "grad_norm": 3.065319061279297,
      "learning_rate": 6.0649292062805494e-05,
      "loss": 2.1736,
      "step": 1339
    },
    {
      "epoch": 0.4413520523693853,
      "grad_norm": 2.943509101867676,
      "learning_rate": 6.059788782416099e-05,
      "loss": 2.0683,
      "step": 1340
    },
    {
      "epoch": 0.44168141957264606,
      "grad_norm": 2.7584707736968994,
      "learning_rate": 6.054647185439814e-05,
      "loss": 2.3206,
      "step": 1341
    },
    {
      "epoch": 0.44201078677590677,
      "grad_norm": 2.789400577545166,
      "learning_rate": 6.049504421043078e-05,
      "loss": 1.8737,
      "step": 1342
    },
    {
      "epoch": 0.44234015397916754,
      "grad_norm": 2.9476754665374756,
      "learning_rate": 6.0443604949185706e-05,
      "loss": 2.2538,
      "step": 1343
    },
    {
      "epoch": 0.44266952118242825,
      "grad_norm": 2.6708343029022217,
      "learning_rate": 6.0392154127602595e-05,
      "loss": 2.156,
      "step": 1344
    },
    {
      "epoch": 0.442998888385689,
      "grad_norm": 2.7900948524475098,
      "learning_rate": 6.0340691802633884e-05,
      "loss": 1.8873,
      "step": 1345
    },
    {
      "epoch": 0.4433282555889497,
      "grad_norm": 2.6233713626861572,
      "learning_rate": 6.028921803124476e-05,
      "loss": 1.7659,
      "step": 1346
    },
    {
      "epoch": 0.4436576227922105,
      "grad_norm": 2.7721614837646484,
      "learning_rate": 6.023773287041308e-05,
      "loss": 2.0038,
      "step": 1347
    },
    {
      "epoch": 0.4439869899954712,
      "grad_norm": 2.7499945163726807,
      "learning_rate": 6.01862363771293e-05,
      "loss": 2.1821,
      "step": 1348
    },
    {
      "epoch": 0.4443163571987319,
      "grad_norm": 2.655479907989502,
      "learning_rate": 6.013472860839642e-05,
      "loss": 1.9229,
      "step": 1349
    },
    {
      "epoch": 0.4446457244019927,
      "grad_norm": 2.927269458770752,
      "learning_rate": 6.008320962122994e-05,
      "loss": 1.8664,
      "step": 1350
    },
    {
      "epoch": 0.4449750916052534,
      "grad_norm": 1.9483531713485718,
      "learning_rate": 6.003167947265777e-05,
      "loss": 2.5558,
      "step": 1351
    },
    {
      "epoch": 0.44530445880851416,
      "grad_norm": 2.5765058994293213,
      "learning_rate": 5.9980138219720125e-05,
      "loss": 2.6047,
      "step": 1352
    },
    {
      "epoch": 0.44563382601177487,
      "grad_norm": 2.483147382736206,
      "learning_rate": 5.992858591946961e-05,
      "loss": 2.324,
      "step": 1353
    },
    {
      "epoch": 0.44596319321503564,
      "grad_norm": 2.8535609245300293,
      "learning_rate": 5.987702262897098e-05,
      "loss": 2.395,
      "step": 1354
    },
    {
      "epoch": 0.44629256041829635,
      "grad_norm": 2.455509662628174,
      "learning_rate": 5.9825448405301175e-05,
      "loss": 2.3842,
      "step": 1355
    },
    {
      "epoch": 0.44662192762155706,
      "grad_norm": 2.347088575363159,
      "learning_rate": 5.977386330554926e-05,
      "loss": 2.2313,
      "step": 1356
    },
    {
      "epoch": 0.4469512948248178,
      "grad_norm": 2.4735326766967773,
      "learning_rate": 5.9722267386816324e-05,
      "loss": 2.1965,
      "step": 1357
    },
    {
      "epoch": 0.44728066202807854,
      "grad_norm": 2.5871574878692627,
      "learning_rate": 5.967066070621541e-05,
      "loss": 2.2823,
      "step": 1358
    },
    {
      "epoch": 0.4476100292313393,
      "grad_norm": 2.4709925651550293,
      "learning_rate": 5.9619043320871494e-05,
      "loss": 2.2467,
      "step": 1359
    },
    {
      "epoch": 0.4479393964346,
      "grad_norm": 2.6086020469665527,
      "learning_rate": 5.956741528792142e-05,
      "loss": 2.2654,
      "step": 1360
    },
    {
      "epoch": 0.4482687636378608,
      "grad_norm": 2.7858760356903076,
      "learning_rate": 5.951577666451379e-05,
      "loss": 2.3658,
      "step": 1361
    },
    {
      "epoch": 0.4485981308411215,
      "grad_norm": 2.6391570568084717,
      "learning_rate": 5.946412750780892e-05,
      "loss": 2.1763,
      "step": 1362
    },
    {
      "epoch": 0.44892749804438226,
      "grad_norm": 2.597666025161743,
      "learning_rate": 5.941246787497884e-05,
      "loss": 2.1193,
      "step": 1363
    },
    {
      "epoch": 0.44925686524764297,
      "grad_norm": 2.591607093811035,
      "learning_rate": 5.9360797823207104e-05,
      "loss": 1.9894,
      "step": 1364
    },
    {
      "epoch": 0.4495862324509037,
      "grad_norm": 2.4428396224975586,
      "learning_rate": 5.930911740968884e-05,
      "loss": 2.1478,
      "step": 1365
    },
    {
      "epoch": 0.44991559965416444,
      "grad_norm": 2.8427088260650635,
      "learning_rate": 5.9257426691630656e-05,
      "loss": 2.0804,
      "step": 1366
    },
    {
      "epoch": 0.45024496685742516,
      "grad_norm": 2.4200117588043213,
      "learning_rate": 5.920572572625056e-05,
      "loss": 2.0263,
      "step": 1367
    },
    {
      "epoch": 0.4505743340606859,
      "grad_norm": 2.4792933464050293,
      "learning_rate": 5.915401457077785e-05,
      "loss": 2.1457,
      "step": 1368
    },
    {
      "epoch": 0.45090370126394663,
      "grad_norm": 3.0544183254241943,
      "learning_rate": 5.910229328245319e-05,
      "loss": 2.476,
      "step": 1369
    },
    {
      "epoch": 0.4512330684672074,
      "grad_norm": 3.122922897338867,
      "learning_rate": 5.90505619185284e-05,
      "loss": 1.8849,
      "step": 1370
    },
    {
      "epoch": 0.4515624356704681,
      "grad_norm": 3.2922286987304688,
      "learning_rate": 5.899882053626646e-05,
      "loss": 2.1517,
      "step": 1371
    },
    {
      "epoch": 0.4518918028737289,
      "grad_norm": 3.095803737640381,
      "learning_rate": 5.8947069192941493e-05,
      "loss": 2.1022,
      "step": 1372
    },
    {
      "epoch": 0.4522211700769896,
      "grad_norm": 3.426305055618286,
      "learning_rate": 5.889530794583855e-05,
      "loss": 2.1558,
      "step": 1373
    },
    {
      "epoch": 0.4525505372802503,
      "grad_norm": 3.2254509925842285,
      "learning_rate": 5.8843536852253745e-05,
      "loss": 2.2135,
      "step": 1374
    },
    {
      "epoch": 0.45287990448351106,
      "grad_norm": 3.3690662384033203,
      "learning_rate": 5.879175596949401e-05,
      "loss": 2.0148,
      "step": 1375
    },
    {
      "epoch": 0.4532092716867718,
      "grad_norm": 2.666074275970459,
      "learning_rate": 5.8739965354877194e-05,
      "loss": 2.445,
      "step": 1376
    },
    {
      "epoch": 0.45353863889003254,
      "grad_norm": 2.3343346118927,
      "learning_rate": 5.8688165065731826e-05,
      "loss": 2.332,
      "step": 1377
    },
    {
      "epoch": 0.45386800609329325,
      "grad_norm": 2.3201119899749756,
      "learning_rate": 5.8636355159397225e-05,
      "loss": 2.5834,
      "step": 1378
    },
    {
      "epoch": 0.454197373296554,
      "grad_norm": 2.493422269821167,
      "learning_rate": 5.858453569322332e-05,
      "loss": 2.5553,
      "step": 1379
    },
    {
      "epoch": 0.45452674049981473,
      "grad_norm": 2.374323606491089,
      "learning_rate": 5.853270672457061e-05,
      "loss": 2.1962,
      "step": 1380
    },
    {
      "epoch": 0.45485610770307544,
      "grad_norm": 2.812485694885254,
      "learning_rate": 5.8480868310810124e-05,
      "loss": 2.3164,
      "step": 1381
    },
    {
      "epoch": 0.4551854749063362,
      "grad_norm": 2.3375439643859863,
      "learning_rate": 5.8429020509323385e-05,
      "loss": 2.4357,
      "step": 1382
    },
    {
      "epoch": 0.4555148421095969,
      "grad_norm": 2.376716136932373,
      "learning_rate": 5.837716337750223e-05,
      "loss": 2.5416,
      "step": 1383
    },
    {
      "epoch": 0.4558442093128577,
      "grad_norm": 2.5697383880615234,
      "learning_rate": 5.8325296972748864e-05,
      "loss": 2.3119,
      "step": 1384
    },
    {
      "epoch": 0.4561735765161184,
      "grad_norm": 2.5520308017730713,
      "learning_rate": 5.827342135247581e-05,
      "loss": 2.3784,
      "step": 1385
    },
    {
      "epoch": 0.45650294371937916,
      "grad_norm": 2.750321865081787,
      "learning_rate": 5.8221536574105694e-05,
      "loss": 2.5948,
      "step": 1386
    },
    {
      "epoch": 0.4568323109226399,
      "grad_norm": 2.980220317840576,
      "learning_rate": 5.816964269507135e-05,
      "loss": 2.5563,
      "step": 1387
    },
    {
      "epoch": 0.45716167812590064,
      "grad_norm": 2.898198366165161,
      "learning_rate": 5.811773977281565e-05,
      "loss": 2.2648,
      "step": 1388
    },
    {
      "epoch": 0.45749104532916135,
      "grad_norm": 3.0173451900482178,
      "learning_rate": 5.806582786479149e-05,
      "loss": 2.2571,
      "step": 1389
    },
    {
      "epoch": 0.45782041253242206,
      "grad_norm": 2.6375420093536377,
      "learning_rate": 5.801390702846171e-05,
      "loss": 2.0654,
      "step": 1390
    },
    {
      "epoch": 0.4581497797356828,
      "grad_norm": 2.8282692432403564,
      "learning_rate": 5.796197732129905e-05,
      "loss": 2.3666,
      "step": 1391
    },
    {
      "epoch": 0.45847914693894354,
      "grad_norm": 2.815434694290161,
      "learning_rate": 5.7910038800786e-05,
      "loss": 2.2048,
      "step": 1392
    },
    {
      "epoch": 0.4588085141422043,
      "grad_norm": 2.911081075668335,
      "learning_rate": 5.7858091524414926e-05,
      "loss": 2.3954,
      "step": 1393
    },
    {
      "epoch": 0.459137881345465,
      "grad_norm": 2.4357879161834717,
      "learning_rate": 5.780613554968777e-05,
      "loss": 1.9379,
      "step": 1394
    },
    {
      "epoch": 0.4594672485487258,
      "grad_norm": 3.2798242568969727,
      "learning_rate": 5.775417093411619e-05,
      "loss": 2.3849,
      "step": 1395
    },
    {
      "epoch": 0.4597966157519865,
      "grad_norm": 2.6205365657806396,
      "learning_rate": 5.770219773522133e-05,
      "loss": 2.2528,
      "step": 1396
    },
    {
      "epoch": 0.4601259829552472,
      "grad_norm": 2.84399676322937,
      "learning_rate": 5.765021601053391e-05,
      "loss": 1.8288,
      "step": 1397
    },
    {
      "epoch": 0.46045535015850797,
      "grad_norm": 2.869101047515869,
      "learning_rate": 5.7598225817594035e-05,
      "loss": 2.3284,
      "step": 1398
    },
    {
      "epoch": 0.4607847173617687,
      "grad_norm": 2.6911323070526123,
      "learning_rate": 5.754622721395119e-05,
      "loss": 2.0477,
      "step": 1399
    },
    {
      "epoch": 0.46111408456502945,
      "grad_norm": 3.472745656967163,
      "learning_rate": 5.74942202571642e-05,
      "loss": 1.8559,
      "step": 1400
    },
    {
      "epoch": 0.46144345176829016,
      "grad_norm": 2.722027063369751,
      "learning_rate": 5.744220500480113e-05,
      "loss": 2.5318,
      "step": 1401
    },
    {
      "epoch": 0.4617728189715509,
      "grad_norm": 2.0918314456939697,
      "learning_rate": 5.739018151443918e-05,
      "loss": 2.1847,
      "step": 1402
    },
    {
      "epoch": 0.46210218617481164,
      "grad_norm": 2.0019867420196533,
      "learning_rate": 5.733814984366474e-05,
      "loss": 2.2064,
      "step": 1403
    },
    {
      "epoch": 0.4624315533780724,
      "grad_norm": 2.3096742630004883,
      "learning_rate": 5.7286110050073194e-05,
      "loss": 2.0878,
      "step": 1404
    },
    {
      "epoch": 0.4627609205813331,
      "grad_norm": 2.7274067401885986,
      "learning_rate": 5.723406219126895e-05,
      "loss": 2.2556,
      "step": 1405
    },
    {
      "epoch": 0.4630902877845938,
      "grad_norm": 2.2362143993377686,
      "learning_rate": 5.718200632486534e-05,
      "loss": 2.0013,
      "step": 1406
    },
    {
      "epoch": 0.4634196549878546,
      "grad_norm": 2.458976984024048,
      "learning_rate": 5.7129942508484556e-05,
      "loss": 2.1049,
      "step": 1407
    },
    {
      "epoch": 0.4637490221911153,
      "grad_norm": 2.278069496154785,
      "learning_rate": 5.707787079975758e-05,
      "loss": 2.0951,
      "step": 1408
    },
    {
      "epoch": 0.46407838939437607,
      "grad_norm": 2.235696315765381,
      "learning_rate": 5.702579125632416e-05,
      "loss": 2.4472,
      "step": 1409
    },
    {
      "epoch": 0.4644077565976368,
      "grad_norm": 2.662332773208618,
      "learning_rate": 5.697370393583269e-05,
      "loss": 2.1896,
      "step": 1410
    },
    {
      "epoch": 0.46473712380089754,
      "grad_norm": 2.677049398422241,
      "learning_rate": 5.692160889594017e-05,
      "loss": 2.0806,
      "step": 1411
    },
    {
      "epoch": 0.46506649100415826,
      "grad_norm": 2.7263095378875732,
      "learning_rate": 5.686950619431215e-05,
      "loss": 2.1826,
      "step": 1412
    },
    {
      "epoch": 0.465395858207419,
      "grad_norm": 2.625284194946289,
      "learning_rate": 5.6817395888622694e-05,
      "loss": 2.4464,
      "step": 1413
    },
    {
      "epoch": 0.46572522541067973,
      "grad_norm": 2.6425976753234863,
      "learning_rate": 5.6765278036554225e-05,
      "loss": 2.0412,
      "step": 1414
    },
    {
      "epoch": 0.46605459261394044,
      "grad_norm": 2.918874502182007,
      "learning_rate": 5.671315269579756e-05,
      "loss": 2.3,
      "step": 1415
    },
    {
      "epoch": 0.4663839598172012,
      "grad_norm": 2.5336811542510986,
      "learning_rate": 5.6661019924051814e-05,
      "loss": 2.4006,
      "step": 1416
    },
    {
      "epoch": 0.4667133270204619,
      "grad_norm": 2.6169357299804688,
      "learning_rate": 5.6608879779024274e-05,
      "loss": 2.1464,
      "step": 1417
    },
    {
      "epoch": 0.4670426942237227,
      "grad_norm": 2.734600305557251,
      "learning_rate": 5.6556732318430437e-05,
      "loss": 2.0906,
      "step": 1418
    },
    {
      "epoch": 0.4673720614269834,
      "grad_norm": 2.7634074687957764,
      "learning_rate": 5.650457759999389e-05,
      "loss": 2.1538,
      "step": 1419
    },
    {
      "epoch": 0.46770142863024416,
      "grad_norm": 2.391335964202881,
      "learning_rate": 5.6452415681446256e-05,
      "loss": 1.7893,
      "step": 1420
    },
    {
      "epoch": 0.4680307958335049,
      "grad_norm": 2.9604604244232178,
      "learning_rate": 5.6400246620527096e-05,
      "loss": 2.1226,
      "step": 1421
    },
    {
      "epoch": 0.4683601630367656,
      "grad_norm": 2.650703191757202,
      "learning_rate": 5.6348070474983905e-05,
      "loss": 1.8164,
      "step": 1422
    },
    {
      "epoch": 0.46868953024002635,
      "grad_norm": 2.603179454803467,
      "learning_rate": 5.629588730257205e-05,
      "loss": 1.6553,
      "step": 1423
    },
    {
      "epoch": 0.46901889744328706,
      "grad_norm": 2.903170108795166,
      "learning_rate": 5.6243697161054584e-05,
      "loss": 1.8419,
      "step": 1424
    },
    {
      "epoch": 0.46934826464654783,
      "grad_norm": 3.1051528453826904,
      "learning_rate": 5.619150010820238e-05,
      "loss": 1.6226,
      "step": 1425
    },
    {
      "epoch": 0.46967763184980854,
      "grad_norm": 1.8390017747879028,
      "learning_rate": 5.613929620179389e-05,
      "loss": 2.2257,
      "step": 1426
    },
    {
      "epoch": 0.4700069990530693,
      "grad_norm": 2.490265130996704,
      "learning_rate": 5.608708549961519e-05,
      "loss": 2.3091,
      "step": 1427
    },
    {
      "epoch": 0.47033636625633,
      "grad_norm": 2.267498254776001,
      "learning_rate": 5.603486805945984e-05,
      "loss": 2.4721,
      "step": 1428
    },
    {
      "epoch": 0.4706657334595908,
      "grad_norm": 2.788170337677002,
      "learning_rate": 5.598264393912891e-05,
      "loss": 2.453,
      "step": 1429
    },
    {
      "epoch": 0.4709951006628515,
      "grad_norm": 2.2585389614105225,
      "learning_rate": 5.593041319643083e-05,
      "loss": 2.2228,
      "step": 1430
    },
    {
      "epoch": 0.4713244678661122,
      "grad_norm": 2.556220531463623,
      "learning_rate": 5.587817588918137e-05,
      "loss": 2.3444,
      "step": 1431
    },
    {
      "epoch": 0.471653835069373,
      "grad_norm": 2.604313611984253,
      "learning_rate": 5.582593207520357e-05,
      "loss": 2.1515,
      "step": 1432
    },
    {
      "epoch": 0.4719832022726337,
      "grad_norm": 2.6779942512512207,
      "learning_rate": 5.577368181232764e-05,
      "loss": 2.1363,
      "step": 1433
    },
    {
      "epoch": 0.47231256947589445,
      "grad_norm": 2.4980621337890625,
      "learning_rate": 5.572142515839098e-05,
      "loss": 2.0295,
      "step": 1434
    },
    {
      "epoch": 0.47264193667915516,
      "grad_norm": 2.331618309020996,
      "learning_rate": 5.5669162171238046e-05,
      "loss": 2.1247,
      "step": 1435
    },
    {
      "epoch": 0.4729713038824159,
      "grad_norm": 3.248292922973633,
      "learning_rate": 5.5616892908720274e-05,
      "loss": 2.2727,
      "step": 1436
    },
    {
      "epoch": 0.47330067108567664,
      "grad_norm": 2.3975675106048584,
      "learning_rate": 5.556461742869609e-05,
      "loss": 2.3504,
      "step": 1437
    },
    {
      "epoch": 0.4736300382889374,
      "grad_norm": 2.7226855754852295,
      "learning_rate": 5.551233578903078e-05,
      "loss": 2.0893,
      "step": 1438
    },
    {
      "epoch": 0.4739594054921981,
      "grad_norm": 2.8404018878936768,
      "learning_rate": 5.5460048047596434e-05,
      "loss": 2.4199,
      "step": 1439
    },
    {
      "epoch": 0.4742887726954588,
      "grad_norm": 2.4289791584014893,
      "learning_rate": 5.540775426227194e-05,
      "loss": 2.4747,
      "step": 1440
    },
    {
      "epoch": 0.4746181398987196,
      "grad_norm": 2.772507905960083,
      "learning_rate": 5.535545449094283e-05,
      "loss": 2.054,
      "step": 1441
    },
    {
      "epoch": 0.4749475071019803,
      "grad_norm": 2.7915074825286865,
      "learning_rate": 5.5303148791501305e-05,
      "loss": 2.3271,
      "step": 1442
    },
    {
      "epoch": 0.47527687430524107,
      "grad_norm": 2.894771099090576,
      "learning_rate": 5.525083722184607e-05,
      "loss": 2.2738,
      "step": 1443
    },
    {
      "epoch": 0.4756062415085018,
      "grad_norm": 2.797031879425049,
      "learning_rate": 5.519851983988239e-05,
      "loss": 2.3225,
      "step": 1444
    },
    {
      "epoch": 0.47593560871176255,
      "grad_norm": 3.053269386291504,
      "learning_rate": 5.514619670352192e-05,
      "loss": 2.1418,
      "step": 1445
    },
    {
      "epoch": 0.47626497591502326,
      "grad_norm": 2.743236780166626,
      "learning_rate": 5.5093867870682725e-05,
      "loss": 1.966,
      "step": 1446
    },
    {
      "epoch": 0.47659434311828397,
      "grad_norm": 2.998206615447998,
      "learning_rate": 5.504153339928914e-05,
      "loss": 2.2283,
      "step": 1447
    },
    {
      "epoch": 0.47692371032154474,
      "grad_norm": 3.0275509357452393,
      "learning_rate": 5.498919334727175e-05,
      "loss": 2.0081,
      "step": 1448
    },
    {
      "epoch": 0.47725307752480545,
      "grad_norm": 3.7755558490753174,
      "learning_rate": 5.4936847772567314e-05,
      "loss": 2.3072,
      "step": 1449
    },
    {
      "epoch": 0.4775824447280662,
      "grad_norm": 3.9976677894592285,
      "learning_rate": 5.488449673311872e-05,
      "loss": 2.2594,
      "step": 1450
    },
    {
      "epoch": 0.4779118119313269,
      "grad_norm": 2.3137824535369873,
      "learning_rate": 5.48321402868749e-05,
      "loss": 2.6206,
      "step": 1451
    },
    {
      "epoch": 0.4782411791345877,
      "grad_norm": 2.003905773162842,
      "learning_rate": 5.477977849179076e-05,
      "loss": 2.221,
      "step": 1452
    },
    {
      "epoch": 0.4785705463378484,
      "grad_norm": 2.37841534614563,
      "learning_rate": 5.4727411405827136e-05,
      "loss": 2.5322,
      "step": 1453
    },
    {
      "epoch": 0.47889991354110917,
      "grad_norm": 2.259744644165039,
      "learning_rate": 5.467503908695073e-05,
      "loss": 1.9622,
      "step": 1454
    },
    {
      "epoch": 0.4792292807443699,
      "grad_norm": 2.1672475337982178,
      "learning_rate": 5.4622661593133996e-05,
      "loss": 2.0246,
      "step": 1455
    },
    {
      "epoch": 0.4795586479476306,
      "grad_norm": 2.359017848968506,
      "learning_rate": 5.457027898235517e-05,
      "loss": 2.0392,
      "step": 1456
    },
    {
      "epoch": 0.47988801515089136,
      "grad_norm": 2.5766847133636475,
      "learning_rate": 5.451789131259814e-05,
      "loss": 2.2329,
      "step": 1457
    },
    {
      "epoch": 0.48021738235415207,
      "grad_norm": 2.3741567134857178,
      "learning_rate": 5.446549864185233e-05,
      "loss": 2.2631,
      "step": 1458
    },
    {
      "epoch": 0.48054674955741283,
      "grad_norm": 2.526066541671753,
      "learning_rate": 5.44131010281128e-05,
      "loss": 2.1493,
      "step": 1459
    },
    {
      "epoch": 0.48087611676067354,
      "grad_norm": 2.551849126815796,
      "learning_rate": 5.4360698529380004e-05,
      "loss": 2.3144,
      "step": 1460
    },
    {
      "epoch": 0.4812054839639343,
      "grad_norm": 2.2339227199554443,
      "learning_rate": 5.4308291203659855e-05,
      "loss": 2.2004,
      "step": 1461
    },
    {
      "epoch": 0.481534851167195,
      "grad_norm": 2.5119829177856445,
      "learning_rate": 5.425587910896357e-05,
      "loss": 2.0944,
      "step": 1462
    },
    {
      "epoch": 0.4818642183704558,
      "grad_norm": 2.8552167415618896,
      "learning_rate": 5.4203462303307685e-05,
      "loss": 2.1718,
      "step": 1463
    },
    {
      "epoch": 0.4821935855737165,
      "grad_norm": 2.6687982082366943,
      "learning_rate": 5.4151040844713886e-05,
      "loss": 2.2164,
      "step": 1464
    },
    {
      "epoch": 0.4825229527769772,
      "grad_norm": 2.963787317276001,
      "learning_rate": 5.409861479120908e-05,
      "loss": 2.22,
      "step": 1465
    },
    {
      "epoch": 0.482852319980238,
      "grad_norm": 2.990370750427246,
      "learning_rate": 5.404618420082521e-05,
      "loss": 2.2234,
      "step": 1466
    },
    {
      "epoch": 0.4831816871834987,
      "grad_norm": 2.695317506790161,
      "learning_rate": 5.39937491315993e-05,
      "loss": 2.2448,
      "step": 1467
    },
    {
      "epoch": 0.48351105438675945,
      "grad_norm": 2.775763988494873,
      "learning_rate": 5.394130964157324e-05,
      "loss": 2.1806,
      "step": 1468
    },
    {
      "epoch": 0.48384042159002016,
      "grad_norm": 2.3641860485076904,
      "learning_rate": 5.388886578879392e-05,
      "loss": 1.927,
      "step": 1469
    },
    {
      "epoch": 0.48416978879328093,
      "grad_norm": 2.720329999923706,
      "learning_rate": 5.383641763131297e-05,
      "loss": 1.9248,
      "step": 1470
    },
    {
      "epoch": 0.48449915599654164,
      "grad_norm": 2.7788572311401367,
      "learning_rate": 5.378396522718683e-05,
      "loss": 1.9824,
      "step": 1471
    },
    {
      "epoch": 0.48482852319980235,
      "grad_norm": 3.20474910736084,
      "learning_rate": 5.373150863447662e-05,
      "loss": 1.5423,
      "step": 1472
    },
    {
      "epoch": 0.4851578904030631,
      "grad_norm": 2.918916940689087,
      "learning_rate": 5.367904791124815e-05,
      "loss": 1.9387,
      "step": 1473
    },
    {
      "epoch": 0.48548725760632383,
      "grad_norm": 3.4597556591033936,
      "learning_rate": 5.3626583115571716e-05,
      "loss": 2.226,
      "step": 1474
    },
    {
      "epoch": 0.4858166248095846,
      "grad_norm": 3.169464588165283,
      "learning_rate": 5.357411430552216e-05,
      "loss": 2.0575,
      "step": 1475
    },
    {
      "epoch": 0.4861459920128453,
      "grad_norm": 2.0502803325653076,
      "learning_rate": 5.352164153917882e-05,
      "loss": 2.3078,
      "step": 1476
    },
    {
      "epoch": 0.4864753592161061,
      "grad_norm": 2.205932855606079,
      "learning_rate": 5.3469164874625345e-05,
      "loss": 2.3107,
      "step": 1477
    },
    {
      "epoch": 0.4868047264193668,
      "grad_norm": 1.9236304759979248,
      "learning_rate": 5.341668436994971e-05,
      "loss": 2.22,
      "step": 1478
    },
    {
      "epoch": 0.48713409362262755,
      "grad_norm": 2.331031322479248,
      "learning_rate": 5.3364200083244175e-05,
      "loss": 2.1931,
      "step": 1479
    },
    {
      "epoch": 0.48746346082588826,
      "grad_norm": 2.279069185256958,
      "learning_rate": 5.3311712072605136e-05,
      "loss": 2.2918,
      "step": 1480
    },
    {
      "epoch": 0.487792828029149,
      "grad_norm": 2.3501081466674805,
      "learning_rate": 5.325922039613316e-05,
      "loss": 2.28,
      "step": 1481
    },
    {
      "epoch": 0.48812219523240974,
      "grad_norm": 2.4179487228393555,
      "learning_rate": 5.320672511193285e-05,
      "loss": 2.4011,
      "step": 1482
    },
    {
      "epoch": 0.48845156243567045,
      "grad_norm": 2.3031511306762695,
      "learning_rate": 5.315422627811278e-05,
      "loss": 2.0346,
      "step": 1483
    },
    {
      "epoch": 0.4887809296389312,
      "grad_norm": 3.2005553245544434,
      "learning_rate": 5.310172395278551e-05,
      "loss": 2.5121,
      "step": 1484
    },
    {
      "epoch": 0.4891102968421919,
      "grad_norm": 2.487902879714966,
      "learning_rate": 5.3049218194067394e-05,
      "loss": 2.0688,
      "step": 1485
    },
    {
      "epoch": 0.4894396640454527,
      "grad_norm": 2.2701635360717773,
      "learning_rate": 5.299670906007866e-05,
      "loss": 2.0644,
      "step": 1486
    },
    {
      "epoch": 0.4897690312487134,
      "grad_norm": 3.022916078567505,
      "learning_rate": 5.294419660894322e-05,
      "loss": 2.3781,
      "step": 1487
    },
    {
      "epoch": 0.49009839845197417,
      "grad_norm": 2.469449758529663,
      "learning_rate": 5.2891680898788665e-05,
      "loss": 1.7029,
      "step": 1488
    },
    {
      "epoch": 0.4904277656552349,
      "grad_norm": 2.625373363494873,
      "learning_rate": 5.283916198774621e-05,
      "loss": 2.0506,
      "step": 1489
    },
    {
      "epoch": 0.4907571328584956,
      "grad_norm": 2.7571139335632324,
      "learning_rate": 5.2786639933950597e-05,
      "loss": 2.1408,
      "step": 1490
    },
    {
      "epoch": 0.49108650006175636,
      "grad_norm": 2.4763495922088623,
      "learning_rate": 5.273411479554008e-05,
      "loss": 1.9425,
      "step": 1491
    },
    {
      "epoch": 0.49141586726501707,
      "grad_norm": 3.169473171234131,
      "learning_rate": 5.2681586630656276e-05,
      "loss": 2.4006,
      "step": 1492
    },
    {
      "epoch": 0.49174523446827784,
      "grad_norm": 2.6018457412719727,
      "learning_rate": 5.262905549744419e-05,
      "loss": 2.0221,
      "step": 1493
    },
    {
      "epoch": 0.49207460167153855,
      "grad_norm": 2.7356503009796143,
      "learning_rate": 5.25765214540521e-05,
      "loss": 2.1555,
      "step": 1494
    },
    {
      "epoch": 0.4924039688747993,
      "grad_norm": 2.443387269973755,
      "learning_rate": 5.2523984558631514e-05,
      "loss": 2.0985,
      "step": 1495
    },
    {
      "epoch": 0.49273333607806,
      "grad_norm": 2.6359920501708984,
      "learning_rate": 5.247144486933706e-05,
      "loss": 2.217,
      "step": 1496
    },
    {
      "epoch": 0.49306270328132074,
      "grad_norm": 2.7415084838867188,
      "learning_rate": 5.241890244432652e-05,
      "loss": 1.8597,
      "step": 1497
    },
    {
      "epoch": 0.4933920704845815,
      "grad_norm": 2.734299421310425,
      "learning_rate": 5.236635734176069e-05,
      "loss": 2.2436,
      "step": 1498
    },
    {
      "epoch": 0.4937214376878422,
      "grad_norm": 2.9822301864624023,
      "learning_rate": 5.231380961980326e-05,
      "loss": 2.0025,
      "step": 1499
    },
    {
      "epoch": 0.494050804891103,
      "grad_norm": 3.2635011672973633,
      "learning_rate": 5.226125933662088e-05,
      "loss": 2.073,
      "step": 1500
    },
    {
      "epoch": 0.4943801720943637,
      "grad_norm": 2.0584778785705566,
      "learning_rate": 5.220870655038308e-05,
      "loss": 2.2952,
      "step": 1501
    },
    {
      "epoch": 0.49470953929762446,
      "grad_norm": 2.4790139198303223,
      "learning_rate": 5.2156151319262045e-05,
      "loss": 2.5564,
      "step": 1502
    },
    {
      "epoch": 0.49503890650088517,
      "grad_norm": 2.687093496322632,
      "learning_rate": 5.2103593701432776e-05,
      "loss": 2.0692,
      "step": 1503
    },
    {
      "epoch": 0.49536827370414593,
      "grad_norm": 2.1168501377105713,
      "learning_rate": 5.2051033755072834e-05,
      "loss": 2.065,
      "step": 1504
    },
    {
      "epoch": 0.49569764090740664,
      "grad_norm": 2.593621253967285,
      "learning_rate": 5.199847153836241e-05,
      "loss": 2.2862,
      "step": 1505
    },
    {
      "epoch": 0.49602700811066736,
      "grad_norm": 2.273329257965088,
      "learning_rate": 5.194590710948419e-05,
      "loss": 1.9901,
      "step": 1506
    },
    {
      "epoch": 0.4963563753139281,
      "grad_norm": 2.755690336227417,
      "learning_rate": 5.189334052662331e-05,
      "loss": 2.1752,
      "step": 1507
    },
    {
      "epoch": 0.49668574251718883,
      "grad_norm": 2.4404118061065674,
      "learning_rate": 5.1840771847967286e-05,
      "loss": 2.1709,
      "step": 1508
    },
    {
      "epoch": 0.4970151097204496,
      "grad_norm": 2.30769681930542,
      "learning_rate": 5.178820113170596e-05,
      "loss": 2.2124,
      "step": 1509
    },
    {
      "epoch": 0.4973444769237103,
      "grad_norm": 2.8207881450653076,
      "learning_rate": 5.1735628436031436e-05,
      "loss": 2.2862,
      "step": 1510
    },
    {
      "epoch": 0.4976738441269711,
      "grad_norm": 2.278991460800171,
      "learning_rate": 5.1683053819137975e-05,
      "loss": 2.0548,
      "step": 1511
    },
    {
      "epoch": 0.4980032113302318,
      "grad_norm": 2.4886574745178223,
      "learning_rate": 5.1630477339221994e-05,
      "loss": 2.3737,
      "step": 1512
    },
    {
      "epoch": 0.49833257853349255,
      "grad_norm": 2.911540985107422,
      "learning_rate": 5.1577899054481996e-05,
      "loss": 2.0253,
      "step": 1513
    },
    {
      "epoch": 0.49866194573675326,
      "grad_norm": 2.7352163791656494,
      "learning_rate": 5.1525319023118434e-05,
      "loss": 2.3057,
      "step": 1514
    },
    {
      "epoch": 0.498991312940014,
      "grad_norm": 2.4180266857147217,
      "learning_rate": 5.147273730333372e-05,
      "loss": 2.3452,
      "step": 1515
    },
    {
      "epoch": 0.49932068014327474,
      "grad_norm": 2.5061450004577637,
      "learning_rate": 5.14201539533321e-05,
      "loss": 1.8195,
      "step": 1516
    },
    {
      "epoch": 0.49965004734653545,
      "grad_norm": 2.4668986797332764,
      "learning_rate": 5.1367569031319715e-05,
      "loss": 1.9367,
      "step": 1517
    },
    {
      "epoch": 0.4999794145497962,
      "grad_norm": 2.661878824234009,
      "learning_rate": 5.1314982595504335e-05,
      "loss": 1.9323,
      "step": 1518
    },
    {
      "epoch": 0.4999794145497962,
      "eval_loss": 2.212698221206665,
      "eval_runtime": 764.8335,
      "eval_samples_per_second": 3.343,
      "eval_steps_per_second": 1.672,
      "step": 1518
    }
  ],
  "logging_steps": 1,
  "max_steps": 3036,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 759,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.328291179772248e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}